(Mulholland, 2020). In contrast, the surveys used by McCabe et al. (2012) and Curtis and Tremayne (2021) mostly require students to recognize a behavior they may have engaged in during an assessment, rather than label that behavior as plagiarism or cheating. People may be more likely to admit to socially undesirable behaviors if they do not label them as deviant (MacDonald and Nail, 2005). In addition, the survey used by Curtis and Tremayne (2021) asks students whether they have ever engaged in behaviors that are arguably less serious, and thus likely more common, than the behaviors examined in the other two time‐lag studies. In contrast, McCabe et al.'s (2012) surveys asked students to report behavior of only the past year. Therefore, it makes sense that Curtis and Tremayne (2021) report the highest prevalence of plagiarism/cheating. Furthermore, the survey used by McCabe et al. (2012) asks about nine different behaviors as compared with only three or four in the surveys used by Stiles et al. (2018). Having more behaviors, any of which would constitute a single instance of plagiarism cheating, means that McCabe et al. (2012) would be more likely to get affirmative responses than Stiles et al. (2018).
Returning to the prevalence trends, an important question to ask is, why might the trend in academic integrity breaches be downward since at least 1994? The authors of the studies themselves offer some differing explanations. Controversially, Stiles et al. (2018) predicted that cheating might be higher among the 2014 sample because the sample is composed of “millennial” students who evince attitudes of “entitlement”. Although Stiles et al. (2018) found that students’ feelings of entitlement correlated positively with their engagement in cheating and plagiarism, this does not explain why cheating was, in fact, lower in the 2014 study, especially as there was an additional item in the 2014 survey that may have led to more cheating being reported than in their earlier surveys. McCabe et al. (2012) suggest that technological changes, including internet‐based plagiarism not covered in their survey, may account for some of the reduced rates of plagiarism and cheating. In addition, McCabe (2016) reports a relationship between institutions having honor codes and lower rates of plagiarism and cheating. Such reports of the impact of honor codes date back to McCabe and colleagues’ earlier work, and therefore, some institutions may have implemented honor codes over time in response to previous research. Curtis and Tremayne (2021) attribute the reduction in plagiarism and cheating observed in their study to specific academic integrity education modules taught at Western Sydney University, the increased use of text‐matching software, and improved assessment practices.
As the contention that millennials are more entitled, and therefore cheat more, is not borne out in Stiles et al.'s (2018) study, two likely explanations remain for why plagiarism and cheating appear to have trended downward. First, it is possible that the forms of plagiarism and cheating examined in measures created between 15 and 60 years ago are less common, and have been replaced by newer forms of cheating and plagiarism. Second, it is possible that as staff and institutional awareness of academic integrity issues has grown over time, various interventions such as citation training, text‐matching software, and honor codes may have had an impact in reducing the prevalence of plagiarism and cheating. The next two sections of this chapter will assess these possibilities as I review data on the prevalence of a newer form of academic misconduct, commercial contract cheating, and then examine the impact of interventions targeted at reducing academic misconduct.
TRENDS IN THE PREVALENCE OF SELF‐REPORTED COMMERCIAL CONTRACT CHEATING
As McCabe et al. (2012) note, internet access became more widespread over the past 30 years. In 2007/2008, they added additional questions about internet use to their surveys and found that nearly 95 percent of students who had engaged in cut‐and‐paste plagiarism had done so from internet sources. It has been argued that just as the internet may facilitate copying and pasting, the internet also has a role in deterring and detecting this behavior with the advent of text‐matching software (Curtis and Vardanega, 2016; Park, 2003). McCabe et al. (2012) argued that “If students know that faculty will be checking … such cut‐and‐paste plagiarism may decline” (p. 71). However, a form of cheating that is potentially undetectable by text‐matching software is outsourced, ghost‐written work.
Various studies throughout the past 30 years have, at least in part, examined students’ use of outsourcing behaviors in assessment, such as getting help from family and friends, accessing essay mills, or paying other people to complete homework, term papers, or exams for them. In 2006, Clarke and Lancaster coined the term contract cheating to describe the situation where students enter into a contractual agreement with another party to complete an assignment for the student (for more on Contract Cheating, see Lancaster's chapter in this book). The definition of what constitutes contract cheating has since varied, with Bretag et al. (2019) classifying all outsourcing, whether paid or unpaid, as contract cheating, while specifically paid contract cheating has been more recently referred to as commercial contract cheating (e.g. Lancaster, 2020)
Charting trends in contract cheating in the past 30 years is complicated by the fact that only one of the three studies reviewed in the previous section included a measure of commercial contract cheating. The studies reported by Curtis and Tremayne (2021) found percentages of commercial contract cheating of 3.1 percent (2004), 3.4 percent (2009), 2.8 percent (2014), and 2.8–3.5 percent (2019). These percentages did not differ significantly over time or between any two pairs of years.
Newton (2018), however, conducted a review and meta‐analysis of the self‐reported prevalence of commercial contract cheating, principally consisting of studies that surveyed students at one point in time. Newton examined 71 studies conducted between 1978 and 2016 and concluded that just over 3.5 percent of students engaged in commercial contract cheating. An identical figure of 3.5 percent was found in an earlier small‐scale meta‐analysis by Curtis and Clare (2017). Importantly, Newton (2018) reported the prevalence of commercial contract cheating engagement increasing, as indicated by a positive correlation between engagement in cheating and the year in which studies were conducted. However, as discussed previously, estimating trends from studies using different methods, measures, and samples should be done with caution because methodological differences may influence the studies’ results. In addition, several large studies have been published since Newton's (2018) review, which provide additional information on the rates of, and possible trends in, commercial contract cheating.
For this chapter, I have updated and re‐analyzed Newton's (2018) data to consider more recent studies. Specifically, I obtained Newton's (2018) table of the studies he analyzed, and their details, and added to these the details of the following studies: Foltýnek and Králíková (2018), Bretag et al. (2019), Rundle et al. (2019), Curtis and Tremayne (2021), and Awdry (2020). These five additional studies surveyed over 25,000 students, and increase the total sample reported by Newton (2018) by over 45 percent. After adding these studies’ data, I then: 1. re‐ran the analysis of commercial contract cheating prevalence and trends; 2. analyzed the data including only studies 1990–2020; and 3. analyzed the 1990–2020 studies that came from majority English‐speaking countries. To explain the third of these analyses, all studies analyzed by Newton (2018) from 1990–2008 were from majority English‐speaking countries; however, half of the studies since 2009 were not. In addition, some of the highest percentages of commercial contract cheating were reported in recent studies from non‐English‐speaking countries, suggesting that these outlier results may be attributable to the peculiarities of the sample, methods, or academic culture. Limiting analyses to the most frequent language group of the nations among the studies allows for closer to like‐with‐like comparisons.
Table 3 shows the analysis of the prevalence of commercial contract cheating, including: 1. Newton's (2018) findings; 2. revised findings including the subsequent surveys through to 2020; 3. studies limited to 1990–2020; and 4. studies limited to English‐speaking countries from 1990–2020. Importantly, in considering trends, the most relevant statistic is the correlation between the year in which the surveys were collected and the percentage of commercial contract cheating reported in the surveys. A positive correlation indicates that as the year of data collection becomes more recent the prevalence of contract cheating becomes higher. Thus, the significant positive correlation, as found by Newton (2018), suggests a trend of increasing student engagement in commercial contract cheating over time. However,