The chronology of collective cheating: a qualitative study of collective dishonesty in academic contexts

Collective cheating can have serious consequences in professional and educational settings. Existing data show that collective cheating is common and that individuals cheat more when they are in groups, highlighting the existence of some collective organisation in the endeavour. However, little is known about the emergence of this behaviour and the group processing involved. Drawing on qualitative data from 20 semi-structured interviews and a thematic analysis, the present study explored collective cheating in academic contexts. Seven themes emerged that retrace the chronology of collective cheating from its beginning to its impact on the group: reasons, birth and organisation of collective cheating, risk management, concealment strategies, justifications and social impact. Participants emphasised the extrinsic benefits, the interdependence of members and the importance of existing social norms, as well as the positive valence of this kind of dishonesty. In participants' recalls, cheating together means cooperating, showing solidarity, helping and supporting each other.


Introduction
In recent years, countless fraud scandals in many contexts, such as business and education have made the headlines.From the most mediatized to the most obscure, these events often show group members, e.g., people from the same company or school, acting in an organised and conscious manner to behave dishonestly.Frauds by major corporations such as Volkswagen (e.g., Hotten, 2015) and copying off during exams at Harvard University (e.g., Pérez-Peña & Bidgood, 2012) are typical examples of collective cheating.
The existing experimental studies in the social, educational, and organisational sciences have shown that cheating and strategic behaviour by groups are common, highlighting the existence of collective organisation in cheating (e.g., Gross et al., 2018).In spite of the evidence documenting the existence and effects of collective cheating, however, this growing literature has not yet revealed where cheating comes from, how it sets in and how it plays out, probably because it is difficult to observe collective cheating in the making.
This study wishes to contribute to a literature in full expansion by a fine-grained qualitative analysis of the processes that make collective cheating possible.Through a rich analysis of retrospective self-reported collective cheating episodes, we aimed to uncover the untold story of why and how people come to cheat together.In so doing, we limited our study to collective cheating in the educational context, arguably the most researched context as far as individual cheating is concerned (e.g., Murdock & Anderman, 2006).

From individual to collective cheating
Researchers' interests in cheating, in particular in academic settings, started since the '80 s (e.g., Covey et al., 1989;Houston, 1983) and developed through the '90 s (e.g., Anderman et al., 1998;Jones & Kavanagh, 1996;Kerkvliet, 1994;Roig & DeTommaso, 1995;Schab, 1991).Most studies focused on individual cheating, a form of dishonesty or breaches of academic integrity that can be defined as "individually bending or breaking rules to gain an unfair advantage for self or others" (van Prooijen & van Lange, 2016, p. 3).
As for mechanisms underlying individual cheating, research showed that psychological and moral costs are more relevant than economic costs and benefits in explaining the phenomenon (e.g., Gino et al., 2009;Mazar et al., 2008).From this point of view, self-concept maintenance theory (Mazar et al., 2008) and moral disengagement theory (Bandura, 1999) both explain how individuals who cheat activate various psychological mechanisms that protect their selfconcept and moral integrity.
Finally, the literature on individual cheating has highlighted the importance of the environment in which individual dishonesty takes place, in particular the influence of peer behaviour and attitudes (e.g., Jordan, 2001;McCabe et al., 2001;O'Rourke et al., 2010;Shmeleva & Semenova, 2019;Zhao et al., 2022).Gino et al., (2009, Experiment 1) have shown that exposure to other people's immorality increased individual cheating if the other person was an in-group (vs.out-group) member.Likewise, Jones and Kavanagh (1996) documented the influence of managers' and peers' behaviour-ethical vs. unethical-on workers' behavioural intentions.These findings highlighted the crucial influence that descriptive norms, i.e. how other individuals behave, have on individual behaviour (e.g., Cialdini et al., 1991;Keizer et al., 2008), even in cheating decisions.
Despite regular examples of collective fraud in various fields of everyday life, academic collective cheating is a relatively recent research area compared to individual cheating.
For this reason, we have included in the present literature review works in all areas of collective cheating, not only academic.Collective cheating is defined as "cheating that occurs when individuals of various backgrounds interact to create, implement, and sustain solutions to problems that violate ethical obligations or norms" (Castille & Fultz, 2018, p. 95).Collective cheating is a form of dishonesty that sees people "cheating together with in-group peers" (Pulfrey et al., 2018, p. 764), which clearly defines a collective behaviour where intra-group dynamics play a facilitating crucial role.Indeed, the existing literature (see Leib et al., 2021 for a meta-analytic review) has highlighted higher levels of dishonesty among those in groups: Individuals randomly assigned to groups cheat more than individuals who are alone (Chytilová & Korbel, 2014;Gross et al., 2018;Kocher et al., 2018;Soraperra et al., 2017).Similarly, groups were found to employ more "sophisticated" deception, advantaging the in-group (Cohen et al., 2009;Sutter, 2009; see also; Bornstein & Yaniv, 1998).Finally, Mazar and Aggarwal (2011, Study 2a) demonstrated that participants primed with a collectivistic mind-set (vs.individualist) were more likely to engage in bribery, i.e., offer an unofficial payment for a personal interest.In conclusion, the literature has shown that collective cheating is frequent, and people are more strategic and dishonest when acting in groups than alone.

Predictors of collective cheating
At the level of ingroup bonds, some studies revealed a positive relation between benevolence values, defined as "the preservation and enhancement of the welfare of people with whom one is in frequent personal contact" (Schwartz et al., 2012, p. 664), and collective cheating (Pulfrey et al., 2018).Manipulating participants' mindset with a focus on perceived self-other similarity (vs.dissimilarity) in collective and individual tasks, Irlenbusch et al., (2020, Study 1) found that, when the cheating was done at the expense of others, the sense of similarity (vs.dissimilarity) increased cheating in dyads (but see Chytilová & Korbel, 2014).Interestingly, studies employing self-report surveys found that students mainly asked other students, friends, or family for help in assignment outsourcing or, in general, academic cheating (Awdry, 2021;Bretag et al., 2019;Đogaš et al., 2014).These findings appear to support the importance of the relationship in collective cheating.
The expected collective utility of cheating was also explored as predictor.Collective cheating was found to be more frequent when cheating was beneficial for oneself but also for the other member (Conrads et al., 2013), especially when both members benefit in equal terms (Weisel & Shalvi, 2015).However, Kocher et al. (2018) found that group members communicated and coordinated in a dishonest manner even if the payoff was not commonly determined, showing the central role of group interaction.
To summarize, the recent literature on collective cheating highlighted the potential role of ingroup-level factors such as values, sense of familiarity with other members and the collective utility of cheating.However, this literature does not illuminate the beginnings of collective cheating: how it emerges and how group members align and organise themselves.

Mechanisms of collective cheating
Diffusion of responsibility, which allows people to feel less accountable for a dishonest action when shared with group members (Bandura, 1999), was presented as one of the most important mechanism promoted by team incentive schemes (Conrads et al., 2013).Likewise, Mazar and Aggarwal (2011) showed that the positive relation between collectivist priming and the propensity for bribery was fully mediated by perceived responsibility (for own actions), which was lower in participants primed with a collectivist mind-set.
As for norms, Soraperra et al. (2017) found that mere social norm exposure (i.e., exposure to rule violations) appeared to be sufficient to increase cheating behaviour.Indeed, the belief and knowledge of other student's cheating was found to be a predictor of students' cheating under the form of outsourcing work to relatives and friends (Awdry & Ives, 2021).Kocher et al. (2018) also found that exchanging arguments and justifications for dishonesty could lead groups members to learn about a new norm and its validity, and to adjust their beliefs about the behaviour.From another perspective, evidence showed that when people can choose to stay with the same dyad member or to switch, dishonest people tended to prefer staying with or looking for a dishonest partner (Gross et al., 2018).
Thus, this literature showed the importance of ingroup interactions and norms exchange, as well as diffusion of responsibility as a moral disengagement mechanism.However, although their importance has been discussed, little is known about how these mechanisms actually lead to collective cheating in the course of a group's interactions.

Environments of collective dishonesty
Some theoretical models, often related to corporate dishonesty and corruption, underlined the impact of organisational and contextual features on the emergence of unethical practices (Ashforth & Anand, 2003;Baucus, 1994;Castille & Fultz, 2018;Palazzo et al., 2012).They emphasized the importance of company culture and leadership, and underlined the role of contextual features in collective cheating.Strong leadership, ideologies and institutions, as well as an ethically permissive culture, routine of decisions, competitiveness, fear and pressure at different levels were highlighted as having an impact on the occurrence of collective cheating.
These models offer a relevant view of the environments in which collective cheating is likely to occur.However, these theoretical perspectives focused on the contextual influences that lead to the emergence of cheating, but investigated to a lesser extent how such contexts actually result in the group processes typically involved in collective cheating.

The present study
The reviewed literature has identified a number of predictors, mechanisms and environments that account for the emergence of collective cheating.However, it is still unclear how they emerge and shape a group's dynamics during collective cheating.Moreover, to our knowledge, no study has so far attempted to reconstruct an entire collective cheating episode.The aim of the present study is to fill these gaps by retracing the whole story: How does collective cheating begin?How does it take place and how is it organised?What processes are involved?
The choice to focus on collective cheating in educational contexts was motivated by the extent of the literature and theoretical bases in the field of individual academic cheating (e.g., Murdock & Anderman, 2006).Moreover, most people have been students and most of the students have cheated at least once (McCabe & Trevino, 1997;Waltzer & Dahl, 2021).This is the main reason for choosing to interview ordinary people: Anyone can potentially have a memory of academic collective cheating.
To collect data, we choose to employ a retrospective qualitative methodology, i.e. asking people to recall an autobiographical episode of collective cheating.We were interested in participants' experiences of collective cheating and how they construe them.Indeed, as Scott and Alwin (1998, p. 104) underlined, retrospective memories are "looking back on or thinking about things past or reviewing/contemplating the past rather than simply recalling or remembering things (…) and, as such, are assumed to be affected not only by past experiences but also by present conditions.(…) Such interpretations of the past draw on things that have happened more recently and explain past happenings in the context of the present".Previous studies have shown the effectiveness of autobiographical memories as an induction technique, particularly for emotional induction (e.g.Gerrards-Hesse et al., 1994;Maner et al., 2007).
The originality of the present study is to constitute a rich analysis of the academic collective cheating episode through a narrative approach.Drawing on interviews and a thematic analysis, our purpose was to explore how the predictors, group processes and mechanisms of collective cheating unfold during group processing.

Participants
Twenty people were recruited to participate in our study.All contacted participants accepted to participate in the study and all of them had a collective cheating episode to report (none of them declared that they never cheated).The study was presented as a short, confidential interview on an episode of collective cheating in the school and academic contexts.The data collection ended after these 20 interviews because the code saturation criterion was reached, i.e. when "no additional issues are identified and the codebook begins to stabilize" (Hennink et al., 2016, p. 593), merging inductive thematic saturation and data saturation (for a discussion see Saunders et al., 2018).Participants ranged from 22 to 45 years old (M = 32.74,SD = 7.40), with 55% of women.All participants were living in the Frenchspeaking part of Switzerland at the time of the interview, which was conducted in French.It is important to note that, although the interviews were conducted in Switzerland, and the majority of the participants recalled an episode that took place in Switzerland, the others talked about events in France, Italy and South America, as a function of their life trajectory.In terms of education, all participants had a university or a high school degree.Interviews ranged in duration from 5′40'' to 28′24'' (M = 13′07'', SD = 7′35'').The extracts of interviews have been translated and are reproduced here in English.

Materials
Each respondent participated in a face-to-face semi-structured interview.The structure of the interview guide (Supplementary Materials A) was the following: one main question, "Remember one time during your school and/or academic career when you cheated with one or more of your classmates.What exactly happened?",complemented, if necessary, by several follow-up questions.We chose this structure because we assumed that a collective cheating event has a "history", but we wanted to leave enough leeway for the participants to confirm or disconfirm this assumption.

Procedure
Participants signed a consent form where they were informed about their voluntary participation, the confidentiality of data and the audio recording of the interview.The Swiss Federal Act on Research involving Human Beings does not require approval from an institutional research ethics board when a study is conducted with adults and does not involve biological measures.Our Ethics Committee provided a written statement certifying this context that is available upon request.
The interview started with the first main question and ended when the participants had recounted the collective cheating episode and answered, if necessary, the follow-up questions.Participants were thanked and fully debriefed.Interviews were audio recorded and transcribed verbatim to ensure accuracy (Poland, 1995).

Data analysis
Given the dataset involved in this study-personal experiences-we chose to analyse data through a thematic analysis, a method popularized and formalized by Braun and Clarke (2013).A thematic analysis consists in "systematically identifying, organising, and offering insight into patterns of meaning (themes) across a data set."(Braun & Clarke, 2012, p. 57).Because we needed to focus on the content of the participants speech-the what-and not on the language-the how-we considered this method as particularly appropriate.Thematic analysis was carried out using the NVivo 12.04 package.
The procedure (Supplementary Materials B for the detailed procedure) used in this study was mainly based on the 6-phase approach to thematic analysis discussed by Braun and Clarke (2013), i.e., 1. Familiarizing with data; 2. Generating initial codes; 3. Searching for themes; 4. Reviewing themes; 5. Defining and naming themes; and 6.Producing the report.After being familiarized with the whole dataset and having allowed the first ideas to emerge, we attributed to each relevant unit of meaning-i.e. a segment of the data relevant for the research question-a code, which reflected the idea contained in the portion of the considered data.For the generation of initial codes, we decided to complement Braun and Clarke's methodology with some guidelines drawn from the qualitative consensual approach research (Hill, 2012;Masdonati et al., 2017), in order to reinforce the analysis' reliability: The first code list was the result of a consensual work between two members of the team, rather than an individual work.Moreover, before starting the systematic coding process of the entire dataset, a third team member verified the clarity and the relevance of each code.Once the coding process was completed for all interviews, codes were collated and combined around a common organising concept that is in fact captured by a theme or a sub-theme, i.e., a specific facet of the pattern reflected in the theme.

Results
Ten themes emerged from this thematic analysis.The analysis showed that seven of these themes could be organised chronologically, giving rise to what we have called the 1 3 chronology of collective cheating (cf.Fig. 1; see Supplementary Materials C for the complete tree of themes and subthemes).Three reflexive themes, that emerged from the ability to reflect on the participant's own experience, -on relations, emotions and cognition-, completed the overview of collective cheating but are beyond the scope of the present article (interested readers may contact the authors).

The chronology of collective cheating
The seven narrative themes were 1) Reasons for collective cheating; 2) Birth of collective cheating; 3) Organisation of collective cheating; 4) Risk management; 5) Concealment strategies; 6) Justifications and 7) Social impact of collective cheating.Themes and sub-themes are presented hereafter and then interpreted in the "Discussion" section.

Reasons for collective cheating
Involvement in collective cheating was described as driven by three types of benefit: academic, financial or social (cf.Fig. 2).

Academic and financial benefit
(...) we thought "yeah, how could we cheat and then get, yeah, get a better grade" (P10, W, 24).1 Unsurprisingly, many participants mentioned a benefit that was strictly related to school and academic performance, where a member or many members of the group wanted to perform well or wanted to avoid a poor or a bad performance.The need for reassurance or being in trouble are other reasons mentioned.
Participants also mentioned a purely financial benefit as a reason for participating in collective cheating.

Social benefit
Cheating was also done in response to a call for help, and presented as a form of altruism, to satisfy a need for belonging or social recognition, or because of some pressure to conform to the group.
It was him who asked me then, well, I was stressed but I was happy to help (...) it was also a time when I needed to integrate so I wasn't going to refuse him (P9, W, 24).

Birth of collective cheating
How does collective cheating begin?Three main sub-themes emerged from participants' recalls: proposing cheating, being proposed cheating and collective construction (cf.Fig. 2).

Proposing to cheat How did people propose collective cheating?
Fig. 1 The chronology of collective cheating Fig. 2 The reasons, birth and organisation of collective cheating But he was the one who, in fact, answered for us (…) I probably said to Claudio2 "you're good at maths (…) you can just quickly write the answer" (P18, M, 35).
Cheating was often proposed in the form of an explicit request.A less bold way of proposing collective cheating involved situations of fun among friends or emerged after an initial phase of investigation.
I think it was in the moment, we were laughing and then uh, I don't know if it was in the moment or the day before we decided to do it (P20, W, 39).
Related to the acceptance of one's proposal, participants also reported a kind of uncertainty.
The choice of the "partner in crime" also emerged as an important topic.For instance, friendship was described as a facilitator of demand or the choice was described as more strategic.
I think mostly... they are friends, so the group you have outside, [outside] the exam, outside the classroom (P3, M, 33).
Cheating was also described as sometimes happening with the same people and, other times, there was a kind of network implied.
(...) there are small groups, sub-groups, and sometimes... yes, there are people a little closer to you and then, depending on the size of the group, a person who knows another person well, a person who knows another person well so (P3, M, 33).
Being proposed to cheat Again, in some situations, a very explicit request was made and, in others, people detected an indirect request for help.
But explicitly (...) it was explicitly asked if I can help, because they couldn't find the solution.(P4, M, 35).
Collective construction Finally, participants perceived and described collective cheating-the intent and/or the initiation-as a collective construction.
I think we decided together, I don't remember, I don't remember anyone suggesting it... uh, it was more of a group dynamic, let's say we're going to do this thing together (P3, M, 33).

Organisation of collective cheating
Several scenarios emerged from the interviews, highlighting a prior discussion or planning, routines and an organisation based on opportunity (cf.Fig. 2).
Prior discussion or planning (...) we were talking about it just before we arrived, like "do we exchange the sheets if there are two series?"and then "ah bah, I turn and then I pass you the sheet, then you pass me yours", because we had to be... we had a few seconds actually to be able to do the exchange (laughs) (P19, W, 35).
Cheating was described as somehow premeditated-discussed and planned-by the group.
Routines A collective cheating routine sometimes emerged that no longer needed to be discussed among group members.
(...) afterwards, that's, there's a sort of pattern, because for example in Latin (…), we always did the same thing, after the first two times maybe not, but afterwards it's true that... they asked me to help, but then we used the same method (P4, M, 35).
Opportunity Collective cheating could also be the result of a combination of favourable circumstances linked to the situation itself.
(...) in this context, we are obliged to work on a computer, so we have access to other tools that potentially shouldn't be... provided to us (P14, M, 45).

Risk management
Risk-taking emerged as a function of various considerations related to facilitators of cheating, the implementation of strategies, a cost-benefit reflection, the protection of one's own performance ("Me first") or the minimization of risks (cf.Fig. 3).

Facilitators of cheating
In order to reduce the likelihood of being caught and, consequently, paying the cost of a dishonest behaviour, several participants reflected on facilitators decreasing risks.

And then, uh, normally, well, the one on the right had the A series and the one on the left had the B series.
And then what we did is, while he was distributing, we always sat in front enough, because he always started with the front tables and ended with the back tables, and while he was distributing, well, we would quickly turn around with the back table, and then we would switch series (P19, W, 35).
Participants referred to the context's predictability, the spatial proximity between group members-the most mentioned facilitator -, the access or the use of technology and the involvement of an authority in the collective cheating.

3
The fact that cheating could be done quickly without too much effort was also a factor reducing the perception of risk.

Implementation of strategies
In other situations, strategies clearly devoted to reducing or eliminating risks of being caught were implemented.

Then I was like, "(…) I'll put the diary a little bit more in the middle with my, my pencil box, then you put your pencil box at the same time as mine on top of my diary and then like that he won't see anything" (P11, W, 35).
Cost-benefit reflection Participants also compared the benefits of cheating with the costs involved at different levels.
And it wasn't really the risk, but... but it was rather the time.(…) for maths, it was really, 2 pages of, of process.So that, that, that, that took quite a long time (P6, M, 35).
"Me first" When cheating is done to benefit another person, the protection of one's own performance emerged as a necessary condition for engaging in cheating.

I gave him an answer to one, for the second one I gave him a tentative answer and then for the third I didn't answer at all. After a while I found it a pain (laughs).
On the one hand because... the, the test in question was complicated, it was long, uh, I also had to be able to answer myself (P1, M, 23).

Minimization of risks
Finally, risks could simply be minimized in how they were perceived.
We hadn't even thought too much about the seriousness of the thing, I think, or the possible negative consequences of the thing, so it was more like a game I think, so we really didn't think much about the possible stakes (P5, M, 31).

Concealment strategy
Reporting and whistleblowing was considered unlikely or rare.The reasons for this pertained to various arguments: personal interest, presence of a code of silence, absence of a (reporting) norm, altruistic action or uselessness of reporting (cf.Fig. 3).

Personal interest
Well, let's say that if I, if I reported him, I also reported myself, so, uh, I wasn't going to report anyway, I wasn't going to report anyone because I was in fact putting myself in an awkward position (P11, W, 35).
Reporting was deemed unlikely because every party involved had a personal interest in concealing cheating.
Code of silence A particular climate where a code of silence was established also emerged.
It's not usual, because it would be a kind of whistleblowing to do it and so it would be very badly seen and all that.(...) it would be a bad, uh, image, a bad reputation of the person who reports, that's it (P4, M, 35).

Absence of a (reporting) norm
There, there was never anyone who reported.In any case, that's the kind of thing I've never seen (P14, M, 45).
Reporting was described as a non-existing norm: This behaviour had never been seen and had never existed.

Altruistic action
The altruistic nature of collective cheating also explained unlikelihood of reporting.
Well, I had no interest in reporting his behaviour at that time.Well, I don't know.It was rather, well, it was rather, it was support (P12, W, 36).Uselessness Finally, reporting cheating was unlikely because this action was considered as useless, blaming the context for keeping the-dishonest-system in place.

Justifications of collective cheating
Several justifications related to morality emerged, like the normalisation of cheating, the minimisation of the behaviour, the ambiguity of the situation, temporal remoteness of the act, and finally a set of justifications that pattern Bandura's moral disengagement theory (Bandura, 1999).Protection of one's own competence is a set of justifications with a different aim, namely protecting, in a clear manner, the representation of the self as a competent person (cf.Fig. 4).

Normalisation of cheating
Participants often described collective cheating as a very common behaviour, a kind of existing norm: "everybody did it".
Most of the people I knew, yes, there were quite a few who were cheating too, so uh, it was easy to say, "well ok, just take your hand off so I can see".(P10, W, 24).

Minimisation of behaviour "
We did it but just a little" is another frequent justification allowing to minimise the intensity or frequency of the dishonest behaviour.
The test we had, there were some questions that were the same, not all of them, but some questions, yes, it was the same, but it only happened twice, eh, at school anyway (P2, W, 22).

Ambiguity of the situation
Participants also mentioned rules that were seen as unclear or ambiguous to justify collective cheating.The context was sometimes considered to leave room for cheating.
The test had not quite started.The teacher had already given us the sheets and then it was an exam where you had to put dates (...) I had started straight away, then I had a friend next to me who, uh, actually wanted to ask me a question just before the test started and then he saw, I put something in, a date he didn't know and then he was able to write (P1, M, 23).

Temporal remoteness of the act
The fact that the cheating took place a long time ago also served as justification.

Now I won't do it because I don't have that, that mentality or, or that sort of, lightness if you can say "no,
we studied then we'll do" (P20, W, 39).

Moral disengagement strategies
The following set of justifications refer to Bandura's moral disengagement theory (Bandura, 1999) and have already been discussed in this framework for individual unethical behaviours.Bandura's theory argues that people can selectivity engage or disengage their moral standards and the associated self-regulatory mechanisms via several psychosocial mechanisms.
The first three mechanisms-moral justifications, euphemistic labelling, and advantageous comparison-were found in our analysis and they aimed to adjust the immoral behaviour into a "good" behaviour.People justified their behaviour as being motivated by a moral or worthy goal, they decreased the immorality of the action according to how the action is called or because of an advantageous selection of the comparison terms.
The second set of mechanisms of moral disengagement we found enabled people to obscure or minimize personal responsibility for the questionable behaviour and felt less accountable.Their own personal agency was decreased by Finally, blaming the victims or the context for the own action, participants described their action as a defensive or reactive behaviour, and they could also feel self-righteous in the process.For example:

Moral Justifications.
Because I'm someone who likes to help people (laughs).Then, who likes, ... if I can help I do it then, uh, well, there you go.Yes, I think that's it (P11, W, 35).

Advantageous Comparison.
(...) apart from getting caught in the act, direct, but it's not something, it's not like I gave him my dissertation and he's going to copy it (P8, M, 22).

Euphemistic Labeling
We have to work on things... we don't necessarily have time, so we helped each other a lot (...) I even think about the exam, it was, it was natural to be able to give each other a hand (P14, M, 45).

Displacement of Responsibility.
Yes, I've never questioned that at any time, well... He's the boss.He was, he was, yes, he was in charge there and then, ok (P12, W, 36).

Diffusion of Responsibility.
For me it was that from the moment we did it together, uh, everyone had the same responsibilities in relation to it... and then, well, they were people, ... we were a group of close friends since, well, since the college, some a little before (P15, W, 39).

Blaming the Victims or the Context.
We really had a teacher who was harsh, who didn't explain, and then, it's not, it's not even that we didn't bother, it's that you could... study, but during the tests, he was, he was, it was really, (…) you could do it, but you could check the answers a bit with this system (P20, W, 39).
The only two mechanisms of moral disengagement theory (Bandura, 1999) that did not emerge from interviews were related to dehumanisation of victims and minimizing, denying, ignoring, or misconstruing the harmful effects of their action.The absence of these two mechanisms can be attributable to the fact that, in academic collective cheating, consequences are not really seen as harmful or detrimental to others who could be perceived as real victims.
Protection of one's competence Emphasizing one's own competence and intelligence appeared as a frequent justification for the involvement in cheating in favour of others.
Well, we were, I wasn't cheating after all, I did the calculations, I mastered maths well enough to... it was more of a game, an exercise to do these maths exams, because I had a much higher level than what they were asking us.(…) And the others knew that I was ok... they knew that I was quite altruistic, they knew me (laughs) at that level, that I would agree to pass on the results to them" (P17, M, 40).

Social impact of collective cheating
Finally, collective cheating appeared to have surprising consequences for the relationship between group members.Three sub-themes emerged from the analysis: no relational change, positive relational change and cheating as a funny memory to share (cf.Fig. 4).
No relational change Participants often remarked that collective cheating did not lead to change in group dynamics.
We, we stayed friends.There were no, no problems, no worries.It was the same relationship at first and it was the same relationship in the end (P6, M, 35).
Positive relational change Some participants mentioned collective cheating as a positive factor in the relationships between group members.

In a positive way, it [the relationship] has changed in a positive way (…) it created even more of a sense of group (…) collaboration and group cohesion (P3, M, 33).
Cheating as a funny memory to share Other participants said to remember the event as something fun, as a memory to share with friends.
It's a story that we remember like a unifying thing, that we laugh about (P15, W, 39)

Discussion
The present results provide for the first time a detailed and chronological overview of the collective cheating episode in academic contexts.The participants' narratives allowed us to retrace the unfolding of collective cheating, from the reasons and the first steps that lead to cheating, through its organization and the reflections about risk, to the justifications used and even the consequences for the group itself.We requested the participants to narrate their retrospective memories, which we argued is an important feature of the present research, since it affords to access the construal that people who have collectively cheated communicate.Despite the complexity and the variety of this frequent group behaviour, several common elements emerged across themes.The first three elements presented in this discussion-extrinsic motivation, descriptive norms, and protection of the self-were anticipated in the literature review.Two other common elements emerged from our analyses that were unexpected -slippery slope or steep cliff, and the cooperative dimension of collective cheating-; they have been theoretically integrated in this discussion.

The role of extrinsic motivation
When talking about reasons for collective cheating, which informs the question of predictors of cheating, participants particularly emphasized the role of extrinsic motivation (Ryan & Deci, 2000a), i.e. performing a behaviour because of external pressure, and performance goals, i.e. goals related to the desire to demonstrate one's own competence (e.g., Elliot & McGregor, 2001).Extrinsic motives (with external regulation, to employ Ryan & Deci's, 2000a taxonomy) such as academic or financial benefits, both for the whole group or for one of its members, were typically associated to competitive settings, steeped in social comparison (Festinger, 1954) and contextual pressures to perform and succeed.Interestingly, research showed that such settings do drive people to cheat in an individual or collective manner (e.g., Bäker & Mechtel, 2019;Cohen et al., 2009;Palazzo et al., 2012;Pulfrey & Butera, 2013;Pulfrey et al., 2018).Another instance of extrinsic motivation (in this case with introjected regulation, Ryan & Deci, 2000a) can be found in the social benefits mentioned by the participants.People tend to develop membership to social groups in order to satisfy their need for relatedness, one of the three psychological human needs (Ryan & Deci, 2000b).Group membership provides a shared positive social identity, which contributes to their self-concept and self-esteem (Tajfel, 1979).Personal need for identification can lead to loyalty to the in-group, which, in turn, has been shown to play a central role in the engagement in corrupt behaviours (Anand et al., 2004).Similarly, motivation to behave to gain social approval increased the adoption of context-specific competitive performance-approach goals and individual cheating (Pulfrey & Butera, 2013).Our results suggest that collective cheating may arise from a similar need.

The importance of descriptive norms
As far as mechanisms of collective cheating are concerned, descriptive norms-the norms inferred from other people's behaviour-yield substantial effects on one's attitudes and behaviours (e.g., Cialdini et al., 1991;Keizer et al., 2008).A typical and common behaviour mentioned in the present scenarios was that nobody reported cheating, regardless of what appears to be moral or immoral, approved or disapproved.Whether as a justification, i.e., everyone did it, or because of the absence of a culture of reporting, our results clearly showed the power of social norms of silence.
In line with this account, the simple exposure to a dishonest norm in collaborative settings was shown to increase (individual) cheating (Soraperra et al., 2017).Moreover, Kocher et al. (2018) have shown the role of communication and norm compliance in collective cheating.Exchanging arguments for dishonesty probably leads participants to learn a new shared norm of dishonesty, which provides normative support for justifications.In terms of environment of collective cheating, such normative support, especially if afforded over long periods of time, may constitute a favourable environment, as revealed by the fact that participants indicated that sometimes cheating did not need any specific organisation as it was rooted in well-practiced routines.

Protection of the self
Our results underlined how important it was for participants to maintain a good self-image.Protecting the self-concept is a topic that emerged in several themes related to mechanisms pertaining to the justification of a moral failure, the maintenance of a good social image and a view of the self as competent.
The rich theme on justifications showed the importance moral failure management, as a number of justifications were mentioned.In addition to the moral disengagement mechanisms (Bandura, 1999) detailed above, another theory deserves to be mentioned in the analysis of these results: self-concept maintenance theory (Mazar et al., 2008), which argued and showed that "ordinary" people cheat but not to a large degree, in order to find a balance between the potential benefit of cheating and the need to reduce the threat to one's sense of self as a moral person.Indeed, some justifications, i.e. minimising the behaviour and mentioning the ambiguity of the situation, could be understood in the light of this theory.As Mazar et al. (2008) explain, cheating "just a little" does not require people to "update" or change their self-image.Moreover, in contexts in which the norms or rules are unclear, ambiguity can be used as a justification in an opportunistic manner, giving people "more room for interpretation of their actions, making the moral implications of dishonesty less accessible" (Mazar et al., 2008, p. 638).
Interestingly, not only did our participants try to maintain a good image in moral terms, but they expressed concerns about social image and competence.Reputation, which strongly influences self-esteem and social identity (Bromley, 1993), can be at stake when proposing collective cheating.Interestingly, sometimes, the need for being free to retract if necessary and manage one's self-image resulted in not being explicit in the request of cheating and test the waters.Moreover, participants often mentioned their own competence and intelligence as a justification for the involvement in cheating in favour of others.This surprising justification may suggest a compensation strategy and brings us to a question: is competence more important than morality?Interestingly, as already mentioned, participants often described collective cheating where they were involved as "helpers", emphasizing the importance of maintaining a view of the self as competent.And indeed, the importance of competence, and the negative consequences of competence threat, have been discussed in social comparison theory (e.g., Butera & Darnon, 2017;Festinger, 1954) and, as a basic human need, in self-determination theory (Ryan & Deci, 2000a, b) to point that individuals resort to numerous mechanisms in oder to maintain a positive image of themselvs in terms of competence (e.g., Tesser, 1988).

Slippery slope or steep cliff?
Participants discussed various scenarios describing how collective cheating started and became organised.They often described a demand-explicit or not-or a collective construction, but also planned events or routines of collective cheatings.
In some scenarios, the request and acceptance of collective cheating can be understood as the first step of the slipperyslope of corruption (e.g., Ashforth & Anand, 2003;Castille & Fultz, 2018;Darley, 2005;Palazzo et al., 2012): A first unethical act starts the chain of other unethical acts.However, according to some other participants, this first step can sometimes be precipitated by contextual circumstances and opportunities.Studying the road to corruption, Köbis et al. (2017) found that bribery was more severe when the opportunity to engage in it was given to individuals in a direct and abrupt manner that could be understood as a favourable circumstance (the steep-cliff metaphor).Beyond these two metaphors, several participants reported being simply engaged in routine behaviours, which parallel those described in the literature on corruption, where unethical behaviour is described as a repetitive institutionalized organisational behaviour, with well-practiced decisions and procedures that participate to normalize corruption (e.g., Castille & Fultz, 2018;Palazzo et al., 2012).

The cooperative dimension of collective cheating
In participants' memories, collective cheating often emerged as a positive feat appealing to friendship, solidarity and cohesion.Indeed, when choosing the "partner in crime", friendship or the habit of working together, are often mentioned as key factors.A parallel can be drawn with findings showing that benevolence values and familiarity between members (Pulfrey et al., 2018), as well as perceived similarity (Irlenbusch et al., 2020, Study 1) are related to the acceptance of collective cheating and cheating behaviour.Moreover, accepting collective cheating seemed to be subject to solidarity and loyalty, but also social pressure (cf. the theme "reasons of collective cheating").Cooperation and loyalty have often been shown to be higher when identification to the group were higher (e.g., Hogg & Reid, 2006), and Hertel and Kerr (2001) found an increase in in-group favouritism and identification when loyalty was primed.
Cooperation and positive goal interdependence have also been shown to increase the perception of bonding among group members (Johnson & Johnson, 2005).This feeling of unity and cohesion could therefore be linked to a common and collective setting, management and justification of cheating.It is important to note that participants, whether they proposed to cheat, or they accepted cheating, often described the behaviour of others.Indeed, participants mainly reported collective cheating by describing their own role as "helpers", academically speaking.Interestingly, Levine and Schweitzer (2014) found that the perception of an individual's morality seems to depend on the intentions of the cheater.Cheaters can be perceived to be moral by others, especially if their behaviour has an altruistic or a prosocial character, and an intention of advantaging and helping others; this is widely mentioned by participants to explain the unlikelihood of reporting.Indeed, reporting and whistleblowing are not a common behaviour in the collected stories of collective cheating.The participants mentioned the two major motivations in favour of concealment that were highlighted by the literature on the code-or lawof silence, known in the context of Italian Mafia under the name "Omertà" (Varese, 2017).The first is the trust between group members (e.g., Chin & Wells, 1998;Skolnick, 2002), aiming to maintain collective well-being (Rothwell & Baldwin, 2007).The second motivation is related to fear of negative consequences or retaliation from the group, such as avoidance, loss of group support or stigmatization (Hersh, 2002;Thau et al., 2015).Also of interest is the impact of cheating on the group members.When looking at the consequences, participants often reported an increase in cohesion and feelings of bonding between group members after the experience of collective cheating.This finding may however not be surprising if analysed from the perspective of cooperation.Indeed, as already mentioned, cooperation and positive social interdependence have shown positive consequences even from an interpersonal point of view (Johnson & Johnson, 1989, 2005, 2009).Interestingly, none of the participants reported a negative impact of the cheating episode on the group.Cheating, in its collective form, is viewed as something rather positive that can have either no impact on group members or a positive impact, strengthening their cohesion and friendship.

Limitations
In the present study, we focused on collective cheating in academic contexts, but we referred to a wide range of literatures rooted in other fields.Future research should study collective cheating in other environments-e.g., at work, in sports, in politics-to be able to draw parallels with regard to the emergence of cheating, its course and the group processes involved.It should also be noted that, even if we restricted our focus to academic collective cheating, our sample size of 20 interviews may be a limitation for generalization.Moreover, it is possible that the mechanisms and group processes are different depending on the number of members in the group, especially if it is a dyad or a group of more than two people.It would then be important to studying these two settings separately and in detail.

Conclusions
Collective cheating is a pervasive and increasingly documented phenomenon in school and academic environments.Through the memories of participants who have experienced collective cheating, we traced the episode of collective cheating from its emergence to its impact on the group.The main contribution of this research is that, for the first time, a qualitative study offers an account of the chronology of collective cheating from an insider's point of view.In addition to providing a detailed overview of the event, this study highlights the positive perception of collective dishonesty.Collective cheating is indeed described as group cooperation, resulting in a positive view of this collective behaviour despite the awareness of dishonesty.

Fig. 3
Fig. 3 Risk management and concealment strategy

Fig. 4
Fig. 4 Justifications and social impact of collective cheating