cheating section in research paper

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Machine learning based approach to exam cheating detection

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliation Department of Electrical Engineering, Canadian University Dubai, Dubai, UAE

Roles Funding acquisition, Writing – review & editing

Affiliation Department of Mathematics and Statistics, American University of Sharjah, Sharjah, UAE

Roles Methodology, Writing – original draft

Affiliation Department of Communication and Media, Canadian University Dubai, Dubai, UAE

• Firuz Kamalov, 
• Hana Sulieman, 
• David Santandreu Calonge

• Published: August 4, 2021
• https://doi.org/10.1371/journal.pone.0254340
• Reader Comments

The COVID-19 pandemic has impelled the majority of schools and universities around the world to switch to remote teaching. One of the greatest challenges in online education is preserving the academic integrity of student assessments. The lack of direct supervision by instructors during final examinations poses a significant risk of academic misconduct. In this paper, we propose a new approach to detecting potential cases of cheating on the final exam using machine learning techniques. We treat the issue of identifying the potential cases of cheating as an outlier detection problem. We use students’ continuous assessment results to identify abnormal scores on the final exam. However, unlike a standard outlier detection task in machine learning, the student assessment data requires us to consider its sequential nature. We address this issue by applying recurrent neural networks together with anomaly detection algorithms. Numerical experiments on a range of datasets show that the proposed method achieves a remarkably high level of accuracy in detecting cases of cheating on the exam. We believe that the proposed method would be an effective tool for academics and administrators interested in preserving the academic integrity of course assessments.

Citation: Kamalov F, Sulieman H, Santandreu Calonge D (2021) Machine learning based approach to exam cheating detection. PLoS ONE 16(8): e0254340. https://doi.org/10.1371/journal.pone.0254340

Editor: Mohammed Saqr, KTH Royal Institute of Technology, SWEDEN

Received: June 20, 2020; Accepted: June 26, 2021; Published: August 4, 2021

Copyright: © 2021 Kamalov et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data used in the experiments consists of simulated and real-life datasets. The description of the simulated data is provided in the text and the code to generate the simulations is uploaded on GitHub: https://github.com/group-automorphism/exam_cheating . The real-life data is described within the manuscript and also uploaded as a Supporting information file.

Funding: This work was supported in part by the American University of Sharjah. No additional external funding was received for this study.

Competing interests: The authors have declared that no competing interests exist.

1 Introduction

The COVID-19 pandemic has brought unprecedented challenges to the way schools and universities facilitate learning. Due to imposed quarantine measures, educational institutions around the world were left with no alternative other than to continue their course delivery in an online format. Although remote teaching has been adopted by many tertiary institutions, it suffers from several significant drawbacks. One of the main challenges is ensuring academic integrity. Students who take their exams at home are expected to work independently without use of any external help. However, in practice, a nontrivial portion of students attempt to circumvent the rules of academic integrity by, for instance, using digital cheating or contract cheating [ 1 , 2 ] i.e., remunerating a third-party to do the work on their behalf. Universities try to combat exam cheating with the use of remote proctoring, webcams, LockDown Browser (Respondus), plagiarism software (e.g., Turnitin, SafeAssign, iThenticate, to cite a few) and monitoring software during the exam. Nevertheless, it remains relatively effortless for a student to receive third-party aid during the exam, as indicated by a University of New South Wales report, which found that 139 science students had “hired ghost writers from the Chinese messaging site WeChat to complete their work” [ 3 ]. Our goal in this paper is to identify cases of cheating on exams based on a post-exam analysis of the scores to detect abnormal scores. The significance of our approach lies in the novel use of machine learning techniques to identify the anomalous grades on the exam.

Studies over the past two decades in various countries have provided important information on the prevalence of academic dishonesty in higher education [ 4 – 8 ]. Results of a study by Lanier [ 7 ] of 1,262 students taking courses in regular and distance learning formats indicated for instance that cheating was much more widespread in the online sessions. Chirumamilla et al. [ 5 ], in a survey of 212 students and 162 teachers in Norway, identified six most commonly used cheating practices: impersonation, forbidden aids, peeking, peer collaboration, outside assistance and student–staff collusion. In the Canadian context, Eaton [ 9 ] investigated cases of academic integrity violations in the higher education sector, covered by the media between 2010 and 2019. The report concluded that academic misconduct was a “large issue” and that deeper investigation into its extent was warranted.

There is a growing body of literature that recognizes post-exam grade analysis as one promising avenue for detecting exam violations [ 10 , 11 ]. By comparing students’ continuous assessment grades to the grades on the final exam, exam violations can be identified. Significant deviations from the expected results may be an indication of potential exam transgression. Sudden and unexpected high scores on the final exam for an average student may raise a few eyebrows, red flags and be considered abnormal. However, the situation is not always as straightforward as it seems. If the final exam is relatively unchallenging and the majority of students receive high grades, any violation would be less observable. In addition, it is important to take into account the sequential nature of course assessments and that the order of the scores matters. The temporal order consideration complicates the situation considerably. One popular algorithm to analyze ordered data is a recurrent neural network [ 12 ]. It allows for information to propagate through a sequence via parameter sharing.

In this study, we attempt to identify anomalous scores on the final exam using a combination of a recurrent neural network and an outlier detection method. The proposed method uses data consisting of student course assessments including quizzes, the midterm exam, and the final exam. Outlier detection methods are effective in identifying points in data that deviate from the general population and are often used in fraud detection. Recurrent neural networks are used to process sequential data and can be applied to chronologically ordered data such as course assessments. First, the data is processed by a recurrent neural network. Then, the output of the neural network is fed into an outlier detection method which determines the potential instances of cheating. The resulting algorithm provides a robust and efficient tool to identify potential cases of academic violation. We are optimistic that the proposed method would be a useful tool in the fight against exam fraud.

This study is crucial as educational institutions around the globe are looking into effective ways of preventing academic misconduct with the use of advanced analytics such as artificial intelligence. According to Senthil Nathan, managing director and co-founder of Edu Alliance Ltd, “Distance education has not been widely accredited in this region, mainly because of authentication issues” [ 13 ]. Recent developments in the field of remote learning have led to a renewed interest in detecting academic dishonesty in online environments, with the use of behavioral biometrics, learning analytics, data forensics [ 14 ] or data mining [ 15 ]. A search of the literature however revealed few studies (apart perhaps from [ 16 ]) which addressed this issue using anomaly detection methods. The experimental work presented here provides an alternative approach for identifying potential cases of exam dishonesty based on student scores using machine learning methods.

Despite the effectiveness of the proposed algorithm, the generalizability of these results is subject to certain limitations. It is certainly plausible and possible for a student to achieve an unusually high score through hard work and study. Therefore, any case identified as a potential violation requires further investigation by a human expert before a final decision is declared. Notwithstanding the limitations, this work offers valuable insights into final exam fraud detection and aims to contribute to this growing area of research.

Our paper is organized as follows. Section 2 provides an overview of existing literature on fraud detection and related topics. In Section 3, we present our methodology. We describe the details of the proposed approach to identify potential cases of exam violations. In Section 4, we carry out a range of numerical experiments to test the performance of our method. We conclude the paper with a closing summary and remarks in Section 5.

2 Literature

“Are Online Exams an Invitation to Cheat?” [ 17 ]. Academic dishonesty in Higher Education during online final exams is prevalent and far from being a new occurrence. A decade ago, Carnevale [ 18 ] had already argued that technology was “offering students new and easier ways to cheat” (para.1). The findings of a study by King, Guyette Jr., and Piotrowski [ 19 ] on the attitude and behavior of business students towards cheating in an exam administered online indicated that 73.6% of the respondents perceived that cheating online was an easy task (p.7). Since the switch to online examinations, as a consequence of the pandemic lockdown, there seems to be an unwelcome recrudescence of student academic misconduct cases [ 20 ]. In the U.S., Boston University reported for instance that students had “… used various means, including websites such as Chegg, to get help during the quizzes given remotely” [ 21 ].

Outlier detection is a well investigated aspect of data science [ 22 ]. Anomaly detection has been used successfully in many applications. For example, medical claims processing involves large volumes of data which necessitates the use of automated screening procedures. Thus, anomaly detection algorithms are well suited to identify potential fraudulent claims [ 23 , 24 ]. Similarly, financial data processing such as credit card transactions requires automatic means of anomaly detection [ 25 ]. In network security, anomaly detection is used to identify malicious signals in the network traffic [ 26 ]. Recently, investigators in education have examined outlier detection methods to efficiently identify students’ irregular learning processes [ 27 , 28 ], typing patterns [ 29 ], or cheating in Massive Open Online Courses [ 30 ].

Outlier detection methods can be divided into two groups: semi-supervised and unsupervised methods. In a semi-supervised outlier detection method, an initial dataset representing the population of negative (non-outlier) observations is available. A machine learning tool such as one-class SVM can be trained to obtain the boundary of the distribution of the initial observations. Then new observations are categorized according to their distance from the boundary. In unsupervised methods, the algorithm is trained without a clean initial dataset of negative observations. The majority of the algorithms fall under the unsupervised category. Unsupervised anomaly detection methods can be grouped into model, distance, and density-based approaches. An up-to-date review and analysis of the modern methods can be found in [ 31 ]. Model-based approaches are the simplest methods for outlier detection. They are based on the assumption that the normal data is generated according to some statistical distribution [ 32 , 33 ]. The distribution parameters such as the mean and standard deviation are calculated based on the sample data. More sophisticated approaches use kernel functions to estimate the underlying distribution of data [ 34 ]. Then the points with low probability are deemed to be outliers. Distance-based approaches operate on the assumption that abnormal points are far from the main cluster of points. A popular distance method uses a heuristically-determined radius δ and percentage p . Then a point x is considered an outlier if at most p percent of all other points have a distance to x less than δ [ 35 ]. Density-based approaches compare the data density at the given point relative to the density at the neighboring points. Points with relatively low density are considered anomalous. One of the popular density methods is the local outlier factor (LOF) which computes the local density at a given point. The points with a lower local density compared to their neighbors are considered outliers [ 36 ]. A variant of the LOF was used in [ 37 ] to identify schools with unusual performance on standardized tests.

Since the number of anomalous scores is relatively low, it leads to the issue of imbalanced class distribution. Traditional machine learning techniques are not well-equipped to handle imbalanced data [ 38 ]. There exists a number of ways to deal with imbalanced data. A popular method called SMOTE achieves data balance by artificially adding new minority points to the original dataset prior to the training phase [ 39 ]. More recently a new method based on gamma distribution was proposed to generate new instances of minority points [ 40 ].

3 Methodology

In this section, we present our algorithm for detecting potential cases of cheating on the final exam. The inputs to the algorithm are sequences of grades—quizzes, midterm exam, the final exam—of an entire class, while the output is a collection of labels—one label per student—indicating whether each student cheated or not. The proposed method consists of two parts: regression and unsupervised outlier detection. First, a recurrent neural network model is trained to predict the final exam scores based on the previous assessment scores. Then an outlier detection model is applied to identify the instances where the difference between the actual and the predicted final exam scores is abnormal. Since the input data is unlabeled, the proposed method is an unsupervised algorithm.

The job of identifying abnormal final exam scores is not a straightforward task. There are several considerations that need to be taken into account. The anomaly of a final exam score depends uniquely on all the other scores in class. An exam score that is anomalous for one class of students may be completely normal for another class. However, further experiments are required to better understand this issue. Second, the existing machine learning techniques must be tailored to the particular features of the problem. We wish to identify the anomalous scores by comparing the scores prior to the final exam to the score on the final exam. It stands to reason that a student with a large gap between the two scores is more likely to have cheated. Implementing this philosophy requires a custom solution. Third, the temporal nature of the assessments must be taken into account during the analysis. Quizzes, term exams, projects, and the final exam are taken in sequence. The order of the assessments contains crucial information. For example, the scores {79, 90, 70, 61, 50, 95} convey different information than the scores {50, 61, 70, 79, 90, 95}. The former sequence of scores is normal whereas the latter is abnormal ( Fig 1 ).

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

The steady progression of grades of Student 2 makes a score of 95 on the final exam seem plausible. On the other hand, the pattern of grades for Student 1 makes a grade of 95 on the final exam highly unexpected.

https://doi.org/10.1371/journal.pone.0254340.g001

Since the scores in different classes have their own particular characteristics the grades of each class must be analyzed individually. The need for individual analysis of each set of exam scores steers us in the direction of outlier detection methods. Traditional outlier detection algorithms are designed to function without any prior training. Each dataset is analyzed on its own and the abnormal instances are flagged by the algorithm. However, the standard outlier detection algorithms suffer from two crucial drawbacks. First, given an n -dimensional feature vector, unsupervised outlier detection methods do not always identify the important features. It is possible that one of the features is more relevant than the rest in identifying outliers. In our case, it is clear that the grade on the final exam is more relevant than, say, the grade on quiz 1. The traditional outlier detection methods do not provide an option of assigning different weights to features. This disadvantage applies to many machine learning algorithms including neural networks. Second, outlier detection methods do not take into account the temporal nature of the sequential data. The temporal information can significantly improve the learning outcomes of algorithms. Given the above deficiencies, outlier detection methods must be complemented with another method that takes into account the outlined considerations. One of the most effective modern machine learning approaches to handling sequential data is recurrent neural networks. These networks use the data from the previous steps to make predictions about the next step. Thus, our approach is based on two key ingredients: neural networks and anomaly detection.

3.1 Anomaly detection

https://doi.org/10.1371/journal.pone.0254340.g002

3.2 Long short-term memory

Recurrent neural networks (RNNs) and their subclass long short-term memory networks (LSTMs) are a natural choice to deal with sequential data. RNNs and LSTMs are designed to deal efficiently with sequence data by allowing previous outputs to be used as inputs. They have been used successfully in various applications including speech and text recognition as well as time series analysis such as daily stock price data [ 42 ]. We use LSTMs in the proposed algorithm. We refer the reader to [ 43 ] for the details of the LSTM architecture.

3.3 Algorithm

Our goal is to design an algorithm that accomplishes the following two fundamental tasks:

• Identify students whose final exam grade is unreasonably high relative to the rest of the class and their prior assessment scores.
• Take into account the sequential nature of the data under consideration.

We employ LSTMs to accomplish the second task. Concretely, we build a regression model using an LSTM network with the aim of predicting the final exam score based on the scores prior to the final exam. The input variables in our model include scores on quizzes, midterm exam, projects, and other pre-final exam assessments. The output of the model is the score on the final exam. The model is trained to minimize the mean squared error of the predictions. The neural network architecture for predicting the final exam scores consists of an input, output, and hidden layers. The size of the input layer depends on the dataset. In our numerical experiments, the input layer has size 5 which corresponds to four quizzes and one midterm exam scores. The output layer has size one which corresponds to the final exam score. The student scores are processed by the three hidden layers to establish the weights of the network to minimize the prediction error. The first hidden layer is a LSTM layer and the other two layers are fully-connected layers. The neural network was constructed using Keras [ 44 ]. The details of the network layers are presented in Table 1 .

https://doi.org/10.1371/journal.pone.0254340.t001

The exam scores predicted by the LSTM regression model are compared to the actual exam scores. Next, we apply a KDE-based outlier detection method described in the preceding section to identify the potential cheating cases on the basis of the predicted and actual exam scores. We apply Scott’s rule to determine the bandwidth parameter for the KDE. To identify the outliers, we select 9% of the instances with the lowest probability based on the estimated density function. The contamination level was determined based on the fact that 10/110 of the instances in the data are outliers. In general, the contamination level is set by the expert based on the expected number of outliers within the data. The synopsis of the proposed detection procedure is given below.

• The input features ( X ) are the scores prior to the final exam, i.e. quizzes, the midterm exam, project, and others.
• The output ( y ) is the final exam score.
• Calculate the error between the scores predicted by the trained model and the actual exam scores. Apply a KDE-based outlier detection method on the set of errors to determine the abnormal scores.

4 Numerical experiments

In this section, we conduct numerical experiments to test the performance of the proposed algorithm against other benchmark methods. The experiments are carried out over a range of scenarios. The data used in the experiments consists of four synthetic and one real-life datasets. We employ a naive as well as more sophisticated anomaly detection strategies as benchmarks against the proposed method. To measure the efficacy of the algorithms, we compute the true positive and false positive rates of the classification results. The results of the experiments show that the proposed method outperforms the benchmark methods in the majority of the scenarios. The proposed algorithm achieves an average of 95% true positive rate and 2.5% false positive rate on the synthetic data. It achieves 100% true positive rate and 4% false positive rate on the real-life data. The results of the experiments are a strong indication that the proposed algorithm may be an effective tool in fraud identification.

4.1 Benchmark methods

The naive strategy employed in practice is to compare the mean of the scores prior to the final exam with the score on the final exam. If the final exam score is significantly higher than the mean score of the preceding assessments, then it would be a sign of a potential violation. It is a simple and intuitive approach commonly used by instructors to identify potential cases of fraud. Despite its simplicity the naive strategy can be a quick and effective tool to detect abnormal scores on the final exam. As such, it is a good baseline strategy that we aim to beat. Further to the baseline method, we benchmarked our proposed algorithms to the following standalone outlier detection techniques: robust covariance [ 33 ], isolation forest [ 45 ], and local outlier factor [ 36 ]. The benchmark methods are implemented in the popular scikit-learn machine learning library. We used the default parameter settings for the models with exception of the contamination parameter which was set to 0.09. Note that in the sklearn implementation, the default value of the contamination parameter is set to 0.1, i.e., 10% of the points are deemed outliers.

It is important to note that in our data, a sequence of grades is labeled based on whether or not the student cheated on the final exam. This information is built into both our data generation and in our cheating detection method, but is wholly unavailable to the baseline methods. In other words, the baseline methods do not make the same assumption about the structure of the data that is available to the proposed method. Thus, the comparison is not entirely fair.

4.2 Datasets

The experimental evaluation of our proposed approach includes four different synthetic and one real-life datasets. The synthetic datasets are designed to simulate real-life scenarios of exam cheating. A representative sample of each synthetic dataset is shown in Fig 3 . Dataset 1 consists of 100 normal grades and 10 cheating cases ( Fig 3a ). The cheating cases are designed to be egregious with a 35-point difference between the average of the regular assessment scores and the final exam score. The normal grades are mostly homogeneous. Approximately 80 of the normal grades remain consistently within a 10-point range across all assessments including the final exam. The remaining 20 normal grades increase during the semester. The cheating cases in Dataset 1 are relatively easy to identify even with the naive strategy of comparing the average grade prior the final exam to the score on the final exam. Dataset 2 is similar to Dataset 1. However, the cheating cases are more masked in Dataset 2 with only a 20-point difference between the average score prior to the final and the score on the final exam ( Fig 3b ). A smaller margin between the final exam and preceding scores would make it more difficult to discern the cases of cheating. As a result, the performance of the anomaly detection algorithms deteriorates. Dataset 3 is similar to Dataset 2. However, Dataset 3 contains normal grades that are rising throughout the semester in such a way that the difference between the average score prior to the final exam and the final exam is the same as in the cheating cases. As shown in Fig 3c , the scores in maroon color are steadily rising indicating a normal pattern. On the other hand, the scores in red color increase abruptly with much larger amount on the final exam. As mentioned, the difference between the average score prior to the final and the score on the final exam is similar in both instances. To distinguish between the former and the latter cases traditionally requires human judgment. Nevertheless, we show that the proposed method can automatically detect cheating cases even under such challenging conditions. The last dataset in our experiment illustrates the scenario of an easy final exam where all the grades increase relative to the regular assessments. The normal grades are simulated to increase by 10 points on the final exam over the average of the preceding scores. The cheating cases are designed to increase by 25 points on the final exam compared to the prior regular semester assessments ( Fig 3d ). Since all grades increase on the final exam it becomes more challenging to identify cases of cheating. Further details about the synthetic datasets together with the code can be found on our public GitHub repository ( https://github.com/group-automorphism/exam_cheating ).

The datasets capture different scenarios for the distribution of the grades. (a) A representative sample of Dataset 1 grades. The dataset consists of 91% normal and 9% anomalous grades. The normal grades consist of three quarters homogeneous grades and one quarter increasing grades. The anomalous grades rise sharply —by 35 points—during the final exam. (b) A representative sample of Dataset 2 grades. The dataset is similar to Dataset 1. However, the anomalous grades rise less sharply —by 20 points—during the final exam. As a result, the outliers are harder to identify. (c) A representative sample of Dataset 3 grades. The dataset is similar to Dataset 2. However, around 10% of the normal grades are increasing at an incremental pace so that the difference between the average prior and final exam scores are same as in the anomalous instances. As a result, it is even more challenging to identify the outlier scores. (d) A representative sample of Dataset 4 grades. The dataset is designed to simulate a scenario when the final exam is easy and everyone receives a relatively high grade. The normal final exam scores 10 points higher than the average on prior assessments. The anomalous final exam scores 25 points higher than the average preceding scores.

https://doi.org/10.1371/journal.pone.0254340.g003

In addition to the synthetic data, we also employ one real-life dataset in our experiments. Although we would have liked to use more real-life data it is difficult to obtain due to a number of reasons. Our dataset consists of 52 observations of which 3 are positive cases. Each observation includes scores from 4 quizzes, a midterm exam, and the final exam.

4.3 Results and discussion

To determine the efficacy of the algorithms, we calculate the true positive rate (TPR) and the false positive rate (FPR) of the classification results. The TPR measures the proportion of actual positives that are correctly identified as such. The FPR is the proportion of all negatives that are identified as positive by the test. For each synthetic dataset, the experiment is run 20 times. We report the mean and standard deviation of the TPRs. The results of the experiments reveal that the proposed algorithm may be an effective tool in identifying cases of cheating on the final exam based on prior scores.

The mean TPRs of the proposed algorithm together with the benchmark methods is presented in Table 2 . Results indicate that the proposed algorithm has the highest overall average TPR of 0.874 among all the tested algorithms. In addition, the proposed method produces the highest individual mean TPR on Datasets 2 and 3. The standard deviations of the TPRs are also presented in Table 2 . The standard deviation of the TPRs range between 0.05 and 0.12.

The results represent experiments on four datasets based on 20 simulated experiments. The proposed method ( NewAlgo ) produces the best overall results.

https://doi.org/10.1371/journal.pone.0254340.t002

The FPR of the proposed algorithm together with the benchmark methods is presented in Table 3 . As can be seen from the table, the proposed algorithm produces the smallest overall average FPR of 0.013 among all tested detection methods. It also produces the lowest FPR on Dataset 2 and Dataset 3.

The results represent experiments on four datasets based on 20 simulated experiments.

https://doi.org/10.1371/journal.pone.0254340.t003

To further test our proposed algorithm, we applied it to simulated data of size 220 which is double the original size used in the preceding experiments. We simulated each dataset 20 times and recorded the mean TPRs. The results of the experiments with the increased class size are largely in line with the original results. As shown in Table 4 , the mean TPR values of the proposed method range between 0.780 and 0.988. The new algorithm achieves the highest overall mean TPR of 0.872 among all the tested methods. The new overall average TPR is close to the original overall average (0.868).

The results represent experiments on four datasets based on 20 simulated experiments and the class size of 220. The proposed method ( NewAlgo ) produces the best overall results.

https://doi.org/10.1371/journal.pone.0254340.t004

The primary factor in identifying the cases of cheating is the score on the final exam relative to the prior assessments. In addition, secondary factors such as the trajectory (momentum) of the scores and the performance relative to the rest of the class must be considered. Although the naive approach does not take into account the trajectory and the relative performance, it does capture the average difference between the prior assessments and the final exam scores. As a result, it produces a robust performance. On the other hand, the standard outlier detection methods are based purely on the geometry of the sample points. The points away from the main cluster or the points in a relatively low-density region of the sample space are labeled as outliers. No special meaning is assigned to any particular feature.

To better understand the results of the numerical experiments (Tables 2 – 4 ) and to illustrate the mechanism of the outlier detection methods consider the dataset of grades consisting of Quiz 1, Quiz 3, and the Final exam scores from DS2 dataset ( Fig 3b ). The details of the cases of cheating and the outliers are provided in Table 5 . As shown in Table 5 , in the cases of cheating the scores on Quiz 1 and Quiz 2 are relatively stable before suddenly jumping on the Final exam. A sudden increase in the score is an indication of a potential case of cheating. The RobustCov method is a model-based method that fits a Gaussian ellipsoid to the dataset using the central data points. As shown in Table 5 , the RobustCov method selects the points with a positive trend in scores. Although it could be argued from a geometric stance that the points with a positive trend could be outliers, it would not be appropriate to make such an assumption from an academic stance. A student who consistently improves his/her marks between assessments and ultimately receives a high grade on the final exam is not surprising. The RobustCov method fails because it is unable to take into account the context of the problem. The IsoForest method is based on the random forest classifier. It operates on the principle that anomalous points have shorter paths from the root to the terminal node. As can be seen from Table 5 , the IsoForest method labels as outliers the points with either high or low-valued elements. It appears that the points with extremely low or extremely high-valued elements require the fewest number of splittings to be isolated. As a result, the IsoForest method misses the cheating cases. The LOF method is based on the idea of selecting the samples that have a substantially lower density than their neighbors. It selects the points that are away from clusters. The cheating cases are not identified because they form a cluster of points with low quiz scores and high final exam score. In addition, the cheating cases are close to the cases where students gradually improve their marks making them even harder to distinguish.

https://doi.org/10.1371/journal.pone.0254340.t005

The results of the experiment on the real-life data are in line with the results on simulated data. As shown in Table 6 , the proposed method attains the highest TPR. Indeed, the proposed method successfully identifies all 3 cases of cheating among 52 cases. The proposed method also attains the lowest FPR. Only 4% of students who did not cheat were flagged as suspicious.

The results represent experiments on a single real-life dataset. The proposed method ( NewAlgo ) produces the best overall results.

https://doi.org/10.1371/journal.pone.0254340.t006

The results of the numerical experiments reveal that the proposed algorithm achieves high levels of accuracy. It can identify almost all cases of cheating while avoiding falsely labeling a normal score. The proposed method significantly outperforms the benchmark methods both in TPR and FPR. The performance of our method is relatively consistent across different datasets.

In the problem of identifying the potential cases of cheating, the final exam scores play a crucial role. The proposed algorithm is tailored to compare the predicted and the actual final exam scores to identify the anomalous grades. The standard outlier detection algorithms have the added burden of learning the importance of the final exam scores from the data and therefore do not perform well.

5 Conclusion

The COVID-19 pandemic has forced most of the schools and universities in the world to switch to online education. In particular, exams have been administered remotely with little supervision. As a result, the likelihood of exam cheating has increased dramatically. The main objective of the present research was to examine a new approach to detect potential cases of cheating using machine learning techniques. In particular, we considered the problem of identifying cases of cheating on the final exam. We propose a novel method for identifying potential cases of cheating on the final exam using a post-exam analysis of the student grades. Our method takes into account student grades prior to the final exam, grades on the final exam, and the overall performance of the class to make a decision. We employ LSTMs in combination with a KDE-based outlier detection technique to identify the potential cases of cheating. The insights gained from this study may be of assistance to academics and administrators interested in preserving the academic integrity of course assessments.

The proposed method yields promising results on various datasets. It achieves an average TPR of 0.95 and FPR of 0.05 in our numerical experiments. Our method significantly outperforms the benchmark methods that were used in the experiments. However, it is important to note that the baseline algorithms have the added burden of learning the importance of the final exam scores from the data. So the comparison may not be entirely adequate. The performance of the proposed method is consistent across different scenarios simulated in our experiments and produces near perfect accuracy in identifying cases of cheating on the data used in the study.

The present study lays the groundwork for future educational research into outlier detection techniques, as it proposes a complement to commercial plagiarism detection software and possibly a non-intrusive deterrent alternative to highly-polemical remotely-invigilated exams. The reader should bear in mind however that the scope of this study was limited in terms of sample size and context, its replication at other universities would undoubtedly be valuable and recommended. Further investigation and experimentation into the proposed method and academic integrity detection is therefore strongly advocated. Remote exam administration poses a great challenge to preserving the academic integrity of the exam. It is relevant issue today and it will remain important in the future. Our method offers a great tool to help address the issue of academic integrity for remotely administered exams.

A natural progression of this work would be to observe the performance of our model by replacing the regression component with alternatives such as linear regression or Gaussian process regression. Indeed, it would be informative to test different regressor/anomaly detection combinations and perhaps even obtain an improvement over the current configuration.

Supporting information

https://doi.org/10.1371/journal.pone.0254340.s001

• View Article
• Google Scholar
• 2. Lancaster, T., & Clarke, R. (2016). Contract cheating: the outsourcing of assessed student work. Handbook of academic integrity, 639–654
• 3. Fellner, C. (2020). Ghost writers helping UNSW students to cheat on assessments, leaked report reveals. The Sydney Morning Herald. May 5. Retrieved from https://www.smh.com.au/national/nsw/cheating-unsw-students-hire-ghost-writers-from-messaging-site-wechat-to-complete-work-20200505-p54q3f.html
• 4. Awdry, R. (2020). Assignment outsourcing: moving beyond contract cheating, Assessment & Evaluation in Higher Education. https://doi.org/10.1080/02602938.2020.1765311
• 9. Eaton, S. E. (2020). An Inquiry into Major Academic Integrity Violations in Canada: 2010–2019. Retrieved from https://prism.ucalgary.ca/handle/1880/111483
• 10. Berman, S., Brandes, G., Carpenter, M. M., & Strauss, A. L. (2020). Secure Online High Stakes Testing: A Serious Alternative as Legal Education Moves Online. Available at SSRN 3567894.
• PubMed/NCBI
• 13. Bollag, B. (2020). Next Steps for New Online Courses: Measure Learning, Prevent Cheating. Al-Fanar Media. May 26. Retrieved from https://www.al-fanarmedia.org/2020/05/next-steps-for-new-online-courses-measure-learning-prevent-cheating/
• 14. Cizek, G. J., & Wollack, J. A. (2016). Handbook of quantitative methods for detecting cheating on tests. Routledge.
• 15. Hernándeza, J. A., Ochoab, A., Muñozd, J., & Burlaka, G. (2006). Detecting cheats in online student assessments using Data Mining. In Conference on Data Mining| DMIN (Vol. 6, p. 205).
• 18. Carnevale, D. (1999). How to Proctor From a Distance. The Chronicle of Higher Education. November 12. Retrieved from https://www.chronicle.com/article/How-to-Proctor-From-a-Distance/19417
• 20. Sheinermann, M.R; & Shevin, Z. (2020). Math TA posted false solution online to catch students in violation of academic integrity. The Daily Princetonian. May 26. Retrieved from https://www.dailyprincetonian.com/article/2020/05/princeton-teaching-assistant-math-department-slader-mat202-academic-integrity-cheating-covid
• 21. Connolly, C. & Moroney, J. (2020). BU Investigates Whether Students Cheated on Online Tests Amid Shutdown. NBC Boston. April 30. Retrieved from https://www.nbcboston.com/news/local/boston-university-investigates-cheating-allegations/2116206/
• 22. Aggarwal, C. C., & Yu, P. S. (2001, May). Outlier detection for high dimensional data. In Proceedings of the 2001 ACM SIGMOD international conference on Management of data (pp. 37–46).
• 25. Ram, P., & Gray, A. G. (2018, January). Fraud Detection with Density Estimation Trees.
• 26. Nguyen, Q. P., Lim, K. W., Divakaran, D. M., Low, K. H., & Chan, M. C. (2019, June). GEE: A gradient-based explainable variational autoencoder for network anomaly detection. In 2019 IEEE Conference on Communications and Network Security (CNS) (pp. 91–99). IEEE.
• 28. Ueno, M., & Nagaoka, K. (2002). Learning Log Database and Data Mining system for e-Learning–On-Line Statistical Outlier Detection of irregular learning processes. In Proceedings of the International Conference on Advanced Learning Technologies (pp. 436–438).
• 29. Leinonen, J., Longi, K., Klami, A., Ahadi, A., & Vihavainen, A. (2016, July). Typing patterns and authentication in practical programming exams. In Proceedings of the 2016 ACM Conference on Innovation and Technology in Computer Science Education (pp. 160–165).
• 30. Alexandron, G., Ruipérez-Valiente, J. A., & Pritchard, D. E. (2019). Towards a general purpose anomaly detection method to identify cheaters in massive open online courses. In Proceedings of the 12th international conference on educational data mining (pp. 480–483).
• 32. Barnett, V., & Lewis, T. (1994). Outliers in statistical data, vol 3. Wiley.
• 37. Simon, M. (2014). Local outlier detection in data forensics: Data mining approach to flag unusual schools. In Test Fraud (pp. 99–116). Routledge.
• 44. Chollet, F., and others (2015). Keras. https://keras.io
• 45. Liu, Fei Tony, Ting, Kai Ming and Zhou, Zhi-Hua. “Isolation forest.” Data Mining , 2008. ICDM‘08 .

• Original article
• Open access
• Published: 27 November 2020

Attitudes towards cheating behavior during assessing students᾽performance: student and teacher perspectives

• Dijana Vučković   ORCID: orcid.org/0000-0003-2129-555X 1 ,
• Sanja Peković 2 ,
• Marijana Blečić 1 &
• Rajka Đoković 3  

International Journal for Educational Integrity volume  16 , Article number:  13 ( 2020 ) Cite this article

16k Accesses

13 Citations

3 Altmetric

Metrics details

Our aim in this study was to determine students’ and teachers’ attitudes towards cheating in assessing students’ performance. We used mixed methodology and the main research method was a case study. We aimed to describe how our respondents: 1. recognize ethical misconduct (EM) in several situations given through case studies, 2. understand the roles of each subject involved, 3. predict consequences of the EM and how they understand its possible causes, 4. create individual answers to EM or resolve problem situations. The research sample of students (120) includes participants from three basic study programs and two postgraduate programs in the field of education. A sample of teachers (42) was obtained from a number of faculties by random selection. Our respondents have identified most forms of EM reasonably well, although in some situations, the respondents recognized other errors (poor organization of time for learning, professors’ strict deadline for paper submission, etc.) as EM. Therefore, the issues of ethics are not completely clear to all respondents, which leads to the conclusion that universities must organize training in this field. Both groups of respondents understand EM in a similar way, and whether it is a professor or a student (or students) who commits EM has not affected their responses. Our results suggest that it is necessary to work on the prevention of fraud by discussing the consequences (especially the long-term ones, which were not considerably discussed in the comments), by learning ethical reasoning, by developing functional strategies of learning for the purpose of preventing fraud.

Introduction

While ethical misconduct (EM) has been studied at world universities since the 1990s (McCabe et al. 2001 ), it has become a popular topic in Montenegro only in recent years. This does not mean that various forms of EM have not been noted in higher education (ETINED Council of Europe Platform on Ethics, Transparency and Integrity in Education, 2018 ). Ethical misconduct in (higher) education in a country experiencing transition Footnote 1 where the education sector is facing a series of reforms represents a most serious disruptive factor to progress. Since there was no research on cheating in assessing students’ performance in Montenegro, we considered it necessary to determine from the outset whether there has been a common understanding of cheating, its causes and consequences. We wanted to find out the implicit ethical norms of our respondents (Colnerud and Rosander 2009 ). Our aim in this study is to determine students’ and teachers’ attitudes towards academic integrity (AI) issues in assessing students’ performance.

We will first present an overview of literature relevant to our research, which also refers to the recognizing and understanding of EM, its social and individual causes and effects, as well as the role of teachers, students and institutions in the prevention and sanctioning of cheating. The results of an empirical research study with students and teachers as respondents follow further on. We have used the case study method, where the respondents evaluated ethical problems presented in case studies.

Literature review

Fraud in assessment – do we consider the same meaning.

The question posed by Ashworth et al. ( 1997 ), which refers to determining the meaning of the term academic fraud, is still relevant. The authors point out that the term does not have the same meaning for all members of the academic community, while the starting point in various studies is that there is a consensus on the meaning of the term fraud (Ashworth et al. 1997 ). Even though the problem is a complex one and while it is quite possible that some students may have difficulties understanding its meaning (Trautner and Borland 2013 ), some authors think that students most certainly know what is expected from them (Burrus et al. 2013 ). Respondents answer the question What do students perceive to be plagiarism/cheating? correctly and precisely (Jones 2011 ), e.g. the majority of students knew what plagiarism/cheating is. What we understand by the term cheating in this paper is “Actions that attempt to get any advantage by means that undermine values of integrity” (Tauginienė et al. 2018 , p.12), while we will be focused on cheating connected with the assessment of students’ performance.

If students have been accustomed to fraud during their previous education and if they live in transitional societies where there is no coherent system of values (Hrabak et al. 2004 ) – which is the case in Montenegro – then it is quite natural to “mix up” the notions of what is allowed, and what is not. For example, the research conducted by Hrabak et al. ( 2004 ) with medicine students (who study ethics as a separate subject) showed that most of them (94%) admitted to cheating in exams, while almost half of respondents would never report cheating. The knowledge of ethics did not appear applicable to this group of students.

Cheating in assessment represents a violation of the rules of AI, by which we understand the expectation of being “honest, trustworthy, responsible, respectful, and fair, by completing work only in ways authorized by... the institution” (Bertram Gallant 2008 , pp.10–11). Coursework and examination cheating has almost become common at universities, i.e. it has acquired the status of normative behavior among students (Chapman et al. 2004 ). Examination cheating has been treated as a more serious type of fraud compared with coursework cheating (Ashworth et al. 1997 ).

A number of types of fraud have been identified (Jones 2011 ; Newman 2020 ), and numerous attempts have been made to classify them (Ashworth et al. 1997 ; Lothringer 2008 ). The Glossary for Academic Integrity (Tauginienė et al. 2018 ) was compiled, and taxonomies contributing to the systematization of the more and more abundant glossary in this field were developed (Lothringer 2008 ; Tauginienė et al. 2019 ). Despite researches, taxonomies and numerous activities whose primary purpose is to preserve ethics in higher education, academic fraud is getting more widespread, its types and forms are multiplying, so that it is obvious that classifications and definitions do not considerably contribute to fraud prevention.

Spreading of cheating, its actors and frequency

We could speak of both individual and collective cheating within the students’ community (Lothringer 2008 ), where collective cheating may be either socially active or socially passive (Thomas 2017 ). Students’ frauds sometimes involve teachers, so that a number of scandals at prestigious universities have been referred to in literature (Aaron and Roche 2013 ; Robin 2004 ). Third parties, i.e. persons from outside the academic community, are also getting more and more involved in cheating, so the contract cheating is growing more and more evident (Steel 2017 ), meaning that ghost writing has become an “industry”, as reported by numerous studies (Glendinning 2016 ; Shahghasemi and Akhavan 2015 ). The circle of the people involved is getting bigger, while the types of fraud are multiplying, regardless of the only acceptable public (desirably, personal as well) attitude that collecting points and the grade itself must be attained exclusively by hard work and knowledge with due respect to all rules which (un) officially apply to knowledge assessment.

Since the 1990s this topic has been drawing particular attention (Ashworth et al. 1997 ) and there are numerous studies which have provided valuable insights, but their impact is obviously not enough because the types and forms of academic fraud have been multiplying incredibly fast mostly by means of the Internet and modern technology (Aaron and Roche 2013 ; Burrus et al. 2013 ). Besides technology, diversity in the classrooms is another important factor that strongly influences students’ cheating (Brodowsky et al. 2019 ).

Cheating poses a problem around the world (Thomas 2017 ) and it represents “a threat to the morality of the students, the integrity of their grades, and the reputation of our institutions of higher education” (Aaron and Roche 2013 , p.161). The problem of academic dishonesty represents a long-term threat to our entire society. The questionable AI of students, and often of professors (Aaron and Roche 2013 ; Thomas 2017 ) is a definite sign that morality of contemporary higher education is not at a satisfactory level. In addition, there is evidence that cheating in education begins long before university (Aaron and Roche 2013 ), as well as that it continues after university, i.e. at work (Jones 2011 ; Teixeira and Rocha 2006 ). If cheating is not prevented in a timely manner, it threatens to reoccur and persist, thus tending to become a common pattern of conduct.

There are frequent researches of cheating in the disciplines of economy (Brodowsky et al. 2019 ), medicine (Hrabak et al. 2004 ), and life sciences (Dömeová and Jindrová 2013 ), while Joshua Newman ( 2020 ) points to a lack of research concerned with studies whose students will work in the public sector. A high degree of self-reporting has been recorded – students admit to having been involved in cheating – but they often do not see it as immoral behavior (Dömeová and Jindrová 2013 ). Education sciences students also admit to having been involved in cheating (Cummings et al. 2002 ). Frequent self-reporting is a certain indicator that cheating has acquired the status of habitual behavior.

Factors determining students’ cheating

Both collective and individual factors determine cheating, and cheating has also been associated with certain contexts (Thomas 2017 ). The cultural context represents the commonest factor which (de) motivates cheating, so that the orientation individualism vs. collectivism has been recognized as that which most influences making ethical decisions (Brodowsky et al. 2019 ). Cultural individualism directs to an individual who does not subjugate his or her interest to the group, while in collectivism everything is subjugated to the interests of the group (Brodowsky et al. 2019 ; Thomas 2017 ). A step further in such researches has been determining such a phenomenon which shows that some cultures and regions in the world more or less tolerate fraud (Brodowsky et al. 2019 ) through the orientation individualism vs. collectivism. Therefore, societies of the Middle East, North Africa and Eastern Europe have been recognized as utilitarian societies, and “utilitarian cultures are the most tolerant of cheating behavior wherein people view cheating as a way to accomplish the greatest good for the greatest number of people” (Brodowsky et al. 2019 , p.26). The authors emphasize an important non-discriminatory idea, i.e. they point out that one should avoid overgeneralizations while determining cultural factors (Brodowsky et al. 2019 ). In addition to this, much as cultures may have powerful influences, it is clear that they are not the only ones. Certain studies indicate similarities in perception among students coming from radically different cultures (Rawwas et al. 2004 ). This similarity of attitudes may be explained by communication through social networks.

Directly related to individualism-collectivism is also the relation or construct agency-communion, where agency represents a tendency by which an individual boosts his or her potentials and forms his or her independence, while communion represents an opposite tendency by which an individual becomes completely devoted to the interests of the collective and directed towards belonging to the collective (Brodowsky et al. 2019 ; Thomas 2017 ). In addition to the aforementioned determinants, Thomas ( 2017 ) further analyzes several more factors, which have also been recognized as culturally modelled, such as: mind-set (fixed vs. growth mind-set), learning environment (transmission vs. active learning) and motivation. In education it is necessary to encourage growth mind-set, active learning approaches (e.g. discussion, problem solving), and students need to be constantly motivated (Thomas 2017 ).

However, individual certainly are culturally modelled, but they also have some individual characteristics (e.g. demographics, capabilities, attitudes and beliefs) which make them more or less liable to cheating (Ashworth et al. 1997 ; Brodowsky et al. 2019 ; Chapman et al. 2004 ). When they find themselves in a peer group, individuals are especially guided by the culture of that group (McCabe et al. 2001 ). In student groups peer loyalty and fellow feeling are highly valued (Aaron and Roche 2013 ; Ashworth et al. 1997 ). Thus a student will rarely report his or her fellow student who is cheating (Aaron and Roche 2013 ), or they will even express a strong disapproval of the one who reports (Brodowsky et al. 2019 ).

Student group culture also implies the fact that students often model their behavior according to what they think other students do. Students cheat because they think their peers do that as well (Engler and Landau 2011 ). Moreover, they find that their fellow students do that more often than they themselves, which is seen “as a justification for their behaviors” (Engler et al. 2008 , p.99). Thus Engler and Landau ( 2011 ) prove that in fraud prevention it is of particular importance to show students what the behavior of an average student is, i.e. that their perception of fraud is unrealistic (Engler and Landau 2011 ).

The application of social norms interventions requires consistency, stability and depth, because if there is no consistency in the message on the part of the university, the student will not thoroughly adopt the value (Engler et al. 2008 ). It is especially important to position learning as a motivating factor for students, because as long as motivation is purely extrinsic (grades), the problem of values is seriously compromised (Aaron and Roche 2013 ). Students can play a major role in maintaining AI by creating the right climate, but also through peer reporting (Burrus et al. 2013 ). In addition, institutions are thought to be able to improve AI with appropriate codes of ethics and their being respected (McCabe et al. 2001 ).

Causes, effects and prevention of fraud

Various causes of fraud have been identified, such as: pressure (mostly from parents), the desire to increase GPA because of a scholarship and/or getting a job, the fact of being overburdened, a lack of self-confidence, etc. (Davis and Ludvigson 1995 ). Academic fraud often occurs because of bad time management, “everybody does that”, “nobody cares”, the material is incomprehensible (Jones 2011 ). Less personal investment in learning also causes cheating (Colnerud and Rosander 2009 ). Plagiarism seems to be unclear to students (unintentional plagiarism is also recognized), there is an alienation from the university, the influence of large groups and an emphasis on group studying – students consider these as factors which encourage cheating and plagiarism (Ashworth et al. 1997 ).

The list of causes is not exhausted, and neither is the list of consequences. It is highly important that students be aware of the consequences of their conduct and understand fully the seriousness of the mistake they make while cheating. When it comes to consequences, it is necessary to point out to students the short-term negative effects (punishments, ignorance of the subject, etc.), but even more the long-term consequences – incompetence in the profession and the resulting dangers. Students themselves are unaware of the consequences for those who commit fraud (Aaron and Roche 2013 ). In that respect students point out that it is of vital importance that faculties take all measures to prevent fraud (Aaron and Roche 2013 ). In addition to strictly controlling of technical conditions where exams take place, it is particularly important that students be taught responsibility and integrity (Rosile 2007 ; Trautner and Borland 2013 ), as well as that they learn to manage their time better (Aaron and Roche 2013 ). Professors are expected to be more devoted, to create exam tasks in a more creative form (avoiding tasks that can be plagiarized), and to encourage higher forms of learning and thinking instead of simply memorizing data (Aaron and Roche 2013 ; Newman 2020 ; Thomas 2017 ). The quality of teaching and assessment must be at a high level (Ashworth et al. 1997 ).

The authors of the pedagogical orientation assert that the moral education of young people poses a major problem in modern society, so that theories on moral development and learning should be put into function of solving this problem (Caro et al. 2018 ). Therefore, one of the suggestions is rereading Lawrence Kohlberg’s theory with an emphasis on its practical side. According to Kohlberg’s theory, it is considerably more important for moral learning to work on the development of students’ moral reasoning through moral dilemmas than to teach them attitudes and values in the form of content (Caro et al. 2018 ). It is precisely the principle of moral reasoning on which Trautner and Borland ( 2013 ) suggest the model of teaching by means of the so-called sociological imagination. The model consists of a series of stories encompassing moral dilemmas which are the topic for discussion and analysis (Trautner and Borland 2013 ). In addition, it is important to keep in mind that certain forms of academic fraud are caused by the fact that it is still necessary to teach some things directly, i.e. in the form of content, because “students may not be as competent at completing academic assignments as educators usually assume that they are” (Newman 2020 , p.73). This phenomenon concerns different contents, such as general academic writing, adequate selection of literature and its proper citation, but also other important aspects. Many students do not have learning skills developed (Perović and Vučković 2019 ).

Anonymous direct question surveys and/or Likert-type scale were the most frequent methods in this research field (Ashworth et al. 1997 ; Thomas 2017 ).

Research context

As for our research context, it is represented by a collectivist culture which is still not oriented enough to growth mind-set, so Thomas’s ( 2017 ) research corresponds to our context very well, regardless of it having been undertaken on the other part of the planet. Namely, Thomas ( 2017 ) study, conducted with 207 university students in Thailand (in an area marked by a collectivist culture), corresponds to our study because Montenegrin culture is also collectivist and belongs to the region of South-Eastern Europe, which is also recognized as an area of collectivist utilitarian cultures (Brodowsky et al. 2019 ). In addition, teaching and learning at the university level in Montenegro, despite a number of changes that have taken place since the signing of the Bologna declaration in 2003, is still marked by traditional transmissive teaching which does not encourage a growth mind-set and agency, as confirmed by research (Perović and Vučković 2019 ; Vučković 2010 ).

The learning environment is mostly marked by traditional transmission teaching, while the motivation for learning is predominantly extrinsic – in the form of points and grades (ETINED Council of Europe Platform on Ethics, Transparency and Integrity in Education, 2018 ). Students are also often demotivated towards studying because securing employment in a particular field is very hard. It is considered that cheating starts considerably before getting to university.

The Council of Europe ETINED platform published a comparative survey on AI in six South-Eastern European countries, including Montenegro. The greatest disadvantage of that research regarding Montenegro is the fact that a small number of respondents participated, especially through the survey (ETINED 2018 ). However, if we even take this disadvantage into consideration, we still think that the results correspond to the real situation quite well. Among other things, the report indicates:

the lack of clear guidelines on how to preserve AI at the level of decision makers;

HEIs do have necessary documentation (codes of ethics), but there are no clear and consistent procedures of their implementation;

students’ cheating is not considered to be a serious problem;

the respondents indicate that prevention is to be provided by the application of external verification measures (e.g. by software for plagiarism detection and by technical prevention of cheating during exams);

some students do get training on academic writing, but they find it insufficient (ETINED 2018 ).

Among other things, the recommendations emphasize the following:

institutions must improve the learning environment;

students should be included in the prevention of academic fraud, and it is necessary to encourage them to report fraud;

academic staff must show responsibility, but also great pedagogical competence (encouraging higher forms of learning, avoiding tasks that can be plagiarized, highly estimating learning, etc.);

getting familiar with every aspect of AI must be intensified (ETINED 2018 ).

Interesting information was obtained from students about their treating fraud as a game, and it was noted that some teachers also denied students’ cheating, which is drastically different from what students stated (ETINED 2018 ).

Some of the difficulties Montenegro is facing are: low component values in the Academic Integrity Maturity Model (see Fig.  1 ); there are examples of unpunished cheating in the media; some students enrol in the university unwillingly, due to the fact that it is hard to get employment; professors do not punish students’ cheating (ETINED 2018 ).

Academic Integrity Maturity Model radar chart for Montenegro. Source: (ETINED Council of Europe Platform on Ethics, Transparency and Integrity in Education, 2018 , p.75)

Certain attempts have been made after ETINED’s report in order to better treat the problems of AI. Montenegro is the first country in the Western Balkans that adopted The Law on Academic Integrity , in March of 2019. The Law was created within the Project Strengthen Academic Integrity and Combat Corruption in Higher Education , co-funded by the Council of Europe and European Commission (2017–2019). The whole project was aimed at understanding questions of ethics in higher education and raising awareness about AI. A lot of different activities have taken place within this project in Montenegrin higher education community since 2017, and that means that the questions about AI are put in the centre of academic discussions, both among members of academia (students, teachers, researchers) and members of the wider society. The greatest amount of the debate in Montenegro is about plagiarism and about software checking of plagiarism, but the questions on AI are much wider and deeper and AI consists of the whole set of academic values.

Methodology

Research design, aims and objectives.

Our aim in this study is to determine students’ and teachers’ attitudes towards cheating in assessing students’ performance. The main research method is case study (Flyvbjerg 2006 ; Yin 1994 ). We aim to describe how our respondents:

Recognize ethical misconduct (EM) in several situations given through case studies,

Understand the roles of each subject involved,

Predict consequences of the EM given in the case study and how they understand its possible causes,

Create individual recommendations to EM or resolve problem situation.

We used mixed methodology (qualitative and quantitative). Each question that was given to the respondents was open-ended, but it also was possible to code, categorize and thematize the answers (Vilig 2016 ) and to include quantitative data analysis.

The study design had a roadmap as follows:

There are many different types of academic dishonesty in assessing students’ performance, so we decided to choose 8 of them from three taxonomies arranged by Lothringer ( 2008 ). These taxonomies consider three domains of cheating: exams, writing assignments and other assignments and actions, and they consist of total 66 possibilities of cheating (Lothringer 2008 ). We chose 8 of them (random choice) and created 8 case studies. These case studies are moral reasoning situations.

Considering the possible “guilty” person in each case study, we decided to have three “guilty” scopes of action: a. guilty student(s), b. guilty teacher(s), c. guilty some other person(s) (other students, persons outside academia etc.). With these three dramatis personae we could have 8 different combinations (2 3 , 3 is a number of actors). It is possible to have 8 combinations as we showed in Table  1 .

Each case study is created according to one horizontal line and one combination, so we have one situation that is in line with ethical principles by all actors, and also the latest case study shows an example where none of the actors behaved ethically. Considering this, our design gets a better connection with reality.

We set up several questions about each case study: Is there any ethical misconduct (EM)? Who (if any) engaged in EM? Why do you think so? Which are the possible consequences of this misconduct? What are the possible causes of this behavior? What would you recommend to be done about this case?

We gave our respondents (120 students and 42 teachers) case studies and questions in the written form. Students and teachers responded to the same questions.

Important limitation of this methodology design was the time needed for individual responses (it varied from half an hour to one hour and 50 min, with the arithmetic mean of 46.50 min), but our respondents reported that they were highly interested and motivated to understand case studies. They rated the questionnaire’s interestingness with a rating of 4.22 on a scale up to five, and the effort invested received an average score of 2.92. So the effort rating is not negligible, but it is very important that the curiosity is valued by a higher score.

The research sample of students (120) includes participants from three basic study programs (pre-school, teacher training and pedagogy), and two postgraduate programs in the field of education. The sample is predominantly female based (114 female and 6 male students), which matches the fact that these studies are almost exclusively “female” studies. The number of students of the first and the second cycle of studies is even (50.1% master’s and 49.9% bachelor). The sample of teachers was obtained from a number of faculties of the University of Montenegro by random selection – respondents who showed interest in this topic were included. The sample consisted of 28 female and 14 male teachers. Mostly younger teachers were included – 38% had up to 10 years of work in education, and 30% had from 11 to 20 years of work in education.

Since there is a large imbalance in the overall sample between the number of female and male respondents (142 female or 87.65% and 20 male or 12.35%), we did not analyze the results according to the gender variable. The case studies were balanced in terms of gender, so realistic female and male “characters” appear in them. Considering that female and male “characters” in the case studies were involved in situations with different types and degrees of EM (from copying texts from paper, cheating through bugs, to registering exams without taking them previously), it was not possible to check the respondents’ attitude towards the female or male perpetrator of EM. Therefore, the design of this study did not allow us to check whether harsher punishments were intended for male or female “character”.

The data we received as answers to questions about causes, effects and recommendations (which we will treat as topics) were processed in accordance with a qualitative analysis, i.e. answers were coded and categorized. We classified 13 codes for effects, 17 codes for causes and 16 codes for recommendations in total. Nine categories were formed in the topics of effects and causes, while 7 were formed in the topic of recommendations. The categories for the three topics are given in Tables  4 , 5 and 6 .

The accuracy and inaccuracy of the assessment was defined in the following way: if a respondent considered the conduct of all who had participated in the event as (un) ethical, the answer correct was recorded; if he or she considered only certain forms of conduct as such, the answer was partially correct ; and if he or she recognized at least one form of conduct as opposite to its real meaning, i.e. if there was a shift between the ethical and the unethical, then the answer was identified as incorrect . The same principle was applied to the respondents’ explanations.

Results and discussion

Our results show good understanding of EM given in the case studies. Each EM was clearly recognized by almost 70% of the respondents (Table  2 ). These data confirm that students mostly know what is expected from them (Burrus et al. 2013 ). Several of the given EM – being incorporated in complex situation in which more than one person was cheating in some way – were not recognized. The only variable which provided us with a few statistically significant differences is the role of the respondent (teacher and student), so that we are delivering the results according to it. Tables  2 , 3 , 4 , 5 and 6 show frequencies and percentages for recognizing EM, explaining EM, suggesting consequences for EM, EM’ causes and recommendations for each case study.

We have used Spearman’s rank correlation coefficient to evaluate the registered effects, causes and recommendations. In three cases we have obtained high coefficient values. In case study 6, in the teachers’ sample it has been noted that a better knowledge of the consequences reduces the cause (r S  = −0,849; p  = 0,016). In case study 3, in the students’ sample a better knowledge of the causes leads to better recommendations (r S  = 0,841; p = 0,036). In the same sample for case study 8, a better knowledge of the causes leads to a better understanding of the consequences (r S  = 0,964; p  < 0,001).

We are presenting the results in accordance with particular stories, whereas we have to point out that most categories which were recognized for a single story in terms of causes, effects and recommendations are practically the same for all case studies. However, differences occur in terms of the frequency of certain categories from one story to another. We are presenting case studies according to the order from Table 1 .

Case study 1

The results of case study 1 that was ethically clear are very interesting. In this case study the student was able to see exam questions which happened to be near him, because the professor left her office for some time and students were left alone there. However, it is clear from the case study that the student was uncertain (“many question went through his mind”), but he resisted the temptation. As in every case study, there was an ethical dilemma with this one, but it was resolved in the correct way. Almost 30% of teacher respondents described the situation as EM, while the same reasoning came from more than 20% of students (Table 2 ). The respondents who identified EM pointed to the professor’s recklessness in their open comments (14%), and they also showed distrust of the student – 19% of the respondents were convinced that the student had actually seen the questions, as well as that he had kept them to himself (there were a few more students in the professor’s office). The respondents’ explanations given in the form of open answers confirmed their estimation (Table 3 ).

Such answers confirm the presence of distrust in academia as well as the assumption that students most certainly cheat if given a chance. The same respondents state that the student was selfish (“he did not tell his fellow students”), curious (“he took a look at the professor’s desk”), and they recommend that “he should not look at other people’s stuff”. Even though it is clearly stated in the case study that the student accidentally spotted the questions on the pile of papers on the desk, these answers indicate that strict prohibitions and forms of control are required, which was confirmed by the answers which followed. Therefore, in our respondents’ opinion, the locus of control is predominantly external, which indicates the need for a number of changes in education. Namely, only when students have developed self-control, is it then possible to speak about interiorized ethical norms.

Other respondents from the teacher group (71%) and the student group (80%) did not observe EM (Table 2 ). Their comments confirmed this – they point out that the student resisted the temptation and wrote comments such as: “May he keep it up!” This group states that the professor was not careless, only that she trusted the students, and trust relationships are what should be built in higher education.

As for the three topics where open comments were asked (effects of conduct, causes and recommendations), few categories were classified. In this case study these are the categories which originate from the group of wrong estimations of EM, so in terms of effects the respondents think that the professor is in danger of losing her authority, because she failed to exercise strict control (4 respondents), and she allowed injustice to occur (3 respondents). The professor’s insufficient pedagogical competencies were stated as the causes of EM (5 respondents), as well as the student’s fear of the exam (5 respondents). Among the recommendations we have noted the need for: strict control (11 teachers and 4 students), learning about AI (5 teachers) and the student’s punishment (2 students). All open comments came from the respondents who had recognized EM in an ethically correct dilemma.

Case study 2

The second case study deals with a student of medicine who cannot pass his anatomy exam, even though he has studied hard. His ambitious mother bribes a person from the faculty administration (her schoolmate), and she records a positive grade. The student does not know what his mother has done. This case study provoked intense and extensive reactions from the respondents. In all likelihood, the studies of medicine must be in particular ethically pure, since they are directly associated with health care.

The way in which our respondents recognized EM in the case study is given in Table 2 , whereas the accuracy of their explanations is stated in Table 3 . The answers which are marked as incorrect labelled the student as the perpetrator of EM, while the partially correct answers excluded the mother, probably guided by the logic that she was not a member of the academic community. However, in the introductory part of the questionnaire it was not stated that the case studies related exclusively to members of the academic community.

The explanations of students and teachers differ considerably in statistical terms (Table 3 ). The students failed to recognize some of the perpetrators of EM and a number of them identified the student from this case study as a responsible one.

The respondents expressed the most reactions for this case study, especially in the part concerning the effects (Table 4 ) of EM (“He will be a lousy doctor”, “He’ll kill somebody”, “Everybody is guilty – the student, his mother, the administration clerk, the professor who does not exercise control”). The number of comments and the intensity of reactions of the respondents point to their opinion that the medical profession is such that people’s lives depend directly on it, so it goes without saying that it needs to be free from all types of EM.

The causes which had led to EM were mostly recognized as part of extrinsic motivation for conduct in academia (the grade is the aim, not learning and knowledge), even though we recorded different reactions as well (Table 5 ). The teachers (15 of them) indicated that this case confirmed the disadvantages of the system of studying.

A third of our sample advocates harsh punishments for all participants (Table 6 ), but there are such respondents who feel it is sufficient to interrogate the student and others involved in the situation. More responses from the respondents who emphasize the necessity of learning about AI are also recorded here. The responsibility of the institution is particularly emphasized in relation to the administration clerk (Table 6 ), who is urgently to be “most severely punished, fired and prosecuted”. The respondents mostly seek the most serious punishment for the student’s mother (prosecution), while they identify the student as a victim of his mother’s ambition. However, he is not completely absolved in their opinion (one respondent wrote “the student should denounce his mother”) and is expected to acknowledge everything to the faculty services and ask for a retake, having realized that he got a grade for the exam he did not pass. The comments also highlight the possible poor learning strategies.

Case study 3

In this case study a student finds out that her former mentor has published a book in which he included her diploma paper, among other students’ papers. The professor published the book under his own name exclusively. This case study opens a dilemma among students whose works have been plagiarized which makes them emotionally hurt and disappointed. They are uncertain whether to denounce the professor or not.

Apart from several (9 in total) answers which indicate that a teacher “is entitled to students’ works, because the mentor is a co-author”, other respondents clearly state that the professor expressed EM (Table 2 ). Their explanations confirm this (Table 3 ). Almost all of them explicitly identified the professor’s conduct as plagiarism or intellectual theft.

The respondents are almost unanimous about this case of plagiarism – the professor should be denounced and he has to bear consequences for such behavior (Table 4 ). Among other things, one of the respondents states: “The book should be declared an act of plagiarism, while all professor’s works should undergo verification. If this was not the single instance of plagiarism, he would be fired.” This comment and the similar ones show that some students are very well acquainted with the regulations of AI, which can be explained by a significant number of activities related to AI having been realized since 2017. In this case study the respondents (23 of them) point out that the professor is in danger of losing his authority and reputation, which in our respondents’ estimation represents a highly serious consequence of his unethical conduct. The reputation of university professors is something which is still highly appreciated, even though a gradual change of attitude has been noted especially due to several scandals which have shocked the public. A few comments which also appear here focus on the hurt emotions of those students whose works were appropriated by the professor.

As for the causes which had led the mentor to the EM situation, the respondents identify the absence of hard work and the lack of teaching competencies by frequency, followed by “the bad character” (Table 5 ). The respondents convincingly recommend very harsh punishments for the professor, from declaring the book a piece of plagiarism to taking away both the academic rank and the job from the professor (Table 6 ).

Case study 4

A student who is good at statistics sends a completed piece of homework to a fellow student. The latter submits the homework as his own, after which the real author of the homework also submits the paper. The professor notices two identical pieces of homework and invites the student who was the first to submit the homework for an interview. Since he does not manage to find out the truth, the professor decides not to deal with it any longer allocating no points to either of them.

We obtained extremely interesting views on EM (Table 2 ) and their explanations (Table 3 ). The respondents disapproved of the student’s behavior and they treated his act as uncongenial. Interestingly enough, in this case study we also obtained the comments which indicated that the behavior of the student who had sent the homework to his friend was unethical – he shouldn’t have sent the homework to a fellow student. A considerable number of the respondents did not recognize the professor’s EM, i.e. they did not treat his decision “not to deal with it any longer” as an unprofessional attitude. The given explanations, however, described the EM somewhat better and more meticulously (Table 3 ).

As for this case, a lot was said about emotions, which was explained by the respondents in terms of peer loyalty being disrupted. They emphasized that the student was hurt, that he had lost his trust in his “friend” (student who copied his answers), so that he would no longer trust anyone (Table 4 ). The respondents thought that the professor “had to solve the situation”, because in this was he let the students themselves solve the problem which objectively related to the very course and the professor as well. Many comments indicated that the students needed to give a precise description of the situation so that the problem could be justly solved. Concerning this case study, the loss of amiable relations was particularly emphasized, followed by the unjust gaining of points as a result of the action of the student who had submitted someone else’s paper under his own name (Table 4 ).

Among the causes of EM the student’s indolence and irresponsibility (Table 5 ) are particularly emphasized as well as his extrinsic motivation (points and grades). What follows is a recognition of the student’s fear or anxiety, which was not discussed in the case study. Very few respondents (only 6 teachers) point out that the professor lacks the competencies to resolve the situation in an ethically acceptable manner.

The respondents recommend rigorous scrutiny, penalties, interviews with the participants (Table 6 ), and a smaller number of them address essential pedagogical issues, such as: the need for quality learning, better teacher competencies, and the lack of knowledge and skills related to AI.

Case study 5

A student is cheating (copying the answers) during the exam and rustling the papers from which she is copying she is making noises which disturb her fellow students. Another student, visibly irritated, tells her aloud to stop doing that because she is disturbing others. She mentions that her fellow student is a “copycat” who does that all the time. The professor was at the other end of the amphitheatre and noticed nothing. This case study refers to mass studies of economics, so it is realistic that 100 or more students take the test in the same amphitheatre. In such numerous groups, it is possible that the professor does not notice two distant students arguing out loud without shouting. Additionally, the emphasis of this case study was on the relationships between students in a situation involving EM, as well as on their reactions to cheating. Therefore, the professor is deliberately “removed” to the other end of the amphitheatre so that students can first try to confront EM on their own. Since the person who is cheating has not stopped, the other student reports her to the professor who takes away her paper without much discussion and orders her to leave the examination room.

In this case study, the students primarily noticed the girl who reported the act of cheating (more than 41%) (Table 2 ). In addition, statistically significant differences were identified as well as a very high contingency coefficient (C = 0.27). Such answers from our respondents indicate that denouncing somebody who has committed EM is not at all desirable in Montenegrin culture, even if that person visibly prevents you from finishing your work. The explanations given mostly focus on the girl who reported the cheating and only after that they focus on the person who committed EM (Table 3 ).

The reactions already reported by other studies emerged among the students – they express a (very) strong tendency towards peer loyalty (Aaron and Roche 2013 ; Ashworth et al. 1997 ), so peer reporting is strongly rejected (Table 3 ). Furthermore, since they belong to a collectivist culture, the students among our respondents label the student who reported cheating as “a snitch”, “a traitor”, i.e. they use derogatory words. Some go even further in their comments stating that the student reported “because she is jealous” (This explanation remained unclear, because the case study did not point out any reason why anyone would be jealous of the girl who is copying.), “obnoxious”, “telltale”, “doesn’t mind her own business”, etc.

The students identified the following key effects of EM: the lack of knowledge, the fact that the student’s cheating was discovered (which is identified as a punishment), as well as the loss of trust among fellow students (Table 4 ). The professors are also oriented towards the lack of knowledge as a serious consequence.

As in most case studies, some of the respondents point to the student’s irresponsibility and indolence as the primary causes (Table 5 ), then to anxiety which students experience during written examination, then to bad learning habits, but also to the disadvantages of the system of higher education (this group contains exclusively the professors’ answers).

In order to avoid such situations, the respondents consider that it is necessary to study harder, and to talk to the student and to clearly indicate to her the seriousness of the mistake she is making while taking her exams in this way (Table 6 ). In addition, some respondents express the need for stricter control of the process of examination and they think that professors should provide the questions “which cannot be copied from the secretly prepared lists”. The professors (13 of them) believe that this case study also indicates the need for learning about AI.

Case study 6

A student is anxious during the exam, and even though she has studied the best she could she is now taking the exam using “bugs”. Another student from her group is dictating the answers. She manages to pass the exam.

This case study was in general correctly perceived in ethical terms, even though a considerable number of students failed to notice the unethical behavior of the student who dictated the answers (Table 2 ). A smaller group of teachers produced similar answers. The situation was even more polarized in the explanations of the two groups of respondents, so that statistically significant differences appeared (Table 3 ). These differences indicate that the students find the EM of the student who dictated the completed questions less obvious.

The most frequent among the effects (Table 4 ) include the development of a bad habit, a lack of knowledge, a bad example for other students, a possible punishment (“if somebody happens to speak out – too many are involved in the act of fraud”), as well as hurt feelings of those students who passed their exam in a regular way.

The case study explicitly states that the student was anxious, which represents the primary cause in our respondents’ opinion (Table 5 ), while they also identify some other, such as: underdeveloped learning habits, the flaws of the system of higher education, indolence, as well as the pressure coming from the family and a grade as motivation. This case study obviously inspired identification among the students, because they expressed a great amount of understanding for the phenomenon of anxiety. Here they also suggest the issue of communication with professors, which is a frequent cause for anxiety in their opinion.

What dominates among the recommendations (Table 6 ) is an interview with the perpetrators of EM, then a strict control during examination, as well as the need to study harder, i.e. in a more organized and quality manner.

Case study 7

A student has forgotten about the deadline of submitting the seminar paper. He asks the professor to extend the term, but the latter rejects, insulting the student as well as his fellow students who are supporting him. Pressured to meet deadlines, the student submits a plagiarized paper.

More than 70% of the respondents recognized EM (Table 2 ), while fewer were accurate in their explanations (Table 3 ). This study inspired the respondents to comment on the fact that students are sometimes overburdened while studying as well as on the real situation in which “one deadline immediately follows another or they even coincide”. It is interesting that some respondents consider the professor’s attitude to be unethical, because he did not alter the term. Partially incorrect interpretation of ethical norms has been noticed in such comments. The professor did not engage in EM not having altered the term for paper submission (in Montenegrin higher education system there is a frequent notion that somebody “does (not) come to students’ assistance”, which implies extremely diverse forms of behavior on the part of professors who meet students’ requests with or without reason), but he expressed it insulting the students – which was correctly identified by most respondents.

The development of a bad habit of the student, a possible punishment, a bad example for others, as well as the professor’s loss of authority and the students’ hurt feelings are the most frequent effect of the discussed EM (Table 4 ). As for the cause, a lack of pedagogical skills of the professor is dominantly emphasized (Table 5 ). A conversation is the preferred solution, rather than punishment, and the respondents justify this by the fact that the student’s act was a direct consequence of the professor’s irresponsible behavior (Table 6 ).

Case study 8

A student hires a person (a ghost-writer) who “writes” for her a diploma paper. A ghost-writer plagiarized a diploma paper because the student said that she was in a hurry to graduate. The professor recognizes plagiarism, but allows the student to defend the thesis, because “a job is waiting for her”.

A small number of respondents fully recognized EM (Table 2 ), and their explanations were similarly successful as well (Table 3 ). The ghost-writer, as a person outside academia, was not identified as a perpetrator of EM, except in terms of “not having done the job for which he/she had been paid properly”. The ghost writer was accused by this comment of direct and unskilled plagiarism, which indicates a serious lack of AI in the student who commented. High solidarity among the responding students was obtained related to this case study. They frequently justified the actions of the characters, explaining that the student’s “life is at stake”.

As for the effects, they are as follows: a lack of knowledge, a possible punishment (a number of persons were involved in EM here as well), and a bad example for others (Table 4 ). Concerning the causes (Table 5 ), our respondents noticed a lack of teaching competencies (“the professor had an opportunity not to allow the defence of such thesis”), the student’s and the professor’s irresponsibility and indolence, and extrinsic motivation (obtaining a diploma at all costs).

The most common recommendations (Table 6 ) suggest punishment, the need to learn about AI, strict control and the use of software for the purpose of authentication.

Recommendations

A great number of our respondents advocate punishments for those who commit EM in assessing students’ performance. Institutions are definitely obliged to have developed procedures for combating every form of EM, meaning procedures that will be clear and familiar to all members of the academic community in advance. Text-matching software is also an important tool. However, from a pedagogical point of view, it is crucially important to learn about academic integrity, which fundamentally implies notions on academic writing, the use of literature, moral reasoning, and an ethical attitude towards teaching and learning. The respondents have also drawn attention to the fact that the lack of learning skills represents a common cause of EM; therefore, they need to be given particular attention. Teachers’ competencies have to be improved as well – a number of respondents have stated in their open comments that “all these situations occur in reality”, which suggests the need that teachers should also be more familiar with the rules of AI, it is important that they be more competent while designing tests (so that they cannot be directly plagiarized) and that they be more responsible while invigilating and grading written papers.

When it comes to the recommendations for future research, this topic is only becoming dominant in Montenegro; so many basic data are still missing. Every segment of academic integrity is to be well known, while it is highly important that the results of research have practical implications, i.e. a direct impact on university policies regarding AI.

Our respondents have recognized most forms of EM well enough, although in some situations, the respondents perceived other errors (poor organization of time for learning, professors’ strict deadline for paper submission, etc.) as EM. Certain manifestations of EM have not been recognized, which may possibly be a consequence of complex situations. For instance, the professor in case study 7 had two activities: 1. he refused to alter the term for the submission of papers, and 2. he insulted the students. Some respondents have identified his first action as EM, while they have failed to pay attention to his second action. Therefore, the issues of ethics are not completely clear to all respondents, which leads to the conclusion that universities must organize trainings in this field.

Both groups of respondents understand EM in a similar way, and whether it is a professor or a student (or students) who commits EM has not affected their responses. We have not found consistent statistically significant differences between them when considering identification of EM and a guilty person in a given situation. Interestingly enough, both groups have emphasized the importance of peer loyalty (case study 5), which is probably a consequence of firm social patterns. This also indicates the fact that in the present circumstances it is hardly expected that students will publicly criticize the one who is cheating in Montenegrin higher education system. However, we think that it is necessary to work with them on the prevention of fraud by discussing the consequences (especially the long-term ones, which were not considerably discussed in the comments), by learning ethical reasoning, and by developing functional strategies of learning for the purpose of preventing fraud. In addition, both groups of respondents have expressed the most intense reactions and given the most extensive responses to the case study dealing with the student of medicine. Furthermore, both groups have expressed weaker reactions to the unethical conduct of persons who are not members of the academic community (for example, the ghost-writer in case study 8). This at least partly indicates an understanding that the responsibility for academic conduct must solely rest on members of the academic community.

Some older students (masters’ level and senior years of study) evaluate EM more rigorously – they anticipate more severe consequences and more frequently seek for causes in individual responsibility, but a few comments recorded have not resulted in statistically significant differences in relation to younger students. Younger students look for causes in overwork (which they associate with the lack of learning skills and/or bad time planning), as well as in poor competencies of professors. The professors sometimes believe that students are lazy, and that they “learned how to cheat, even before coming to university.”

Within the third research objective (predicting consequences and understanding possible causes) we have found some differences which we could describe as colored by personal role. Namely, when the students are describing consequences of EM that was committed by students within the case study, they are slightly permissive while for teachers they tend to be stricter. The teachers respondents are strict in both situations, and their commonest recommendation is punishment or strict control. The students also frequently mention punishment, yet a bit lower in percentage terms. This group has frequently recommended the conversation with the perpetrator of EM. It is expected that punishment and strict control will represent a way of preventing EM, but we think it is equally important, if not more important, to raise awareness of non-ethics and its consequences for an individual and for society, as well as to learn about AI as a teaching goal, and to develop the manner of thinking which is ethical (i.e. learn ethical reasoning). When considering the causes, the students tend to be more diverse in their ideas (all 17 codes were registered in their group), while the teachers’ group is more homogenous (9 codes).

Taking into consideration that our respondents showed great interest in participating in a study that had a rather complicated structure composed of eight case studies, we believe that students and teachers from our sample are ready and motivated to seriously confront the problems produced by EM. Many of our respondents’ reactions lead to the conclusion that the academic community, or at least part of it, perceives the serious threats emerging as a result of EM. Identifying the problem is the first step towards solving it, so we believe that this first step in the field of AI has been made.

Availability of data and materials

The original research data upon which the Tables in this article are based is held by the Lead Author. The data is available on the request. The data is in local language(s).

Montenegro was part of the united state of the South Slavic peoples, whose last name was the Socialist Federal Republic of Yugoslavia (SFR Yugoslavia), for most of the twentieth century (from 1919 to 1990). After the 1990s and the civil war in the territory of several former members of SFR Yugoslavia, Montenegro remained in a state union with Serbia until 2006, when the population voted for state independence through a referendum. After the restoration of state independence, Montenegro expressed its commitment to European and transatlantic integration, so it joined NATO in 2017. Negotiations on joining the European Union have been very intensive. All these numerous changes have been turbulent and testify to the transitional position of the country.

Abbreviations

Ethical misconduct

The Council of Europe Platform on Ethics, Transparency and Integrity in Education

• Academic integrity

Grade Point Average

Aaron LS, Roche C (2013) Stemming the tide of academic dishonesty in higher education: it takes a village. J Educ Technol Syst 42:161–196. https://doi.org/10.2190/FET.42.2.h

Article   Google Scholar  

Ashworth P, Bannister P, Thorne P (1997) Guilty in whose eyes? University students’ perceptions of cheating and plagiarism in academic work and assessment. Stud High Educ 22:187–203. https://doi.org/10.1080/03075079712331381034

Bertram Gallant T (2008) Academic integrity in the twenty-first century: a teaching-learning imperative. Jossey-Bass, San Francisco

Google Scholar  

Brodowsky GH, Tarr E, Ho FN, Sciglimpaglia D (2019) Tolerance for cheating from the classroom to the boardroom: a study of underlying personal and cultural drivers. J Marketing Educ 42:23–36. https://doi.org/10.1177/0273475319878810

Burrus RT, Jones AT, Sackley B, Walker M (2013) It’s the students, stupid: how perceptions of student reporting impact cheating. Am Econ 58:51–59. https://doi.org/10.1177/056943451305800106

Caro S, Ahedo Ruiz JC, Esteban Bara F (2018) Kohlberg’s moral education proposal and its legacy at university: present and future. Revista Española de Pedagogía 76:85–100. https://doi.org/10.22550/REP76-1-2018-12

Chapman KJ, Davis R, Toy D, Wright L (2004) Academic integrity in the business school environment: I’ll get by with a little help from my friends. J Marketing Educ 26:236–249 Accessed 12 Jan 2020 from https://www.learntechlib.org/p/64906/

Colnerud G, Rosander M (2009) Academic dishonesty, ethical norms and learning. Assess Eval High Edu 34:505–517. https://doi.org/10.1080/02602930802155263

Cummings R, Maddux CD, Harlow S, Dyas L (2002) Academic misconduct in undergraduate teacher education students and its relationship to their principled moral reasoning. J Instr Psychol 29:286–296. https://doi.org/10.12691/education-2-11-9

Davis SF, Ludvigson HW (1995) Additional data on academic dishonesty and a proposal for remediation. Teach Psychol 22:119–121. https://doi.org/10.1207/s15328023top2202_6

Dömeová L, Jindrová A (2013) Unethical behavior of the students of the Czech University of life sciences. Int Educ Stud 6:77–85. https://doi.org/10.5539/ies.v6n11p77

Engler JN, Landau JD (2011) Source is important when developing a social norms campaign to combat academic dishonesty. Teach Psychol 38:46–48. https://doi.org/10.1177/0098628310390848

Engler JN, Landau JD, Epstein M (2008) Keeping up with the joneses: students’ perceptions of academically dishonest behavior. Teach Psychol 35:99–102. https://doi.org/10.1080/00986280801978418

ETINED Council of Europe Platform on Ethics, Transparency and Integrity in Education (2018) South-east European project on policies for academic integrity – Vol 5, Council of Europe, Strasbourg

Flyvbjerg B (2006) Five misunderstandings about case-study research. Qual Inq 12:219–245. https://doi.org/10.1177/1077800405284363

Glendinning I (2016) European perspectives of academic integrity. In: Bretag T (ed) Handbook of academic integrity. Science+Business Media Springer, Singapore, pp 55–74

Chapter   Google Scholar  

Hrabak M, Vujaklija A, Vodopivec I, Hren D, Marušić M, Marušić A (2004) Academic misconduct among medical students in a post-communist country. Med Educ 38:276–285. https://doi.org/10.1111/j.1365-2923.2004.01766.x

Jones DR (2011) Academic dishonesty: are more students cheating? Bus Commun Q 74:141–150. https://doi.org/10.1177/1080569911404059

Lothringer LB (2008) Evaluation of the use of an academic integrity training course as a proactive measure encouraging academic honesty. Doctoral dissertation 15714. Iowa State University. https://doi.org/10.31274/rtd-180813-7895

McCabe DL, Trevino LK, Butterfield KD (2001) Cheating in academic institutions: a decade of research. Ethics Behav 13:219–232. https://doi.org/10.1207/S15327019EB1103_2

Newman J (2020) Academic integrity in public administration programmes: practical reflections on prevention and response. Teaching Public Administration 38:63–77. https://doi.org/10.1177/0144739419864128

Perović Đ, Vučković D (2019) Success in studying at the University of Montenegro: is there hyper-production of diplomas? Interdiscip Descr Complex Syst 17:385–402. https://doi.org/10.7906/indecs.17.2.13

Rawwas MYA, Al-Khatib JA, Vitell SJ (2004) Academic dishonesty: a cross-cultural comparison of U.S. and Chinese marketing students. J Marketing Educ 26:89–100. https://doi.org/10.1177/0273475303262354

Robin RT (2004) Scandals and scoundrels: seven cases that shook the academy. University of California Press, Ewing, NY

Book   Google Scholar  

Rosile GA (2007) Cheating: making it a teachable moment. J Manag Educ 31:582–613. https://doi.org/10.1177/1052562906289225

Shahghasemi E, Akhavan M (2015) Confessions of academic ghost authors: the Iranian experience. SAGE Open:1–7. https://doi.org/10.1177/2158244015572262

Steel A (2017) Contract cheating: will students pay for serious criminal consequences? Alternative Law Journal 42:123–129. https://doi.org/10.1177/1037969X17710627

Tauginienė L, Gaižauskaitė I, Glendinning I, Kravjar J, Ojsteršek M, Ribeiro L, Odiņeca T, Marino F, Cosentino M, Sivasubramaniam S (2018) Glossary for academic integrity. ENAI Report 3G Accessed 10 Feb 2020. http://www.academicintegrity.eu/wp/wp-content/uploads/2018/02/GLOSSARY_final.pdf

Tauginienė L, Gaižauskaitė I, Razi S, Glendinning I, Sivasubramaniam S, Marino F, Cosentino M, Anohina-Naumeca A, Kravjar J (2019) Enhancing the taxonomies relating to academic integrity and misconduct. J Acad Ethics 17:345–361. https://doi.org/10.1007/s10805-019-09342-4

Teixeira AAC, Rocha MF (2006) Academic cheating in Austria, Portugal, Romania and Spain: a comparative analysis. Res Comp Int Educ 1:198–209. https://doi.org/10.2304/rcie.2006.1.3.198

Thomas D (2017) Factors that explain academic dishonesty among university students in Thailand. Ethics Behav 27:140–154. https://doi.org/10.1080/10508422.2015.1131160

Trautner MN, Borland E (2013) Using the sociological imagination to teach about academic integrity. Teach Sociol 41:377–388. https://doi.org/10.1177/0092055X13490750

Vilig K (2016) Kvalitativna istraživanja u psihologiji. Clio, Beograd

Vučković D (2010) Organizacija nastave na Studijskom programu za obrazovanje učitelja, [Organization of teaching at the Teacher Training Department]. Sociološka luča 4:146–172

Yin RK (1994) Case study research: design and methods. Sage, Beverly Hills, CA

Download references

Acknowledgements

This manuscript has been produced in the framework of the national project 01-792 entitled “Strengthening Academic Integrity – Interdisciplinary Research-based Approach to Ethical Behavior in Higher Education” which was financed by the Ministry of Science of Montenegro.

Originality and review status

The authors confirm that this work is original and has not been published elsewhere, nor is it currently under consideration for publication elsewhere.

Financed by the Ministry of Science of Montenegro.

Author information

Authors and affiliations.

Faculty of Philosophy, University of Montenegro, Cetinjski put bb, Podgorica, Montenegro

Dijana Vučković & Marijana Blečić

Faculty of Tourism and Hotel Management, University of Montenegro, Cetinjski put bb, Podgorica, Montenegro

Sanja Peković

Faculty of Law, University of Montenegro, Cetinjski put bb, Podgorica, Montenegro

Rajka Đoković

You can also search for this author in PubMed   Google Scholar

Contributions

This article was written by the lead author with support from the second, third and fourth author who helped in data collection, data processing, and disscusion. The percentage of contributions is indicated by the order of authorship provided. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Dijana Vučković .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Vučković, D., Peković, S., Blečić, M. et al. Attitudes towards cheating behavior during assessing students᾽performance: student and teacher perspectives. Int J Educ Integr 16 , 13 (2020). https://doi.org/10.1007/s40979-020-00065-3

Download citation

Received : 12 August 2020

Accepted : 09 November 2020

Published : 27 November 2020

DOI : https://doi.org/10.1007/s40979-020-00065-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Causes of cheating
• Consequences of cheating
• Non-ethical behavior

International Journal for Educational Integrity

ISSN: 1833-2595

• Submission enquiries: Access here and click Contact Us
• General enquiries: [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• HHS Author Manuscripts

Once a Cheater, Always a Cheater? Serial Infidelity Across Subsequent Relationships

Kayla knopp.

University of Denver

Shelby Scott

Veteran Affairs Eastern Colorado Health Care System

Lane Ritchie

Galena k. rhoades, howard j. markman, scott m. stanley.

Although there is a large body of research addressing predictors of relationship infidelity, no study to our knowledge has specifically addressed infidelity in a previous relationship as a risk factor for infidelity in a subsequent relationship. The current study addressed risk for serial infidelity by following adult participants ( N = 484) longitudinally through two mixed-gender romantic relationships. Participants reported their own extradyadic sexual involvement (i.e., having sexual relations with someone other than their partner; abbreviated ESI) as well as both known and suspected ESI on the part of their partners in each romantic relationship. Findings from logistic regressions showed that those who reported engaging in ESI in the first relationship were three times more likely to report engaging in ESI in their next relationship compared to those who did not report engaging in ESI in the first relationship. Similarly, compared to those who reported that their first-relationship partners did not engage in ESI, those who knew that their partners in the first relationships had engaged in ESI were twice as likely to report the same behavior from their next relationship partners. Those who suspected their first-relationship partners of ESI were four times more likely to report suspicion of partner ESI again in their next relationships. These findings controlled for demographic risk factors for infidelity and held regardless of respondent gender or marital status. Thus, prior infidelity emerged as an important risk factor for infidelity in next relationships. Implications for novel intervention targets for prevention of serial relationship infidelity are discussed.

Although the vast majority of romantic relationships in the United States include expectations of monogamy ( Conley, Moors, Matsick, & Ziegler, 2013 ; Treas & Giesen, 2000 ), infidelity is widespread, with estimates of lifetime engagement in extra-relational affairs around 20% for married couples ( Blow & Hartnett, 2005b ) and up to 70% for unmarried couples ( Wiederman & Hurd, 1999 ). Relationship infidelity is usually damaging ( Allen et al., 2005 ), frequently leading to psychological distress both for those who engage in infidelity and for their partners ( Cano & O’Leary, 2000 ), as well as to relationship distress or dissolution ( Allen & Atkins, 2012 ; Johnson et al., 2002 ). Indeed, infidelity is one of the most commonly reported causes of divorce ( Amato & Previti, 2003 ; Scott, Rhoades, Stanley, Allen, & Markman, 2013 ) and one of the most difficult issues for couple therapists to treat ( Whisman, Dixon, & Johnson, 1997 ). The current study sought to address gaps in the literature about risks of serial infidelity by assessing the degree to which infidelity in one romantic relationship predicted similar experiences in participants’ next relationships.

Researchers have examined a variety of individual and contextual risks for becoming involved in an extradyadic relationship. Cross-sectional data suggest that risk factors include low relationship commitment ( Drigotas, Safstrom, & Gentilia, 1999 ), declining sexual and relationship satisfaction ( Mark, Janssen, & Milhausen, 2011 ; Scott et al., 2016 ), certain personality characteristics ( Barta & Kiene, 2005 ; Dewall et al., 2011 ; Mark et al., 2011 ), permissive attitudes about sex or infidelity ( Fincham & May, 2017 ; Treas & Giesen, 2000 ), and exposure to approving social norms ( Buunk, Bakker, & Taylor, 1995 ). Dating relationships are also typically thought to have a substantially higher risk of infidelity than marriages ( Blow & Hartnett, 2005b ; McAnulty & Brineman, 2007 ). Further, some research has investigated individual differences in motivations for engaging in infidelity. For example, Allen (2001) found that those high in avoidant attachment were more likely to report ESI for reasons related to independence, whereas those higher in attachment anxiety were more likely to report ESI for reasons related to intimacy and self-esteem. Mark et al. (2011) reported that approximately 20% of the variance in infidelity motivation was explained by different patterns of sexual inhibition and excitement.

In addition to these process variables, associations between individual demographic characteristics and predispositions toward infidelity have also been widely studied ( Allen et al., 2005 ; Green & Sabini, 2006 ). The most well-established demographic finding has been that men tend to be more likely to engage in infidelity than women, possibly due to greater social power or evolutionary motivations (e.g., Hughes, Harrison, & Gallup, 2004 ; Lalasz & Weigel, 2011 ; Lammers, Stoker, Jordan, Pollmann, & Stapel, 2011 ). Women and men may also vary in their emotional responses to perceived partner ESI; men tend to report a greater degree of jealousy and distress in response to partner infidelity and to be more threatened by sexual rather than emotional infidelity of their female partners, whereas women report more distress in response to emotional infidelity of their male partners ( Edlund, Heider, Scherer, Farc, & Sagarin, 2006 ; Frederick & Fales, 2016 ; Harris & Christenfeld, 1996 ). However, some research has demonstrated that age and prior experiences with partner ESI moderate these findings ( Varga, Gee, & Munro, 2011 ), and that gender discrepancies in general may be decreasing in younger cohorts ( Atkins, Baucom, & Jacobson, 2001 ; Fincham & May, 2017 ; Mark et al., 2011 ). Thus, age may be another key factor in understanding risk for infidelity. Black or African American populations typically report higher rates of infidelity among men in mixed-gender relationships (e.g., Whisman, Gordon, & Chatav, 2007 ), which is likely attributable to scarcity of desirable male partners in Black or African American communities due to incarceration and other social contextual factors ( Pinderhughes, 2002 ). Some studies have found that socioeconomic variables related to opportunity for infidelity, including more education, higher income, and employment, tend to be positively associated with both engagement in and approval of infidelity ( Atkins et al., 2001 ; Treas & Giesen, 2000 ), although this association is inconsistent in the literature ( Fincham & May, 2017 ); socioeconomic risk factors may be further moderated by gender ( Munsch, 2012 ) and by relationship history ( Atkins et al., 2001 ).

Although cross-sectional studies are descriptively useful, they do not necessarily provide information about how risks for infidelity can be understood over time. In particular, these studies do not address whether a person who engages in infidelity in one relationship is likely to engage in infidelity again in a subsequent romantic relationship, nor do they address the magnitude of the increased risk. This question is important, given that most people in the U.S. have multiple dating relationships before entering into a marriage or long-term commitment with a partner ( Sassler, 2010 ), and research suggests that individuals’ earlier romantic experiences may have consequences that can persist into later relationships or marriage. For example, living with more than one different romantic partner before marriage is associated with reduced marital quality and stability ( Lichter & Qian, 2008 ), and having more sexual or relationship partners predicts poorer outcomes in later relationships, including sexual infidelity ( Maddox Shaw, Rhoades, Allen, Stanley, & Markman, 2013 ) and lower marital satisfaction ( Rhoades & Stanley, 2014 ). Thus, research regarding risks from prior relationship experiences may have important implications for researchers and clinicians who are interested in helping people develop healthy relationship patterns. An understanding of serial infidelity patterns could be clinically useful in therapy and relationship education contexts by helping individuals who are at risk of becoming entrenched in unhealthy relationship patterns recognize and address risk factors in themselves and their partners (e.g., Rhoades & Stanley, 2009 ).

Building an understanding of serial infidelity risk is a logical extension of existing theories about the factors that put people at risk of engaging in infidelity in general. Although numerous motivations for infidelity have been identified in existing theories, the two of most relevance to serial infidelity are (1) the quality and availability of alternative partners, and (2) attitudes about the acceptability of infidelity ( Drigotas & Barta, 2001 ). First, regarding alternative partners, models of commitment and social exchange ( Rusbult, 1983 ; Thibaut & Kelley, 1959 ) suggest that infidelity is likely to occur when individuals perceive having desirable alternatives to their current relationship partner ( Drigotas et al., 1999 ). Individuals who have already had emotional affairs or sexual encounters outside of their current relationship have firsthand knowledge that such alternatives exist, and may subsequently believe that such alternatives remain available to them, thus creating a higher risk of engaging in infidelity again in future relationships.

Second, regarding attitudes about ESI, models of infidelity risk often incorporate the reciprocal effects of people’s attitudes. The theory of reasoned action claims that people tend to behave in accordance with their attitudes and with widespread social norms ( Fishbein & Ajzen, 1975 ). Consistent with this theory, research has demonstrated that within a social context of widespread disapproval of infidelity, individuals with more approving or permissive personal beliefs regarding infidelity are more likely to cheat ( Hackathorn, Mattingly, Clark, & Mattingly, 2011 ; Treas & Giesen, 2000 ). At the same time, past engagement in infidelity also predicts having more approving attitudes about infidelity, consistent with cognitive dissonance theory ( Foster & Misra, 2013 ; Jackman, 2015 ; Sharpe, Walters, & Goren, 2013 ; Solstad & Mucic, 1999 ; Wiederman, 1997 ). Engaging in infidelity in a past relationship may therefore increase the risk of infidelity in future romantic relationships by increasing one’s acceptance of engaging in infidelity.

Despite strong theoretical rationale, few studies have evaluated how actual infidelity experiences persist across different romantic relationships. Studies by Banfield and McCabe (2001) and Adamopolou (2013) each demonstrated that a recent history of engaging in infidelity increased the risk of future infidelity, but these studies were ambiguous with regard to whether the repeated infidelity occurred within the same relationship or across different relationships. This distinction is important, given that some risk factors for infidelity are relationship-specific (e.g., commitment) whereas others are linked to individual characteristics that are likely to persist into future relationships as well (e.g., personality). Recent work by Martins et al. (2016) more directly showed that infidelity in a previous relationship increased risk of infidelity in a later relationship, but was limited by the use of retrospective reports of prior infidelity that did not specify in which previous relationship or how long ago the infidelity occurred. Thus, the existing literature does not provide clear information about whether and to what degree engaging in infidelity in a previous relationship impacts the likelihood that an individual will engage in infidelity in the next relationship. The current study aims to fill that gap.

In addition to a person’s own behavior, serial infidelity risk may include actual or suspected infidelity on the part of an individual’s romantic partner. Research taking an interpersonal perspective has identified relationship-specific factors, in addition to involved-partner factors, that contribute to risk of infidelity ( Mark et al., 2011 ). Couple-based approaches are careful to avoid blaming an individual for their partner’s infidelity ( Baucom, Snyder, & Gordon, 2011 ); at the same time, researchers acknowledge that both partners may play a role in creating the relationships characteristics that could potentially increase the chance that a partner will cheat ( Allen et al., 2008 ). Thus, individuals with previous partners who have engaged in infidelity may be at increased risk for partnering with individuals in later relationships who also engage in infidelity because these individuals may be more likely to contribute to relationship contexts associated with higher risk of infidelity ( Allen et al., 2005 ). It may also be the case that individuals who have learned about a previous partner’s infidelity have developed expectations that infidelity is more common and/or acceptable in subsequent relationships (e.g., Glass & Wright, 1992 ). If this is the case, individuals who have known about a previous partner’s infidelity may be more likely to tolerate infidelity in a subsequent relationship as well, leading to persistent risk of partner infidelity across relationships over time.

Further, some research indicates that one’s own past engagement in infidelity can increase the likelihood of suspecting infidelity from a relationship partner ( Whisman et al., 2007 ). Subjective responses to hypothetical partner infidelity also differ based on whether a person has experienced actual infidelity in their own lives ( Confer & Cloud, 2011 ; Harris, 2002 ; Ritchie & van Anders, 2015 ). Although research on this topic is quite limited, these studies suggest that perception of partner infidelity in a current relationship may be more likely if someone has personally engaged in infidelity previously. Therefore, we may expect risk of serial infidelity to include cross-partner effects as well.

One widely acknowledged issue in studies of infidelity is ambiguity or inconsistency in definitions of infidelity and the terminology used to refer to it ( Blow & Hartnett, 2005a ). Terms such as infidelity, unfaithfulness, cheating, extra-marital or extra-relational affairs, extra-dyadic involvement, and extra-dyadic sexual involvement are commonly used in the literature. Although each term has particular nuances in connotation or in which behaviors are included, they all attempt to assess the same underlying construct, which we refer to as infidelity.

The current study measures infidelity as extra-dyadic sexual involvement (ESI) – i.e., whether a person in a romantic relationship has had sexual relations with someone other than their relationship partner. This measure of infidelity has a few distinct limitations. First, it may not capture all behaviors that a couple may consider to be infidelity, such as kissing or an emotional affair, and the term “sexual relations” can be ambiguous. Second, not all ESI should be categorized as infidelity, because some couples may agree that ESI is acceptable under certain conditions, and other people may have more than one committed relationship partner (i.e., consensually non-monogamous or CNM couples). At the final wave of data collection, fewer than 2% of participants in the current sample reported being “in an open relationship”; however, this term may not capture the wide array of CNM agreements that may exist (e.g., swinging, threesomes, etc.). Therefore, we believe that the vast majority of behaviors captured by the current ESI measure are probably accurately labeled as infidelity, but we are not able to know for certain whether participants in our study or their partners considered ESI to be allowed or not in their relationships.

Current Study

Taken together, existing theoretical and empirical research supports the idea that prior experiences of one’s own or a partner’s relationship infidelity may be associated with increased risk of those same experiences in future relationships. At the same time, research has been limited in empirically evaluating to what extent infidelity in a previous relationship, either on the part of oneself or one’s partner, predicts increased risk of infidelity in a subsequent relationship. The current study utilized longitudinal measurements of individuals’ and their partners’ infidelity behaviors across two different romantic relationships in order to provide specific information about the magnitude of increased risk associated with previous infidelity. The focus on proximal risk (from one relationship to the next) may be particularly relevant for identifying effective points of intervention, as well as for understanding the aspects of infidelity risk that persist into different relationship contexts.

Our first research question addressed whether one’s own prior engagement in infidelity predicted a higher risk of engaging in infidelity again in the future. We hypothesized that those who engaged in ESI in one relationship would be more likely to engage in ESI again in the next relationship compared to those who did not engage in ESI in the first relationship.

Our second research question addressed serial partner infidelity: we asked whether individuals who had previous partners who engaged in ESI were more likely to experience partner infidelity again in a later relationship, and whether this applied to those who felt certain about previous partner ESI as well as those who only suspected, but were not certain about, previous partner ESI. Based on research on dyadic or relational contributions to infidelity risk, we hypothesized that individuals who felt certain that their partners engaged in ESI in one relationship would be more likely to report known partner ESI again in the next relationship. Similarly, we hypothesized that those who suspected their partners of ESI in one relationship would also be more likely to report suspected partner ESI again in the next relationship.

Finally, to expand the limited existing literature on perceptions of a partner’s infidelity, our third research question evaluated how personal engagement in ESI predicted subsequent perceptions of partner engagement in ESI. We hypothesized that individuals who had engaged in ESI themselves in previous relationships would be more likely to report either known or suspected ESI on the part of their next relationship partners.

Because we were interested in the process of serial infidelity, analyses in the current study controlled for a set of demographic variables relevant to cross-sectional infidelity risk. Specifically, we controlled for participants’ age, gender, socioeconomic status, and race.

Further, we identified two variables that may impact the processes underlying engagement in infidelity and could therefore act as moderators of a serial infidelity effect: gender and marital status. As previously discussed, dating relationships tend to involve different kinds of commitment than marital relationships and have higher rates of infidelity. Further, women and men tend to report different motivations for and differential reactions to infidelity. Therefore, we explored whether gender or marriage moderated the persistence of infidelity from one relationship to the next.

Participants and Procedure

Participants in the current study were drawn from a larger longitudinal study of romantic relationship development, which recruited a nationwide sample representative of English-speaking young adults in the U.S. who were in unmarried romantic relationships lasting at least two months at baseline (see Rhoades, Stanley, & Markman, 2010 ). Participants across the U.S. were recruited using a targeted-listed telephone sampling strategy. Participants who were eligible and interested in enrolling ( N = 1294) completed surveys by mail every four to six months for eleven waves of data collection, spanning approximately 5 years. Surveys were estimated to take 75 minutes to complete, and participants earned $40 per completed survey. The sample for the current study ( N = 484) consisted of all participants who answered questions about at least two different romantic relationships over the course of the study (total range = 1 to 7 relationships, M = 1.6, SD = 0.95). On average, the 484 participants selected for the current study sample completed 10.0 out of the 11 survey waves. All study procedures were approved by the principal investigator’s university Institutional Review Board.

The current sample reflected the demographic distribution of the larger sample. The current sample included 329 women (68%) and 155 men. In terms of race, this sample was, 0.8% American Indian or Alaskan Native, 2.9% Asian, 15.3% Black or African American, 0.3% Native Hawaiian or Pacific Islander, and 76.0% White; 4.7% of participants either did not report or reported more than one race. In terms of ethnicity, this sample was 7.0% Hispanic or Latino and 93.0% not Hispanic or Latino. At the time of the first wave of data collection, the sample ranged in age from 18 to 35 years old ( M = 24.8, SD = 4.73). Seventy-six percent of participants were employed at baseline. Participants had a median income of $10,000 – $14,999 per year and a median 14 years of education, both of which were representative of unmarried adults in the sample age range when the study began in the mid-2000s (U.S. Census Bureau, 2000).

Participants’ own engagement in ESI was assessed at each wave with the question, “Have you had sexual relations with someone other than your partner since you began seriously dating?” A response of “No,” was coded as 0, and responses of either “Yes, with one person,” or “Yes, with more than one person,” were coded as 1. When respondents completed more than one survey wave within the same relationship, own ESI was coded as a 1 for the relationship if respondents ever reported engaging in ESI during the relationship. Forty-four percent of the sample reported their own involvement in ESI at some point over the course of the study. Of note, this question did not assess whether the ESI was considered to be allowed or consensual in the relationship.

Known or suspected partner ESI

Perceived partner involvement in ESI was assessed at each wave with the question, “Has your partner had sexual relations with someone other than you since you began seriously dating?” Responses of “No,” and “Probably not,” were coded as 0. A response of “Yes, I think so,” was coded as 1 for the suspected partner ESI variable, and a response of “Yes, I know for sure” was coded as 1 for the known partner ESI variable. Thirty percent of participants reported known partner ESI during the study, and 18% reported suspected partner ESI. Suspected and known partner ESI were mutually exclusive categories within time points; that is, participants could report either suspected or known ESI, but not both, at each survey wave. Therefore, analyses on known partner ESI excluded participants who reported suspected partner ESI in that relationship, and vice versa. When respondents completed more than one survey wave within the same relationship, known partner ESI was coded as a 1 for the relationship if respondents ever reported known partner ESI within that relationship, at any time point. We coded suspected partner ESI as a 1 for the relationship if respondents ever reported suspected partner ESI at any time point within the relationship, but never reported known partner ESI within the relationship.

Demographics

A set of demographic control variables were used in the current study. At the baseline survey, participants reported their age, their gender, whether they were employed, their annual income, and the number of years of education they had completed. Self-identified race and ethnicity were also measured at baseline; consistent with prior research showing increased risk for infidelity among Black or African American populations, a race dummy variable coded each participant as Black/African American (1) or not (0). All participants began the study in unmarried relationships, but one-third of the sample got married during the course of the study. In order to model marriage as a potential moderator of serial infidelity, we included a variable coding whether participants married during the study (1) or not (0).

Data Analytic Plan

Data were utilized from the first two relationships that participants in our sample reported over the course of data collection. A series of separate models used logistic regression to test our first two research questions about whether a particular infidelity experience in the first relationship (own ESI, known partner ESI, or suspected partner ESI) predicted a greater likelihood of having the same infidelity experience again in the second relationship. Specifically, we tested whether those who reported their own ESI in their first relationships were more likely to report their own ESI again in their second relationships compared to those with no reported own ESI in their first relationships (first research question). Similarly, we tested whether those reporting known partner ESI in first relationships were more likely to report known partner ESI in second relationships, and whether those reporting suspected partner ESI in first relationships were more likely to report suspected partner ESI in second relationships (second research question). All models controlled for a set of demographic control variables relevant to infidelity risk, including age, gender, race, education, employment status, and income.

Next, we evaluated our third research question, which asked whether one’s own previous engagement in ESI changed the likelihood of knowing about or suspecting partner ESI in the next relationship. In separate analyses, we tested whether those reporting their own ESI in their first relationships were more likely to report either known or suspected partner ESI in their second relationships compared to those who did not report their own ESI. Again, all models controlled for relevant demographic variables.

Finally, we conducted follow-up analyses to determine whether gender or marriage moderated the persistence of infidelity across relationships. We tested whether the interaction of these variables with first-relationship ESI behaviors predicted second-relationship ESI behaviors in each model described previously.

Basic relationship characteristics indicated that participants tended to have been in their first relationships longer than their second relationships: first relationships lasted an average of 38.8 months before they ended, and second relationships had lasted an average of 29.6 months by the conclusion of the study. There were also differences in likelihood of living together. Sixty-five percent of participants reported living together with their first-relationship partners at some point, whereas only nineteen percent of participants reported living with their second-relationship partners.

Table 1 shows correlations between demographic control variables and whether participants reported their own ESI, known partner ESI, or suspected partner ESI during either relationship over the course of the study. Black or African American participants were more likely to report own ESI as well as both known and suspected partner ESI than non-Black participants. More educated participants in this sample were less likely to report their own ESI and known or suspected partner ESI. Reporting suspicion of partner ESI was more likely for older participants and those who were not employed. Finally, neither gender nor income was associated with ESI.

Associations Between Demographic Variables and ESI During Either Relationship

Primary results from logistic regression models are presented in Table 2 . Participants who reported their own ESI in the first relationship were significantly more likely to report their own ESI in the second relationship, by 3.4 times compared to those who did not report engaging in ESI in the first relationship. Specifically, of the participants who reported engaging in ESI in the first relationship, 45% also reported engaging in ESI in the second relationship, whereas only 18% of the participants who did not report engaging in ESI in the first relationship reported engaging in ESI in the second relationship. Thus, the hypothesis for our first research question was supported.

Logistic Regression Results Predicting ESI Across Relationships.

Note: All analyses control for participant age, gender, race, and socioeconomic status.

Partners’ known or suspected ESI behavior was also significantly associated across relationships, supporting the hypotheses for our second research question. When compared to those who reported no partner ESI in the first relationship, participants who reported known partner ESI in the first relationship were 2.4 times more likely to report known partner ESI in the second relationship (22% compared to 9%). Further, participants who reported suspected partner ESI in the first relationship were 4.3 times more likely to suspect their second relationship partners of ESI (37% compared to 6%).

Hypotheses regarding our third research question were not supported. We found no evidence that engagement in ESI in first relationships predicted any differences in the likelihood of reporting one’s partner’s known or suspected ESI in the second relationship.

Finally, in follow-up analyses, neither gender nor marriage significantly moderated the link between ESI in first relationships and ESI in second relationships for any model. Thus, we found no evidence that the persistence of ESI from one relationship to the next differed for women versus men or for couples who married compared to those who did not.

The current study addressed an important gap in the literature on infidelity in romantic relationships by examining persistent or serial risk of infidelity across subsequent romantic relationships over time. Results from this study indicated that people who engaged in infidelity themselves, knew about a partner’s infidelity, or suspected a partner of infidelity had a higher risk of having those same infidelity experiences again in their next romantic relationships. These findings controlled for many demographic variables that are predictive of engaging in infidelity, and they did not vary based on gender or marital status.

General Infidelity Characteristics

Overall rates of infidelity in this sample were toward the high end of the range of previous estimates, with 44% of participants reporting engaging in infidelity themselves during the relationships captured by this study, 30% reporting having at least one partner who they knew engaged in infidelity, and 18% reporting that they suspected a partner of engaging in infidelity. These higher rates are expected, given that this was an unmarried sample at baseline, and unmarried samples tend to have higher rates of infidelity than married samples ( Treas & Giesen, 2000 ). Two notable departures from the prior literature were that there was no difference in the prevalence of reporting one’s own or a partner’s infidelity for women and men in this sample, and that participants with more years of education were less likely to report infidelity. These findings suggest that the existing understanding of gender and education differences in infidelity is nuanced; it likely reflects a complex interplay of social forces (e.g., power, privilege, and opportunity) that is not easily captured by simple demographic characteristics and that may be changing rapidly along with larger societal changes. Previous descriptions of demographic risk factors for infidelity do not necessarily accurately characterize the younger, unmarried population represented in the current study.

Magnitude of Serial Infidelity Risk

Our results indicated a three-fold increase in the likelihood that a person will engage in infidelity if they already have a history of engaging in ESI, and a two- to four-fold increase in the likelihood of having an partner engage in ESI if a person knew about or suspected infidelity from a past relationship partner. Thus, effects in the current study were generally medium in size.

These findings suggests that previous engagement in infidelity is an important risk factor predicting engagement in infidelity in a subsequent relationship, even after accounting for key demographic risk factors. At the same time, it is important to interpret these effects in the context of their base rates, which suggest that most people who reported either their own or their partner’s infidelity during their first relationship in this study did not report having that same experience again in their second relationship during the study timeframe. That is, although a history of infidelity may be an important risk factor of which to be aware, it is not necessarily true that someone who is “once a cheater” is “ always a cheater.” Understanding what distinguishes those who experience repeated infidelity from those who do not remains an important next step, both for understanding the development of infidelity risk and for designing effective interventions for individuals who would like to stop negative relationship behaviors and experiences from carrying over into their future relationships.

One important consideration is that first relationships and second relationships were somewhat different in the current study. First relationships were longer and more likely to involve living together. This makes sense in our sample, given that first relationships began before the study timeframe, whereas second relationships were newer simply by virtue of our data collection procedure. First relationships also necessarily ended during the study, but not all second relationships did. These differences likely explain the differences in rates of infidelity in first and second relationships. However, we do not believe these differences alter the conclusions reached from the current analyses. It may be the case that even more participants with infidelity in first relationships would have gone on to report infidelity again in the second relationships that were still ongoing, which would have strengthened the effects found in our analyses. The only circumstance that would threaten the validity of our primary conclusions about serial infidelity risk would be if participants without first-relationship infidelity were to “catch up” to those with first-relationship infidelity by reporting a greater rate of infidelity later on in second relationships, and we do not know of any reason to expect that to be the case.

Influences on Partner Infidelity

We found no evidence that reported suspected or known partners’ infidelity was related to a person’s own past history of engaging in infidelity. These null results belie the common wisdom that those who are suspicious of their partners’ fidelity have likely engaged in infidelity themselves, at least within the context of the two subsequent young adult romantic relationships captured in the present study. On the other hand, our results did indicate that even when they left one relationship and began another, people who suspected previous partner ESI were much more likely to be suspicious of their new relationship partners as well. Individual differences in trait suspiciousness or jealousy, independent of relationship context, may play a role in suspecting a relationship partner of infidelity; for example, parent relationship models (e.g., Rhoades, Stanley, Markman, & Ragan, 2012 ) and stable relationship attachment styles (e.g., Dewall et al., 2011 ) may impact persistent attitudes or beliefs about fidelity. Further, little is known about the accuracy of suspicions of infidelity. Future research investigating how frequently individuals are correct when suspecting partner infidelity could shed light on the rationale people may have for being suspicious of their partners.

Perhaps most intriguing, we found that participants who said they were certain that their previous relationship partners engaged in ESI were more than twice as likely to go on to report feeling certain that their current partner had engaged in ESI in their next relationship. We cannot make assertions about causality using data from the current study. It may be that some individuals have persistent relationship styles that tend to create a relationship context in which a partner’s infidelity is likely ( Allen et al., 2005 ). Alternatively, some people may learn that these types of behaviors are more acceptable or expected after experiencing them once (e.g., Glass & Wright, 1992 ; Simon et al., 2001 ), and thus may become more tolerant of signs of infidelity in future relationship partners. This explanation is consistent with theories that posit a bidirectional link between infidelity experiences and attitudes. It may also be the case that socioeconomic constraints, cultural values, or limited partner pools make certain individuals more likely to select or tolerate infidelity in partners again and again. For example, scholars of race and relationships posit that social factors causing an unequal gender ratio in Black communities create a context in which male infidelity is ignored, tolerated, or even considered normative ( Bowleg, Lucas, & Tschann, 2004 ; Pinderhughes, 2002 ). Finally, because we did not assess how or why participants knew about their partners’ infidelity, we must consider the possibility that even participants who reported being “certain” about their partners’ infidelity could be reporting subjective perceptions related to their own suspicion, jealousy, or other personality traits. These individuals may be more likely to repeatedly report certain knowledge of partner infidelity despite lacking definitive evidence.

Clinical Implications

In addition to filling an important gap in our understanding of serial infidelity, results from this study may be relevant for clinical interventions as well. Prevention efforts such as relationship education may help individuals interrupt the tendency to repeatedly engage in ESI in different relationships. Other researchers have identified a need for infidelity prevention to identify the people who are most at risk, and to address the contextual and situational risks that may then lead to infidelity ( Markman, 2005 ). This study demonstrated that past infidelity is an important indicator to identify those who are at continued risk of engaging in infidelity, over and above common demographic risk factors. Moreover, this study points to an opportunity for clinicians to help individuals identify the circumstances that led to past infidelity in order to avoid repeating similar patterns again in future relationships.

The findings regarding serial partner similarities may indicate an important additional target for preventative relationship education aimed at helping individuals make better decisions in their romantic lives. Relationship interventions can encourage participants to make informed choices about selecting potential partners based on those partners’ romantic histories. Interventions can also teach skills appropriate for mitigating the particular risks that may accompany having a relationship with someone who has engaged in infidelity during a previous relationship. For example, professionals have noted that couples’ abilities to discuss the individual and relationship factors that led up to infidelity is a strong indicator of successful relationship recovery ( Gordon, Baucom, & Snyder, 2004 ), and there is a growing body of research in support of explicit conversations aimed at defining relationships as they undergo transitions ( Stanley, Rhoades, & Markman, 2006 ). Preliminary research suggests that such conversations may be particularly important with regard to managing infidelity ( Knopp, Vandenberg, Rhoades, Stanley, & Markman, 2016 ). It may be that conversations aimed at reaching a mutual definition of relationship fidelity and anticipating potential barriers to maintaining fidelity could be beneficial for couples who are at risk due to past experiences or other risk factors, though this intervention remains to be empirically evaluated.

Limitations

The current study has limitations that are important to consider. First, the sample is not likely to represent all people in the United States equally well. The ages of eligible participants at the time of initial recruitment were restricted to between 18 and 35 years old; the effects of serial infidelity may be different among younger adolescents or older adults, particularly considering the different relationship structures and expectations that exist throughout the lifespan. Although the sample from the current study was a subset of a larger group of participants that was representative of the U.S. in terms of geographic location, race, and ethnicity, the smaller sample used here is not likely to reliably represent all racial and ethnic minorities, because non-White participants comprise a small number of the sample participants. In addition, inclusion criteria for participants in this study involved being in a relationship with “someone of the opposite sex” at the time of recruitment. Thus, findings may not generalize to people who have same-gender relationships.

As previously discussed, the measurement of infidelity in the current study has some limitations. Although most research on infidelity to date, including the current study, has defined infidelity as ESI (c.f. Allen et al., 2005 ; Blow & Hartnett, 2005a ), ESI is an imperfect proxy for infidelity. In particular, it is unclear in this study whether the ESI was considered to be allowed in the relationship (e.g., as in consensually non-monogamous relationships). Our data indicate that at the end of the study, 2% of our sample reported being in an “open relationship,” but we do not know this information about the majority of the relationships included in the current analysis; further, the term “open relationship” may not capture all different forms of consensual ESI. Future research could define infidelity in more precise ways to distinguish it from consensual non-monogamy and could also measure different facets of infidelity (e.g., Luo, Cartun, & Snider, 2010 ).

Finally, all data in the current study are self-reported and are therefore subject to reporting bias and shared method variance. This issue is exacerbated by the fact that infidelity is a sensitive topic that is subject to social desirability effects in research (e.g., Whisman & Snyder, 2007 ). Thus, these results may be partially explained by consistency in willingness to report infidelity: individuals who are unwilling to report their own (or a partner’s) infidelity in one relationship are probably also not willing to report infidelity at any point in the future. The current study’s procedure – collecting surveys by mail rather than in person – may have helped to ameliorate this issue, but future research could try alternative methods of collecting data about infidelity.

Conclusions

The current study provides novel contributions to established notions of infidelity across serial relationships, including that personal engagement in ESI and perceptions of partner engagement in ESI predict increased risk of serial infidelity in subsequent relationships. Infidelity can harm individuals and relationships, and these results can inform prevention or intervention efforts by targeting risk factors based on previous relationship patterns in addition to the various individual, relational, and contextual factors demonstrated to predict infidelity in previous work. Although intervention research will be necessary to explore which risk factors are most useful to address and through what mechanisms, our findings clearly demonstrate the need for researchers and clinicians to take into account previous infidelity patterns while developing an understanding of how to predict and intervene with regard to risk for serial infidelity.

Acknowledgments

Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health [award number R01HD047564]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Declaration of Conflicting Interests:

The authors declare that they have no conflicts of interest with respect to authorship or the publication of this article.

Contributor Information

Kayla Knopp, University of Denver.

Shelby Scott, Veteran Affairs Eastern Colorado Health Care System.

Lane Ritchie, University of Denver.

Galena K. Rhoades, University of Denver.

Howard J. Markman, University of Denver.

Scott M. Stanley, University of Denver.

• Adamopoulou E. New facts on infidelity. Economics Letters. 2013; 121 (3):458–462. https://doi.org/10.1016/j.econlet.2013.09.025 . [ Google Scholar ]
• Allen ES. Unpublished doctoral dissertation. University of North Carolina; Chapel Hill: 2001. Attachment styles and their relation to patterns of extradyadic and extramarital involvement. [ Google Scholar ]
• Allen ES, Atkins DC. The association of divorce and extramarital sex in a representative U.S. sample. Journal of Family Issues. 2012; 33 (11):1477–1493. https://doi.org/10.1177/0192513X12439692 . [ Google Scholar ]
• Allen ES, Atkins DC, Baucom DH, Snyder DK, Gordon KC, Glass SP. Intrapersonal, interpersonal, and contextual factors in engaging in and responding to extramarital involvement. Clinical Psychology: Science and Practice. 2005; 12 (2):101–30. https://doi.org/10.1093/clipsy.bpi014 . [ Google Scholar ]
• Allen ES, Rhoades GK, Stanley SM, Markman HJ, Williams T, Melton J, Clements ML. Premarital precursors of marital infidelity. Family Process. 2008; 47 (2):243–259. https://doi.org/10.1111/j.1545-5300.2008.00251.x . [ PubMed ] [ Google Scholar ]
• Amato PR, Previti D. People’s reasons for divorcing: Gender, social class, the life course, and adjustment. Journal of Family Issues. 2003; 24 (5):602–26. https://doi.org/10.1177/0192513X03254507 . [ Google Scholar ]
• Atkins DC, Baucom DH, Jacobson NS. Understanding infidelity: Correlates in a national random sample. Journal of Family Psychology. 2001; 15 (4):735–749. https://doi.org/10.1037/0893-3200.15.4.735 . [ PubMed ] [ Google Scholar ]
• Banfield S, McCabe MP. Extra relationship involvement among women: Are they different from men? Archives of Sexual Behavior. 2001; 30 (2):119–142. https://doi.org/10.1023/A:1002773100507 . [ PubMed ] [ Google Scholar ]
• Barta WD, Kiene SM. Motivations for infidelity in heterosexual dating couples: The roles of gender, personality differences, and sociosexual orientation. Journal of Social and Personal Relationships. 2005; 22 (3):339–360. https://doi.org/10.1177/0265407505052440 . [ Google Scholar ]
• Baucom DH, Snyder DK, Gordon KC. Helping couples get past the affair: A clinician’s guide. Guilford Press; 2011. [ Google Scholar ]
• Blow AJ, Hartnett K. Infidelity in committed relationships I: A methodological review. Journal of Marital and Family Therapy. 2005a; 31 (2):183–216. https://doi.org/10.1111/j.1752-0606.2005.tb01555.x . [ PubMed ] [ Google Scholar ]
• Blow AJ, Hartnett K. Infidelity in committed relationships II: A substantive review. Journal of Marital and Family Therapy. 2005b; 31 (2):217–33. https://doi.org/10.1111/j.1752-0606.2005.tb01556.x . [ PubMed ] [ Google Scholar ]
• Bowleg L, Lucas KJ, Tschann JM. “The ball was always in his court”: An exploratory analysis of relationship scripts, sexual scripts, and condom use among African American women. Psychology of Women Quarterly. 2004; 28 (1):70–82. https://doi.org/10.1111/j.1471-6402.2004.00124.x . [ Google Scholar ]
• Buunk BP, Bakker AB, Taylor P. Extradyadic sex: The role of descriptive and injunctive norms. Journal of Sex Research. 1995; 32 (4):313–18. https://doi.org/http://dx.doi.org/10.1080/00224499509551804 . [ Google Scholar ]
• Cano A, O’Leary KD. Infidelity and separations precipitate major depressive episodes and symptoms of nonspecific depression and anxiety. Journal of Consulting and Clinical Psychology. 2000; 68 (5):774–81. https://doi.org/10.1037/0022-006X.68.5.774 . [ PubMed ] [ Google Scholar ]
• Confer JC, Cloud MD. Sex differences in response to imagining a partner’s heterosexual or homosexual affair. Personality and Individual Differences. 2011; 50 (2):129–34. https://doi.org/10.1016/j.paid.2010.09.007 . [ Google Scholar ]
• Conley TD, Moors AC, Matsick JL, Ziegler A. The fewer the merrier? Assessing stigma surrounding consensually non-monogamous romantic relationships. Analyses of Social Issues and Public Policy. 2013; 13 (1):1–30. https://doi.org/10.1111/j.1530-2415.2012.01286.x . [ Google Scholar ]
• Dewall CN, Lambert NM, Slotter EB, Pond RS, Deckman T, Finkel EJ, Fincham FD. So far away from one’s partner, yet so close to romantic alternatives: Avoidant attachment, interest in alternatives, and infidelity. Journal of Personality and Social Psychology. 2011; 101 (6):1302–16. https://doi.org/10.1037/a0025497 . [ PubMed ] [ Google Scholar ]
• Drigotas SM, Barta W. The cheating heart: Scientific explorations of infidelity. Current Directions in Psychological Science. 2001; 10 (5):177–80. https://doi.org/10.1111/1467-8721.00143 . [ Google Scholar ]
• Drigotas SM, Safstrom CA, Gentilia T. An investment model prediction of dating infidelity. Journal of Personality and Social Psychology. 1999; 77 (3):509–24. https://doi.org/10.1037/0022-3514.77.3.509 . [ Google Scholar ]
• Edlund J, Heider J, Scherer C, Farc M, Sagarin B. Sex differences in jealousy in response to actual infidelity. Evolutionary Psychology. 2006; 4 :462–70. https://doi.org/10.1037/a0019378 . [ Google Scholar ]
• Fincham FD, May RW. Infidelity in romantic relationships. Current Opinion in Psychology. 2017; 13 :70–74. https://doi.org/10.1016/j.copsyc.2016.03.008 . [ PubMed ] [ Google Scholar ]
• Fishbein M, Ajzen I. Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley; 1975. [ Google Scholar ]
• Foster JD, Misra TA. It did not mean anything (about me): Cognitive dissonance theory and the cognitive and affective consequences of romantic infidelity. Journal of Social and Personal Relationships. 2013; 30 (7):835–57. https://doi.org/10.1177/0265407512472324 . [ Google Scholar ]
• Frederick DA, Fales MR. Upset over sexual versus emotional infidelity among gay, lesbian, bisexual, and heterosexual adults. Archives of Sexual Behavior. 2016; 45 (1):175–91. https://doi.org/10.1007/s10508-014-0409-9 . [ PubMed ] [ Google Scholar ]
• Glass SP, Wright TL. Justifications for extramarital relationships: The association between attitudes, behaviors, and gender. The Journal of Sex Research. 1992; 29 (3):361–87. https://doi.org/http://dx.doi.org/10.1080/00224499209551654 . [ Google Scholar ]
• Gordon KC, Baucom DH, Snyder DK. An integrative intervention for promoting recovery from extramarital affairs. Journal of Marital and Family Therapy. 2004; 30 (213):213–31. https://doi.org/10.1111/j.1752-0606.2004.tb01235.x . [ PubMed ] [ Google Scholar ]
• Green MC, Sabini J. Gender, socioeconomic status, age, and jealousy: Emotional responses to infidelity in a national sample. Emotion. 2006; 6 (2):330–34. https://doi.org/10.1037/1528-3542.6.2.330 . [ PubMed ] [ Google Scholar ]
• Hackathorn J, Mattingly BA, Clark EM, Mattingly MJB. Practicing what you preach: Infidelity attitudes as a predictor of fidelity. Current Psychology. 2011; 30 (4):299–311. https://doi.org/10.1007/s12144-011-9119-9 . [ Google Scholar ]
• Harris CR. Sexual and romantic jealousy in heterosexual and homosexual adults. Psychological Science. 2002; 13 (1):7–12. https://doi.org/10.1111/1467-9280.00402 . [ PubMed ] [ Google Scholar ]
• Harris CR, Christenfeld N. Gender, jealousy, and reason. Psychological Science. 1996; 7 (6):364–6. https://doi.org/http://www.jstor.org/stable/40062981 . [ Google Scholar ]
• Hughes SM, Harrison MA, Gallup GG. Sex differences in mating strategies: Mate guarding, infidelity and multiple concurrent sex partners. Psychology, Evolution & Gender. 2004; 6 (1):3–13. https://doi.org/10.1080/14616660410001733588 . [ Google Scholar ]
• Jackman M. Understanding the cheating heart: What determines infidelity intentions? Sexuality and Culture. 2015; 19 (1):72–84. https://doi.org/10.1007/s12119-014-9248-z . [ Google Scholar ]
• Johnson CA, Stanley SM, Glenn ND, Amato PR, Nock SL, Markman HJ, Dion MR. Marriage in Oklahoma: 2001 baseline statewide survey on marriage and divorce (S02096 OKDHS) Oklahoma City, OK: Oklahoma Department of Human Services; 2002. [ Google Scholar ]
• Knopp K, Vandenberg P, Rhoades GK, Stanley SM, Markman HJ. “DTR” or “WTF”: The role of infidelity in defining the relationship. Poster presented at the meeting of the American Psychological Association; Denver, CO. 2016. [ Google Scholar ]
• Lalasz CB, Weigel DJ. Understanding the relationship between gender and extradyadic relations: The mediating role of sensation seeking on intentions to engage in sexual infidelity. Personality and Individual Differences. 2011; 50 (7):1079–83. https://doi.org/10.1016/j.paid.2011.01.029 . [ Google Scholar ]
• Lammers J, Stoker JI, Jordan J, Pollmann M, Stapel DA. Power increases infidelity among men and women. Psychological Science. 2011; 22 (9):1191–7. https://doi.org/10.1177/0956797611416252 . [ PubMed ] [ Google Scholar ]
• Lichter DT, Qian Z. Serial cohabitation and the marital life course. Journal of Marriage and Family. 2008; 70 :861–78. https://doi.org/10.1111/j.1741-3737.2008.00532.x . [ Google Scholar ]
• Luo S, Cartun MA, Snider AG. Assessing extradyadic behavior: A review, a new measure, and two new models. Personality and Individual Differences. 2010; 49 (3):155–63. https://doi.org/10.1016/j.paid.2010.03.033 . [ Google Scholar ]
• Maddox Shaw AM, Rhoades GK, Allen ES, Stanley SM, Markman HJ. Predictors of extradyadic sexual involvement in unmarried opposite-sex relationships. Journal of Sex Research. 2013; 50 (6):598–610. https://doi.org/10.1080/00224499.2012.666816 . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Mark KP, Janssen E, Milhausen RR. Infidelity in heterosexual couples: Demographic, interpersonal, and personality-related predictors of extradyadic sex. Archives of Sexual Behavior. 2011; 40 (5):971–82. https://doi.org/10.1007/s10508-011-9771-z . [ PubMed ] [ Google Scholar ]
• Markman HJ. The prevention of extramarital involvement: Steps toward “affair proofing” marriage. Clinical Psychology: Science and Practice. 2005; 12 (2):134–38. https://doi.org/10.1093/clipsy/bpi016 . [ Google Scholar ]
• Martins A, Pereira M, Andrade R, Dattilio FM, Narciso I, Canavarro MC. Infidelity in dating relationships: Gender-specific correlates of face-to-face and online extradyadic involvement. Archives of Sexual Behavior. 2016; 45 (1):193–205. https://doi.org/10.1007/s10508-015-0576-3 . [ PubMed ] [ Google Scholar ]
• McAnulty R, Brineman J. Infidelity in dating relationships. Annual Review of Sex Research. 2007; 18 (1):94–114. https://doi.org/10.1080/10532528.2007.10559848 . [ Google Scholar ]
• Munsch CL. The science of two-timing: The state of infidelity research. Sociology Compass. 2012; 6 (1):46–59. https://doi.org/10.1111/j.1751-9020.2011.00434.x . [ Google Scholar ]
• Pinderhughes EB. African American marriage in the 20th century. Family Process. 2002; 41 (2):269–82. https://doi.org/10.1111/j.1545-5300.2002.41206.x . [ PubMed ] [ Google Scholar ]
• Rhoades GK, Stanley SM. Relationship education for individuals: The benefits and challenges of intervening early. In: Benson H, Callan S, editors. What works in relationship education: Lessons from academics and service deliverers in the United States and Europe. Doha, Qatar: Doha International Institute for Family Studies and Development; 2009. pp. 45–54. [ Google Scholar ]
• Rhoades GK, Stanley SM. Before “I Do”: What do premarital experiences have to do with marital quality among today’s young adults? National Marriage Project, University of Virginia; 2014. Retrieved from http://before-i-do.org . [ Google Scholar ]
• Rhoades GK, Stanley SM, Markman HJ. Should I stay or should I go? Predicting dating relationship stability from four aspects of commitment. Journal of Family Psychology. 2010; 24 (5):543–50. https://doi.org/http://dx.doi.org/10.1037/a0021008 . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Rhoades GK, Stanley SM, Markman HJ, Ragan EP. Parents’ marital status, conflict, and role modeling: Links with adult romantic relationship quality. Journal of Divorce & Remarriage. 2012; 53 (5):348–67. https://doi.org/10.1080/10502556.2012.675838 . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Ritchie LL, van Anders SM. There’s jealousy…and then there’s jealousy: Differential effects of jealousy on testosterone. Adaptive Human Behavior and Physiology. 2015; 1 (2):231–46. https://doi.org/10.1007/s40750-015-0023-7 . [ Google Scholar ]
• Rusbult CE. A longitudinal test of the investment model: The development (and deterioration) of satisfaction and commitment in heterosexual involvements. Journal of Personality and Social Psychology. 1983; 45 (1):101–17. [ Google Scholar ]
• Sassler S. Partnering across the life course: Sex, relationships, and mate selection. Journal of Marriage and Family. 2010; 72 (3):557–75. https://doi.org/10.1111/j.1741-3737.2010.00718.x . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Scott SB, Parsons A, Post KM, Stanley SM, Markman HJ, Rhoades GK. Changes in the sexual relationship and relationship adjustment precede extradyadic sexual involvement. Archives of Sexual Behavior. 2016; 46 (2):395–406. https://doi.org/10.1007/s10508-016-0797-0 . [ PubMed ] [ Google Scholar ]
• Scott SB, Rhoades GK, Stanley SM, Allen ES, Markman HJ. Reasons for divorce and recollections of premarital intervention: Implications for improving relationship education. Couple & Family Psychology. 2013; 2 (2):131–45. https://doi.org/10.1037/a0032025 . [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Sharpe DI, Walters AS, Goren MJ. Effect of cheating experience on attitudes toward infidelity. Sexuality and Culture. 2013; 17 (4):643–58. https://doi.org/10.1007/s12119-013-9169-2 . [ Google Scholar ]
• Simon TR, Anderson M, Thompson MP, Crosby AE, Shelley G, Sacks JJ. Attitudinal acceptance of intimate partner violence among U.S. adults. Violence and Victims. 2001; 16 (2):115–26. [ PubMed ] [ Google Scholar ]
• Solstad K, Mucic D. Extramarital sexual relationships of middle-aged Danish men: Attitudes and behavior. Maturitas. 1999; 32 (1):51–9. https://doi.org/10.1016/S0378-5122(99)00012-2 . [ PubMed ] [ Google Scholar ]
• Stanley SM, Rhoades GK, Markman HJ. Sliding versus deciding: Inertia and the premarital cohabitation effect. Family Relations. 2006; 55 :499–509. https://doi.org/10.1111/j.1741-3729.2006.00418.x . [ Google Scholar ]
• Thibaut JW, Kelley HH. The social psychology of groups. New Brunswick: Transaction Books; 1959. [ Google Scholar ]
• Treas J, Giesen D. Sexual infidelity among married and cohabiting americans. Journal of Marriage and Family. 2000; 61 (1):48–60. https://doi.org/10.1111/j.1741-3737.2000.00048.x . [ Google Scholar ]
• Varga CM, Gee CB, Munro G. The effects of sample characteristics and experience with infidelity on romantic jealousy. Sex Roles. 2011; 65 (11–12):854–66. https://doi.org/10.1007/s11199-011-0048-8 . [ Google Scholar ]
• Whisman MA, Dixon AE, Johnson B. Therapists’ perspectives of couple problems and treatment issues in couple therapy. Journal of Family Psychology. 1997; 11 (3):361–6. https://doi.org/10.1037/0893-3200.11.3.361 . [ Google Scholar ]
• Whisman MA, Gordon KC, Chatav Y. Predicting sexual infidelity in a population-based sample of married individuals. Journal of Family Psychology. 2007; 21 (2):320–4. https://doi.org/10.1037/0893-3200.21.2.320 . [ PubMed ] [ Google Scholar ]
• Whisman MA, Snyder DK. Sexual infidelity in a national survey of American women: Differences in prevalence and correlates as a function of method of assessment. Journal of Family Psychology. 2007; 21 (2):147–54. https://doi.org/10.1037/0893-3200.21.2.147 . [ PubMed ] [ Google Scholar ]
• Wiederman MW. Extramarital sex: Prevalence and correlates in a national survey. Journal of Sex Research. 1997; 34 (2):167–74. https://doi.org/10.1080/00224499709551881 . [ Google Scholar ]
• Wiederman MW, Hurd C. Extradyadic involvement during dating. Journal of Social and Personal Relationships. 1999; 16 (2):265–74. https://doi.org/10.1177/0265407599162008 . [ Google Scholar ]

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

Sulfur isotope characteristics in the qian-3 4 section of the qianjiang depression and its implications for the paleoenvironment.

1. Introduction

2. geological settings.

Click here to enlarge figure

3. Materials and Methods

3.1. elemental composition analyses, 3.2. bulk sample pyrite’s sulfur isotope analyses, 5. model of pyrite’s sulfur isotopes, 5.1. one-dimensional diffusion–advection–reaction model, 5.2. modeling results, 5.3. sensitivity test, 6. conclusions, author contributions, data availability statement, conflicts of interest.

• Wei, H.Y.; Wei, X.M.; Qiu, Z.; Song, H.Y.; Shi, G. Redox conditions across the G-L boundary in South China: Evidence from pyrite morphology and sulfur isotopic compositions. Chem. Geol. 2016 , 440 , 1–14. [ Google Scholar ] [ CrossRef ]
• Canfield, D.E.; Teske, A. Late Proterozoic rise in atmospheric oxygen concentration inferred from phylogenetic and sulphur-isotope studies. Nature 1996 , 382 , 127–132. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Fike, D.A.; Bradley, A.S.; Rose, C.V. Rethinking the Ancient Sulfur Cycle. Annu. Rev. Earth Planet. Sci. 2015 , 43 , 593–622. [ Google Scholar ] [ CrossRef ]
• Gill, B.C.; Lyons, T.W.; Saltzman, M.R. Parallel, high-resolution carbon and sulfur isotope records of the evolving Paleozoic marine sulfur reservoir. Palaeogeogr. Palaeoclimatol. Palaeoecol. 2007 , 256 , 156–173. [ Google Scholar ] [ CrossRef ]
• Ding, X.D.; Li, D.W.; Zheng, L.W.; Bao, H.Y.; Chen, H.F.; Kao, S.J. Sulfur Geochemistry of a Lacustrine Record from Taiwan Reveals Enhanced Marine Aerosol Input during the Early Holocene. Sci. Rep. 2016 , 6 , 38989. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Berner, R.A. Sedimentary Pyrite Formation: An Update. Geochim. Cosmochim. Acta 1984 , 48 , 605–615. [ Google Scholar ] [ CrossRef ]
• Gomes, M.L.; Fike, D.A.; Bergmann, K.D.; Jones, C.; Knoll, A.H. Environmental insights from high-resolution (SIMS) sulfur isotope analyses of sulfides in Proterozoic microbialites with diverse mat textures. Gebiology 2018 , 16 , 17–34. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Halevy, I.; Peters, S.E.; Fischer, W.W. Sulfate Burial Constraints on the Phanerozoic Sulfur Cycle. Science 2012 , 337 , 331–334. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Shawar, L.; Halevy, I.; Said-Ahmad, W.; Feinstein, S.; Boyko, V.; Kamyshny, A.; Amrani, A. Dynamics of pyrite formation and organic matter sulfurization in organic-rich carbonate sediments. Geochim. Cosmochim. Acta 2018 , 241 , 219–239. [ Google Scholar ] [ CrossRef ]
• Strauss, H. The isotopic composition of sedimentary sulfur through time. Palaeogeogr. Palaeoclimatol. Palaeoecol. 1997 , 132 , 97–118. [ Google Scholar ] [ CrossRef ]
• Lang, X.; Tang, W.; Ma, H.; Shen, B. Local environmental variation obscures the interpretation of pyrite sulfur isotope records. Earth Planet. Sci. Lett. 2020 , 533 , 116056. [ Google Scholar ] [ CrossRef ]
• Leavitt, W.D.; Halevy, I.; Bradley, A.S.; Johnston, D.T. Influence of sulfate reduction rates on the Phanerozoic sulfur isotope record. Proc. Natl. Acad. Sci. USA 2013 , 110 , 11244–11249. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Rickard, D. Kinetics of pyrite formation by the H 2 S oxidation of iron (II) monosulfide in aqueous solutions between 25 and 125 °C: The rate equation. Geochim. Cosmochim. Acta 1997 , 61 , 115–134. [ Google Scholar ] [ CrossRef ]
• Sim, M.S.; Bosak, T.; Ono, S. Large Sulfur Isotope Fractionation Does Not Require Disproportionation. Science 2011 , 333 , 74–77. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Wing, B.A.; Halevy, I. Intracellular metabolite levels shape sulfur isotope fractionation during microbial sulfate respiration. Proc. Natl. Acad. Sci. USA 2014 , 111 , 18116–18125. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Liu, X.T.; Fike, D.; Li, A.C.; Dong, J.; Xu, F.J.; Zhuang, G.C.; Rendle-Bühring, R.; Wan, S.M. Pyrite sulfur isotopes constrained by sedimentation rates: Evidence from sediments on the East China Sea inner shelf since the late Pleistocene. Chem. Geol. 2019 , 505 , 66–75. [ Google Scholar ] [ CrossRef ]
• Gomes, M.L.; Hurtgen, M.T. Sulfur isotope fractionation in modern euxinic systems: Implications for paleoenvironmental reconstructions of paired sulfate-sulfide isotope records. Geochim. Cosmochim. Acta 2015 , 157 , 39–55. [ Google Scholar ] [ CrossRef ]
• Pósfai, M.; Dunin-Borkowski, R.E. Sulfides in Biosystems. Rev. Miner. Geochem. 2006 , 61 , 679–714. [ Google Scholar ] [ CrossRef ]
• Raiswell, R.; Whaler, K.; Dean, S.; Coleman, M.L.; Briggs, D.E.G. A simple three-dimensional model of diffusion-with-precipitation applied to localised pyrite formation in framboids, fossils and detrital iron minerals. Mar. Geol. 1993 , 113 , 89–100. [ Google Scholar ] [ CrossRef ]
• Nielsen, J.K.; Shen, Y. Evidence for sulfidic deep water during the Late Permian in the East Greenland Basin. Geology 2004 , 32 , 1037–1040. [ Google Scholar ] [ CrossRef ]
• Wilkin, R.T.; Arthur, M.A.; Dean, W.E. History of water-column anoxia in the Black Sea indicated by pyrite framboid size distributions. Earth Planet. Sci. Lett. 1997 , 148 , 517–525. [ Google Scholar ] [ CrossRef ]
• Xing, C.C.; Lang, X.G.; Ma, H.R.; Peng, Y.; Peng, Y.B.; Liu, Y.R.; Wang, R.M.; Ning, M.; Cui, Y.X.; Yu, X.; et al. Predominant microbial iron reduction in sediment in early Cambrian sulfidic oceans. Glob. Planet. Chang. 2021 , 206 , 103637. [ Google Scholar ] [ CrossRef ]
• Zhu, D.Y.; Liu, Q.Y.; Zhou, B.; Jin, Z.J.; Li, T.Y. Sulfur isotope of pyrite response to redox chemistry in organic matter-enriched shales and implications for components of shale gas. Interpretation 2018 , 6 , SN71–SN83. [ Google Scholar ] [ CrossRef ]
• Wu, L.; Mei, L.; Paton, D.A.; Guo, P.; Liu, Y.; Luo, J.; Wang, D.; Li, M.; Zhang, P.; Wen, H. Deciphering the origin of the Cenozoic intracontinental rifting and volcanism in eastern China using integrated evidence from the Jianghan Basin. Gondwana Res. 2018 , 64 , 67–83. [ Google Scholar ] [ CrossRef ]
• Kong, X.; Jiang, Z.; Ju, B.; Liang, C.; Cai, Y.; Wu, S. Fine-grained carbonate formation and organic matter enrichment in an Eocene saline rift lake (Qianjiang Depression): Constraints from depositional environment and material source. Mar. Pet. Geol. 2022 , 138 , 105534. [ Google Scholar ] [ CrossRef ]
• Li, M.; Chen, Z.; Qian, M.; Ma, X.; Jiang, Q.; Li, Z.; Tao, G.; Wu, S. What are in pyrolysis S1 peak and what are missed? Petroleum compositional characteristics revealed from programed pyrolysis and implications for shale oil mobility and resource potential. Int. J. Coal Geol. 2020 , 217 , 103321. [ Google Scholar ] [ CrossRef ]
• Huang, C.; Hinnov, L. Evolution of an Eocene–Oligocene Saline Lake Depositional System and Its Controlling Factors, Jianghan Basin, China. J. Earth Sci. 2014 , 25 , 959–976. [ Google Scholar ] [ CrossRef ]
• Fang, Q.; Li, Y. Exhaustive brine production and complete CO 2 storage in Jianghan Basin of China. Environ. Earth Sci. 2014 , 72 , 1541–1553. [ Google Scholar ] [ CrossRef ]
• Hou, Y.G.; Wang, F.R.; He, S.; Dong, T.; Wu, S.Q. Properties and shale oil potential of saline lacustrine shales in the Qianjiang Depression, Jianghan Basin, China. Mar. Pet. Geol. 2017 , 86 , 1173–1190. [ Google Scholar ] [ CrossRef ]
• Dai, S. Petroleum Geology of Jianghan Saline Basin ; Petroleum Industry Publishing House: Beijing, China, 1997. [ Google Scholar ]
• Xu, L.X.; Yan, C.D.; Yu, H.L.; Wang, B.Q.; Yu, F.Q.; Wang, D.F. Chronology of Paleogene volcanic rocks in Jianghan Basin. Oil Gas Geol. 1995 , 16 , 132–137. [ Google Scholar ] [ CrossRef ]
• Wei, R.; Ma, H.; Jin, Z.; Wang, T.; Zhang, C.; Wang, Y.; Dong, L. New insights into the carbon cycle and depositional models of the Eocene saline lake, Jianghan basin, China. Mar. Pet. Geol. 2023 , 149 , 106079. [ Google Scholar ] [ CrossRef ]
• Huang, K.-J.; Shen, B.; Lang, X.-G.; Tang, W.-B.; Peng, Y.; Ke, S.; Kaufman, A.J.; Ma, H.-R.; Li, F.-B. Magnesium isotopic compositions of the Mesoproterozoic dolostones: Implications for Mg isotopic systematics of marine carbonates. Geochim. Cosmochim. Acta 2015 , 164 , 333–351. [ Google Scholar ] [ CrossRef ]
• Lang, X.G.; Shen, B.; Peng, Y.B.; Xiao, S.H.; Zhou, C.M.; Bao, H.M.; Kaufman, A.J.; Huang, K.J.; Crockford, P.W.; Liu, Y.G.; et al. Transient marine euxinia at the end of the terminal Cryogenian glaciation. Nat. Commun. 2018 , 9 , 3019. [ Google Scholar ] [ CrossRef ]
• Schulz, H.D.; Zabel, M. Marine Geochemistry ; Springer: Berlin/Heidelberg, Germany, 2006. [ Google Scholar ]
• Rees, C.E. A steady-state model for sulphur isotope fractionation in bacterial reduction processes. Geochim. Cosmochim. Acta 1973 , 37 , 1141–1162. [ Google Scholar ] [ CrossRef ]

Share and Cite

Wang, T.; Wei, R.; Ling, K.; Dong, L. Sulfur Isotope Characteristics in the Qian-3 4 Section of the Qianjiang Depression and Its Implications for the Paleoenvironment. Minerals 2024 , 14 , 626. https://doi.org/10.3390/min14060626

Wang T, Wei R, Ling K, Dong L. Sulfur Isotope Characteristics in the Qian-3 4 Section of the Qianjiang Depression and Its Implications for the Paleoenvironment. Minerals . 2024; 14(6):626. https://doi.org/10.3390/min14060626

Wang, Tianyu, Ren Wei, Kun Ling, and Lin Dong. 2024. "Sulfur Isotope Characteristics in the Qian-3 4 Section of the Qianjiang Depression and Its Implications for the Paleoenvironment" Minerals 14, no. 6: 626. https://doi.org/10.3390/min14060626

Article Metrics

Further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals

Will China’s historic paper on Chang’e-6 lunar far side samples be in English or Chinese?

• The Chang’e-6 mission’s cargo is expected to yield a wealth of research but debate is growing about what language it will be published in first

Within China’s scientific community, there persists a notion that publishing in English is not only a medium of communication, but also a bridge to global recognition. The use of Chinese remains taboo, a silent sacrifice at the altar of international acceptance.

When China’s Chang’e-5 mission in 2020 retrieved the first lunar samples in decades from the moon’s near side, the first research was carried out by a joint team of Chinese and Western scientists and appeared in Science magazine in October 2021.

This was followed by three more scientific papers published by Nature in the same month, according to an editor with the Science China Press, a scientific journal publishing company of the Chinese Academy of Sciences (CAS), recalling the global sensation.

The rocks collected in 2020 led to a number of surprising discoveries, as they turned out to be much younger than the samples brought back by the US Apollo and Soviet Luna missions in the 1960s and 1970s.

“We certainly hope that some of our country’s groundbreaking scientific and technological achievements can appear in China’s top journals, so that we can expand our influence,” said the editor, who asked not to be named.

It was not always so. Tu Youyou, who won China’s first Nobel Prize for science in 2015, published her paper on the discovery of artemisinin in the Chinese Science Bulletin in 1977.

The journal, co-sponsored by CAS and the National Natural Science Foundation of China, once published many major discoveries but since the 1990s has suffered from a lack of quality manuscripts.

Speaking at a conference in 2018, George Gao Fu – a leading scientist in the field of virology and immunology and former head of the Chinese Centre for Disease Control and Prevention (CDC) – said Chinese as a language of academic communication “used to be glorious”.

Breakthroughs, including Tu’s achievement and the discovery of high-temperature iron-based superconducting materials, had been published first in Chinese-language journals and then recognised by the world, he said.

However, for three decades China’s important scientific research results were “basically first reported by foreign journals”, Gao noted.

Interestingly, just a few years later, Gao led a landmark study by a Chinese CDC team on the epidemiology of Covid-19 that was first published in January 2020 by the New England Journal of Medicine.

The move caused controversy in China, where the public was eager for any information about the new coronavirus that causes Covid-19 as the country grappled with the early stages of the pandemic.

The response to the overseas publication of the study reflected a broader, uncomfortable dilemma for Chinese researchers: while they recognise the importance or necessity of writing in their native language, it is difficult on a practical level.

China’s Chang’e-6 touches down on far side of moon on mission to bring rock samples back to Earth

Newton’s Principia Mathematica was written in Latin. Einstein’s first influential papers were written in German. Marie Curie’s work was published in French.

Yet, since the middle of the last century, there has been a shift in the global scientific community, with most scientific research now published in a single language – English, which is spoken by only about 18 per cent of the world’s population.

While it is estimated that up to 98 per cent of global scientific research is published in English, the number of papers by Chinese scholars has been climbing.

As early as 2010, Chinese biologist Zhu Zuoyan, a CAS academician, observed that the number of papers published by scholars from China had risen from 0.2 per cent of the world’s total to 10 per cent within a decade, second only to the United States.

But China’s academic evaluation system encourages the flow of excellent papers to foreign journals, which had partly led to the country’s lack of international academic impact, despite having the second largest number of academic journals – more than 4,800 – in the world, he said.

In late 2019, Li Zhimin, former director of the Ministry of Education’s Science and Technology Development Centre, called for papers to be published in the country’s official language if the research is funded by the government.

The requirement would make it easier for funders to review research projects, facilitate exchanges with their domestic counterparts and improve the nation’s scientific literacy, he said.

A CAS physicist, who declined to be named, stressed that the proposal to “write research results on the soil of the motherland” could not simply be understood as submitting and publishing articles in domestic journals and in Chinese.

That would be “parochial”, he said. Rather, the key is to focus research on solving crucial issues or problems in China’s development, rather than blindly following global research hotspots and wasting research funds and resources.

But at an individual level, there are plenty of pragmatic reasons and incentives for researchers to do just that. Under China’s evaluation system, getting articles published in prestigious English-language journals often brings rewards.

In addition to promotion opportunities and academic honours, there is also fame, with overseas scientific recognition tending to attract wide media and public attention.

Last month, for example, biologist Zhu Jiapeng earned a prize from Nanjing University of Traditional Chinese Medicine for his “outstanding contribution” and a grant of 1 million yuan (US$138,000) as one of the lead authors in a study published by Nature.

Astronomer Deng Licai, with the National Astronomical Observatories under CAS, believes that research results from national missions such as the Chang’e programme should be prioritised for publication in domestic journals.

Deng, who has been a team leader on China’s giant telescope project since its development in the 1990s and 2000s, said he insisted that the first batch of studies to emerge from the Large Sky Area Multi-Object Fibre Spectroscopic Telescope (Lamost) appeared in domestic journals.

“This can firstly highlight the nationality of these independent and cutting-edge major scientific projects, and also help to enhance the international impact of domestic academic journals,” he said.

But English has become the international scientific community’s lingua franca and should be used as a medium of communication, Deng said, adding that it had nothing to do with politics.

China’s space plans: lunar GPS, a 3D-printed moon base and soil samples from Mars

According to Deng, the scientists who study the Chang’e-6 lunar samples could consider publishing in some of China’s English-language journals, such as Research in Astronomy and Astrophysics (RAA).

“All of our pre-research articles on the Lamost programme published in RAA have made it into international lists of highly cited articles,” he said.

Chinese Academy of Social Sciences researcher Zhu Rui, who prefers to publish in Chinese journals – partly because the academy encourages it – said that using his own language when writing academic papers is not an obstacle, as long as the scientific community maintains substantive communication.

Unmasking academic cheating behavior in the artificial intelligence era: Evidence from Vietnamese undergraduates

• Open access
• Published: 05 February 2024

Cite this article

You have full access to this open access article

• Hung Manh Nguyen   ORCID: orcid.org/0000-0003-2640-6650 1 &
• Daisaku Goto   ORCID: orcid.org/0000-0002-9117-3761 2 , 3  

4045 Accesses

Explore all metrics

The proliferation of artificial intelligence (AI) technology has brought both innovative opportunities and unprecedented challenges to the education sector. Although AI makes education more accessible and efficient, the intentional misuse of AI chatbots in facilitating academic cheating has become a growing concern. By using the indirect questioning technique via a list experiment to minimize social desirability bias, this research contributes to the ongoing dialog on academic integrity in the era of AI. Our findings reveal that students conceal AI-powered academic cheating behaviors when directly questioned, as the prevalence of cheaters observed via list experiments is almost threefold the prevalence of cheaters observed via the basic direct questioning approach. Interestingly, our subsample analysis shows that AI-powered academic cheating behaviors differ significantly across genders and grades, as higher-grade female students are more likely to cheat than newly enrolled female students. Conversely, male students consistently engage in academic cheating throughout all grades. Furthermore, we discuss potential reasons for the heterogeneous effects in academic cheating behavior among students such as gender disparity, academic-related pressure, and peer effects. Implications are also suggested for educational institutions to promote innovative approaches that harness the benefits of AI technologies while safeguarding academic integrity.

Similar content being viewed by others

Online Academic Cheating in the Twenty-First Century

Ethics of Artificial Intelligence in Academic Research and Education

Avoid common mistakes on your manuscript.

1 Introduction

Artificial intelligence (AI) has emerged as a transformative technology, reshaping how businesses and individuals interact, communicate, and access services (Kutyauripo et al., 2023 ; Olan et al., 2022 ; Phan et al., 2023 ; Wang et al., 2023 ). The rapid adoption of these intelligent virtual applications has occurred across many sectors such as business, agriculture, transportation, and healthcare services (Ali et al., 2023 ; Du et al., 2023 ; Kulkov, 2021 ; Kumar et al., 2023 ; Wang et al., 2022 ). In a similar vein, the field of education has undergone significant transformation with the incorporation of AI applications (Mubin et al., 2020 ; Qu et al., 2022 ; Udupa, 2022 ). Specifically, AI virtual assistants are altering teacher-student interactions, content delivery, and learning methods (Aung et al., 2022 ; Dai et al., 2023 ). By providing detailed instruction, instantaneous assistance, greater interactivity, and streamlined administration, AI-powered chatbots are revolutionizing the educational system (Ratten & Jones, 2023 ). Education is improved in terms of accessibility, efficiency, and engagement through the use of AI virtual assistants. AI-powered chatbots render lectures more accessible and productive for all educational stakeholders (Kasneci et al., 2023 ).

While AI-powered applications offer many valuable outcomes in the field of education, there are also a lot of potential drawbacks regarding data privacy, accuracy, overreliance, and ethical concerns (Guo et al., 2023 ; Kasneci et al., 2023 ; Koo, 2023 ; Sollosy & McInerney, 2022 ). Importantly, academic misconduct issues have been raised by the intervention of AI-powered chatbots, which present challenging problems for educational institutions (Fyfe, 2023 ; Sweeney, 2023 ). AI-powered chatbots, which are outfitted with sophisticated algorithms and capabilities, provide students with a wide range of assistance during assignments or exams (Ansari et al., 2023 ; Cotton et al., 2023 ; Currie, 2023 ; Dalalah & Dalalah, 2023 ; Moisset & Ciampi De Andrade, 2023 ). With the assistance of AI chatbots, students can quickly and easily access auto-generated answers, responses, or plagiarized content, pushing them to break the fundamental regulations of academic integrity (Bakar-Corez & Kocaman-Karoglu, 2023 ; Li et al., 2023 ). Importantly, students might intentionally use AI-generated responses for academic cheating purposes that appear highly credible but may not be easily detectable by any anti-plagiarism applications (Choi et al., 2023 ; Livberber & Ayvaz, 2023 ; Sweeney, 2023 ). The intricate interplay between AI chatbots and academic cheating raises emerging concerns among educational institutions in preserving the principles of academic integrity (Guo & Wang, 2023 ; Kasneci et al., 2023 ).

Although previous studies have provided valuable insights into academic cheating in the digital age, noticeable research gaps remain. First, most existing studies rely on the direct questioning approach in their data collection method to examine academic cheating behavior. For instance, Ossai et al. ( 2023 ) examined the relationship between academic performance and academic integrity among 3,214 Nigerian high school students via the direct questioning approach in a paper survey. Footnote 1 Similarly, Park ( 2020 ) examined a sample of 2,360 Korean college students by employing direct questions to measure the frequency of cheating behaviors on a 5-point Likert scale. Footnote 2 Regarding the differences in academic cheating behavior in online education and face-to-face education, Ababneh et al. ( 2022 ) used online questionnaires to investigate 176 UAE undergraduates. Footnote 3 However, examining highly sensitive issues such as academic cheating via the direct questioning approach may raise concerns about the reliability of outcomes due to the effect of social desirability bias. Specifically, social desirability bias is a widely observed phenomenon wherein individuals provide untruthful responses to align with societal norms or expectations, thus helping them to positively present themselves rather than revealing accurate or precise information (Blair & Imai, 2012 ). Biased responses can arise from the predilection to pursue social validation or an aversion to criticism. Importantly, social desirability bias potentially manifests in diverse settings, encompassing interviews, surveys, or other data collection methods that focus on self-reports, notwithstanding the anonymity afforded by these approaches (Larson, 2019 ). As a result, social desirability bias can significantly compromise the credibility and accuracy of research outcomes. The skewing of data resulting from untruthful participants can bias the findings and produce erroneous conclusions (Ahmad et al., 2023 ; Latkin et al., 2017 ; Ried et al., 2022 ). In the context of the education sector, direct responses to academic cheating might be biased, as students might conceal academic cheating behavior for a variety of reasons, often rooted in a complex interplay of academic and social reasons. For academic reasons, cheating is typically considered a violation of academic integrity regulations and can result in disciplinary actions ranging from failing a specific assignment to even expulsion from the institution. In terms of social reasons, admitting to academic dishonesty might negatively affect students’ self-esteem and reputation. As such, students may conceal their cheating behavior in basic direct questioning to avoid unexpected consequences.

Second, numerous studies have extensively examined the heterogeneity in cheating behavior by gender. For instance, Yazici et al. ( 2023 ) indicate that females report a lower prevalence of academic cheating in face-to-face education. In a similar vein, Mohd Salleh et al. ( 2013 ) highlighted that male students are more likely to violate academic integrity than their counterparts. Conversely, Ezquerra et al. ( 2018 ) and Ip et al. ( 2018 ) revealed that no difference in academic cheating exists between males and females. In addition to valuable findings related to heterogeneity in academic cheating behavior by gender, the disparity in academic cheating behavior by gender across different grades remains understudied.

Addressing these gaps is essential for developing a comprehensive understanding of academic cheating in the era of AI. This study seeks to answer the following research question: To what extent do undergraduates conceal AI-powered academic cheating behaviors when investigated using direct questioning and indirect questioning? Regarding the scope of cheating behaviors in our study, we focused on cheating history (students who had cheated) and cheating intention (students who intend to cheat in the future). By delving into this question, our study aims to uncover not only the current situation of AI-powered academic cheating among undergraduates but also the heterogeneity of AI-powered academic cheating observed among students from diverse individual characteristics. To do so, we examine a sample of 1,386 Vietnamese undergraduates to unveil academic cheating behaviors by using ChatGPT (Generative Pretrained Transformer), which is an AI-powered language model developed by OpenAI. In terms of popularity, ChatGPT reached 100 million monthly active users just two months after its launch in November 2022 and became the fastest-growing consumer application in history (UBS, 2023 ). Based on the reliable outcomes of the list experiment, our study contributes valuable insights that inform policy formulation and management strategies, ultimately striving for academic integrity in the Fourth Industrial Revolution.

The remainder of this paper is structured as follows: Section 2 provides data descriptions. Section  3 describes the research methodology and the experiment design to investigate academic cheating behaviors among undergraduates. Section  4 presents the main findings. Section  5 provides a discussion. The last section provides conclusions and explores the potential implications of preventing AI-powered academic cheating.

Our study was conducted in May 2023. We focused on one of three Vietnam regional universities, Thai Nguyen University. The experiment included three stages. In the first stage, we sent the collaboration invitations to all 9 graduate schools of Thai Nguyen University, as these administrative formalities are mandatory in Vietnam. Consequently, we obtained acceptance letters from 4 graduate schools as follows: Graduate School of Education, Graduate School of Medicine and Pharmacy, Graduate School of Engineering, and Graduate School of Information Technology. We then confirmed the total number of undergraduates in all participating graduate schools and selected an initial sample of 1,450 participants. The number of participants in each graduate school was proportionally limited to the total number of undergraduates in all four schools. In the second stage, we transferred survey invitations attached with QR code access to the online survey powered by Qualtrics to participating graduate schools. In the last stage, each graduate school distributed survey invitations to all their undergraduates via internal management systems. The number of responses in each graduate school was proportionally limited by the system according to the total number of students in all 4 graduate schools. From 9 May 2023 to 12 May 2023, we received a total of 1,386 valid responses. The distribution of respondents across the four universities is shown in Appendix Table 6 .

Regarding the awareness among undergraduates about punishment for academic misconduct, all participating graduate schools regularly inform their students about the punishment policy for academic cheating (including AI-powered academic cheating) at the beginning of each academic semester. All academic misconduct is strictly prohibited, and offenders have to face strict punishments including expulsion from educational institutions. Footnote 4

Table 1 shows descriptive statistics of respondents in our study. On average, students are approximately 20.3 years old. Male students are dominant, as they account for 57.3% of respondents. In terms of grade, newly enrolled students represent more than one-third of the sample. Footnote 5 Regarding ethnicity, 26.6% of respondents were minority ethnic students. In terms of after-school activities, nearly three-fourths of the students were members of social associations, while 26.3% of students reported that they engaged in part-time jobs.

3.1 List experiment

The list experiment, also referred to as the item count technique or unmatched count technique, is a survey method used in social sciences and polling to collect sensitive or confidential information from respondents while maintaining their anonymity (Blair & Imai, 2012 ; Li & Van den Noortgate, 2022 ; Igarashi & Nagayoshi, 2022 ). The indirect questioning method is especially effective for examining sensitive topics that respondents may be reluctant to admit openly, such as illegal activities, socially undesirable behaviors, or stigmatized beliefs (Hinsley et al., 2019 ). While maintaining respondent anonymity, list experiments enable researchers to collect more precise and trustworthy data on sensitive topics. The list experiment method has been used in a wide range of social topics, including political science, public health, discrimination, consumer behavior, and food security (Eriksen et al., 2018 ; Harris et al., 2018 ; Lépine et al., 2020 ; Nicholson & Huang, 2022 ; Song et al., 2022 ; Tadesse et al., 2020 ).

The basic design of the list experiment includes two distinct groups: the control group and the treatment group. The control group is presented with a list containing n nonsensitive statements. The treatment group contains the same n nonsensitive statements as the control group, plus an additional sensitive statement. Respondents are then required to report only the total number of statements that are associated with them without specifying exactly specific statements (Blair & Imai, 2012 ). The prevalence of sensitive behavior is measured by comparing the average number of statements reported between the control group and the treatment group. The difference in averages is used to infer the prevalence of the sensitive item without revealing individual responses—an indirect questioning approach. The key assumption in the list experiment is that respondents in both groups will, on average, provide truthful answers about nonsensitive statements (Imai, 2011 ). Therefore, any difference in the average counts between the treatment and groups can be attributed to the prevalence of respondents who are associated with the sensitive statements.

3.2 Experiment design

We adopt the basic design of the list experiment with a few adjustments to reveal responses to multiple academic cheating-related statements based on our sample. Specifically, we designed a control group and two separate treatment groups. The respondents were randomly allocated to one of the three groups. Table 2 describes the detailed experimental design. Our experiment included two separate phases. Phase 1 (list experiment) aimed to investigate AI-powered academic cheating behaviors via indirect questioning. On the other hand, Phase 2 (direct questioning) helps to investigate AI-powered academic cheating behaviors via the direct questioning approach.

Phase 1 includes the participation by all groups. Respondents in the control group received a list containing four nonsensitive statements. Treatment group 1 received a list that included a similar list of nonsensitive statements as the control group and an additional sensitive statement that helps to measure the prevalence of students who had cheated by using the ChatGPT (cheating history). Similarly, the list for treatment group 2 is equipped with an additional sensitive statement along with four nonsensitive statements from the control group to measure the prevalence of students who intend to cheat by using the ChatGPT (cheating intention). In Phase 1, all the respondents were required to indicate only the total number of statements that they agreed with. Consequently, we can calculate the average response value of each group. We then captured the prevalence of cheating by calculating the difference in the average response value between the control group and treatment group 1. Similarly, the prevalence of students who intend to cheat is calculated by the difference in the average response value between the control group and treatment group 2.

Next, we investigated academic cheating behaviors via direct questioning (Phase 2). Only respondents in the control group participated in this phase. To guarantee the accuracy of the outcomes, Phase 2 includes participation only by respondents in the control group because these respondents did not engage with sensitive statements during the list experiment, as opposed to respondents in the treatment groups. In Phase 2, respondents in the control group were required to answer only “yes” or “no” for two direct academic cheating-related questions (cheating history and cheating intention). By doing so, we can observe the prevalence of respondents who are associated with cheating history and cheating intention via direct questioning.

3.3 List experiment assumptions

To estimate the prevalence of sensitive behaviors, list experiments must satisfy three key assumptions: (1) random assignment, (2) no liars, and (3) no design effect (Imai, 2011 ). These three assumptions are empirically validated in this subsection.

First, we ran balance tests to confirm whether respondents were allocated randomly to the treatment regardless of demographic variables. Accurate causal analysis, reduced bias, increased statistical power, and generalizability all depend on list experiments having guaranteed randomization of treatment (Imai, 2011 ). Individuals are assigned to different treatment groups at random when randomization is used. It is crucial in any experimental design to keep the control and treatment groups similar in terms of respondent characteristics. Table 3 depicts the outcomes of the balance tests. Since no significant difference in respondent characteristics exists, we can confirm that random assignment was well guaranteed in our list experiment.

Second, the concept of "no liars", validated through the absence of floor and ceiling effects, plays a pivotal role within the framework of the list experiment. The floor effect manifests when certain groups of respondents consistently express disagreement with all survey statements, while the ceiling effect occurs when respondents consistently report affirmative responses to all statements. Such deceptive response patterns often stem from concerns about privacy among respondents, and these effects can undermine the reliability of estimates derived from a list experiment. If a significant number of respondents consistently select extreme response options, the accuracy of the estimated prevalence of sensitive attitudes is questioned (Blair & Imai, 2012 ). To counteract these effects, we applied the design method of Glynn ( 2013 ) by including at least one nonsensitive statement predicted to be rejected by the majority of respondents and another nonsensitive statement predicted to be accepted by the majority. Based on the distribution of response values presented Appendix in Table 7 , it is evident that there were no instances of ceiling or floor effects, as the proportions of entirely affirmative or entirely negative responses in our list experiment were all below 9% of all responses.

Finally, we examine whether the design effect appears in our list experiment. A design effect exists when the presence of a sensitive item alters respondents' tendencies to select nonsensitive items. Since list experiments rely on differences in the average value of statements chosen between treatment and control groups, the selection of nonsensitive items should not be affected by the presence of sensitive statements (Blair & Imai, 2012 ). Design effects pertain to alterations in an individual's responses to innocuous statements due to the inclusion of sensitive statements. They impact the dependability of results derived from a list experiment, signifying a heightened influence of intricate sampling or design elements on the estimates. Consequently, the reliability of these estimates may diminish, posing challenges for drawing precise conclusions. To address this, we applied the design effect test package of Tsai ( 2019 ) to ascertain the presence of design effects. Based on the outcomes described in Appendix Table 8 , no design effects existed in our list experiment.

3.4 Empirical strategy

Our primary objective is to examine the magnitude of misreporting about AI-powered academic cheating behaviors among respondents. To do so, we first estimate the prevalence of academic cheating behaviors among undergraduates via list experiment by employing the estimation model of Lépine et al. ( 2020 ), with modifications by controlling for multiple covariates and school-level fixed effects Footnote 6 as follows:

\({Y}_{is}\) represents the response value (number of statements that the respondent agrees with) reported by respondent i in school s . \({\alpha }_{1}\) is the intercept, indicating the constant term in the model. \({T}_{is}\) represents binary treatment variables of respondent i in school s ( \({T}_{is}\) = 0 for the control group and \({T}_{is}\) = 1 for the treatment group) . \({\tau }_{1}\) corresponds to the prevalence of sensitive cheating behavior elaborated in Section 3.1 , which is equivalent to the difference in average response value between the control and treatment groups. \({\mathbf{X}}_{{\varvec{i}}{\varvec{s}}}\) is a vector of student-level covariates of respondent i in the school s , including age, gender, ethnicity, grade, social association membership, and part-time job engagement while \(\delta\) is the coefficient associated with these covariates. \({\theta }_{s}\) denotes the school-level fixed effects, which capture unobserved school-specific characteristics, and \({\varepsilon }_{is}\) is the error term that represents unobserved factors or random variations in the dependent variable \({Y}_{is}\) .

To measure the misreporting magnitude in responses among respondents between direct questioning and indirect questioning, we consequently compare the differences in outcomes obtained via list experiment and direct questioning. To quantify this, we use the immediate form of a two-sample t-test with the unequal variances option to compare the estimated prevalence of academic cheating behaviors obtained from the list experiment with the prevalence of affirmative responses to academic cheating behavior obtained from direct questioning.

We further examine heterogeneity in AI-powered academic cheating behaviors across different subsamples. Equation  2 represents our estimation model to evaluate the heterogeneous effects in the subsamples:

in which \({G}_{is}\) is the subsample dummy for respondent i in school s for potential factors . For instance, when we examine the heterogeneous effects of academic cheating behaviors by gender, \({G}_{is}\) is equal to 1 for male respondents and 0 for female respondents (i.e., male dummy). \({\alpha }_{2}\) is the intercept, indicating the constant term in the model. \({\tau }_{2}\) indicates the prevalence of academic cheating behavior among the subsample when \({G}_{is}\) = 0, which is equivalent to the difference in average response value between the control and treatment groups in that subsample. \({\tau }_{2}+ \gamma\) indicates the prevalence of sensitive cheating behavior in the subsample when \({G}_{is}\) = 1. Hence, \(\gamma\) corresponds to the difference in the prevalence of academic cheating behavior among subsamples. \({v}_{is}\) is the error term that represents unobserved factors or random variations in the dependent variable \({Y}_{is}\) .

Our main findings are highlighted in this section. First, we present the results of both the list experiment and direct questioning, as well as the misreporting magnitude observed from these two questioning techniques. Next, we investigate the heterogeneous effects of AI-powered academic cheating behaviors among subsamples.

4.1 Misreporting magnitude

The prevalence of students who reported that they had cheated by using ChatGPT increased significantly according to the list experiment. Table 4 depicts the prevalence of academic cheating behaviors and the magnitude of misreporting between the two questioning methods. Regarding the outcomes of direct questioning, only 9.6% of respondents reported that they had cheated. However, the prevalence of cheaters rose nearly threefold to 23.7% via the list experiment. The results suggest that confessing to cheating was an especially sensitive issue among students, as the misreporting magnitude between indirect and direct questioning was 14 percentage points (significant at the 5% level). In terms of cheating intention, no significant differences exist between the two questioning methods, as the prevalence of students reporting that they have the intention to cheat between the list experiment and the direct questioning method remains similar (21.6% and 22.5%, respectively).

4.2 Subsample analysis

Subsample analysis effectively detects differential responses or outcomes among diverse demographic, social, or contextual groups. By rigorously examining heterogeneous effects among subsamples, our study found disparities in AI-powered academic cheating behavior across different subsamples.

In terms of the heterogeneous effects of academic cheating behavior by gender, male students are more likely to use ChatGPT to cheat than female students in terms of cheating history. Figure  1 , shows the disparity in cheating history among respondents by gender. In the pooled sample, 35.1% of the male students reported that they had cheated, which is more than triple the prevalence of their counterparts showing the same behavior. The magnitude of the difference between the two genders is approximately 25 percentage points, which is significant at the 10% level. Furthermore, the difference in cheating history by gender is even higher among newly enrolled students (40.1 percentage points, significant at the 5% level). Conversely, no significant differences exist in cheating history by gender in higher grades.

Heterogeneous effects of the cheating history by gender. Note: Fig. 1a represents the estimated prevalence of respondents who reported affirmative responses to cheating history by gender. Figure 1b represents the disparity in cheating history by gender (male dummy). Robust standard errors in parenthesis. *** p  <  0.01, ** p  <  0.05, * p  <  0.1

Heterogeneous effects of cheating behavior by grade among majority ethnic group. Note: Fig. 2a represents the estimated prevalence of respondents who reported affirmative to sensitive statements by grade. Figure 2b represents the disparity in cheating behaviors by grade (higher-grade dummy). Robust standard errors in parenthesis. *** p  <  0.01, ** p  <  0.05, * p  <  0.1

Importantly, the cheating history of each gender differs significantly across grades. Among female students, higher-grade female students are more likely to cheat than newly enrolled female students. As shown in Appendix Table 9 , approximately 33% of female students in higher grades reported that they had used ChatGPT to cheat, while no proof of cheating was found among newly enrolled female students. The difference in cheating history among female students across grades is 43 percentage points (significant at the 5% level). Conversely, no difference exists in cheating history among male students across grades, as male students consistently engage in academic cheating in all grades. In particular, approximately 42.5% of higher-grade male students admitted that they had cheated in comparison with 30.1% of newly enrolled male students who reported the same behavior. However, the differences in cheating history among male students across grades are not statistically significant.

With regard to the heterogeneous effects of cheating intention by gender, male and female students show no disparity in cheating intention in the pooled sample (23% and 22.4%, respectively). Correspondingly, no heterogeneous effect on academic cheating intention was found by gender across grades (as shown in Appendix Fig. 3 ).

Regarding the heterogeneous effects of academic cheating behavior by ethnicity, higher-grade students are more likely to cheat than newly enrolled students within the majority ethnic group. Figure  2 represents the heterogeneous effects of cheating behavior between newly enrolled students and higher-grade students in the majority ethnic group. Specifically, 38.3% of higher-grade students admitted that they had used ChatGPT to cheat, which is more than fourfold the prevalence of newly enrolled students reporting the same behavior. Concerning cheating intention among majority ethnic students, both newly enrolled students and higher-grade students had the intention to cheat using ChatGPT, but the difference in cheating intention between these two groups is not statistically significant.

Relevant to the heterogeneous effects of academic cheating behavior by major, only information technology students reported engagement with both cheating history and cheating intention (38.0% and 33.9%, respectively). However, there is no significant difference in cheating history between information technology majors and other majors. Furthermore, information technology students are more likely to have the intention to cheat than medicine and pharmacy students (as shown in Appendix Fig. 4 ).

4.3 Robustness tests

To examine the stability and reliability of our results, we conducted additional robustness tests by controlling for multiple covariates and fixed effects at the school level. Based on the outcomes of the robustness tests presented in Table  5 , we confirm that our results are strongly consistent with those indicated in the previous sections. In addition, we further examine the consistency of our findings regarding heterogeneous effects across subsamples. As shown in Appendix Fig. 5 , the results of robustness tests validate the consistency of the subsample analysis results.

5 Discussion

By using the indirect questioning approach via a list experiment, our findings show that students conceal academic cheating behavior under direct questioning. Any confession of academic cheating may subject the student to negative consequences. Cheating is often punishable by failing assignments or exams, academic probation, or even expulsion from academic institutions. Furthermore, students may be concerned about how their peers, teachers, and parents will perceive them if they are identified as cheaters. Admitting to academic cheating can harm their reputation as honest and capable students. Cheating is frequently associated with moral and ethical stigma. Students conceal their cheating to avoid feelings of shame, guilt, or remorse associated with their dishonest behavior. Consequently, respondents understandably conceal truthful answers when directly questioned.

Our subsample analysis highlighted the heterogeneity in AI-powered academic cheating behavior by gender, as male students are more likely to cheat than female students. In terms of pooled sample analysis, our results align with the findings of previous studies (e.g., Mohd Salleh et al., 2013 ; Yazici et al., 2023 ). Gender disparities in moral attitudes and risk-taking tendencies possibly cause heterogeneous effects in cheating behavior between male and female students. Regarding the moral attitude, Ip et al. ( 2018 ) highlight that male students hold a more forgiving perspective toward acts of academic cheating than their female counterparts. Gender disparities in academic cheating may be attributed to the notion that women, who tend to prioritize social harmony, are less inclined to violate regulations, while men, who often exhibit greater competitiveness, may be more inclined to transgress rules in pursuit of success (Fisher & Brunell, 2014 ). In a similar vein, Zhang et al. ( 2018 ) reveal that female students exhibit considerably more negative attitudes toward academic misconduct and demonstrate greater levels of discomfort when they are detected as cheaters. In terms of risk-taking tendencies, Chala ( 2021 ) suggested that, on average, the propensity for risk-taking behaviors is greater for males than for females. Male students may be inclined to engage in academic dishonesty as a means to attain their academic objectives due to their greater propensity for taking risks.

In terms of heterogeneity in cheating behavior by grade, higher-grade students are more likely to cheat than newly enrolled students in the majority ethnic group. Our findings contrast with some previous studies. For instance, Bakar-Corez and Kocaman-Karoglu ( 2023 ) found a higher level of academic dishonesty among master’s students than among Ph.D. students. In a similar vein, Lord Ferguson et al. ( 2022 ) highlighted that the prevalence of academic dishonesty is higher among undergraduates than graduates. Importantly, we found that the cheating history of each gender differs substantially across grades. Although male students are more likely to cheat by using ChatGPT in the pooled sample, our subsample analysis shows that no significant difference in cheating history by gender exists among higher-grade students. Conversely, there was a substantial difference in cheating history by gender among newly enrolled students, as the prevalence of cheating among males is strongly dominant. Specifically, female students seem to change their cheating behaviors over time, as they are more likely to cheat in higher grades, as opposed to male students who consistently report cheating history across grades.

Academic-related pressure and peer effects might lead higher-grade students to be more likely to cheat than their counterparts. First, academic-related pressure is usually high for juniors and seniors, particularly in their final academic years. Higher-grade students may engage in academic dishonesty because they perceive it as a band-aid solution to achieve their goals, which are heightened expectations and future career prospects (Ababneh et al., 2022 ). Additionally, the final academic years are often especially stressful due to the accumulation of coursework, exams, and deadlines. To meet academic requirements, students might cheat to alleviate the stress of managing multiple courses and assignments (Amigud & Lancaster, 2019 ; Costley, 2019 ). Specifically, Orok et al. ( 2023 ) revealed that fear of failure is the most common reason for engaging in academic dishonesty, as 77% of respondents reported. Second, higher-grade students might be more likely to engage in academic cheating due to the peer dishonesty effect. For instance, Zhao et al. ( 2022 ) reveal that the peer dishonesty effect has a strong positive relationship with academic cheating, as observing peers engaging in academic misconduct potentially reinforces the idea that cheating is an effective solution to achieve academic objectives without the detection of educational institutions. In a similar vein, Lucifora and Tonello ( 2015 ) found that peer effects have a significant influence on academic cheating behaviors among students as the likelihood of cheating increases if educational institutions loosen the level of class monitoring systems. During the academic journey, the probability of witnessing peer cheating might increase among higher-grade students, potentially influencing them to follow their peers to violate academic integrity with the assistance of AI.

6 Conclusions and implications

This study has provided valuable insights into academic cheating in the era of AI growth. Although AI applications can be valuable educational tools, they also pose associated risks to academic integrity. By exploring a sample of 1,386 Vietnamese undergraduates via the list experiment to minimize social desirability bias, we found a significant magnitude of misreporting in response to AI-powered academic cheating behaviors among undergraduates. Specifically, the prevalence of cheaters observed via list experiments is almost threefold the prevalence of cheaters observed via direct questioning. Regarding the heterogeneous effect of AI-powered academic cheating behaviors among subsamples, we observed that female students are more likely to cheat in the later grades, while male students engage in academic cheating in all grades. In addition, academic cheating is more common in the final academic years among the majority ethnic group.

Based on our findings, we suggest potential implications that safeguard academic integrity. In terms of theoretical implications, academic cheating should be measured via the indirect questioning method, as students reasonably conceal their truthful answers due to the sensitivity of cheating issues. Educational policies for promoting academic integrity are effective only if cheating behaviors are accurately examined. In terms of practical implications, male students and higher-grade students of majority ethnicity must be well managed, as these groups showed a greater prevalence of AI-powered academic cheating. In addition, our subsample analysis shows that female students are also more likely to engage in academic dishonesty in higher grades; therefore, educational institutions should implement stringent management policies for these students during their final academic years. To prevent AI-powered cheating while leveraging the advantages of AI in education, it is necessary to apply concurrently supportive solutions and prevention solutions. Regarding supportive solutions, educational institutions should, for instance, offer counseling services to students dealing with stress, anxiety, or other personal issues that may facilitate academic dishonesty, in addition to more intensive orientation programs designed to educate students about the proper use of AI to harness the potential of AI-powered academic cheating but keep improving the learning effectiveness for students. Regarding prevention solutions, educational institutions should further consider investing in advanced monitoring systems to detect AI-powered academic cheating. Simultaneously, the implementation of adaptive assessment methods including randomization, dynamic question generation, and algorithmic modifications is necessary to mitigate the possibility of academic dishonesty facilitated by AI.

While this study contributes to the understanding of AI-powered academic cheating in education, it is important to acknowledge the remaining limitations. Because several graduate schools refused to participate, our study is limited to only four specific graduate schools. The generalizability of findings to other student populations, educational backgrounds, or major contexts may be restricted. To address these limitations, further research, methodological improvement, and cross-disciplinary cooperation are needed to deeply investigate academic cheating behavior in the era of accelerated AI.

Data Availability

The dataset of this study is available in the Mendeley Data at: https://data.mendeley.com/datasets/jyrw4wrtjr/1 .

Ossai et al. ( 2023 ) used the following direct statement to measure cheating behavior: “I sometimes copy already prepared assignments from my friends”.

Park ( 2020 ) used the following direct question to measure cheating behavior: “How often did you conduct the following behaviors in the past semester?”.

Ababneh et al. ( 2022 ) used the following direct question to measure cheating behavior: “During the past year, how frequently did you cheat on online tests/exams at your university”.

As members of Thai Nguyen University, all four participating graduate schools have applied Circular No.10/2016/TT-BGDĐT (Regulations for Student Affairs in Formal Higher Education programs) issued by the Vietnam Ministry of Education and Training to treat academic offenders. Following this circular, first-time offenders will fail the subjects in which they engage in academic cheating and receive caution. For repeated offenders, enhanced punishment (expulsion from the academic institution) will be applied.

We separate grades into 2 distinct groups: newly enrolled students (including freshmen and sophomores) and higher-grade students (including juniors and seniors).

To estimate the prevalence of academic cheating behaviors, the t-test for difference-in-mean estimator is qualified to compare the average response value between the control group and treatment groups. However, we followed the model of Lépine et al. ( 2020 ) supplemented by Ordinary Least Squares (OLS) regressions while controlling for multiple variables and school-level fixed effects, which have advantages in statistical analysis, particularly in addressing potential biases and improving the robustness of the model.

Ababneh, K. I., Ahmed, K., & Dedousis, E. (2022). Predictors of cheating in online exams among business students during the Covid pandemic: Testing the theory of planned behavior. The International Journal of Management Education, 20 (3), 100713. https://doi.org/10.1016/j.ijme.2022.100713

Article   PubMed Central   Google Scholar  

Ahmad, S., Lensink, R., & Mueller, A. (2023). Religion, social desirability bias and financial inclusion: Evidence from a list experiment on Islamic (micro-)finance. Journal of Behavioral and Experimental Finance, 38 , 100795. https://doi.org/10.1016/j.jbef.2023.100795

Article   Google Scholar  

Ali, Md. A., Dhanaraj, R. K., & Nayyar, A. (2023). A high performance-oriented AI-enabled IoT-based pest detection system using sound analytics in large agricultural field. Microprocessors and Microsystems, 103 , 104946. https://doi.org/10.1016/j.micpro.2023.104946

Amigud, A., & Lancaster, T. (2019). 246 reasons to cheat: An analysis of students’ reasons for seeking to outsource academic work. Computers & Education, 134 , 98–107. https://doi.org/10.1016/j.compedu.2019.01.017

Ansari, A. N., Ahmad, S., & Bhutta, S. M. (2023). Mapping the global evidence around the use of ChatGPT in higher education: A systematic scoping review. Education and Information Technologies . https://doi.org/10.1007/s10639-023-12223-4

Aung, Z. H., Sanium, S., Songsaksuppachok, C., Kusakunniran, W., Precharattana, M., Chuechote, S., Pongsanon, K., & Ritthipravat, P. (2022). Designing a novel teaching platform for AI : A case study in a Thai school context. Journal of Computer Assisted Learning, 38 (6), 1714–1729. https://doi.org/10.1111/jcal.12706

Bakar-Corez, A., & Kocaman-Karoglu, A. (2023). E-dishonesty among postgraduate students and its relation to self-esteem. Education and Information Technologies . https://doi.org/10.1007/s10639-023-12105-9

Blair, G., & Imai, K. (2012). Statistical analysis of list experiments. Political Analysis, 20 (1), 47–77. https://doi.org/10.1093/pan/mpr048

Chala, W. D. (2021). Perceived seriousness of academic cheating behaviors among undergraduate students: An Ethiopian experience. International Journal for Educational Integrity, 17 (1), 2. https://doi.org/10.1007/s40979-020-00069-z

Choi, E. P. H., Lee, J. J., Ho, M.-H., Kwok, J. Y. Y., & Lok, K. Y. W. (2023). Chatting or cheating? The impacts of ChatGPT and other artificial intelligence language models on nurse education. Nurse Education Today, 125 , 105796. https://doi.org/10.1016/j.nedt.2023.105796

Article   PubMed   Google Scholar  

Costley, J. (2019). Student perceptions of academic dishonesty at a cyber-University in South Korea. Journal of Academic Ethics, 17 (2), 205–217. https://doi.org/10.1007/s10805-018-9318-1

Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International , 1–12. https://doi.org/10.1080/14703297.2023.2190148

Currie, G. M. (2023). Academic integrity and artificial intelligence: Is ChatGPT hype, hero or heresy? Seminars in Nuclear Medicine, 53 (5), 719–730. https://doi.org/10.1053/j.semnuclmed.2023.04.008

Dai, Y., Lin, Z., Liu, A., & Wang, W. (2023). An embodied, analogical and disruptive approach of AI pedagogy in upper elementary education: An experimental study. British Journal of Educational Technology , bjet.13371. https://doi.org/10.1111/bjet.13371

Dalalah, D., & Dalalah, O. M. A. (2023). The false positives and false negatives of generative AI detection tools in education and academic research: The case of ChatGPT. The International Journal of Management Education, 21 (2), 100822. https://doi.org/10.1016/j.ijme.2023.100822

Du, P., He, X., Cao, H., Garg, S., Kaddoum, G., & Hassan, M. M. (2023). AI-based energy-efficient path planning of multiple logistics UAVs in intelligent transportation systems. Computer Communications, 207 , 46–55. https://doi.org/10.1016/j.comcom.2023.04.032

Eriksen, S., Lutz, C., & Tadesse, G. (2018). Social desirability, opportunism and actual support for farmers’ market organisations in Ethiopia. The Journal of Development Studies, 54 (2), 343–358. https://doi.org/10.1080/00220388.2017.1299138

Ezquerra, L., Kolev, G. I., & Rodriguez-Lara, I. (2018). Gender differences in cheating: Loss vs. gain framing. Economics Letters, 163 , 46–49. https://doi.org/10.1016/j.econlet.2017.11.016

Article   MathSciNet   Google Scholar  

Fisher, T. D., & Brunell, A. B. (2014). A bogus pipeline approach to studying gender differences in cheating behavior. Personality and Individual Differences, 61–62 , 91–96. https://doi.org/10.1016/j.paid.2014.01.019

Fyfe, P. (2023). How to cheat on your final paper: Assigning AI for student writing. AI & Society, 38 (4), 1395–1405. https://doi.org/10.1007/s00146-022-01397-z

Glynn, A. N. (2013). What can we learn with statistical truth serum? Public Opinion Quarterly, 77 (S1), 159–172. https://doi.org/10.1093/poq/nfs070

Guo, K., & Wang, D. (2023). To resist it or to embrace it? Examining ChatGPT’s potential to support teacher feedback in EFL writing. Education and Information Technologies . https://doi.org/10.1007/s10639-023-12146-0

Guo, K., Zhong, Y., Li, D., & Chu, S. K. W. (2023). Effects of chatbot-assisted in-class debates on students’ argumentation skills and task motivation. Computers & Education, 203 , 104862. https://doi.org/10.1016/j.compedu.2023.104862

Harris, A. S., Findley, M. G., Nielson, D. L., & Noyes, K. L. (2018). The economic roots of anti-immigrant prejudice in the global south: Evidence from South Africa. Political Research Quarterly, 71 (1), 228–241. https://doi.org/10.1177/1065912917734062

Hinsley, A., Keane, A., St. John, F. A. V., Ibbett, H., & Nuno, A. (2019). Asking sensitive questions using the unmatched count technique: Applications and guidelines for conservation. Methods in Ecology and Evolution , 10 (3), 308–319. https://doi.org/10.1111/2041-210X.13137

Igarashi, A., & Nagayoshi, K. (2022). Norms to be prejudiced: List experiments on attitudes towards immigrants in Japan. Social Science Research, 102 , 102647. https://doi.org/10.1016/j.ssresearch.2021.102647

Imai, K. (2011). Multivariate regression analysis for the item count technique. Journal of the American Statistical Association, 106 (494), 407–416. https://doi.org/10.1198/jasa.2011.ap10415

Article   MathSciNet   CAS   Google Scholar  

Ip, E. J., Pal, J., Doroudgar, S., Bidwal, M. K., & Shah-Manek, B. (2018). Gender-based differences among pharmacy students involved in academically dishonest behavior. American Journal of Pharmaceutical Education, 82 (4), 6274. https://doi.org/10.5688/ajpe6274

Article   PubMed   PubMed Central   Google Scholar  

Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences , 103 , 102274. https://doi.org/10.1016/j.lindif.2023.102274

Koo, M. (2023). Harnessing the potential of chatbots in education: The need for guidelines to their ethical use. Nurse Education in Practice, 68 , 103590. https://doi.org/10.1016/j.nepr.2023.103590

Kulkov, I. (2021). The role of artificial intelligence in business transformation: A case of pharmaceutical companies. Technology in Society, 66 , 101629. https://doi.org/10.1016/j.techsoc.2021.101629

Kumar, P., Sharma, S. K., & Dutot, V. (2023). Artificial intelligence (AI)-enabled CRM capability in healthcare: The impact on service innovation. International Journal of Information Management, 69 , 102598. https://doi.org/10.1016/j.ijinfomgt.2022.102598

Kutyauripo, I., Rushambwa, M., & Chiwazi, L. (2023). Artificial intelligence applications in the agrifood sectors. Journal of Agriculture and Food Research, 11 , 100502. https://doi.org/10.1016/j.jafr.2023.100502

Larson, R. B. (2019). Controlling social desirability bias. International Journal of Market Research, 61 (5), 534–547. https://doi.org/10.1177/1470785318805305

Latkin, C. A., Edwards, C., Davey-Rothwell, M. A., & Tobin, K. E. (2017). The relationship between social desirability bias and self-reports of health, substance use, and social network factors among urban substance users in Baltimore, Maryland. Addictive Behaviors, 73 , 133–136. https://doi.org/10.1016/j.addbeh.2017.05.005

Lépine, A., Treibich, C., & D’Exelle, B. (2020). Nothing but the truth: Consistency and efficiency of the list experiment method for the measurement of sensitive health behaviours. Social Science & Medicine, 266 , 113326. https://doi.org/10.1016/j.socscimed.2020.113326

Li, J., & Van den Noortgate, W. (2022). A meta-analysis of the relative effectiveness of the item count technique compared to direct questioning. Sociological Methods & Research, 51 (2), 760–799. https://doi.org/10.1177/0049124119882468

Li, L., Ma, Z., Fan, L., Lee, S., Yu, H., & Hemphill, L. (2023). ChatGPT in education: A discourse analysis of worries and concerns on social media. Education and Information Technologies . https://doi.org/10.1007/s10639-023-12256-9

Livberber, T., & Ayvaz, S. (2023). The impact of artificial intelligence in academia: Views of Turkish academics on ChatGPT. Heliyon, 9 (9), e19688. https://doi.org/10.1016/j.heliyon.2023.e19688

Lord Ferguson, S., Flostrand, A., Lam, J., & Pitt, L. (2022). Caught in a vicious cycle? Student perceptions of academic dishonesty in the business classroom. The International Journal of Management Education, 20 (3), 100677. https://doi.org/10.1016/j.ijme.2022.100677

Lucifora, C., & Tonello, M. (2015). Cheating and social interactions. Evidence from a randomized experiment in a national evaluation program. Journal of Economic Behavior & Organization, 115 , 45–66. https://doi.org/10.1016/j.jebo.2014.12.006

MohdSalleh, M. I., Alias, N. R., Hamid, H. A., & Yusoff, Z. (2013). Academic dishonesty among undergraduates in the higher education. International Journal of Academic Research, 5 (2), 222–227. https://doi.org/10.7813/2075-4124.2013/5-2/B.34

Moisset, X., & Ciampi De Andrade, D. (2023). Neuro-ChatGPT? Potential threats and certain opportunities. Revue Neurologique, 179 (6), 517–519. https://doi.org/10.1016/j.neurol.2023.02.066

Article   CAS   PubMed   Google Scholar  

Mubin, O., Cappuccio, M., Alnajjar, F., Ahmad, M. I., & Shahid, S. (2020). Can a robot invigilator prevent cheating? AI & Society, 35 (4), 981–989. https://doi.org/10.1007/s00146-020-00954-8

Nicholson, S. P., & Huang, H. (2022). Making the list: Reevaluating political trust and social desirability in china. American Political Science Review , 1–8. https://doi.org/10.1017/S0003055422000946

Olan, F., OgiemwonyiArakpogun, E., Suklan, J., Nakpodia, F., Damij, N., & Jayawickrama, U. (2022). Artificial intelligence and knowledge sharing: Contributing factors to organizational performance. Journal of Business Research, 145 , 605–615. https://doi.org/10.1016/j.jbusres.2022.03.008

Orok, E., Adeniyi, F., Williams, T., Dosunmu, O., Ikpe, F., Orakwe, C., & Kukoyi, O. (2023). Causes and mitigation of academic dishonesty among healthcare students in a Nigerian university. International Journal for Educational Integrity, 19 (1), 13. https://doi.org/10.1007/s40979-023-00135-2

Ossai, M. C., Ethe, N., Edougha, D. E., & Okeh, O. D. (2023). Academic integrity during examinations, age and gender as predictors of academic performance among high school students. International Journal of Educational Development, 100 , 102811. https://doi.org/10.1016/j.ijedudev.2023.102811

Park, S. (2020). Goal contents as predictors of academic cheating in college students. Ethics & Behavior, 30 (8), 628–639. https://doi.org/10.1080/10508422.2019.1668275

Phan, Q. N., Tseng, C.-C., Thi Hoai Le, T., & Nguyen, T. B. N. (2023). The application of chatbot on Vietnamese misgrant workers’ right protection in the implementation of new generation free trade agreements (FTAS). AI & Society , 38 (4), 1771–1783. https://doi.org/10.1007/s00146-022-01416-z

Qu, J., Zhao, Y., & Xie, Y. (2022). Artificial intelligence leads the reform of education models. Systems Research and Behavioral Science, 39 (3), 581–588. https://doi.org/10.1002/sres.2864

Ratten, V., & Jones, P. (2023). Generative artificial intelligence (ChatGPT): Implications for management educators. The International Journal of Management Education, 21 (3), 100857. https://doi.org/10.1016/j.ijme.2023.100857

Ried, L., Eckerd, S., & Kaufmann, L. (2022). Social desirability bias in PSM surveys and behavioral experiments: Considerations for design development and data collection. Journal of Purchasing and Supply Management, 28 (1), 100743. https://doi.org/10.1016/j.pursup.2021.100743

Sollosy, M., & McInerney, M. (2022). Artificial intelligence and business education: What should be taught. The International Journal of Management Education, 20 (3), 100720. https://doi.org/10.1016/j.ijme.2022.100720

Song, J., Iida, T., Takahashi, Y., & Tovar, J. (2022). Buying votes across Borders? A list experiment on mexican immigrants in the United States. Canadian Journal of Political Science, 55 (4), 852–872. https://doi.org/10.1017/S0008423922000567

Sweeney, S. (2023). Who wrote this? Essay mills and assessment – considerations regarding contract cheating and AI in higher education. The International Journal of Management Education, 21 (2), 100818. https://doi.org/10.1016/j.ijme.2023.100818

Tadesse, G., Abate, G. T., & Zewdie, T. (2020). Biases in self-reported food insecurity measurement: A list experiment approach. Food Policy, 92 , 101862. https://doi.org/10.1016/j.foodpol.2020.101862

Tsai, C. (2019). Statistical analysis of the item-count technique using stata. The Stata Journal, 19 (2), 390–434. https://doi.org/10.1177/1536867X19854018

UBS. (2023). Let's chat about ChatGPT . https://secure.ubs.com/public/api/v2/investment-content/documents/XILxY9V9P5RazGpDA1Cr_Q?apikey=Y8VdAx8vhk1P9YXDlEOo2Eoco1fqKwDk&Accept-Language=de-CH . Accessed 6 Aug 2023.

Udupa, P. (2022). Application of artificial intelligence for university information system. Engineering Applications of Artificial Intelligence, 114 , 105038. https://doi.org/10.1016/j.engappai.2022.105038

Wang, Z., Li, M., Lu, J., & Cheng, X. (2022). Business innovation based on artificial intelligence and blockchain technology. Information Processing & Management, 59 (1), 102759. https://doi.org/10.1016/j.ipm.2021.102759

Wang, Z., Liu, Y., & Niu, X. (2023). Application of artificial intelligence for improving early detection and prediction of therapeutic outcomes for gastric cancer in the era of precision oncology. Seminars in Cancer Biology, 93 , 83–96. https://doi.org/10.1016/j.semcancer.2023.04.009

Yazici, S., YildizDurak, H., AksuDünya, B., & Şentürk, B. (2023). Online versus face-to-face cheating: The prevalence of cheating behaviours during the pandemic compared to the pre-pandemic among Turkish University students. Journal of Computer Assisted Learning, 39 (1), 231–254. https://doi.org/10.1111/jcal.12743

Zhang, Y., Yin, H., & Zheng, L. (2018). Investigating academic dishonesty among chinese undergraduate students: Does gender matter? Assessment & Evaluation in Higher Education, 43 (5), 812–826. https://doi.org/10.1080/02602938.2017.1411467

Zhao, L., Mao, H., Compton, B. J., Peng, J., Fu, G., Fang, F., Heyman, G. D., & Lee, K. (2022). Academic dishonesty and its relations to peer cheating and culture: A meta-analysis of the perceived peer cheating effect. Educational Research Review, 36 , 100455. https://doi.org/10.1016/j.edurev.2022.100455

Download references

Acknowledgements

We are grateful to the managing board, staff, and students of Thai Nguyen University for their excellent assistance with our survey. We would like to thank anonymous reviewers for taking the time and effort necessary to review the manuscript, which helped us to improve the quality of the manuscript.

Open Access funding provided by Hiroshima University.

Author information

Authors and affiliations.

Graduate School of Humanities and Social Sciences, Hiroshima University, 1-5-1 Kagamiyama, Higashi-Hiroshima, 739-8529, Japan

Hung Manh Nguyen

The IDEC Institute, Hiroshima University, 1-5-1 Kagamiyama, Higashi-Hiroshima, 739-8529, Japan

Daisaku Goto

Network for Education and Research On Peace and Sustainability (NERPS), Hiroshima University, 1-5-1 Kagamiyama, Higashi-Hiroshima, 739-8529, Japan

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Hung Manh Nguyen .

Ethics declarations

Informed consent.

Informed consent was obtained from all participants in the experiment.

Conflict of interests

The authors have no relevant financial or nonfinancial interests to disclose.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Heterogeneous effects of the cheating intention by gender. Note: Fig. 3a represents the estimated prevalence of respondents who reported affirmative for cheating intention by gender. Figure 3b represents the disparity in cheating intention by gender (male dummy). Robust standard errors in parenthesis. *** p  <  0.01, ** p  <  0.05, * p  <  0.1

Heterogeneous effects of cheating behavior by major. Note: Fig. 4a represents the estimated prevalence of respondents who reported affirmative to sensitive statements. Figure 4b represents the disparity in cheating behaviors by major (the major base is Information Technology). Covarites include age, gender, ethnicity, grade, social, and part-time job. Robust standard errors in parenthesis. *** p  <  0.01, ** p  <  0.05, * p  <  0.1

Subsample robustness tests. Note: Fig. 5a represents the heterogeneous effects of cheating history by gender (male dummy). Figure 5b represents the heterogeneous effects of cheating history by gender among newly enrolled students (male dummy). Figure 5c represents the heterogeneous effects of cheating history by grade among students with majority ethnicity (higher-grade dummy). Covarites include age, gender, ethnicity, grade, social, and part-time job. Fixed effects at the school level. Robust standard errors in parenthesis. *** p  <  0.01, ** p  <  0.05, * p  <  0.1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Nguyen, H.M., Goto, D. Unmasking academic cheating behavior in the artificial intelligence era: Evidence from Vietnamese undergraduates. Educ Inf Technol (2024). https://doi.org/10.1007/s10639-024-12495-4

Download citation

Received : 07 November 2023

Accepted : 18 January 2024

Published : 05 February 2024

DOI : https://doi.org/10.1007/s10639-024-12495-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Academic cheating
• Misreporting
• Artificial intelligence
• Vietnamese undergraduates
• Find a journal
• Publish with us
• Track your research

Introducing Apple’s On-Device and Server Foundation Models

At the 2024 Worldwide Developers Conference , we introduced Apple Intelligence, a personal intelligence system integrated deeply into iOS 18, iPadOS 18, and macOS Sequoia.

Apple Intelligence is comprised of multiple highly-capable generative models that are specialized for our users’ everyday tasks, and can adapt on the fly for their current activity. The foundation models built into Apple Intelligence have been fine-tuned for user experiences such as writing and refining text, prioritizing and summarizing notifications, creating playful images for conversations with family and friends, and taking in-app actions to simplify interactions across apps.

In the following overview, we will detail how two of these models — a ~3 billion parameter on-device language model, and a larger server-based language model available with Private Cloud Compute and running on Apple silicon servers — have been built and adapted to perform specialized tasks efficiently, accurately, and responsibly. These two foundation models are part of a larger family of generative models created by Apple to support users and developers; this includes a coding model to build intelligence into Xcode, as well as a diffusion model to help users express themselves visually, for example, in the Messages app. We look forward to sharing more information soon on this broader set of models.

Our Focus on Responsible AI Development

Apple Intelligence is designed with our core values at every step and built on a foundation of groundbreaking privacy innovations.

Additionally, we have created a set of Responsible AI principles to guide how we develop AI tools, as well as the models that underpin them:

• Empower users with intelligent tools : We identify areas where AI can be used responsibly to create tools for addressing specific user needs. We respect how our users choose to use these tools to accomplish their goals.
• Represent our users : We build deeply personal products with the goal of representing users around the globe authentically. We work continuously to avoid perpetuating stereotypes and systemic biases across our AI tools and models.
• Design with care : We take precautions at every stage of our process, including design, model training, feature development, and quality evaluation to identify how our AI tools may be misused or lead to potential harm. We will continuously and proactively improve our AI tools with the help of user feedback.
• Protect privacy : We protect our users' privacy with powerful on-device processing and groundbreaking infrastructure like Private Cloud Compute. We do not use our users' private personal data or user interactions when training our foundation models.

These principles are reflected throughout the architecture that enables Apple Intelligence, connects features and tools with specialized models, and scans inputs and outputs to provide each feature with the information needed to function responsibly.

In the remainder of this overview, we provide details on decisions such as: how we develop models that are highly capable, fast, and power-efficient; how we approach training these models; how our adapters are fine-tuned for specific user needs; and how we evaluate model performance for both helpfulness and unintended harm.

Pre-Training

Our foundation models are trained on Apple's AXLearn framework , an open-source project we released in 2023. It builds on top of JAX and XLA, and allows us to train the models with high efficiency and scalability on various training hardware and cloud platforms, including TPUs and both cloud and on-premise GPUs. We used a combination of data parallelism, tensor parallelism, sequence parallelism, and Fully Sharded Data Parallel (FSDP) to scale training along multiple dimensions such as data, model, and sequence length.

We train our foundation models on licensed data, including data selected to enhance specific features, as well as publicly available data collected by our web-crawler, AppleBot. Web publishers have the option to opt out of the use of their web content for Apple Intelligence training with a data usage control.

We never use our users’ private personal data or user interactions when training our foundation models, and we apply filters to remove personally identifiable information like social security and credit card numbers that are publicly available on the Internet. We also filter profanity and other low-quality content to prevent its inclusion in the training corpus. In addition to filtering, we perform data extraction, deduplication, and the application of a model-based classifier to identify high quality documents.

Post-Training

We find that data quality is essential to model success, so we utilize a hybrid data strategy in our training pipeline, incorporating both human-annotated and synthetic data, and conduct thorough data curation and filtering procedures. We have developed two novel algorithms in post-training: (1) a rejection sampling fine-tuning algorithm with teacher committee, and (2) a reinforcement learning from human feedback (RLHF) algorithm with mirror descent policy optimization and a leave-one-out advantage estimator. We find that these two algorithms lead to significant improvement in the model’s instruction-following quality.

Optimization

In addition to ensuring our generative models are highly capable, we have used a range of innovative techniques to optimize them on-device and on our private cloud for speed and efficiency. We have applied an extensive set of optimizations for both first token and extended token inference performance.

Both the on-device and server models use grouped-query-attention. We use shared input and output vocab embedding tables to reduce memory requirements and inference cost. These shared embedding tensors are mapped without duplications. The on-device model uses a vocab size of 49K, while the server model uses a vocab size of 100K, which includes additional language and technical tokens.

For on-device inference, we use low-bit palletization, a critical optimization technique that achieves the necessary memory, power, and performance requirements. To maintain model quality, we developed a new framework using LoRA adapters that incorporates a mixed 2-bit and 4-bit configuration strategy — averaging 3.5 bits-per-weight — to achieve the same accuracy as the uncompressed models.

Additionally, we use an interactive model latency and power analysis tool, Talaria , to better guide the bit rate selection for each operation. We also utilize activation quantization and embedding quantization, and have developed an approach to enable efficient Key-Value (KV) cache update on our neural engines.

With this set of optimizations, on iPhone 15 Pro we are able to reach time-to-first-token latency of about 0.6 millisecond per prompt token, and a generation rate of 30 tokens per second. Notably, this performance is attained before employing token speculation techniques, from which we see further enhancement on the token generation rate.

Model Adaptation

Our foundation models are fine-tuned for users’ everyday activities, and can dynamically specialize themselves on-the-fly for the task at hand. We utilize adapters, small neural network modules that can be plugged into various layers of the pre-trained model, to fine-tune our models for specific tasks. For our models we adapt the attention matrices, the attention projection matrix, and the fully connected layers in the point-wise feedforward networks for a suitable set of the decoding layers of the transformer architecture.

By fine-tuning only the adapter layers, the original parameters of the base pre-trained model remain unchanged, preserving the general knowledge of the model while tailoring the adapter layers to support specific tasks.

We represent the values of the adapter parameters using 16 bits, and for the ~3 billion parameter on-device model, the parameters for a rank 16 adapter typically require 10s of megabytes. The adapter models can be dynamically loaded, temporarily cached in memory, and swapped — giving our foundation model the ability to specialize itself on the fly for the task at hand while efficiently managing memory and guaranteeing the operating system's responsiveness.

To facilitate the training of the adapters, we created an efficient infrastructure that allows us to rapidly retrain, test, and deploy adapters when either the base model or the training data gets updated. The adapter parameters are initialized using the accuracy-recovery adapter introduced in the Optimization section.

Performance and Evaluation

Our focus is on delivering generative models that can enable users to communicate, work, express themselves, and get things done across their Apple products. When benchmarking our models, we focus on human evaluation as we find that these results are highly correlated to user experience in our products. We conducted performance evaluations on both feature-specific adapters and the foundation models.

To illustrate our approach, we look at how we evaluated our adapter for summarization. As product requirements for summaries of emails and notifications differ in subtle but important ways, we fine-tune accuracy-recovery low-rank (LoRA) adapters on top of the palletized model to meet these specific requirements. Our training data is based on synthetic summaries generated from bigger server models, filtered by a rejection sampling strategy that keeps only the high quality summaries.

To evaluate the product-specific summarization, we use a set of 750 responses carefully sampled for each use case. These evaluation datasets emphasize a diverse set of inputs that our product features are likely to face in production, and include a stratified mixture of single and stacked documents of varying content types and lengths. As product features, it was important to evaluate performance against datasets that are representative of real use cases. We find that our models with adapters generate better summaries than a comparable model.

As part of responsible development, we identified and evaluated specific risks inherent to summarization. For example, summaries occasionally remove important nuance or other details in ways that are undesirable. However, we found that the summarization adapter did not amplify sensitive content in over 99% of targeted adversarial examples. We continue to adversarially probe to identify unknown harms and expand our evaluations to help guide further improvements.

In addition to evaluating feature specific performance powered by foundation models and adapters, we evaluate both the on-device and server-based models’ general capabilities. We utilize a comprehensive evaluation set of real-world prompts to test the general model capabilities. These prompts are diverse across different difficulty levels and cover major categories such as brainstorming, classification, closed question answering, coding, extraction, mathematical reasoning, open question answering, rewriting, safety, summarization, and writing.

We compare our models with both open-source models (Phi-3, Gemma, Mistral, DBRX) and commercial models of comparable size (GPT-3.5-Turbo, GPT-4-Turbo) 1 . We find that our models are preferred by human graders over most comparable competitor models. On this benchmark, our on-device model, with ~3B parameters, outperforms larger models including Phi-3-mini, Mistral-7B, and Gemma-7B. Our server model compares favorably to DBRX-Instruct, Mixtral-8x22B, and GPT-3.5-Turbo while being highly efficient.

We use a set of diverse adversarial prompts to test the model performance on harmful content, sensitive topics, and factuality. We measure the violation rates of each model as evaluated by human graders on this evaluation set, with a lower number being desirable. Both the on-device and server models are robust when faced with adversarial prompts, achieving violation rates lower than open-source and commercial models.

Our models are preferred by human graders as safe and helpful over competitor models for these prompts. However, considering the broad capabilities of large language models, we understand the limitation of our safety benchmark. We are actively conducting both manual and automatic red-teaming with internal and external teams to continue evaluating our models' safety.

To further evaluate our models, we use the Instruction-Following Eval (IFEval) benchmark to compare their instruction-following capabilities with models of comparable size. The results suggest that both our on-device and server model follow detailed instructions better than the open-source and commercial models of comparable size.

We evaluate our models’ writing ability on our internal summarization and composition benchmarks, consisting of a variety of writing instructions. These results do not refer to our feature-specific adapter for summarization (seen in Figure 3 ), nor do we have an adapter focused on composition.

The Apple foundation models and adapters introduced at WWDC24 underlie Apple Intelligence, the new personal intelligence system that is integrated deeply into iPhone, iPad, and Mac, and enables powerful capabilities across language, images, actions, and personal context. Our models have been created with the purpose of helping users do everyday activities across their Apple products, and developed responsibly at every stage and guided by Apple’s core values. We look forward to sharing more information soon on our broader family of generative models, including language, diffusion, and coding models.

[1] We compared against the following model versions: gpt-3.5-turbo-0125, gpt-4-0125-preview, Phi-3-mini-4k-instruct, Mistral-7B-Instruct-v0.2, Mixtral-8x22B-Instruct-v0.1, Gemma-1.1-2B, and Gemma-1.1-7B. The open-source and Apple models are evaluated in bfloat16 precision.

Related readings and updates.

Advancing speech accessibility with personal voice.

A voice replicator is a powerful tool for people at risk of losing their ability to speak, including those with a recent diagnosis of amyotrophic lateral sclerosis (ALS) or other conditions that can progressively impact speaking ability. First introduced in May 2023 and made available on iOS 17 in September 2023, Personal Voice is a tool that creates a synthesized voice for such users to speak in FaceTime, phone calls, assistive communication apps, and in-person conversations.

Apple Natural Language Understanding Workshop 2023

Earlier this year, Apple hosted the Natural Language Understanding workshop. This two-day hybrid event brought together Apple and members of the academic research community for talks and discussions on the state of the art in natural language understanding.

In this post, we share highlights from workshop discussions and recordings of select workshop talks.

Discover opportunities in Machine Learning.

Our research in machine learning breaks new ground every day.

Work with us

• Open access
• Published: 02 December 2022

University students’ understanding of contract cheating: a qualitative case study in Kuwait

• Inan Deniz Erguvan   ORCID: orcid.org/0000-0001-8713-2935 1  

Language Testing in Asia volume  12 , Article number:  56 ( 2022 ) Cite this article

3857 Accesses

2 Citations

Metrics details

Contract cheating, or students outsourcing their assignments to be completed by others, has emerged as a significant threat to academic integrity in higher education institutions around the world. During the COVID-19, when traditional face-to-face instruction became unsustainable, the number of contract cheating students increased dramatically. Through focus group interviews, this study sought the perspectives of 25 students enrolled in first year writing in a private higher education institution in Kuwait during the pandemic in 2020–2021, on their attitudes towards contract cheating. MAXQDA 2020 was used to examine the data. The participants believe that the primary motivations for engaging in contract cheating are mainly the opportunities presented by online learning and the psychological and physical challenges they experienced during online learning. Those who did not cheat had some shared traits, such as a competitive spirit, confidence in their talents, and a strong desire to learn. Additionally, those with high moral values avoided cheating. To combat contract cheating, students believe that teaching and evaluation techniques should be drastically altered and that students should be educated about plagiarism, while institutions should impose tougher sanctions on repeat offenders.

Introduction

Universities met their teaching commitments through remote teaching because traditional face-to-face teaching temporarily became unsustainable due to the COVID-19 pandemic that hit the world between 2020 and 2022. Because students were not physically present in the classroom during this time, many higher education institutions conducted tests through digital platforms or replaced exams with essays and other types of written tasks (Gamage et al., 2020 ). The technology and infrastructure required to join online sessions, concerns about privacy as information technology devices demanded access to students' cameras, microphones, and desktop, and, most importantly, questions about academic integrity have been raised as a result of alternate methods to assessment. Replacement of exams with writing and/or take-home assignments constituted a danger to academic integrity, necessitating the adoption of fraud-free methods (Almeida & Monteiro, 2021 ).

Academic dishonesty, which refers to committing or contributing to dishonest acts in teaching, learning, research, and related academic activities (Cizek, 2003 ; Whitley & Keith-Spiegel, 2001 ), has long been a source of concern in higher education and has been on the rise in recent years. According to some estimates, up to 75% of university students have engaged in some type of academic misconduct during their academic career (Brimble & Stevenson-Clarke, 2005 ; McCabe & Bowers, 1994 ).

Plagiarism is one of the most persistent issues confronting higher education institutions, and it can take many forms, including “copy and paste” without citing the source; patch-writing; providing incorrect or incomplete citations or references; presenting or citing a secondary source as a primary source; ghost-writing; and contract cheating (De Jager & Brown, 2010 ; Ellery, 2008 ; Ellis et al., 2018 ; Park, 2003 ; Zafarghandi et al., 2012 ). Clarke and Lancaster ( 2006 ) coined the term “contract cheating” to characterize the unacknowledged usage of materials prepared by another person or entity involved in the sale of academic resources. Students outsource their coursework to others to do, usually for a fee, and then present it as their own. Contract cheating, according to many people, is a growing problem that most academic institutions are dealing with. To make matters worse, it is difficult to spot ghostwritten work because it is a new piece of writing tailored to course requirements and specific assignments (Ali & AlHassan, 2021 ).

Although contract cheating is common in both traditional face-to-face and online settings, it is more likely to take place in the latter. There are some strong indications that contract cheating went rampant during the pandemic and became a significant COVID-19 side effect for higher education institutions. According to Lancaster and Cotarlan ( 2021 ), the number of requests for answers to academic questions on a popular student website jumped by over 200% during the pandemic. Lancaster ( 2020 ) found that a website providing essay writing services expanded the number of tutors they recruited and was able to offer discounts due to the increased profitability. Similarly, during the first COVID-interrupted semester, additional research found that university students believed their colleagues cheated when classes went online (Daniels et al., 2021 ) and that their willingness and pressure to cheat were stronger online than in-person (Walsh et al., 2021 ). Likewise, King and Case ( 2014 ) discovered that throughout a 5-year period, the number of students who self-reported academic cheating increased, and around 75% of students said it was easier to cheat in online assessment.

There are several possible explanations for why students engaged in contract cheating in online education more often than in person education. Psychological distance adversely affected interpersonal relationships; and the internet obscured the line between academically honest and dishonest behavior (Eshet, 2022 ). Sudden campus closures and abrupt transition to online teaching modals provided more opportunities for students to complete assignments with online assistance. Furthermore, essay mills saw the lack of face-to-face interaction and proctoring on campus as an opportunity and used aggressive marketing methods to attract students. Through social media, students quickly became aware of the possibilities of a wide variety of options to carry out contract cheating; as a result, contract cheating has emerged as a real threat to academic integrity (Erguvan, 2021 ; Hill et al., 2021 , Bautista & Pentang, 2022 ).

Review of literature

Academic dishonesty is a complicated system including a variety of components that interact in unanticipated ways. Due to the vast number of paper mills, full-text databases, and collaborative web pages, many researchers attribute the rise in academic dishonesty to the increased usage of the internet, which generates “opportunities” for cheating (Peytcheva-Forsyth, et al., 2018 ; Townley & Parsell, 2004 ). Students engage in academic dishonesty for a variety of reasons, according to researchers: desire for high grades, procrastination, time pressure to complete assignments or study for tests, lack of organizational skills, fear of failing a course (loss of time and money), lack of understanding of academic dishonesty, and plagiarism not being considered a serious offense (Eshet et al., 2012 ; Jone, 2011 ; McGee, 2013 ). Academic dishonesty is influenced by social factors including peer pressure, social attitudes and norms about academic dishonesty, or domestic job market circumstances (Carpenter, et al., 2006 ; Gallant & Drinan, 2006 ). A “competitive culture” to earn excellent grades or succeed in school (Roberts & Hai-Jew, 2009 ). Furthermore, if the dominant culture does not regard academic dishonesty as a significant problem that requires attention, such situations will be handled on an individual basis, and formal consequences will rarely be pursued. The existence of an institutional policy on academic integrity, code of honor, and effective disciplinary procedures performed by educational institutions are all institutional elements that may affect academic dishonesty (Roberts & Hai-Jew, 2009 ; Vilchez & Thirunarayanan, 2011 ).

Many theories have been developed to describe why and how students engage in plagiarism and what factors play a role in this. Plagiarism has been commonly understood using theoretical frameworks originating from criminology literature which conceptualizes students as delinquents; however, that is rather problematic and ineffective in the long run (DiPietro, 2010 ). Some other theories could be listed as deterrence theory, rational choice theory, neutralization theory, planned behavior theory, situational ethics, social learning theory, self-control, and rational choice theory (DiPietro, 2010 ; Sattler, et al., 2013 ). This research was guided by social learning theory and rational choice theory.

Social learning theory by Albert Bandura could be applied to explain the learners’ plagiarism behavior. Bandura ( 1997 ) theorizes that learning is a cognitive-process that takes place in a social-context and can occur through “the influence of example” by observing a behavior and/or the consequences of the behavior. Therefore, if learners discover their fellow classmates plagiarizing and getting high grades and receiving nominal or no punishment at all for these acts, they will also feel inclined to adopt cheating. If a behavior is learned with a perceived negative consequence associated with it, then an individual is more likely to inhibit that behavior for him- or herself. However, positive reinforcement, which can also mean not having a negative consequence associated with the behavior, may encourage behaviors, whether they are positive or negative. (Denler et al., 2014 ).

Students choose to plagiarize in their assignments or tests for a range of reasons, and it is possible to examine the students’ motivation within the framework of rational choice theory, according to which every individual follows the principle of maximum utilization when they have to make a decision (Hawes, 2020 ). Individuals compare potential advantages to possible costs entailed by their decision and the course of action is chosen after weighing the advantages and disadvantages of all possible alternatives. Therefore, the decision to cheat and plagiarize results from a cost–benefit analysis. On the one hand, plagiarism offers some benefits: allowing students to finish the work quickly and save time and improve their grades; on the other hand, there are some counter-factors such as the risk of being caught. In case of plagiarism, potential losses would become real if the fact of plagiarism were to be discovered, which is not always likely. The consequences of plagiarism for students might include unsatisfactory marks for the assignment, reproach by the teachers, or other disciplinary punishments. Nevertheless, such measures do not seem to be significant enough to have the possibility to prevent students from plagiarizing. The risk of being caught has a medium negative effect on cheating, the fear of punishment has a small negative effect, and the importance of the outcome has a medium positive effect (Whitley, 1998 ).

Even though large numbers of students are claimed to partake in contract cheating in Kuwaiti higher education, such as purchasing papers from shops on campus that seemingly provide only printing and photocopying services (Al Jiyyar, 2017 ), there is little research on contract cheating in Kuwait. Indeed, a literature review by Ahsan et al. ( 2021 ) identifies research deficits outside of Australia, the UK, and Canada, as well as in contexts of contract cheating such as society, culture, and religion. Contract cheating during and after COVID-19 is another dimension that has received insufficient attention.

As a result, the purpose of this study is to investigate students' opinions of contract cheating occurring in first year writing classes in a private university context in Kuwait, a country that has had little research done in this area. The questions that will guide the study are as follows:

Why did more students engage in contract cheating during the pandemic?

What stopped students from engaging in contract cheating?

What consequences did contract cheating students face, if any?

What should be done to curb contract cheating?

In this exploratory case study, participants’ perspectives were acquired through focus group interviews, which is a popular strategy for acquiring qualitative data (Morgan, 1996 ). The strength of focus groups is that they allow participants to interact in groups which can provide insights into the causes of complicated actions as a result of group interaction (Carey & Smith, 1994 ; Morgan & Krueger, 1993 ). Because members simultaneously question and explain themselves to one another, focus groups are more than the sum of separate individual interviews. Because of the sensitive nature of the subject, the researcher determined that a group discussion, rather than individual interviews, would yield more insightful data.

Participants

Regarding the number and size of focus groups, different authors have varied advice and references. Various researchers have noted that sizes of focus groups may range from four to five, six to eight, and even eight to twelve people, and some have even suggested that there are no universal standards for the best number of focus groups (as cited in Gundumogula, 2020 ).

The researcher recruited twenty-five students for this study and five focus group interviews were scheduled, with each interview containing five to six students. There were eleven females and fourteen males among the participants. Purposeful sampling was used as the sampling method. Other faculty members in the department were contacted and asked to provide a list of potential participants. The list of participants suggested by faculty members was screened for eligibility to see if they met the following inclusion criteria:

Fluent in English.

Currently registered in a course during the 2021–2022 Fall term in the university.

Enrolled in a university and attended online classes in the previous academic year, 2020–2021.

Diversity in gender, discipline, and academic performance was observed in recruitment. The potential participants who were selected were contacted by e-mail. Official invitations to the online meeting were sent via emails to those expressed interest in attending.

The informed consent form which was approved by the Institutional Review Board included a detailed description of the interview process and confidentiality information. The participants were asked to read and sign these forms before the interview took place.

Researcher’s role

In qualitative research, because the researcher is the instrument in semi structured or unstructured qualitative interviews, unique researcher attributes have the potential to influence the collection of empirical materials (Pezalla et al., 2012 ); therefore, explicitly identifying oneself is more important than it is in quantitative research. In this study, the researcher is a faculty member in the same institution where the research took place, thus, familiar with the plagiarism and contract cheating habits and attitudes of Kuwaiti undergraduate students. She also included some of her own as well as her colleagues’ students in the study. She stated clearly at the beginning of the study that student responses will not be used for any course related assessment and evaluation and the collected data will be limited to research only. She only joined the focus group meetings as a listener, and another trained colleague moderated these meetings. Although every effort has been made to ensure objectivity, certain biases may remain, and these biases may shape the way the data is collected and the participants’ experiences are interpreted in this study.

Data collection

All sessions were conducted online, due to the pandemic restrictions still in place, and in English, between November and December 2021. The sessions were recorded and simultaneously transcribed using the built-in function of the online meeting platform. The participants were informed of recording at the time of recruitment, as well as the beginning of each session.

Each online focus group lasted between 60 and 90 min that allowed in depth discussions. The focus group sessions were moderated by a trained Education Department faculty member. A script was developed for the moderator to guide the discussion. The moderator used the script that explained the purpose of the focus group, went over the focus group rules, and reinforced the confidentiality of all the information shared.

Although there are no general rules as to the optimal number of focus group discussions, researchers state that four focus groups are generally sufficient, but that consideration of response saturation should be made after the third focus group discussion (Nyamathi & Shuler, 1990 ). Guest et al. ( 2017 ) stated that within two to three more than 80%, and within three to six focus groups 90% of all themes were discoverable. Three focus groups were also enough to identify the most prevalent themes within the data set. The number of focus groups in this study was guided by theme saturation. After conducting five focus group sessions, it has been observed that the information collected was becoming repetitive, as no new themes were emerging. Therefore, it was decided that data saturation has been reached.

Data analysis

Content and thematic analysis methodologies were used to analyze the data collected in this study. Content analysis refers to the process in which presentations of behavior or qualitative data from self-reports are analyzed (Karataş, 2015 ). Content analysis is more related to the initial analysis and the coding process, where researchers look for redundant and similar codes (Humble & Mozelius, 2022 ). The thematic analysis occurs after the coding process as researchers aggregate similar codes to form major concepts or themes. Basically, thematic analysis converts qualitative data into quantitative data. Once data is transcribed, it is reviewed repeatedly so that the researcher can identify trends in the meaning conveyed by language. The themes identified are re-analyzed so that they become more refined and relevant and given codes (Boyatzis, 1998 ; Braun & Clarke, 2006 ). The researcher can then annotate the transcript with the codes that have been identified. In this study, we started with the content analysis as a more basic way of approaching the data material, and then proceeded with the thematic analysis to detect, analyze, and report themes, as well as organize and describe data in dimensions. As distinct and fundamental qualitative approaches, the two should be used by qualitative researchers as transparent structures provide researchers with clear and user-friendly methods for analyzing data (Vaismoradi et al, 2013 ).

Each student was assigned a code to safeguard their anonymity and confidentiality during the study. The name of the institution was also taken out of the transcribed focus group sessions. The researcher ran a pilot focus group with some students who were not in the sample to verify the understandability of the questions for the focus group interviews’ reliability and validity.

When there were any discrepancies, the meeting platform’s transcriptions were compared to the audio recordings and modified. The written data was afterwards uploaded to the MAXQDA 2020 program, which allows for systematic data analysis (Kuckartz & Rädiker, 2019 ). The earliest codes were constructed using an inductive approach, and codes that were connected to each other were grouped together and assigned names. Following that, the emerging themes were explained in a way that readers could comprehend. Finally, the researcher provided an interpretation of the findings as well as supporting images.

Research question 1

The first question analyzed students’ perceptions regarding why more students got engaged in contract cheating during the pandemic. In line with the statements of the participants, the motivators category was defined with ten different codes in order of frequency: wanting to get easy grades, having more opportunities to cheat, challenges of online education and difficult assignments, culture/pressure, and insecurity/lack of ability emerged as some major motivators (Fig.  1 ).

Motivators for contract cheating

The participants’ statements regarding the major motivators are below:

“For starters, the online education sort of opened the window of opportunity for students. Now when you see a before and after image, you would think that before we didn’t have access to these sorts of resources. People usually were very busy going to day to day classes. They didn’t have time to research these sort of things. I believe that because people had a bit more free time to do many activities or do whatever during the pandemic because of online education or online learning, they were able to come across these resources through search engines and were able to practice how to use these facilities.” (FG3-1).

Among the motivators, the challenges of online education and the difficulty of assignments in online learning were also mentioned as a reason for resorting to contract cheating. Participants mentioned that students had difficulty accessing information and could not focus during online education:

“Although there are office hours, maybe they need face to face with the professor, in order to learn how to write it properly, because personally when I had an essay writing class, it was easy for me because when we wrote one paragraph, we would review it one on one with the professor. So I feel like that’s why when it came to online, the percentage got a lot higher than when we were on campus.” (FG5-3).

An interesting code that came out of student responses was the culture in Kuwait. Participants mentioned that Kuwaiti children grow up in an environment where everything is done for them. Another element of the culture is the pressure on students to get good grades and graduate with a high GPA to be eligible for government jobs, therefore cheating is considered almost acceptable.

“In Kuwait, in particular, culturally speaking, because of how Kuwaitis were raised or brought up or how they live through an environment of luxury and lack of hardship, to say the very least, it led them to this mindset where they could do these things because they have the option of doing so because it’s so easy to them.” (FG3-1).

Research question 2

When participants were asked what stopped some students from contract cheating during the online education, their responses revealed six different codes. The major deterrents of cheating emerged as moral and religious values and having certain personality traits. Some minor deterrents were fear of getting caught, not wanting to risk future job prospects, and getting trapped in a vicious cycle, also seeing contract cheating as a waste of money (Fig.  2 ).

Reasons for not misconducting

One of the most popular responses was the student’s moral values as a reason for not cheating in their assignments.

“Some students, have strong moral values. So, no matter what challenges they face, they would not resort to cheating, because they believe that it’s wrong.” (FG1-2).

Kuwait is officially an Islamic country and Kuwaitis are quite religious people. This was also perceived as part of the non-cheating students’ set of moral values.

“Religion definitely does play a role because in Islam we know that if you cheat to get yourself success, everything you earn from that success is going to be forbidden upon you, so you won’t benefit in the end. But nowadays the religious commitment is not that big.” (FG4-2).

Another code that the participants’ responses revealed was certain personality traits. Under this code participants mentioned three different subcodes: self-confidence, motivation to learn, and competitiveness (Fig.  3 ).

Hierarchical code—subcodes model of reasons for not misconducting

Participants mentioned that when students have enough self-confidence, motivation, and competitiveness, they do not need any help, they are excited about their achievements, and they cannot trust anyone else to do their work for them.

“I think some students do not like to depend on other people to do their work. Also they are not trustworthy like they can’t trust them to do their work as they feel more confident doing their own work. In this way they will improve themselves to become a better person.” (FG2-4). “Eventually people we will come back to fully on a real life and we’ll be in a position where you cannot cheat. So, in order to enhance our knowledge, or to do better in future classes, they wouldn’t cheat, so they can actually learn something.” (FG5-4).

Another reason why some students did not cheat was the fear of getting caught, according to students’ perceptions. Participants mentioned that some students did not resort to cheating because they were afraid of the outcomes in case they get caught. This was also similar to another deterrent mentioned by students as not wanting to risk future job prospects. Some quotes below exemplify such perceptions:

“I think one of the biggest things that most students fear when it comes down to plagiarizing or cheating is getting caught. But I also think what would devastate a student is if the teacher or the instructor make an example out of the student. Because if you pull out their assignment in front of the entire class and say that ‘this is plagiarized, and because of that, I will give you a zero’, you know you would be set as an example, and I think that would break a student, and so I think that thought or the fact that you’re getting caught. And then being exposed is what really scares or that fear that set a lot of students aside from wanting to plagiarize or cheat.” (FG1-2). “This would affect them in long term, and they wouldn’t be able to do stuff that normal person would be able to do and complete their assignments and all of that stuff. They will have problems in their jobs later on their lives.” (FG4-3). “I think it’s all a certain mindset. Some students don’t cheat because they realize there’s no meaning to it in a sense that if you do cheat your whole life, you’ll keep cheating… if you cheat now, you’re going to cheat in your whole life and there’s no point to it.” (FG3-1).

Research question 3

Students were asked about their perceptions towards the consequences cheating students faced. Did cheaters get caught? Did they get penalized sufficiently when they were caught? Did cheating ever go unnoticed? The responses revealed six different codes, which could be summarized as most of the time cheating went unnoticed, and when it was noticed, they received certain punishments ranging from failing the assignment and/or the course or being suspended from the university (Fig.  4 ).

Punishments when students get caught

The most striking response as analysis revealed was most participants thought instructors did not even realize that cheating took place.

“I think it’s less likely that the professor will catch the student until and unless they know the student and his past assignments. Because the professors don’t know anything about them. They don’t know how they’ve been doing. I think they get their assistants to check the paper so they are very less likely they can catch the culprit.” (FG4-3).

In case students got caught contract cheating, the most common consequence they faced was getting a zero and failing that specific assignment. Failing the entire course or being suspended from the university for a semester and blacklisted on a list that circulates among faculty members were also mentioned by some participants was another repercussion mentioned by some participants.

“I had no experience with people getting caught with cheating, but like usually if they get caught they just get a zero for the essay. For that particular assignment, not for the whole course.” (FG4-1).

Students also talked about being interrogated by the instructor as a consequence of cheating. This interrogation sometimes took place in private, but sometimes in public, in front of their peers, which was a big source of embarrassment for the student:

“To make it even better, they should do the punishments publicly so other students can see this person is being punished for this reason so they can scare everyone else from being humiliated in front of the class. Well, what we did in our old school if someone cheats, they rip the paper on the spot and then kick the student out the class.” (FG5-1).

Research question 4

The final research question of the study focused on solutions to this academic misconduct and asked students to make suggestions to prevent this problem. Students were reminded to consider all the stakeholders in their suggestions, such as students, instructors, and the administration of the institution in which they are studying. Their responses revealed nine codes as seen in Fig.  5 . The solutions could be classified as positive and negative ones, with the positive, more nurturing solutions being changing teaching and assessment methods, educating students about cheating, giving students second chances, offering more learning support services, conducting face-to-face education and raising awareness on social media and on campus. Some students were in favor of more punitive solutions, such as applying harsher punishments and stricter control, and using anti-cheating software and equipment.

Suggestions for combatting contract cheating

The suggestion with the highest frequency was changing the teaching and assessment methods. Participants mentioned that education system relies too much on rote learning, memorization and should include more hands-on assignments and projects. As a result, assessment types should move from multiple choice, written exams to more applied methods and performance assessment.

“That’s the only way I could see this work is if the entire like education industry changed the way they move forward with their teaching and their learning, make it more practical with experience… more than just about grades. The fact that you know your grades are on the line and students compare one of their grades, their GPA over the other, the pressure is intense, and that’s where you know people resort to things that are much easier. But if it’s more about, having fun, learning, experiencing, things are quite practical to the world out there specifically tailored to what they want to do in the future, that would completely eliminate the. You know, the problem of plagiarism and cheating or whatnot, because then they’re doing something they love doing.” (FG1-2).

The next common code was educating students about cheating. Participants mentioned that students need to be educated about cheating and given clear warnings about the outcomes. Some participants expressed the need for a more nurturing environment for students and that they should be given a second chance when caught. Offering more learning support services was also proposed as a solution as some participants to encourage students to seek more help from legal sources.

“I do believe that a severe punishment would refrain the students from cheating, but what would be an even better learning objective to make them understand that cheating would be wrong, plagiarizing would be wrong is to let them understand how severe it is beforehand so they wouldn’t do it in the first place, not by punishing them if they cheated. Making it sink down deep end that this is how severe cheating is… this is what will happen. Because some students don’t really understand the full gravity of what they’re doing and what would happen if they’re caught.” (FG3-1).

“Maybe they’re lazy, but in most cases they need help, they need people to help them with things they don’t understand. They need help with things that they probably don’t know. Or maybe they weren’t focused in class on a particular day. Students need help, like education is not easy. It’s a learning process. We as people learn through mistakes and experiences, but we also need a guiding hand in order for us to succeed in life and for us we should not focus solely on punishments because at the end of the day students or these young people are the future of the nation of Kuwait of this country. If we can guide them to a better path instead of punishing them and then going on a very darker or very negligent path, it would have been much better. Not only for us,

Other codes that came out of our analysis were conducting face-to-face education, using anti-cheating software and equipment in exams to detect cheating, warning classmates about the outcomes of cheating, and raising awareness on social media. Some participants asked for harsher punishments to combat cheating.

The first research question on the motivators of contract cheating revealed the fact that some students outsourced their tasks because they just wanted to get easy grades and online learning made this a possibility by providing more opportunities to cheat. Many students were just looking to get by and pass the course because the shift to online education has drastically affected their ability to learn and retain information, and they only intended to cheat in the short-term.

According to Gallant (cited in Dey, 2021 ), there was probably increased cheating because there were more temptations and opportunities. When colleges shut down or restricted in-person access, students were taking exams in their bedrooms, with unrestricted access to their phones and other information technologies. This spurred cheating to take on new and different forms. Regarding students cheating in online courses, if students feel anonymous and unlikely to be adequately monitored, they may assume that the likelihood of being caught cheating is virtually zero and cheat more in online classes using online resources. Previous research has shown that participants had a higher propensity to cheat when chances of being caught were less likely (Kajackaite & Gneezy, 2017 ). Despite being supervised through a web camera, the teachers cannot control the surroundings or the computer screens of the students. In class, students are regularly monitored and watched and thus are less free to consult sources of information, but with the physical distance, those odds decrease, and so cheating increases.

College students could not help but want to be a part of the herd since they did not want to be left out when their peers were earning good scores without putting in any effort. This is further reinforced by research findings that students are less inclined to cheat when they believe their peers are trustworthy and the misconception of “everyone else is doing it” encourages cheating (Carpenter et al., 2006 ; Daniels et al., 2021 ; Turner & Uludag, 2013 ). Observing their peers’ cheating activities in online classes through group chats, as our participants’ responses reveal, encourages more students to cheat, especially after the initial shock has worn off and they felt more at ease, in the second semester of online education.

The difficulties of online education have been cited in various research as a factor that contributed to contract cheating. Along with the opportunities online learning provided, stress and pressure started building up and the pandemic essentially intensified a feeling of potential loss among college students. Asking questions during exams was difficult without the in-person experience. Students were able to ask questions via email or attend virtual office hours, but many missed the ease of raising a hand and getting a question answered in real time. According to a study conducted in Vietnam (Tran, et al., 2021 ), students had generally negative feelings toward online education, with 63.31% of respondents stating that they disliked online exams and 64.8 percent stating that online learning was only marginally effective and only a temporary solution. The difficulties in assessing and testing online, as well as not understanding the course and communicating with peers, were identified as negatives.

With the outburst of the pandemic, many students found their surroundings transformed completely. Such a change probably caused an increased fear of loss with students being away from their friends and normal social environment, away from the usual learning atmosphere and resources they are used to. They developed a fear of losing social connections, falling behind in class, losing internship and career opportunities, etc. In a study that surveyed students from all over the USA (Hoyt et al., 2021 ), students reported that the loss of their social life had a major influence on their mental state during the pandemic. When these results are considered with the established idea that increased fear of loss can cause a biological reaction that increases dishonest behavior, it can reasonably be assumed that one of the primary reasons colleges all over the world detected abnormally high cheating rates among their students is an increase in a fear of loss (Arie, 2021 ). Similar findings were seen in other research, including as in Hong Kong, where students said they struggled to maintain self-discipline when studying alone on online platforms (Mok et al, 2021 ); students experienced stress, worry, and pressure as a result of the pandemic (Sahu, 2020 ), and they did not find online learning to be totally rewarding, particularly when they experienced disruptions during online classes due to insufficient educational and institutional assistance (Fauzi & Sastra Khusuma, 2020 ; Xie & Yang, 2020 ).

The purpose of our second study question was to examine students’ perspectives of the motivations for not cheating. Students’ responses emphasize the importance of students’ own moral compass, as well as particular personality traits like self-confidence, ambition to learn, and competition, as key deterrents to cheating. This is consistent with research that emphasizes the role of attitudes and beliefs in preventing academic misconduct and promoting an ethical culture (Rundle et al., 2019 ; Grym and Liljander, 2016 ) Strong individual views and ideals regarding integrity, according to Reedy et al. ( 2021 ), minimize the likelihood of students cheating. Following these two major deterrents, fear of being caught emerges as a third code, which is corroborated by the findings of an Australian study by Rundle et al. ( 2019 ). Three significant predictors of fear of detection and punishment were identified in Rundle’s regression analysis (Machiavellianism, narcissism, and consistency of interest), implying that students who scored high on these are more likely to report fear of detection and punishment as a reason for not engaging in contract cheating. These findings imply that appealing to students’ values and beliefs while conveying clear messages about academic integrity could be an effective method for improving the integrity of online and offline exams.

The study’s third finding concerned the implications of cheating. Students were asked what the consequences and punishments were when cheaters were caught. Surprisingly, the highest frequency was observed in the code that cheaters were not generally caught, and contract cheating went unnoticed. This finding is intriguing in a way as we are not sure whether instructors really fail to recognize cheating or tend to ignore it and not take any action as it is difficult to present hard evidence to prove contract cheating. Research found that faculty were able to identify 62% of contract cheating when they were advised to specifically look for it (Dawson & Sutherland-Smith, 2019 ); but when they were unaware of the possible presence of contract cheating, they could not detect any (Lines, 2016 ). Although Erguvan’s study ( 2021 ) on faculty awareness of contract cheating found that faculty members are confident in their ability to spot it even when cases are detected, teaching staff are concerned that proving cheating may be difficult (Walker & Townley, 2012 ). Faculty members frequently express concerns about cheating during online education, but they have not always been able to detect and punish cheating as they would like to, due to a lack of security measures, reliable plagiarism detection tools, and training on online assessment and cheating prevention measures (Meccawy et al, 2021 ). Because of the problem’s complexity and the difficulties to solving it, faculty members may simply choose to ignore it (Coren, 2011 ; McCabe, 2005 ).

When faculty members detect students cheating on an assignment, the most typical repercussions include failing the work and being interrogated by the teacher, sometimes in private and sometimes in public. Students, on the other hand, stated that failing an assignment is not a strong enough deterrent to cheating because these assignments are often so little in proportion that they have little impact on students’ total grade in the course. Another study found that academic dishonesty is frequent among Kuwaiti university students because the danger of detection and severity of sanctions for academic misconduct is minimal (Alsuwaileh et al., 2016 ). Participants believe academic dishonesty remains widespread because sanctions are not enough. According to most of the participants, embarrassment is the only informal sanction for academic dishonesty, and they would be embarrassed more by the lecturer than by their friends or families.

According to the findings of the fourth research question, students believe that the existing educational system merely promotes students to memorize and does not teach them the real skills they need in their jobs. Students expressed their desire for a change in the university's educational and assessment techniques. Indeed, there is a growing body of research on the function of evaluation in contract cheating prevention and detection. Some suggest increasing the use of invigilated (Clarke & Lancaster, 2006 ; Lines, 2016 ) and in-class viva voce examinations (Carroll and Appleton, 2001 ) to reduce the potential for cheating. Others focus on reducing motivations to cheat through increasing student engagement by choosing personalized topics (Sutherland-Smith, 2013 ), and using authentic assessment, which aims to engage students in “real-world” tasks (Collins et al, 2007 ; QAA, 2020 ).

However, some researchers are skeptical about the impact of changing the assessment in curbing contract cheating and suggest that authentic assessment does not necessarily assure academic integrity and that educators need to be aware that cheating may take place even in applied and “authentic” exams such as oral exam/viva or practical exam (Harper et al., 2021 ; Ellis et al, 2020 ).

Text-rich forms of assessment, according to Harper et al. ( 2021 ), should keep their place in university assessment strategies, not because they are impervious to contract cheating, but because faculty get more proficient in detecting cheating in written assignments like research papers, which enable them to develop more personalized relations with students. This supports the students’ belief that, in addition to assessment, a shift in pedagogy could play a role in minimizing contract cheating.

Our research suggests that students want to be educated about academic integrity and given explicit warnings about the consequences, but they also believe punishments should be severe enough to work as deterrents. The rational choice theory may offer hints about how to curb the plagiarism problem in this regard. Universities should increase the benefits associated with non-plagiarized papers and publicly circulate information about plagiarism; otherwise, any punishments or sanctions will not be deterrent to plagiarism. Academic integrity values may be fostered, and students can become familiar with this culture through course objectives and activities. A system of progressive educational punishment might likewise be implemented (Cinali, 2016 ; Mervis, 2012 ). Faculty members and university administrators should diminish the benefits of plagiarism and increase the costs and the probability of detection. If students still choose to plagiarize, they must take higher risks into account; otherwise, they need to either be experts in hiding plagiarism or make greater effort in producing plagiarism that is hard to detect, which will reduce the time-saving benefits of plagiarism (Collins et al., 2007 ) and in turn reduce the number of students committing plagiarism.

Conclusion and recommendations

The findings of this study reveal that as per students’ perceptions, online learning has driven more students to contract cheat, primarily by approaching an essay mill or a tutor and paying them to do the work for them. Students expressed they were tempted by the opportunities presented by online learning, such as not having any proctoring or an obligation to turn on their webcams during exams, ease of finding a tutor to do the work at a very affordable rate, and not having the motivation and skills to cope with the challenges of online classes, therefore choosing the easy way. Some students described the feeling as “being part of the herd,” which could be compared to the new trendy acronym FOMO (fear of missing out), that basically refers to the feeling or perception that others are having more fun, living better lives, or experiencing better things than you are, which is often exacerbated by social media.

To summarize, academic integrity violations have been on the rise as a result of COVID-19-mandated online or hybrid education systems which may tempt many students to continue using their tried and tested methods of cheating when they return to face-to-face instruction. Therefore, violations of academic integrity necessitate a rethinking of teaching and evaluation methodologies. Higher education institutions must adapt to the changing contract cheating marketplace and ensure that the faculty are aware of contract cheating and can recognize the indicators of contract cheating. Students should be given the message that their tutors are aware of contract cheating services. To keep up with the constant changes in technology, academic integrity processes must be current, resilient, and assessed on a regular basis.

If we do not take immediate action, contract cheating will likely reach epidemic proportions. We need to take a comprehensive approach that includes a focus on assessment design, a strengthened culture of integrity, and robust technical tools. We should also urge academics to perform ongoing research on ways to improve academic integrity during and post pandemic higher education instruction.

Limitations

The author is confident this paper will add significant value to the body of existing literature; however, we cannot be sure that the focus groups have captured a representative sample of students studying in higher education institutes in Kuwait. It is also important to note that the study is limited to the experiences and assumptions of students who participated in the study and therefore the findings should not be generalized.

Availability of data and materials

The qualitative data that support the findings of this study are available on request from the corresponding author, DE. The data are not publicly available as they contain information that could compromise the privacy of research participants.

Abbreviations

Coronavirus disease

Max Qualitative Data Analysis

Ahsan, K., Akbar, S., & Kam, B. (2021). Contract cheating in higher education: A systematic literature review and future research agenda. Assessment & Evaluation in Higher Education . https://doi.org/10.1080/02602938.2021.1931660

Article   Google Scholar  

Ali, H.I. & Alhassan, A. (2021) Fighting contract cheating and ghostwriting in higher education: Moving towards a multidimensional approach. Cogent Education , 8 (1). https://doi.org/10.1080/2331186X.2021.1885837

Al Jiyyar, A. (2017). Cheated education: Research papers for sale at Kuwaiti University. Arab Reporters for Investigative Journalism (ARIJ). Retrieved November 17, 2021. https://en.arij.net/investigation/cheated-education-research-papers-for-sale-at-kuwait-university

Almeido, F., & Monteiro, J. (2021). The challenges of assessing and evaluating the students at distance. Journal of Online Higher Education , 5 (1), 3- 10. https://doaj.org/article/eaab3e55f98346eeaa410db7528b7f59

AlSuwaileh, B. G., Russ-Eft, D. F., & Alshurai, S. R. (2016). Academic dishonesty: a mixed-method study of rational choice among students at the College of Basic Education in Kuwait. Journal of Education and Practice , 7(30), 139–151. Retrieved November 17, 2021, https://www.iiste.org/Journals/index.php/JEP/article/view/33629/34573 .

Arie, R. (2021). Academic dishonesty and COVID-19: a biological explanation. University Writing Program, Brandeis University. https://www.brandeis.edu/writingprogram/write-now/2020-2021/arierotem/arie-rotem.pdf

Bandura, A. (1997). Self-efficacy: The exercise of control . W H Freeman/Times Books/ Henry Holt & Co.

Google Scholar  

Bautista, R. M., & Pentang, J. T. (2022). Ctrl C + Ctrl V: Plagiarism and knowledge on referencing and citation among pre-service teachers. International Journal of Multidisciplinary: Applied Business and Education Research, 3 (2), 245–257. https://doi.org/10.11594/ijmaber.03.02.10

Boyatzis, R. E. (1998). Transforming qualitative information: Thematic analysis and code development . Sage.

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3 (2), 77–101. https://doi.org/10.1191/1478088706qp063oa

Brimble, M., & Stevenson-Clarke, P. (2005). Perceptions of the prevalence and seriousness of academic dishonesty in Australian universities. Australian Educational Researcher, 32 , 19–44. https://doi.org/10.1007/BF03216825

Carey, M. A., & Smith, M. W. (1994). Capturing the group effect in focus groups: A special concern in analysis. Qualitative Health Research, 4 (1), 123–127. https://doi.org/10.1177/104973239400400108

Carpenter, D. D., Harding, T. S., Finelli, C. J., Montgomery, S. M., & Passow, H. J. (2006). Engineering students’ perception of and attitudes towards cheating. Journal of Engineering Education, 95 (3), 181–194.

Carroll, J., & Appleton, J. (2001). Plagiarism: a good practice guide. https://i.unisa.edu.au/siteassets/staff/tiu/documents/plagiarism---a-good-practice-guide-by-oxford-brookes-university.pdf .

Cinali, G. (2016). Middle Eastern perspectives of academic integrity: a view from the Gulf region. In T. Bretag (Ed.), Handbook of Academic Integrity. Singapore: Springer. https://doi.org/10.1007/978-981-287-098-8_8

Chapter   Google Scholar  

Cizek, G. (2003) Detecting and preventing classroom cheating. Corwin Press, Thousand Oaks ISBN 0–7619–4655-1

Clarke, R. & Lancaster, T. (2006). Eliminating the successor to plagiarism? Identifying the usage of contract cheating sites. Proceedings of the 2nd International Plagiarism Conference available at: https://marketing-porg-statamic-assets-uswest-2.s3-us-west-2.amazonaws.com/main/Clarke2_fullpaper2006.pdf

Collins, A., Judge, G., & Rickman, N. (2007). On the economics of plagiarism. European Journal of Law and Economics., 24 (2), 93–107. https://doi.org/10.1007/s10657-007-9028-4

Coren, A. (2011). Turning a blind eye: Faculty who ignore student cheating. Journal of Academic Ethics., 9 (4), 291–305. https://doi.org/10.1007/s10805-011-9147-y

Daniels, L. M., Goegan, L. D., & Parker, P. C. (2021). The impact of COVID-19 triggered changes to instruction and assessment on university students’ self-reported motivation, engagement, and perceptions. Social Psychology of Education., 24 (1), 299–318. https://doi.org/10.1007/s11218-021-09612-3

Dawson, P., & Sutherland-Smith, W. (2019). Can training improve marker accuracy at detecting contract cheating? A multi-disciplinary pre-post study. Assessment & Evaluation in Higher Education, 44 (5), 715–725. https://doi.org/10.1080/02602938.2018.1531109

Denler, H., Walters, C., & Benzon, M. (2014). Social cognitive theory. http://www.education.com/reference/article/social-cognitive-theory/

De Jager, K., & Brown, C. (2010). The tangled web: Investigating academics’ views of plagiarism at the University of Cape Town. Studies in Higher Education, 35 (5), 513–528. https://doi.org/10.1080/03075070903222641

Dey, S. (2021). Reports of cheating at colleges soar during the pandemic. https://www.npr.org/2021/08/27/1031255390/reports-of-cheating-at-colleges-soar-during-the-pandemic .

DiPietro, M. (2010). 14: Theoretical frameworks for academic dishonesty. To Improve the Academy, 28(1), 250–262.  https://doi.org/10.1002/j.2334-4822.2010.tb00606.x

Ellery, K. (2008). An investigation into electronic-source plagiarism in a first-year essay assignment. Assessment and Evaluation in Higher Education, 33 (6), 607–617. https://doi.org/10.1080/02602930701772788

Ellis, C., van Haeringen, K., Harper, R., Bretag, T., Zucker, I., McBride, S., Rozenberg, P., Newton, P. & Saddiqui, S. (2020). Does authentic assessment assure academic integrity? Evidence from contract cheating data. Higher Education Research & Development, 39(3), 454–469, https://doi.org/10.1080/07294360.2019.1680956

Ellis, C., Zucker, I. M., & Randall, D. (2018). The infernal business of contract cheating: Understanding the business processes and models of academic custom writing sites. International Journal for Educational Integrity, 14(1). https://doi.org/10.1007/s40979-017-0024-3

Erguvan, I. D. (2021). The rise of contract cheating during the COVID-19 pandemic: A qualitative study through the eyes of academics in Kuwait. Language Testing in Asia, 11(34). https://doi.org/10.1186/s40468-021-00149-y

Eshet, Y., Grinautski, K., & Peled, Y. (2012). Learning motivation and student academic dishonesty: a comparison between face-to-face and online courses. Proceedings of the Chais conference on instructional technology research 2012: Learning in technology era. (pp. 22–29). The Open University of Israel. Raanana

Eshet, Y. (2022). Contract cheating in Israel during the COVID-19 pandemic. In European Conference. https://philpapers.org/archive/ESHCCI-2.pdf

Fauzi, I., & Sastra Khusuma, I. H. (2020). Teachers’ elementary school in online learning of COVID-19 pandemic conditions. Jurnal Iqra’: Kajian Ilmu Pendidikan, 5 (1), 58–70. https://doi.org/10.25217/ji.v5i1.914

Gallant, T. B., & Drinan, P. (2006). Organizational theory and student cheating: Explanation, responses, and strategies. The Journal of Higher Education, 77 (5), 839–860.

Gamage, K. A., Silva, E. K., & Gunawardhana, N. (2020). Online delivery and assessment during COVID-19: Safeguarding academic integrity. Education Sciences, 10 (11), 301. https://doi.org/10.3390/educsci10110301

Grym, J., & Liljander, V. (2016). To cheat or not to cheat? The effect of a moral reminder on cheating. Nordic Journal of Business , 65(3–4), 18–37. http://njb.fi/wp-content/uploads/2017/01/Grym_Liljander.pdf

Guest, G., Namey, E., & McKenna, K. (2017). How many focus groups are enough? Building an evidence base for nonprobability sample sizes. Field Methods, 29 (1), 3–22. https://doi.org/10.1177/1525822X16639015

Gundumogula, M. (2020). Importance of focus groups in qualitative research. International Journal of Humanities and Social Science (IJHSS) , 8 (11):299–302. ff10.24940/theijhss/2020/v8/i11/HS2011–082ff. ffhal-03126126f

Harper, R., Bretag, T., & Rundle, K. (2021). Detecting contract cheating: Examining the role of assessment type. Higher Education Research & Development, 40 (2), 263–278. https://doi.org/10.1080/07294360.2020.1724899

Hawes, D. (2020). Rational choice and political power. Journal of Contemporary European Studies, 28 (1), 132. https://doi.org/10.1080/14782804.2019.1684744

Hill, G., Mason, J., & Dunn, A. (2021). Contract cheating: An increasing challenge for global academic community arising from COVID-19. Research and Practice in Technology Enhanced Learning, 16(1). https://doi.org/10.1186/s41039-021-00166-8

Hoyt, L. T., Cohen, A. K., Dull, B., Maker Castro, E., & Yazdani, N. (2021). “Constant stress has become the new normal”: Stress and anxiety inequalities among U.S. college students in the time of COVID-19. Journal of Adolescent Health, 68 (2), 270–276. https://doi.org/10.1016/j.jadohealth.2020.10.030

Humble, N., & Mozelius, P. (2022). Content analysis or thematic analysis: Doctoral students' perceptions of similarities and differences. Electronic Journal of Business Research Methods, 20(3), 89–98. https://doi.org/10.34190/ejbrm.20.3.2920

Jone, D. L. R. (2011). Academic dishonesty: Are more students cheating? Business Communication Quarterly, 74 (2), 141–150.

Kajackaite, A., & Gneezy, U. (2017). Incentives and cheating. Games and Economic Behavior, 102 , 433–444. https://doi.org/10.1016/j.geb.2017.01.015

Kuckartz, U., & Rädiker, S. (2019). Analyzing qualitative data with MAXQDA: Text, audio, and video . Springer. https://doi.org/10.1007/978-3-030-15671-8

Book   Google Scholar  

Karataş, Z. (2015). Sosyal bilimlerde nitel araştırma yöntemleri. Manevi Temelli Sosyal Hizmet Araştırmaları Dergisi, 1(1), 62–80. https://www.academia.edu/33009261/Sosyal_Hizmet_E_Dergi_SOSYAL_B%C4%B0L%C4%B0MLERDE_N%C4%B0TEL_ARA%C5%9ETIRMA_Y%C3%96NTEMLER%C4%B0 .

King, D. L., & Case, C. J. (2014). E-cheating: incidence and trends among college students. Issues in Information Systems, 15 (1), 20–27. https://iacis.org/iis/2014/4_iis_2014_20-27.pdf

Lancaster, T. (2020). Commercial contract cheating provision through micro-outsourcing web sites. International Journal for Educational Integrity, 16(1). https://doi.org/10.1007/s40979-020-00053-7

Lancaster, T., & Cotarlan, C. (2021). Contract cheating by STEM students through a file sharing website: A COVID-19 pandemic perspective. International Journal for Educational Integrity, 17(3). https://doi.org/10.1007/s40979-021-00070-0

Lines, L. (2016). Ghostwriters guaranteeing grades? The quality of online ghostwriting services available to tertiary students in Australia. Teaching in Higher Education, 21(8), 889–914. https://doi.org/10.1080/13562517.2016.1198759

McCabe, D. L., & Bowers, W. (1994). Academic dishonesty among male college students: A thirty-year perspective. Journal of College Student Development, 35 , 3–10.

McCabe, D. L. (2005). Cheating among college and university students: A North American perspective. International Journal for Educational Integrity, 1 (1). https://doi.org/10.21913/ijei.v1i1.14

McGee, P. (2013). Supporting academic honesty in online courses. The Journal of Educators Online, 10 (1), 1–31.

Meccawy, Z., Meccawy, M. & Alsobhi, A. (2021). Assessment in ‘survival mode’: student and faculty perceptions of online assessment practices in HE during Covid-19 pandemic. International Journal for Educational Integrity. 17(16). https://doi.org/10.1007/s40979-021-00083-9

Mervis, J. (2012). Growing pains in the desert. Science, 338 (7), 1276–1281. https://doi.org/10.1126/science.338.6112.1276

Mok, K. H., Xiong, W., & Bin Aedy Rahman, H. N. (2021). COVID-19 pandemic’s disruption on university teaching and learning and competence cultivation: Student evaluation of online learning experiences in Hong Kong. International Journal of Chinese Education, 10(1), 221258682110070. https://doi.org/10.1177/22125868211007011

Morgan, D. L. (1996). Focus groups. Annual Review of Sociology, 22 , 129–152. https://doi.org/10.1146/annurev.soc.22.1.129

Morgan, D. L., & Krueger, R. A. (1993). When to use focus groups and why. In D. L. Morgan (Ed.), Successful focus groups: advancing the state of the art (pp. 3–19). Sage Publications Inc. https://doi.org/10.4135/9781483349008.n1

Nyamathi, A., & Shuler, P. (1990). Focus group interview: A research technique for informed nursing practice. Journal of Advanced Nursing, 15 (11), 1281–1288. https://doi.org/10.1111/j.1365-2648.1990.tb01743.x

Park, C. (2003). In other (people’s) words: Plagiarism by university students–literature and lessons. Assessment and Evaluation in Higher Education, 28 (5), 471–488.

Peytcheva-Forsyth, R., Aleksieva, L., & Yovkova, B. (2018). The impact of technology on cheating and plagiarism in the assessment – the teachers’ and students’ perspectives. AIP Conference Proceedings, 2048 , 020037. https://doi.org/10.1063/1.5082055

Pezalla, A. E., Pettigrew, J., & Miller-Day, M. (2012). Researching the researcher-as-instrument: An exercise in interviewer self-reflexivity. Qualitative Research: QR, 12 (2), 165–185. https://doi.org/10.1177/1487941111422107

Reedy, A., Pfitzner, D., Rook, L., & Ellis, L. (2021). Responding to the COVID-19 emergency: Student and academic staff perceptions of academic integrity in the transition to online exams at three Australian universities. International Journal for Educational Integrity, 17(9). https://doi.org/10.1007/s40979-021-00075-9

Roberts, C. J., & Hai-Jew, S. (2009). Issues of academic integrity: An online course for students addressing academic dishonesty. MERLOT Journal of Online Learning and Teaching, 5 (2), 182–196.

Rundle, K., Curtis, G. J., & Clare, J. (2019). Why students do not engage in contract cheating. Frontiers in Psychology, 10 (2229). https://doi.org/10.3389/fpsyg.2019.02229

Sahu, P. (2020). Closure of universities due to coronavirus disease 2019 (COVID-19): Impact on education and mental health of students and academic staff. Cureus, 12 (4), e7541. https://doi.org/10.7759/cureus.7541

Sattler, Sebastian, Graeff, Peter, & Willen, Sebastian. (2013). Explaining the decision to plagiarize: An empirical test of the interplay between rationality, norms, and opportunity. Deviant Behavior, 34 (6), 444–463.  https://doi.org/10.1080/01639625.2012.735909

Sutherland-Smith, W. (2013). Crossing the line: collusion or collaboration in university group work? Australian Universities Review, 55(1). https://files.eric.ed.gov/fulltext/EJ1004398.pdf

The Quality Assurance Agency for Higher Education. (2020). Contracting to cheat in higher education. https://www.qaa.ac.uk/docs/qaa/guidance/contracting-to-cheat-in-higher-education-2nd-edition.pdf

Townley, C., & Parsell, M. (2004). Technology and academic virtue: Student plagiarism through the looking glass. Ethics and Information Technology, 6 (4), 271–277. https://doi.org/10.1007/s10676-005-5606-8

Tran, T. K., Dinh, H., Nguyen, H., Le, D.-N., Nguyen, D.-K., Tran, A. C., … Tieu, S. (2021). The impact of the COVID-19 pandemic on college students: an online survey. Sustainability , 13:10762. https://doi.org/10.3390/su131910762

Turner, S. W., & Uludag, S. (2013). Student perceptions of cheating in online and traditional classes. In 2013 IEEE frontiers in education conference (FIE 2013): Oklahoma City, Oklahoma, USA, 23 - 26 October 2013. IEEE. https://doi.org/10.1109/FIE.2013.6685007

Vaismoradi, M., Turunen, H., & Bondas, T. (2013). Content analysis and thematic analysis: Implications for conducting a qualitative descriptive study. Nursing & Health Sciences, 15 (3), 398–405. https://doi.org/10.1111/nhs.12048

Vilchez, M., & Thirunarayanan, M.O. (2011). Cheating in online courses: A qualitative study. Department of Teaching and Learning. 14. https://digitalcommons.fiu.edu/tl_fac/14

Walker, M., & Townley, C. (2012). Contract cheating: a new challenge for academic honesty? Journal of Academic Ethics, 10 (1), 27–44. https://doi.org/10.1007/s10805-012-9150-y

Walsh, L. L., Lichti, D. A., Zambrano-Varghese, C. M., Borgaonkar, A. D., Sodhi, J. S., Moon, S., Wester, E. R., & Callis-Duehl, K. L. (2021). Why and how science students in the United States think their peers cheat more frequently online: Perspectives during the COVID-19 pandemic. International Journal for Educational Integrity, 17(23). https://doi.org/10.1007/s40979-021-00089-3

Whitley, B. E. (1998). Factors associated with cheating among college students: A review. Research in Higher Education, 39 (3), 235–274.

Whitley, B. E., & Keith-Spiegel, P. (2001). Academic dishonesty: An educator's guide. Psychology Press. https://doi.org/10.4324/9781410604279

Xie, Z., & Yang, J. (2020). Autonomous learning of elementary students at home during the COVID-19 epidemic: a case study of the second elementary school in Daxie, Ningbo, Zhejiang province, China. Best Evidence of Chinese Education, 4 (2), 535–541. https://doi.org/10.15354/bece.20.rp009

Zafarghandi, A., Khoshroo, F., & Barkat, B. (2012). An investigation of Iranian EFL Masters students’ perceptions of plagiarism. International Journal for Educational Integrity, 8 , 69–85. https://doi.org/10.21913/IJEI.v8i2.811

Download references

Acknowledgements

This study has been funded by Gulf University for Science and Technology in Kuwait, internal seed grant number 234553.

Author information

Authors and affiliations.

Gulf University for Science and Technology, West Mishref, Kuwait

Inan Deniz Erguvan

You can also search for this author in PubMed   Google Scholar

Contributions

Deniz Erguvan as the sole author of this manuscript wrote the literature review, collected and analyzed the data, and produced the discussion section of the manuscript. The author read and approved the final manuscript.

Author’s information

DE has been teaching undergraduate students in a private university in Kuwait for the past 12 years. She is quite familiar with contract cheating practices and shes observed a significant surge in contract cheating among students during the pandemic.

Corresponding author

Correspondence to Inan Deniz Erguvan .

Ethics declarations

Competing interests.

The author declares no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Erguvan, I.D. University students’ understanding of contract cheating: a qualitative case study in Kuwait. Lang Test Asia 12 , 56 (2022). https://doi.org/10.1186/s40468-022-00208-y

Download citation

Received : 29 August 2022

Accepted : 19 November 2022

Published : 02 December 2022

DOI : https://doi.org/10.1186/s40468-022-00208-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Contract cheating
• Online learning
• Academic integrity
• Student perspectives
• Focus group

The Solar System May Have Passed through Dense Interstellar Cloud 2 Million Years Ago, Altering Earth’s Climate

In a new bu-led paper, astrophysicists calculate the likelihood that earth was exposed to cold, harsh interstellar clouds, a phenomenon not previously considered in geologic climate models.

For a brief period of time millions of years ago, Earth may have been plunged out of the sun’s protective plasma shield, called the heliosphere, which is depicted here as the dark gray bubble over the backdrop of interstellar space. According to new research, this could have exposed Earth to high levels of radiation and influenced the climate. Photo courtesy of Opher, et al., Nature Astronomy

Jessica Colarossi

Around two million years ago, Earth was a very different place, with our early human ancestors living alongside saber-toothed tigers, mastodons, and enormous rodents . And they may have been cold: Earth had fallen into a deep freeze , with multiple ice ages coming and going until about 12,000 years ago. Scientists theorize that ice ages occur for a number of reasons , including the planet’s tilt and rotation, shifting plate tectonics, volcanic eruptions, and carbon dioxide levels in the atmosphere. But what if drastic changes like these are not only a result of Earth’s environment, but also the sun’s location in the galaxy?

In a new paper published in Nature Astronomy , BU-led researchers find evidence that some two million years ago, the solar system encountered an interstellar cloud so dense that it could have interfered with the sun’s solar wind. They believe it shows that the sun’s location in space might shape Earth’s history more than previously considered. 

Our whole solar system is swathed in a protective plasma shield that emanates from the sun, known as the heliosphere. It’s made from a constant flow of charged particles, called solar wind, that stretch well past Pluto, wrapping the planets in what NASA calls a “a giant bubble.” It protects us from radiation and galactic rays that could alter DNA, and scientists believe it’s part of the reason life evolved on Earth as it did. According to the latest paper, the cold cloud compressed the heliosphere in such a way that it briefly placed Earth and the other planets in the solar system outside of its influence. 

“This paper is the first to quantitatively show there was an encounter between the sun and something outside of the solar system that would have affected Earth’s climate,” says BU space physicist Merav Opher , an expert on the heliosphere and lead author of the paper.

Her models have quite literally shaped our scientific understanding of the heliosphere, and how the bubble is structured by the solar wind pushing up against the interstellar medium— the space in our galaxy between stars and beyond the heliosphere. Her theory is that the heliosphere is shaped like a puffy croissant , an idea that shook the space physics community. Now, she’s shedding new light on how the heliosphere, and where the sun moves through space, could affect Earth’s atmospheric chemistry. 

“Stars move, and now this paper is showing not only that they move, but they encounter drastic changes,” says Opher, a BU College of Arts & Sciences professor of astronomy and member of the University’s Center for Space Physics. She worked on the study during a yearlong Harvard Radcliffe Institute fellowship. 

Opher and her collaborators essentially looked back in time, using sophisticated computer models to visualize where the sun was positioned two million years in the past—and, with it, the heliosphere and the rest of the solar system. They also mapped the path of the Local Ribbon of Cold Clouds system, a string of large, dense, very cold clouds mostly made of hydrogen atoms. Their simulations showed that one of the clouds close to the end of that ribbon, named the Local Lynx of Cold Cloud, could have collided with the heliosphere. 

If that had happened, says Opher, Earth would have been fully exposed to the interstellar medium, where gas and dust mix with the leftover atomic elements of exploded stars, including iron and plutonium. Normally, the heliosphere filters out most of these radioactive particles. But without protection, they can easily reach Earth. According to the paper, this aligns with geological evidence that shows increased 60Fe (iron 60) and 244Pu (plutonium 244) isotopes in the ocean, Antarctic snow, and ice cores—and on the moon—from the same time period. The timing also matches with temperature records that indicate a cooling period.

“Only rarely does our cosmic neighborhood beyond the solar system affect life on Earth,” says Avi Loeb , director of Harvard University’s Institute for Theory and Computation and coauthor on the paper. “It is exciting to discover that our passage through dense clouds a few million years ago could have exposed the Earth to a much larger flux of cosmic rays and hydrogen atoms. Our results open a new window into the relationship between the evolution of life on Earth and our cosmic neighborhood.”

The outside pressure from the Local Lynx of Cold Cloud could have continually blocked out the heliosphere for a couple of hundred years to a million years, Opher says—depending on the size of the cloud. “But as soon as the Earth was away from the cold cloud, the heliosphere engulfed all the planets, including Earth,” she says. And that’s how it is today. 

It’s impossible to know the exact effect the cold cloud had on Earth—like if it could have spurred an ice age. But there are a couple of other cold clouds in the interstellar medium that the sun has likely encountered in the billions of years since it was born, Opher says. And it will probably stumble across more in another million years or so.

Opher and her collaborators are now working to trace where the sun was seven million years ago, and even further back. Pinpointing the location of the sun millions of years in the past, as well as the cold cloud system, is possible with data collected by the European Space Agency’s Gaia mission , which is building the largest 3D map of the galaxy and giving an unprecedented look at the speed stars move. 

“This cloud was indeed in our past, and if we crossed something that massive, we were exposed to the interstellar medium,” Opher says. The effect of crossing paths with so much hydrogen and radioactive material is unclear, so Opher and her team at BU’s NASA-funded SHIELD (Solar wind with Hydrogen Ion Exchange and Large-scale Dynamics) DRIVE Science Center are now exploring the effect it could have had on Earth’s radiation, as well as the atmosphere and climate. 

“This is only the beginning,” Opher says. She hopes that this paper will open the door to much more exploration of how the solar system was influenced by outside forces in the deep past. 

This research was supported by NASA.

Explore Related Topics:

• Data Sciences
• Share this story
• 1 Comments Add

Science Writer Twitter Profile

Jessica Colarossi is a science writer for The Brink . She graduated with a BS in journalism from Emerson College in 2016, with focuses on environmental studies and publishing. While a student, she interned at ThinkProgress in Washington, D.C., where she wrote over 30 stories, most of them relating to climate change, coral reefs, and women’s health. Profile

Comments & Discussion

Boston University moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (EST) and can only accept comments written in English. Statistics or facts must include a citation or a link to the citation.

There is 1 comment on The Solar System May Have Passed through Dense Interstellar Cloud 2 Million Years Ago, Altering Earth’s Climate

Hi Jessica, this paper was extremely incredible with lots of sense. I always love to explore space and very convinced that life somewhere outside of our solar system exists. I know that the nearest solar system is 4 light years away from us. All the time I think that how we can make it possible to get there within our lifetime span. I know it is impossible but we can still keep thinking about it.

Post a comment. Cancel reply

Your email address will not be published. Required fields are marked *

Latest from The Brink

Being open about lgbtq+ identities in the classroom creates positive learning environments, should schools struggling with classroom management clamp down with more suspensions or turn to restorative justice, getting police officers to trust implicit bias training, air quality sensors could be coming to a bicycle near you, airplane noise may be bad for your health, bu researcher named a 2024 hertz fellow; award honors “innovators with the greatest potential”, is ai biased against some groups and spreading misinformation and extreme views, 3d printing robot uses ai machine learning for us army research, permanent birth control—vasectomies and tubal ligations—up following supreme court abortion ruling, explain this how do planets form, bu’s innovator of the year has pioneered devices to advance astronomy, microscopy, eye exams, bu study shows a correlation between social media use and desire for cosmetic procedures, covid-19 photo contest winners capture moments of joy, sorrow, meaning in crisis, how high-level lawsuits are disrupting climate change policies, the h5n1 bird flu is a growing threat for farm animals and humans—how serious is it, why is a bu researcher so fascinated with the diets of dung beetles, we are underestimating the health harms of climate disasters, three bu researchers elected aaas fellows, should people be fined for sleeping outside.