Introduction

As higher education becomes more and more globalized, more and more students in the U.S. and around the world find themselves study alongside their foreign-born classmates. According to an estimate from the Association of International Educators (NAFSA 2019), the number of U.S. students studying abroad has continued rising over the past decade with Europe as the most popular destination with about 181,000 students as of 2017 (nafsa.org). Likewise, the U.S. has been the top choice of international students. According to the Institute of International Education (2019), the number of students from the European Union studying in the U.S. in 2018 remains strong at 92,655, a slight decrease of 0.2% compared to 2017 (2018 Open Door report). Given the trend of increasing internationalization across the world campuses, besides the documented benefits by internationalization (e.g., cultural diversity, reduced stereotyping), researchers have explored the downside associated with internationalization such as increased immoral behavior (Lu et al. 2017). It is therefore important to understand cultural differences in academic dishonesty to improve outcome and procedural fairness because native born students and their foreign-born classmates compete for the same educational resources (e.g., scholarship).

Academic dishonesty, defined as any dishonest behavior or action taking place during an academic exercise, has long been a subject of interest among academic researchers in the U.S. (See McCabe et al. 2006; Whitley 1998 for a review). Although research comparing academic dishonesty engaged by students in the U.S. and European countries have emerged in recent years (e.g., Chudzicka-Czupała et al. 2016; Grimes 2004; Lupton et al. 2000); the question as to what extent academic dishonesty exists in colleges in France and Greece remain elusive.

Academic dishonesty was found to exist in schools among French pre-teen and teenagers based on prior research using self-report and overclaiming technique data (e.g., Fell & König 2020; Gentina et al. 2017; Guibert & Michaud 2009). However, research is scarce concerning academic dishonesty among college students in France. Hendy and Montargot (2019) found that academic dishonesty was prevalent within a sample of 178 business students at a southwestern business school in France. Similarly, academic dishonesty is poorly understood and researched in Greece. In our literature review, we found two studies that examined business students’ attitudes toward academic dishonesty at Aristotle University of Thessaloniki (Karassavidou & Glaveli 2006, 2007) and one study exploring the perception of penalty for academic dishonesty among a sample of graduate students at the University of Athens Dental School (Koletsi-Kounari et al. 2011).

The scarcity of academic dishonesty statistics for undergraduate and graduate students at the national level in France and Greece was further confirmed in the respective country reports filed with the European Union under the Impact of Policies for Plagiarism in Higher Education Across Europe (IPPHEA) in which more systematic research to identify the extent to which academic dishonesty exists was called for (Glendinning et al. 2013; Kokkinaki 2013). This study was intended to fill this gap in the literature by investigating academic dishonesty among college students in France and Greece using Bandura’s (1986) social learning theory. The results were then compared to a sample of college students from the U.S.

Literature Review

Academic dishonesty research over the past sixty years has focused on examining individual and situational characteristics of students in explaining academic dishonesty. Individual variables as antecedents of academic dishonesty include demographic variables (e.g., age, sex), cognitive and non-cognitive variables. Many studies including a meta-analysis showed that older and female students were found to have a more negative attitude toward academic dishonesty and lower frequency of academic dishonesty behavior than did younger and male students (e.g., Harris et al. 2019; Niiya et al. 2008; Scrimpshire et al. 2017; Whitley 1998; Whitely et al. 1999; Yang 2012). Across our life span, as people age, one tends to become more responsible and hardworking than when they were younger (e.g., Bazana & Stelmack 2004). Because hardworking students were found to engage in lower frequency of academic dishonesty than those not as hardworking based on a meta-analysis (Giluk & Postlethwaite 2015), it may explain why chronological age has an inverse relationship with academic dishonesty.Footnote 1

The gender difference in attitude toward academic dishonesty favoring women can be explained using gender role expectation theory (Eagly 1995). Specifically, women are expected to be more harmonious, rule bound, caring, and kind relative to men. This may explain why women may have higher ethical standards than men and therefore, tended to view academic dishonesty as more unacceptable compared to men (Karassavidou & Glaveli 2007; Whitley et al. 1999). However, this gender difference has diminished over time. For example, a recent study showed that female students from gender egalitarian countries/cultures engaged in academic faking (measured as frequency of overclaimed statements) more so than did male students (Fell & König 2020). A plausible explanation for the reverse gender difference in academic cheating behavior might be the increased participation of females in the workforce may necessitate male oriented role expectation in females.

Whereas there is a lot of research examining non-cognitive variables as antecedents of academic dishonesty, only a handful studies investigated cognitive ability as a predictor of academic dishonesty with mixed findings. For example, cognitive ability measured by the Wonderlic Personnel Test (WPT) was found to have no relationship with self-reported academic dishonesty among a sample of surveyed business and non-business college students in the U.S. (Hendy & Biderman 2019). However, in an experimental study, perceived cognitive ability was negatively related to academic dishonesty such that those who perceived themselves as intelligent engaged the least in academic dishonesty behavior and those seeing themselves as having a lower level of cognitive ability engaged the most in academic dishonesty (Bing et al. 2012).

Various non-cognitive variables as antecedents of academic dishonesty have been studied including personality traits and individual attitudes. Based on a meta-analysis, of the five personality traits including agreeableness, extraversion, conscientiousness, neuroticism, and openness to experience, only conscientiousness was a significant predictor (Giluk & Postlethwaite 2015). Specifically, students scoring higher on conscientiousness were less likely to engage in academic dishonesty than those scoring lower on conscientiousness (Giluk & Postlethwaite 2015).

Other attitudinal variables examined in previous research include Ajzen’s (1991) theory of planned behavior variables including attitude toward cheating, subjective norms, perceived behavioral control, and cheating intention (e.g., Chudzicka-Czupała et al. 2016; Harding et al. 2007; Hendy & Montargot 2019; Stone et al. 2009; 2010). Moral obligation not to cheat or the reverse of cheating justification, defined as one’s belief that cheating is wrong, is another attitudinal variable shown to be a negative predictor of academic dishonesty in previous research (Hendy & Montargot 2019; Scrimpshire et al. 2017; Whitley 1998).

Situational or contextual variables previously studied as predictors of academic dishonesty include whether there is an honor code within the school or university (e.g., McCabe & Treviño 1993; Bing et al. 2012); disciplinary action/penalty for academic dishonesty, faculty attitude toward cheating, learning environment (e.g., McCabe et al. 2001). Most of the above cited studies showed that the presence of an honor code; faculty who were strict about enforcing the honor code, a well-enforced system of penalty for honor code violation, and a supportive learning environment all contributed to lowering incidents of academic dishonesty (McCabe & Treviño 1993; McCabe et al. 2001; Bing et al. 2012).

Despite a plethora of research explaining academic dishonesty, very few studies have examined academic dishonesty under the lens of Bandura’s (1986) social learning theory, especially in a cross-cultural context. We are aware of one study that empirically examined unethical behavior using a large sample of U.S. business students (O’Fallon & Butterfield 2012). A series of focused group interviews were utilized in another study (Burnett et al. 2016) using a sample of 39 U.S. students majoring in health sciences. Whereas the focus of theories used in prior research to explain academic dishonesty is on the individual, Bandura’s (1986) social learning theory focuses on the situation as the driver of academic dishonesty behavior. As discussed in detail in later paragraphs, Bandura’s (1986) assumes that learning to engage in academic dishonesty is determined by the situation (e.g., whether peers also cheat, whether there is a consequence associated with academic cheating) rather than by individual factors such as ability to cheat and/or attitude toward cheating. Also known as social cognitive theory, Bandura (1990, 2001) later proposed an individual characteristic of moral disengagement as a driver, in combination with the situation, of unethical behavior. In this study, we examined the combined influence of individual and situational characteristics on academic dishonesty.

There are at least two reasons why it is important for researchers and policy makers to have a deeper understanding of student academic dishonesty from a social learning perspective. First, observational learning, an underlying principle of Bandura’s (1986) social learning theory, defined as the extent to which students imitate their peers’ behavior (both positive and negative) in the classroom, is under the control of faculty. Whitley (1998) in his extensive meta-analysis of academic cheating, called for more research examining why students engage in academic dishonesty from the situational point of view (e.g., observational learning) so that early intervention can be developed to reduce academic dishonesty. Second, given the effectiveness of social learning in cross-cultural training programs, it is relevant and important to understand how social learning can be adapted and implemented to reduce academic cheating as a negative consequence of increased globalization in higher education (Lu et al 2017). Our study was designed to address the gap in extant literature on academic dishonesty by explaining it through the lens of Bandura’s (1986) social learning theory and across three cultures of the U.S., France, and Greece.

Hypotheses Development

Bandura’s (1986) social learning theory states that observational learning occurs indirectly through the process of behavioral modeling. This modeling process includes four subprocesses of attentional, retention, motoric, and reinforcement and motivational processes. In the attentional process, Bandura stated “The people with whom one regularly associates delimit the types of behavior that one will repeatedly observe and hence learn most thoroughly” (Bandura 1986: 6). This suggests that students will learn to engage in academic dishonesty by observing their peers cheat and not getting caught. Next, in the retention process, students will need to memorize the behavior through “symbolic transformation and cognitive organization of modeling stimuli and covert rehearsal” (Bandura 1986: 7). The students will then reproduce the learned behavior, i.e., academic cheating, in the motoric reproduction process. Last, the learned behavior will be performed repeatedly with the accompanying feedback and/or rewards that helps with the reinforcement and motivational process.

Similarly, Akers’ (1985) social learning theory of deviant behavior posits that students learn to refrain from performing deviant behavior through interactions with others including their friends and family. One study found support for Akers’ (1985) social learning theory such that the level of peer association in academic dishonesty was positively related to self-reported academic dishonesty (Lersch 1999). Peer involvement, defined as the extent to which students share textbooks, notes, and coursework materials, was found to be positively related to academic cheating among a sample of French adolescents (Gentina et al. 2017). Students who admitted engaging in academic cheating also reported having observed their peers cheat and that motivated them to cheat in at least one study (e.g., Jordan 2001).

Hypothesis 1: Peer Involvement in Academic Dishonesty will be Positively Related to Self-reported Academic Dishonesty Across French, Greek, and U.S. students.

According to social learning theory (Bandura 1986), observational learning will be strengthened by informing learners in advance of the reward or lack thereof (i.e., penalty). Thus, it is reasonable to expect that if the students know in advance of the penalty associated with academic dishonesty, they will be likely deterred from engaging in such behaviors. In an experimental study, students identified avoiding the severe penalty or consequence of cheating if being caught as a reason not to cheat (Miller et al. 2011). In another experimental study, faculty discussion of an honor code and the consequences associated with violating such code was found to be associated with lower frequency of academic cheating than just the presence of an honor code alone (Bing et al. 2012). In addition, one study reported that the perception of peer behavior was positively related to academic cheating across three types of campuses – those without an honor code, a modified honor code, and a traditional honor code – with the traditional honor code being associated with the lowest level of academic cheating (McCabe et al. 2002). The following hypothesis is formed:

Hypothesis 2: Penalty for Academic Dishonesty will be Negatively Related to Self-reported Academic Dishonesty Across French, Greek, and U.S. students.

As discussed in the literature review, moral obligation not to cheat or cheating justification has been found to be an important antecedent of academic cheating in prior research. For example, Whitley (1998) conducted a review of 107 studies published between 1970 and 1996 reported a strong negative effect size (d = -0.76, k = 3, N = 204) for moral obligation to avoid academic cheating. Moral obligation, operationalized as the extent to which students believed it is consistent with their moral values to cheat, was found to be positively related to cheating intention and past cheating behavior among a sample of engineering and humanities students in the U.S. (Harding et al. 2007). In addition, one study found a positive correlation between justification of cheating and self-reported academic dishonesty among a French student sample (Hendy & Montargot 2019). Roberts et al. (2018) found that moral/cheating justification, a component of moral disengagement (Bandura, 1990; Detert et al. 2008), to be positively related to unethical decision making.

Hypothesis 3: Cheating Justification will be Positively Related to Self-reported Academic Dishonesty Across French, Greek, and U.S. students.

According to Hofstede (1998), survey data can be used to compare cultural differences in attitudinal outcomes when the criteria for comparison are specified, an appropriate unit of analysis for comparison is used, and the observations are equivalent. The above conditions were satisfied in this study because we measured the perception of students concerning academic dishonesty, an attitudinal outcome. The criteria for comparison were specifically student perception and the unit of analysis is individual students nested within each of three countries included in the study. We utilized Hofstede’s cultural framework based upon its popularity as a foundation for cross-cultural human resource management training and development (Taras & Steel 2009). Hofstede’s cultural scores have also been replicated and updated using longitudinal meta-analysis (Taras et al. 2012).

Power Distance. People in high power distant cultures were found to have a lower level of life satisfaction and more corruption relative to those in low power distant cultures (Carl et al. 2004; Seleim & Bontis 2009). In academic settings, students might justify their academic cheating behavior as one way to restore equity and reduce power distance if they perceive themselves to be unfairly treated (McCabe 2001). Thus, we expect that students from a higher power distant culture (i.e., France) would engage in more academic dishonesty than those from a lower power distant culture (i.e., Greece, the U.S.).

Individualism/Collectivism. People living in collectivistic cultures tend to rely on the group for sharing resources whereas people in individualistic cultures tend to rely on their own individual resources. In collectivistic cultures, individual achievements are considered important, even if such achievements were obtained by cheating (Hofstede 1986). Thus, we expect that students living in collectivistic cultures (e.g., France, Greece) might be more inclined to cheat by obtaining information on an exam and sharing that information to their peers than students from a more individualistic culture (e.g., the U.S.). A qualitative study conducted by Hayes and Introna (2005) revealed that graduate students from a more collectivistic culture (e.g., Asian culture) were more likely to view plagiarism on an academic assignment as more acceptable than students from an individualistic culture (e.g., the U.K.)

Masculinity/Femininity. Achievement, assertiveness, and competition are some of the highly prized values in masculine cultures whereas caring for the weak and building relationships are important values in feminine cultures (Hofstede 1991). Engaging in academic dishonesty such as letting others copy your own work or taking the exam for someone else or let someone else take the exam for you might be one way to help the weak and build relationships with one’s peers. In addition, using data from the Program for International Student Assessment (PISA), students from more gender egalitarian (i.e., feminine) cultures were found to engage in academic faking more so than students from less gender egalitarian (i.e., masculine) cultures (Fell & König 2020). Given the more feminine culture of France, we expect that French students would engage in more academic cheating than Greek and U.S. students respectively.

Uncertainty Avoidance. According to Hofstede’s (1991) cultural dimensions research, people in high uncertainty avoidance cultures tend to rely on religion, law, and technology to give them a sense of structure. In addition, social norms and rules are relied upon to reduce future uncertainty (House & Javidan 2004). Because academic cheating is a form of risk taking, which deviates from the social rules and norms, it is reasonable to expect that students from a high uncertainty avoidant culture (Greece, France) will engage in less academic dishonesty relative to students from a low uncertainty culture (U.S.).

Long-term Orientation. France scores relatively high on the long-term orientation (63) cultural dimension compared to the U.S. (45) and Greece (26), which means French people on average prefer a more pragmatic approach toward education than do American and Greeks (Hofstede Insights 2018). A pragmatic approach to education emphasizes efficiency at the expense of effectiveness. One study showed that self-reported cheating intention was higher among Ukrainian students (Ukraine also scores relatively high on long-term orientation) compared to that of American and New Zealand students (Chudzicka-Czupała et al. 2016). Because French students were more pragmatic than were U.S. and Greek students, we expect that French students would engage in more academic dishonesty than would Greek and U.S. students.

As an individual/personal factor and an antecedent of academic dishonesty, conscientiousness has been established as a positive and most valid predictor of the Big Five personality traits (agreeableness, extraversion, conscientiousness, emotional stability, and openness to experience) when it comes to academic dishonesty based on a meta-analysis (Giluk & Postlethwaite 2015). Highly conscientious students were found to be less likely engaged in academic dishonesty. Further, prudence, a component of conscientiousness, was found to be a distal predictor of academic dishonesty in prior research such that situational factors including attitude toward cheating added incremental variance beyond that of conscientiousness (e.g., Stone et al. 2009; 2010). In this study, we wanted to replicate prior research and included conscientiousness as a control variable. Figure 1 shows a conceptual model of the current study.

Fig. 1
figure 1

– A Hypothesized Model of Academic Dishonesty in the current study

Full size image

Method

Participants and Procedure

Data were collected online anonymously via Qualtrics from a total of seven hundred and forty-two university students from France, Greece, and the United States who participated in this study on a voluntary basis during 2017 and 2018. Each of the authors served as principal investigator for data collection in each of the countries included in this study. The research protocol was approved by each author’s Institutional Review Board or ethics committee at each of the participating universities. Students were invited to participate in the study on a voluntary basis via an email invitation in which a link to the survey on Qualtrics was inserted. Students read and signed a consent form prior to completing the survey. Students were able to withdraw from the study or discontinue their participation in this study without penalty. The research questionnaire was translated to French and Greek by the second and third authors who are native speakers of the respective languages and certified by local academics. Back translation into English was followed per Brislin’s (1986) suggestion.

An email was sent out to students with a link to the survey soliciting their participation at the beginning of the semester. An email reminder was sent after two weeks to improve response rates. The U.S. data were collected during both years whereas for the French sample, data were collected in 2017 and in 2018 for the Greek sample. The U.S. data were collected from a mid-Atlantic university whereas data were collected at a private business school in southwestern France and a university in northern Greece. Students were given extra credit in exchange for their participation in the study in the U.S. No extra credit was given to the French and Greek students because of the respective policy at those institutions. After removing 32 international students from the U.S. and French samples to maintain the homogeneity of each country for comparison purposes (the Greek sample consists of native-born students only), as well as cases with missing dataFootnote 2; the final sample consists of six hundred and seventy-three students. Table 1 shows the descriptive and demographic statistics of the study participants by country.

Table 1 Descriptive statistics of student sample by country
Full size table

Measures

Conscientiousness

Ten items from the International Personality Inventory Pool (IPIP) were used to measure conscientiousness (Goldberg 1999). Sample items include “I am exacting in my work”; “I follow a schedule”; and “I get chores done right away”. Participants were asked to indicate how accurately each statement described them on a 5-point scale with “1” being “very inaccurate” and “5” being “very accurate”. The Cronbach’s alphas for this variable were 0.67, 0.78, and 0.78 for the French, Greek, and U.S. samples respectively. High scores on this variable reflect a higher level of conscientiousness than do low scores.

Peer Involvement in Academic Dishonesty

Two items were written to measure the extent to which students perceived their peers engage in academic dishonesty. Participants were asked to rate the likelihood that their peers engaged in academic dishonesty during the previous academic year. Items include “How likely is it that your classmates engaged in a major academic dishonesty behavior (e.g., plagiarize a research paper, took an exam for somebody else or have someone else take an exam?)” and “How likely is it that your classmates engaged in a minor academic dishonesty behavior (e.g., cheat on a quiz or a homework assignment?”. Anchors include 1 “very unlikely” to 5 “very likely”. Cronbach’s alphas for this variable were 0.78, 0.59, and 0.78 in the French, Greek, and U.S. samples respectively. High scores on this variable reflects a more frequent peer involvement in academic dishonesty than do low scores.

Perceived Penalty for Academic Dishonesty

Two items were written to measure this variable. Participants were asked to rate the severity of penalty for a minor vs. major academic dishonesty behavior. Anchors ranged from 1 “getting a failing grade on an assignment”; 2 “getting a failing grade from a course”; 3 “getting a failing grade and the infraction being recorded in transcript”; 4 “getting a failing grade and being suspended from school for a semester”; to 5 “getting a failing grade and being expelled from school”. High scores on this variable indicate a more severe penalty for being caught cheating than do low scores. Cronbach’s alphas for this variable were 0.60, 0.66, and 0.66 for the French, Greek, and U.S. samples respectively.

Justification of Academic Dishonesty

Four items were created to measure student rationalization of their cheating behavior. They include “academic cheating is in my best interest”; “no one will ever know about it”; “my friends will stand by me in case I get caught cheating”; and “my family will stand by me in case I get caught cheating”. High scores on this variable indicate a low level of moral obligation and high level of justification of academic cheating. Cronbach’s alphas for this variable were 0.74, 0.70, and 0.57 for the U.S., Greek, and French samples respectively.

Academic Dishonesty

We utilized the Academic Dishonesty Inventory (ADI) developed by Newstead, Franklyn-Stokes, and Armstead (1996) and later modified by Koljatic and Silva (2002) to measure academic dishonesty behavior. Sixteen Yes–No items constitute this scale. Participants were asked to indicate whether in the previous two years they had engaged in any of the 16-listed behavior at least once. Scale items were coded as “0” for not yet engaged in the behavior and “1” for having engaged in unethical behavior at least once over the previous two years. Sample items include “paraphrased material from a book without acknowledging the source”; “fabricated reference or a bibliography”; “copied from a neighbor during an examination”. We note that the behaviors are neither mutually exclusive nor represent a comprehensive list of academic dishonesty. High scores on this variable indicate a high frequency of academic dishonesty. Internal consistency estimates for this variable were 0.72, 0.66, and 0.76 for the French, Greek, and U.S. samples, respectively.

Analyses

All data analyses were conducted using SPSS version 23 (IBM corp. 2015). We used Pearson correlation and hierarchical regression analyses to test our hypotheses, following previous research on cross-cultural comparison of academic dishonesty among college students (e.g., Chudzicka-Czupała et al. 2016; Tormo-Carbó et al. 2019). For mean differences in academic dishonesty among countries, we used dummy coded variables to represent this nominal scale of country, a practice recommended by Cohen and Cohen’s (1983).

Results

Table 2 shows the frequency of academic dishonesty behaviors broken down by country. Of the sixteen behaviors, the frequencies of three behaviors were statistically non-significant across three countries based on chi-square statistics. The three behaviors were: “fabricating references or a bibliography” (#4); “having taken an exam for someone else” (#12) and “having had someone else take an exam for you” (#13). Table 3 shows the descriptive statistics and intercorrelations of variables in the study broken down by country.

Table 2 Frequency of Academic Dishonesty broken down by country
Full size table
Table 3 Descriptive statistics and intercorrelations among variables in the study by country (N = 673)
Full size table

As shown in Table 2, the most frequently admitted behaviors were those that involved a lower risk of detection (e.g., allowing own coursework to be copied by others, signing for someone on the attendance sheet or had someone sign for you) relative to those involving a higher risk of detection (e.g., taking an exam for someone else or having someone else take an exam for you). This finding was consistent with prior research using a U.S. student sample (e.g., Hendy 2017). An investigation of Table 3 shows a significant negative relation between age and self-reported academic dishonesty across two out of three countries. Specifically, older students reported as having engaged in significantly less academic dishonesty than did younger students (rs = -0.23 and -0.14) among Greek and U.S. students respectively. This finding was consistent with prior research showing older students being more ethical than younger students (e.g., Tormo-Carbó et al. 2019; Williams et al. 2010). Among French students, the negative relation was not statistically significant (r = -0.04, ns). Sex was not significantly related to self-reported academic dishonesty in any of the samples. Therefore, we included age as a control variable in subsequent regression analyses. Also shown in Table 3 is the negative correlation coefficients between conscientiousness and self-reported academic dishonesty across three countries included in the study. Therefore, controlling for this variable in subsequent analyses was justified.

Prior research reported that academic dishonesty varied as a function of discipline. For example, Harding and colleagues found that that engineering students were more likely to engage in academic cheating than other students even after controlling for opportunities (Harding et al. 2007). Graduate business students were found to cheat more than non-business students based on a large sample of more than five thousand MBA students (McCabe et al. 2006). In this study, there were no significant differences in mean self-reported academic dishonesty as well as peer dishonesty, penalty, and academic dishonesty justification between business and non-business students using a series of independent sample t-tests. Business students were a bit more conscientious than were non-business students in the U.S. sample only (3.72 vs. 3.59, t = 2.19, p = 0.03). Thus, we combined the data for business and non-business students in subsequent analyses.

Hypothesis 1 states that peer involvement in academic dishonesty will be positively related to self-reported academic dishonesty. As shown in Table 3, peer dishonesty was significantly positively related to self-reported academic dishonesty across three countries. Specifically, the correlation coefficients between peer dishonesty and self-reported dishonesty were 0.30, 0.22, and 0.30 in the French, Greek, and U.S. samples respectively with all being statistically significant at p < 0.01 level. These findings provide preliminary support to Hypothesis 1. Hypothesis 2 states that perceived penalty for academic dishonesty will be negatively related to self-reported academic dishonesty. An inspection of Table 3 reveals a significant inverse correlation between penalty and self-reported dishonesty for the U.S. sample only (r = -0.13, p < 0.01). For the French and Greek samples, the correlations were negative, but failed to reach statistical significance.

Hypothesis 3 states that justification of academic dishonesty will be positively related to self-reported academic dishonesty. As shown in Table 3, justification was positively related to self-reported dishonesty across 3 samples of French, Greek, and U.S. students respectively (rs = 0.43, 0.38, and 0.25, p < 0.001). Thus, Hypothesis 3 received preliminary support.

We further conducted a hierarchical regression analysis to examine the cultural differences in academic dishonesty. Two dummy variables were created representing the French – U.S. cultural difference and Greek – U.S. difference with the U.S. as a reference group. We regressed self-reported academic dishonesty onto these dummy variables in the first step. In the second step, we entered age as a control variable. In the third step, conscientiousness was added as another control variable. In the fourth step, the social learning theory variables, i.e., peer dishonesty and penalty, were entered. In the fifth and last step, cheating justification was added to the equation. Table 4 presents the hierarchical regression results.

Table 4 Hierarchical regression analyses predicting self-reported academic dishonesty
Full size table

As shown in Table 4, the dummy variable associated with French – U.S. culture was a significant predictor of self-reported academic dishonesty (β = 0.32, p < 0.01) whereas the Greek – U.S. culture was non-significant (β = -0.04, p = 0.26). This means that on average, French business students reportedly engaged in academic dishonesty more frequently than did U.S. and Greek students. A one-way ANOVA with Bonferroni post-hoc test confirmed this finding. The mean difference in self-reported academic dishonesty between French and U.S. students was significant (4.82 vs. 2.36, 95% confidence interval of 1.9 to 2.96). The same finding applied to French and Greek students in academic dishonesty (4.82 vs. 2.43, 95% confidence interval of 1.7 to 3.02) suggesting that French students engaged in more academic dishonesty than did Greek students. Further, as shown in Table 4, the two country dummy variables explained 14.4% of variation in self-reported academic dishonesty, which is almost 50% of the 30.7% adjusted variance explained by all predictors in the regression equation.

Next, we found that age was marginally significant as a predictor of academic dishonesty after controlling for the country variables. Specifically, the standardized regression coefficient of age was slightly negative and marginally significant (β = -0.06, p = 0.07). This suggests that older students were slightly less likely to report engaging in academic dishonesty regardless of the three cultures included in this study. This finding was consistent with prior research using U.S. and non-U.S. student samples (e.g., Gentina et al. 2017; Whitley 1998; Williams et al. 2010). In terms of relative importance as a predictor, age explained 1.3% of unique variation in academic dishonesty beyond the country effect.

Next, conscientiousness was a significant negative predictor after controlling for age and country (β = -0.18, p < 0.001). In addition, conscientiousness explained an additional 5.8% of variation in self-reported dishonesty beyond country and age. Next, peer dishonesty was a significant positive predictor of self-reported academic dishonesty (β = 0.21, p < 0.001) after controlling for country, age, and conscientiousness. This provides support for Hypothesis 1.

Contrary to our expectation, the standardized regression coefficient associated with penalty was not statistically significant, albeit in the right direction (β = -0.04, p = 0.18). A series of hierarchical regression analysis by country showed that penalty was a negative and significant predictor of academic dishonesty in the U.S. sample only (β = -0.12, t = -2.48, p < 0.05). For the French students’ sample, the regression coefficient was negative, but not significant (β = -0.004, t = -0.06, p = 0.95). For the Greek students, the standardized regression coefficient was positive, but not significant (β = 0.17, t = 1.83, p = 0.07). Altogether, peer dishonesty and penalty explained 5.6% of unique variance in self-reported academic dishonesty in the combined sample of three countries. Thus, Hypothesis 2 received mixed support.

Last, as shown in Table 4, justification of cheating was a positive and significant predictor of academic dishonesty after controlling for countries, age, conscientiousness, and two social learning theory variables (β = 0.22, p < 0.001). Further, cheating justification explained 4.4% unique variance in self-reported academic dishonesty beyond all other predictors combined. Therefore, hypothesis 3 received full support.

Discussion

In this study, we examined academic dishonesty among college  students in France, Greece, and the U.S. under the lens of Bandura’s (1986) social learning theory. We found support for the theory in general, which is consistent with the few prior studies using focused group interviews (Burnet et al. 2016). Specifically, peer involvement in academic cheating was a significant positive predictor of self-reported academic cheating among university students across three cultures of the U.S., France, and Greece. This finding was consistent with prior research including U.S. and Eastern European college students (e.g., Grimes 2004; McCabe et al. 2002; O’Fallon & Butterfield 2012) as well as studies comparing U.S. and French high school students (Gentina et al. 2017). Our study was the first in reporting that peer involvement in cheating was predictive of academic cheating among French and Greek university students.

We found cultural differences accounted for nearly half of the variance in academic dishonesty. Specifically, the French-U.S. difference in academic dishonesty supports our expectations that college students in the French culture of high-power distance, long-term orientation, high uncertainty avoidance, collectivism, and femininity, engaged in academic dishonesty more so than those from the U.S. culture of lower power distance, shorter term orientation with less uncertainty avoidance, individualism, and masculinity. However, our expectation concerning Greek – U.S. difference received mixed support. Specifically, although Greek students reported a slightly higher level of academic dishonesty than did U.S. students (2.43 vs. 2.36); the difference failed to reach statistical significance. This non-significant finding might be explained by the differences in Greek student perception concerning plagiarism. According to one qualitative study, Greek students perceived that copying words or sentences without acknowledging the source was acceptable if it was less than a paragraph (Hayes & Introna 2005). Additionally, another quantitative study showed that Greek business students viewed copying answers or using hidden notes during exams and plagiarism as more acceptable than reporting a classmate for cheating on an exam (Karassavidou & Glaveli 2007). Based on the above findings in previous literature, Greek students may have underreported the extent of plagiarism in this study (item #1) as well as underreported the extent and frequency of academic dishonesty behaviors such as copying from neighbors during exams (item #8) and bringing in unauthorized materials to the exam room (item #10).

In terms of prevalence of cheating, as shown in Table 2, we had complete data for frequency of academic dishonesty from 404 students in the U.S. sample, 146 in the Greek sample, and 192 in the French sample. Of these, we found that 69.3%% of U.S. students; 75.7% of Greek students and 96.4% of French students reported having engaged in academic cheating at least once over the previous two years. The result for U.S. students was consistent with Whitley’s (1998) meta-analytic estimate showing 70.4% of U.S. students reportedly having cheated at least once across forty-six studies published during the period of 1971 and 1996 (Whitley 1998) but higher than 50.23% as reported in Grimes’ (2004).

According to a large-scale review of French student cheating in college (Guibert & Michaud 2009), averaging six cheating behaviors including (1) asking someone to do the work for you, (2) duplicating another student’s work without mentioning it, (3) requesting additional time to turn in your work and faked an excuse, (4) copying a test or part of it and presenting it as own work, (5) referencing a book or articles that you did not read, and (6) using a synthesis or reading a summary rather than the original book, based on a sample of 1,485 students, 30.98% reported having performed the above behavior at least once (Guibert & Michaut 2009, Table 2: 47). Our finding was considerably higher than what was reported in the above study probably because almost a decade has passed since Guibert and Michaut’s (2009) study was published. There was some evidence that academic dishonesty was on the rise given the explosion of the Internet, which makes it easier to cheat (Roberts & Wasieleski 2012). The nature of the French student sample may have contributed to the high level of academic dishonesty because it consists entirely of business students given the fact that prior research showed business students reported a higher level of academic cheating than did non-business students (McCabe et al. 2006). It is also possible that French students were more honest in responding to the survey in this study relative to Greek and U.S. students because they did not receive any extra credit in exchange for participation in this study as did U.S. students. Therefore, French student responses may not have been influenced by social desirability to the same degree as did those of U.S. students.

The reason we did not find any significant differences in three behaviors “fabricating references or a bibliography” (#4); “having taken an exam for someone else” (#12) and “having had someone else take an exam for you” (#13) across three countries can be explained as follows. First, the above behaviors are considered “major” academic cheating behavior. Prior research showed that “major” cheating behaviors were less frequent and less prevalent than “minor” cheating behavior (Scrimphire et al. 2017). Second, most students were found to game the system by cheating just a little, meaning they would feel okay to engage in “minor” cheating such as copying from a neighbors the correct answers to a few items on an exam. However, having someone else take an exam for them would be considered “major” cheating and students would not be comfortable doing so (Ariely 2012; Scrimpshire et al. 2017).

An examination of Table 2 shows that the frequency distribution of the sixteen academic dishonesty behavior examined in this study varies between countries. For example, no French and Greek students in our study reported that they had taken an exam for someone else. Only one Greek student reported having someone else take an exam for them during the previous two years. The proportion of U.S. students reported having taken an exam for someone else in our study or having someone else take an exam for them, albeit very low and not significantly different from the French and Greek samples, suggests that the intensity of academic dishonesty might be higher among U.S. students relative to French and Greek students even though the frequency of academic dishonesty was the lowest.

According to the country report on the effectiveness of academic integrity policy, the higher education system in France relies heavily on rote learning, including learning at the master’s level, which discourages innovative and original ideas and encourage academic cheating such as plagiarism (Glendinning et al. 2013). This was confirmed by the high proportion of French students who admitted to having paraphrased textbook materials without acknowledging the source (52.1%) relative to the same in the U.S. sample (38.3%) and 29.3% in the Greek sample (χ2 = 31.85(2), p < 0.001).

The reinforcement and motivational process, a subprocess within observational learning process received support for the U.S. sample only. This finding was consistent with previous research showing that the harsh penalties associated with cheating were a reason cited by students not to cheat (e.g., Miller et al. 2011). One reason that might explain why penalty was a deterrent to scholastic cheaters in the U.S. is the well-established policies on academic integrity at the U.S. university included in this research. For example, all students must sign a code of ethics during orientation prior to starting their college education at this university. During their academic study, ranging from four to six years, they are reminded of the consequences and penalty for violating the code of ethics. Students had to take a required business ethics course in which codes of ethics were discussed. Empirically, the presence of a code of ethics was found to be inversely associated with academic cheating (McCabe et al. 2002). The reminder coupled with the presence of an ethics code were found to be more effective than just having a code of ethics present  in at least one prior study (Bing et al. 2012).

The reason we did not find penalty to be a valid predictor of academic cheating among Greek and French students is probably due to the following. First, it was shown that Greek students were more relaxed when it comes to academic cheating and their perceived penalty was more lenient than the same from the faculty point of view as shown in one study of the University of Athens Dental School student (Koletsi-Kounari et al. 2011). In addition, there were no accepted penalties for academic dishonesty such as plagiarism evidenced by a low penalty score given to plagiarism by both faculty and students (Koletsi-Kounari et al. 2011). Students who were caught plagiarizing during an exam or a class may be able to take that class or exam again multiple times (Kokkinaki 2013), therefore, there was no deterrent to offenders of academic dishonesty. It was also uncommon for French students to be punished for plagiarism even though 46% of students and 50% of teachers reporting that they might have plagiarized accidentally or deliberately according to the country report under IPPHEA for France (Glendinning et al. 2013). Second, the honor code was found to be lacking in many Greek universities (Koletsi-Kounari et al. 2011). Third, both French and Greek students were not taught about what constituted plagiarism or academic dishonesty on campus as well as informed about consequences related to plagiarism as stated in the country report filed with the European Union under IPPHEA (Glendinning et al. 2013; Kokkinaki 2013). In addition, many Greek students felt they had to cheat because they had a high level of anomia, or a negative world view (Karassavidou & Glaveli 2007). Greek students had a low level of trust in the integrity of the education system based upon a previous qualitative study (Hayes & Introna 2005). The above discussion may explain why French and Greek students in our study were ambivalent about the penalty for academic cheating, which explained the non-significant association with self-reported academic dishonesty as shown in this study.

We also found support for cheating justification, which is consistent with prior research using U.S. students as well as Eastern and Latin European students (e.g., Chudzicka-Czupała et al. 2016; Hendy & Montargot 2019). Further, this justification to commit academic dishonesty or this lack of a moral obligation to uphold honesty and integrity, explained unique variance in academic dishonesty beyond the two constructs of peer involvement in dishonesty and penalty in Bandura’s (1986) social learning theory. As a component of moral disengagement construct (Detert et al. 2008), moral justification was found to mediate the relationship of moral conviction and unethical decision-making behavior (Roberts et al. 2018). Because peers contribute to the shaping of moral conviction, it is a question worthy of future research endeavor to examine whether cheating justification mediates the relationship of peer involvement in academic dishonesty and academic dishonesty.

Contributions and future research

Our study was among the first to document the prevalence of academic dishonesty among French and Greek college students relative to U.S. students under the lens of Bandura’s (1986) social learning theory. Given the increased internationalization and globalization of higher education, our findings might be cause for concern to college administrators as they balance the need to diversity their campuses with the need to reduce unethical behavior among students. Because peer dishonesty behavior was found to be influential above and beyond the country effect, the importance of having peers or buddies served as role models cannot be overemphasized.

Our research provides sorely needed data of French and Greek college students’ frequency of academic dishonesty behavior as prior research in those two countries either investigated student attitude toward academic dishonesty (e.g., Karassavidou & Glaveli 2006, 2007) or academic dishonesty behavior of high school students (Guibert & Michaut 2009). In addition, our measure of academic dishonesty covers a wider range of behavior (16) than what was typically utilized in previous research including 10 behaviors (McCabe & Treviño 1993). This suggests that academic dishonesty may include a wider range of behavior than traditionally thought. Alternatively, it might be worthy of future research to move the field forward to identify which of the 16 behaviors is truly unethical universally and which behavior may be acceptable relativistically. Mudrack and Mason (2020) provided a thought-provoking review and analysis of relativism and questioned the well-established finding of a negative relationship between relativism and unethical behavior.

Our research provided data on French business students. Future researchers may want to collect the same data on non-business students in France to facilitate comparable analysis. In addition, future research should examine other situational factors that explains the motivation of academic dishonesty. For example, student perception of faculty caring, and approachability was found to be a predictor of academic cheating among Chinese students such that the more students found that faculty cared about their studies, the less likely the students were engaged in academic cheating (Tsui & Ngo 2016). Because faculty attitude is within faculty control, early intervention can be easily developed to prevent students from engaging in academic dishonesty.

Additionally, the prevalence of academic dishonesty as shown in this study should be considered an underestimation of the true level of academic dishonesty due to socially desirable responding. Even though the responses were collected anonymously, prior research showed a low to moderate level of social desirability associated with self-reported academic cheating using U.S. student samples (e.g. Lucas & Friedrich 2005) as well as European samples (e.g., Trost 2009).

Finally, an area worthy of future research is to examine student collaborative academic dishonesty, rather than individual academic dishonesty as examined in this study. Emerging studies in this area (Zhang & Yin 2020) showed that peer involvement had a positive relationship with both collaborative student cheating attitude and behavior among a sample of Chinese college students and that this relationship was moderated by Hofstede’s individualism such that students scoring higher on individualism reportedly engaged more in collaborative cheating.

Practical Implications

As demonstrated in this study, academic dishonesty was a result of social learning or observational learning from peers. This suggests that a group norm of honesty and integrity should be cultivated while students attend universities. If students perceive that their peers do not engage in cheating, they will be less likely to perform this behavior based on this study’s findings. This is particularly important given the recent finding of U.S. healthcare workers who were found to be hesitant to report their coworkers whom they considered as friends for misconduct due to their ethics of care (Hess et al. 2019). Our finding reiterates the need voiced by previous researchers in nurturing a culture in which peers value honesty above efficiency and pragmatism (Gentina et al. 2017; McCabe et al. 2002).

Faculty should include in their lectures information about what constitutes academic dishonesty to increase student understanding of this concept. For example, previous research showed that students viewed submitting the same paper to two different courses without instructor approval as “fair game” and did not see that as “cheating” (McKay et al. 2019; Miller et al. 2011). Also, class discussion about consequences of academic dishonesty should reduce cheating as shown in this study in the U.S. sample. We suggest that the above class discussion should be integrated across curricula, rather than delivered in one course such as business ethics because previous research showed a negligible gain, if any, in improving student ethical decision making after being exposed to one course of business ethics (e.g., Nguyen et al. 2008; Tormo-Carbó et al. 2019). Based upon our finding of French students reportedly engaging in academic cheating behavior at a significantly higher frequency compared to Greek and U.S. students, faculty are cautioned of these cultural differences if they are to respond to incidents of academic dishonesty among international students.

Limitations

Our study is not without limitations. First, the cross-sectional nature of this study prevents us from fully testing Bandura’s (1986) social learning theory. A controlled experiment with randomization would be needed to ensure a comprehensive testing of the theory with causation. Second, the convenience sampling nature of the study may limit its internal validity while awaiting further research using a representative sample of each respective country. Third, the sample sizes for the Greek and French samples were considerably smaller than that of the U.S. sample, giving us less confidence in generalizing the results to the respective countries. However, the non-significant t-test results comparing students with complete vs. incomplete data within those samples suggest that our results were not biased due to their small sample size or data missingness. Fourth, some of the scales used in this study had lower reliability estimates than traditionally considered acceptable based on the 0.7 cutoff proposed by Nunnally (1978: 245). The perceived penalty scale had low internal consistency estimate, which resulted in low statistical power to detect a relationship should one exists. Lastly, we used the country dummy coded variable as the proxy variable to represent national culture based on Hofstede’s framework without measuring the respective cultural dimensions. This prevented us from testing the potential moderation effect of culture on academic dishonesty.

Conclusion

Academic dishonesty among students can damage both the reputation of the colleges as well as the larger society because of reduced student learning. In addition, students who cheated in college would be likely to cheat in the workplace (e.g., Nonis & Swift 2001; Sims 1993). Our study shows that academic dishonesty behavior can be reduced through social learning, something that is within the control of higher education administrators and faculty relative to more stable personality traits of students (e.g., conscientiousness). We hope that this study paves the way for future research to examine additional cross-cultural differences in academic dishonesty using untapped theories to benefit colleges around the world as we continue our research for policy making and practices to reduce academic dishonesty.