Why they do and why they don’t: a combined criminological approach to understanding assignment outsourcing in higher education

Abstract

Assignment outsourcing is an intractable challenge for higher education. While various academic and legislative approaches have sought to explain/respond to this problem, recent media, community, and government concerns suggest students are increasingly outsourcing assessments. This paper reports on the qualitative findings of an international survey (n = 7000 +) of students’ perceptions and experiences of outsourcing, to test the utility of a multi-theoretical criminological explanation for this behaviour. The results reveal a complex learning environment where students’ knowledge is shaped by an assemblage of social, cultural, and institutional influences; a learning environment where engagement in, or avoidance of, cheating are dynamic. Interdependent behaviours are learnt, shared or observed, and reinforced through peer networks. Interrogation of students’ qualitative comments emphasises that intervention is possible, albeit equally complex.

Introduction

Assignment outsourcing involves students getting others to complete their assignments using ‘…essay mills, bespoke assignment services, essay bidding services, peer-to-peer file sharing sites (peer-sharing sites), and obtaining work from other students, colleagues, friends and family members’ (Awdry et al. 2020:1). While much has been written on this type of cheating (often referred to as contract cheating), in higher education (HE) in the last two decades (Bretag et al. 2019; Dawson & Sutherland-Smith 2018; Ellis et al. 2018; Lancaster & Clarke 2016), relatively little has come from the discipline of criminology, though this base of knowledge is growing. Detailed theoretical analyses and some empirical studies have emerged, employing methods and/or epistemologies drawn from criminal justice or criminology perspectives looking at university cheating, contract cheating or assignment outsourcing more broadly (Awdry et al. 2022; Clare & Rundle 2022; Draper & Newton 2017; Hodgkinson et al. 2016; Nagy & Groves 2021; Rundle et al. 2019).

Criminological research has demonstrated that its theories can be used to account for, and understand, cheating behaviours with scholars focussing on its determinants, reduction in opportunities for, and increase in penalties in response to such ‘offending’ (Baird & Clare 2017; Clare et al. 2017; Nagy & Groves 2021; Walker & Holtfreter 2015). Despite this, subsequent legislation has had little deterrent value for students who continue to engage in outsourcing behaviours even when it is, or they believe the behaviour is, criminalised (Amigud & Dawson 2020; Awdry et al. 2022). Rather, the incidence of this serious form of academic misconduct continues to increase (Newton 2018). Scholars have consequently articulated a need or responsibility for the academy to employ interdisciplinary, comprehensive and more pragmatic approaches that include both educational and criminal justice responses, guided by the techniques and principles of crime prevention (Clare 2022; Hodgkinson et al. 2016), to examine and offer potential ‘solutions’.

While discussion of the harms associated with outsourcing and subsequent punishment is important, there is greater pragmatism in using the tools and philosophy of criminology to prevent such cheating, rather than simply criminalise it. The challenge becomes how – given the relatively limited theoretical interrogation of cheating behaviour – we can identify and utilise criminological theory to explain outsourcing and develop situational crime prevention techniques. There is a complex milieu of reasons for why students cheat, which comprises both internal and external determinants. This means that theoretical explanations for this behaviour are similarly diverse such that no one theory can sufficiently explain cheating. Indeed, much of the extant literature has explored individual or dual perspectives; this research seeks to push beyond this to provide a more comprehensive interpretation.

This paper critically evaluates the application of a multifaceted theoretical approach, innovatively comprising five complementary criminological theories, drawn from a detailed empirical investigation. It outlines how we might better understand the reasons for assignment outsourcing in HE. This represents an important and evidence-based approach, which draws from principles of criminology, recognising that the threat of and opportunities for outsourcing emerge in social, familial, and institutional contexts beyond assessment and curriculum design. We examine qualitative data from an international survey of more than 7000 students that sought to examine how and why students are outsourcing their assignments at university. The Global Essay Mills Survey (GEMS) was disseminated in three continents, in multiple languages (n = 22), and was designed to capture a picture of the problem of cheating across the international tertiary education sector. We use the term ‘assignment outsourcing’ to capture both commercial and non-commercial cheating and evaluate both informal/formal methods through which students may outsource (see Awdry 2021).

Background

There is a long history of research examining students’ motivations for cheating (Brimble 2016; McCabe 2005), its prevalence in diverse global contexts (Awdry 2021; Curtis et al. 2021; McCabe et al. 2006; Newton 2018), various demographic and predictor variables (Bretag et al. 2019; Ives & Giukin 2020; McCabe & Trevino 1997), and potential legal consequences, including victimisation (e.g., blackmail) (Steel 2017; Yorke et al. 2020). Comparatively, fewer studies have examined criminological theory to explain cheating and/or offer applied strategies to reduce/prevent cheating behaviour in HE settings.

Most scholars who have examined this problem have focused on opportunities for ‘offending’ and students’ decision-making, investigating Routine Activities Theory (RAT) (Cohen & Felson 1979) and Rational Choice Theory (RCT) (Clarke & Cornish 1985, 1986), linking patterns of cheating behaviour with those typically associated with traditional forms of crime (Baird & Clare 2017; Clare et al. 2017). The theory states that offending occurs when a motivated offender, a suitable target and lack of guardianship coalesce in a given space/time (RAT) and that offenders make context-specific rational choices and decisions, weighing up the risks and rewards (RCT). However, evidence suggests some students will ‘never enter the essay market regardless of market or institutional conditions’ (Rigby et al. 2015:25), no matter how beneficial the context. While RAT and RCT are useful theoretical tools explored across much criminological scholarship, this paper adds to the growing call (see Walker & Holtfreter 2015) for further complementary or intermediary theories to explain students’ motivations, both for why they cheat and why they do not, and the attendant processes that shape such decision-making.

Consistent across much of the wider criminological literature is recognition of the need to examine individuals’ perceived or actual levels of self-control, social learning and/or influences, and ways in which they rationalise or neutralise deviance in making their decisions (Akers 1990; Gottfredson & Hirschi 1990; Sykes & Matza 1957). Various scholars have argued that students make rational decisions based upon perceptions and experiences relating to risks and rewards, and so act accordingly. Asking individuals then, about why they do and why they do not cheat is an important step in extending criminological theory beyond the well-trodden ground of RAT and RCT. Students’ motivations and learning of what various risks and rewards are or indeed how to behave in certain situations is crucial to understand how cheating behaviour is learned and from whom, to extend existing knowledge and practice, including prevention efforts.

A complementary approach

We argue then, that evaluation of the role of learning in understanding cheating behaviours requires careful unpacking of the intersections between several key, complementary theoretical perspectives that lay beyond or, indeed, between RAT and RCT. Building upon recent scholarship (see Walker & Holtfreter 2015 for detailed discussion of the theoretical framing), this paper specifically examines Differential Association (Sutherland 1947), Differential Reinforcement (Jeffery 1965), and Social Learning (Akers 1990; Bandura 1977) theories. Furthermore, given the criticism of both RAT, RCT, and strain perspectives that they do not effectively account for why offenders do not offend/cheat, Self-control theory (Gottfredson & Hirschi 1990) and Sykes & Matza's (1957) Techniques of Neutralisation are offered to bridge this gap. Similar to recent applications (see Walker & Holtfreter 2015) we argue that bringing these five theories together within the one approach provides a more balanced model, with each addressing one or more criticism of another theory. Our approach considers who individuals associate with and their respective influence, how behaviours are reinforced, how these interactions occur within a broader social milieu, how individuals reconcile them with their internal values, and what strategies might be needed to neutralise deviance. Consequently, understanding how cheating behaviour is learned and then implemented (or not) can indicate opportunities where targeted intervention may be applied to prevent outsourcing.

Differential Association (DA)

DA suggests that behaviour is learnt from who an individual associates with and that their subsequent behaviour – whether deviant or conformist – will be shaped by the nature and intensity of these networks, where influence is greatest from intimate/close social bonds, such as friends or peer groups (Sutherland 1947). Notably, DA describes the processes by which individuals learn from their networks both the skills and techniques to offend, as well as how to act and in which contexts (Cressey 1960). In the educational setting, there is considerable empirical evidence accounting for why those with(in) deviant networks are more likely to cheat and those with conformist influences do not (Awdry & Ives 2021; McGloin & Thomas 2016; Megehee & Spake 2008; O’Rourke et al. 2010). For example, knowledge or perception of others’ cheating behaviours is strongly associated with student cheating, where in times of stress (e.g., multiple assignments due, illness, etc.) outsourcing can be understood as a legitimate alternative enabling achievement of key goals (e.g., passing assignments, units, courses). This may be exacerbated if culturally normative, evidenced by the commercialisation of the essay assistance industry and number of ‘ads’ students are exposed to (Amigud 2020; Lancaster 2020; Rowland et al. 2018). This is simultaneously a feature and criticism of DA (Burgess & Akers 1966; Cressey 1960); anyone can cheat, but not everyone does and/or not all the time, suggesting others influences on students’ decision-making, beyond simply associating with ‘bad influences’.

Differential Reinforcement (DR)

Building on this foundation, DR usefully focusses more on the external incentives and anticipated rewards/punishments that influence behavioural choices (Jeffery 1965). DR concerns the balance of direct and indirect benefits and costs, past, current, and likely future outcomes, both formal and informal sanctions, and the effects of positive and negative reinforcement (Akers 1990; Burgess & Akers 1966). Drawing from wider criminological and behavioural scholarship, DR extends understanding of social learning processes beyond mere association, emphasising that behaviours are reinforced through observation and interaction within social networks that create opportunities for learning how to behave in certain situations and for specific purposes (Akers 1990; Jeffery 1965). Importantly, these interactions are then shaped by the learning environment, identifying wider structural influences on student behaviour, challenging claims that only deviant or lazy students cheat. Academic cheating is in fact linked to high rates of recidivism (Clare et al. 2017), which some have attributed to its positive reinforcement, learnt either from direct personal experience or seeing others successfully cheat (O’Rourke et al. 2010). Equally, it can be negatively reinforced, whereby students who have cheated are caught and receive institutional penalties (e.g., zero for task/unit), are condemned or shamed by others, or otherwise experience undesirable outcomes (e.g., exposure to blackmail by third-parties). Both internal (e.g., social networks) and external (e.g., tertiary institutions) influences can positively or negatively reinforce students’ experiences of cheating, emphasising the need to examine cheating situationally to identify where/how to intervene.

Social Learning Theory (SLT)

SLT is a socio-psychological approach that explains forms and contexts of learning behaviour. Initially applied in educational settings, Bandura, (1977), posited that behaviour can be explained by how individuals learn through observation and modelling within both personal and environmental interactions. Specifically, individuals observe others’ behaviour and learn to model/imitate the relevant skills, capabilities, and responses, including rewards and punishments in given contexts, which is reinforced through external and self-reinforcement (Bandura 1977). Significantly, SLT recognises that learning occurs in social contexts, where the learner is not passive but actively interacts with others and their environment (Bandura 1977). SLT then, enables greater investigation of the specific processes by which students learn to cheat, in terms of when, how, and why. Aligning with/in existing models (e.g., RCT), this integrated approach may expose students’ motivations for their decision-making, noting that individuals do not model everything they learn, but are instead selective and enact behaviour that is most rewarding (Akers 1990; Bandura 1977), which may partly explain the variability observed in rates of cheating. Burgess & Akers, (1966) combined the principles of SLT and operant conditioning to build upon Sutherland's (1947) Differential Association theory to create a more comprehensive theory of deviance. Specifically, their interpretation pushed beyond the limitations of DA to argue that effective learning involves developing the tools and knowing the appropriate settings and ways in which to use them (Akers 1990; Burgess & Akers 1966). SLT offers a complementary perspective to students’ outsourcing, revealing how behaviour is learnt and reinforced through social networks within, as well as outside of, the tertiary education context.

While the aforementioned theories may explain how people learn behavioural norms, reinforcements and methods for behaviours (honest or otherwise), they have been criticised for not adequately explaining why people may not engage in deviant behaviour (Akers 1990). To account for this, the paper incorporates analysis of Self-control Theory (SCT) and Techniques of Neutralisation (ToN).

Self-control theory

SCT posits that lack of/low self-control is central to the commission of all crime, where deviance occurs because a person has not learned social values/norms or otherwise lacks the control to avoid engaging in crime (Gottfredson & Hirschi 1990). SCT examines individuals’ core values, developed primarily through childhood, and shaped by key social institutions (e.g., family, school), and suggests people try to avoid deviance because of fear of consequences, shame on self/others, or acknowledgment that it is morally wrong (Gottfredson & Hirschi 1990). In contrast to DA, DR, and SLT, which largely explain why individuals commit crime, SCT examines why they do not, highlighting the role and influence of formal controls (e.g., laws), social bonds (e.g., jobs, or friendships), and beliefs (e.g., social norms) in constraining deviant behaviour (Gottfredson & Hirschi 1990). This moral code helps to establish and implement self-control practices that guide individuals towards normative behaviour. Several cheating studies have linked SCT to engagement in academic misconduct (Awdry & Sarre 2013; Muraven et al. 2006; Stone et al. 2010), and assignment outsourcing (Rundle et al. 2019). Rundle and colleagues (2019:11) reported that students primarily do not cheat because of their “sense of morals, perception of norms, and their motivation to learn” but that others may be tempted as self-control is a resource that can be depleted by demands such as stress. However, SCT has been criticised for focussing too heavily on opportunity, with critics arguing that it assumes that all individuals with low self-control are opportunistic and would fall into offending if presented with the ‘right conditions’ (Downes & Rock 2007). This position is not empirically supported, however, given that most students do not cheat (Newton 2018), and even those that do, do not cheat consistently, including when reporting low self-control or periods of significant stress (Rundle et al. 2019), which suggests there are other perspectives needed to explain students’ outsourcing behaviour.

Techniques of neutralisation

Developed initially by Sykes and Matza (1957), ToN argues that deviant behaviours are learned and justified through social interactions. These interactions provide individuals with vocabularies of ‘motive’ or ‘adjustment’ that enable them to commit deviant acts (e.g., cheating) and override internalised norms or social expectations to conform (i.e., not cheat), even when it is recognised that the behaviour is wrong (Brent & Atkisson 2011; Kaptein & van Helvoort 2019). Essentially, individuals can render inoperative the social bonds and controls that would typically inhibit deviance, by neutralising their responsibility or blaming others and maintaining a positive self-image (Sykes & Matza 1957). ToN was first utilised to explain juvenile delinquency, and includes five techniques:

• Denial of responsibility: ‘it’s not my fault’ – cheating caused by forces outside/beyond individual control (e.g., too many assignments, having to work multiple jobs while studying, etc.).

• Denial of injury: ‘it’s not a big deal’ – cheating does not cause injury or victimise (e.g., a ‘victimless crime’ that does not hurt anyone, it is merely another tool).

• Denial of the victim: ‘they deserve it’ – the victim deserves the consequences and cheating is viewed as a legitimate response, whereby the student was forced to cheat by the victim (i.e., the university) (e.g., limited tutor availability or support, or punitive institutional policies).

• Condemnation of the condemners: ‘you are just as bad’ – students attempt to shift the blame to those condemning the cheating, arguing universities or their staff do not act with integrity, so question apparent hypocrisy (e.g., tutors did not teach properly, recycling of online materials, focus on making money and not education).

• Appeal to higher loyalties: ‘the greater good’ – cheating enables students to prioritise and/or address the needs of smaller social groups over the demands of the larger society (e.g., helping a struggling friend by completing their work overrides the responsibility for academic integrity).

Research has regularly employed ToN to explain cheating and academic misconduct more broadly (Beasley 2014; O’Rourke et al. 2010). These studies have shown students neutralising their actions when caught cheating, blaming others or their situation as creating the need to cheat. However, in a meta-analysis of diverse student motivations for academic misconduct, Ives' (2020) findings challenged ToN’s utility, at least as a sole perspective. Ives, (2020) offered several recommendations: to investigate broader theories of motivation; more open-ended opportunities for students to share their motivations for cheating; and models broader than ToN when designing interventions, which individually and collectively speak to the comprehensive model proposed in the current paper.

Using a multi-theory model

As this research demonstrates, various theories can, and have, explained student cheating behaviours. However, many studies have explored theories in isolation and/or merely sought to quantify cheating, viewed dichotomously as either a problem of HE or of students, with the latter more frequently and intensely blamed. Yet, HE is more expansive, comprised of students’ perceptions, backgrounds, cultural upbringing, moral values, and broader socio-cultural expectations of what university study is or should be and cannot be easily explained through individual theories. We argue that more than one theory, or indeed all, should be used concurrently, in local contexts to understand how students decide to outsource. Exploration of how and why behaviours are learnt, their dynamic shaping through both moral and peer influences, and their reinforcement and justification will help to increase knowledge of assignment outsourcing and develop targeted prevention strategies making this more than a theoretical exercise. To understand the extent of applicability of overlapping theories, we examined whether student responses could be coded to key criminological theories. This paper presents our qualitative analysis of student responses to open-ended, optional survey items.

Methods

This project employed a large-scale, mixed-method international survey to garner university students’ reasons for outsourcing. The survey sought to assess the prevalence, behaviours and motivations of students engaged in contract cheating and assignment outsourcing across multiple countries. The items were influenced by the surveys of (Bowers, (1964), Foltýnek & Králíková (2018), and McCabe & Trevino, (1993). Though not initially built with the aim of a criminological study, the survey tool enabled statistical and thematic analyses, applied inductively and deductively, to evaluate the power of criminological theory in explaining outsourcing. Items asked students to describe their study behaviours, as well as that of other students. This was purposeful to increase validity of the responses, as respondents often represent their own behaviour through reports of what they know/believe others to be doing (O’Rourke et al. 2010).

There were both qualitative and quantitative items in the survey, and a mix of optional and mandatory questions (to view the full survey, please refer to (Awdry et al. 2020). The survey was piloted in the different languages to test terminology and validity of definitions used, to try to control for differences in terms and descriptions of behaviours and academic conventions. Limitations of the survey design are acknowledged, as well as the use of a self-report survey (Krásničan et al. 2022). However, for the purposes of this paper, the researchers were interested in coding of given responses, rather than a realistic presentation of cheating rates, for example.

This paper focuses on qualitative responses which were returned in multiple languages and translated into English, in line with Deakin University ethics approval (2018:085). Comments were then collated (by question) and uploaded into NVivo for analysis, requiring iterative coding due to the quantum of comments (see Table 1, below, for final codes/rate). Initially, inductive thematic analysis was undertaken by each author to ensure familiarity with the data and guide interpretation of responses (as recommended by Guest et al. 2011; Varpio et al. 2017). Collaborative review followed, regarding identification of theories present in the data, which were categorised as related to students’ learning and reasoning of decisions to outsource. Social Learning Theory (SLT), Differential Association (DA) and Differential Reinforcement (DR) were identified as key learning theories, while Techniques of Neutralisation (ToN) and Self-control Theory (SCT) concerned reasoning, which collectively represented whether students would or would not cheat. Comments were separated between the authors and coded deductively, applying these theories to the data, with some single-coded and others assigned multiple codes, where each step was carefully recorded (including coding rules, theory descriptions and analysis notes) to ensure methodological integrity (Berg 2004; Guest et al. 2011).

Table 1 Distibution of qualitative responses coded against theory

Each category had detailed, theory-specific coding rules; for example, DR had conditions describing behaviours as being either positively or negatively reinforced, as it is otherwise directionless regarding the process of reinforcement (i.e., does not require specific relationships or identification of who comments related to). Conversely, the coding rules for SLT precisely articulated ‘who’ comments concerned (e.g., general culture, university culture, or non-specific learning), in ‘what’ way (how it was learned or modelled), ‘why’ (motivations for/against outsourcing), and ‘how’ (how behaviour/learning was enacted). For DA, codes related to comments describing internalised norms and/or behaviour learnt from close social networks (e.g., friends, family), as well as definitions of cheating, learning about acceptable behaviours, and personal integrity. For the theories concerned with reasoning, ToN was coded by sub-type (i.e., ‘Denial of Responsibility’, Denial of Injury’, etc.), with broad or contextless excuses (e.g., ‘stress’) coded generically to ‘ToN’. Comments coded to SCT included any discussion of not cheating through fear of impacting others or their own situation (e.g., bringing shame, or losing scholarship/job).

Further thematic rules were created to capture data outliers. For example, comments condemning others without context were coded against DA due to their intrinsic moral implications, while the notion of causing harm to others (e.g., impact of unqualified graduates in the workplace, inflation of average grades, etc.) were coded to SLT. Specific rules established double-coding practices, as comments often related to more than one category/theory; comments conflating outsourcing with plagiarism without context were dual-coded to DA and SLT, for example, as it was unknown if they were driven by internalised moral values (that it is wrong) or represented behaviour taught by their institution.

Following initial coding, preliminary results were discussed highlighting data that did not fit the five theories, requiring creation of an ‘Other’ category/code. Further coding was undertaken independently by each author and then compared, which revealed need for greater clarity in coding and led to refinement of coding rules, specifically adding type of behaviour/reasoning (e.g., why they cheated), and the person/context it related to (e.g., the university, or friends) to the coding. In addition to those who did not respond/provide an answer (‘Irrelevant/no answer’), five ‘Other’ sub-categories were also created (see Table 1).

It was apparent that the ‘Other—Consequences for cheating’ category could be further refined, with three child codes established: ‘Legality’ (questions of the legality of outsourcing); ‘Outcomes for cheating’ (what other things do/should happen when someone is caught cheating at university); ‘Student punishment’ (references to known outcomes for cheating, or what should occur specifically to students). Anything coded to the parent code was re-coded accordingly. A high number of comments were coded to DA where, due to the specificity of the theory and variable implications of learnt behaviour (i.e., conforming vs non-conforming), suggesting requirement for further subdivision. This was categorised into three child codes: ‘Personal/intrinsic’ (not cheating due to personal integrity/morality); ‘Pragmatic’ (justifying cheating for a purpose); ‘Learnt’ (learnt behaviours from close networks, including comments about parents buying students’ way through university by any means). All coding was quality-checked by both authors throughout this process and any inconsistencies rectified collaboratively. Thematic analysis of the data then commenced, with the findings detailed below.

Results

Overall, 16.9% of students reported some form of outsourcing (different behaviours were defined and included in this, such as getting an assignment from a friend or family member, downloading an assignment from an essay mill or obtaining a bespoke assignment).

In the survey tool, qualitative fields followed several quantitative items, designed to add explanatory data to the descriptive statistics. Most qualitative items were optional but experienced good response rates, receiving a total of 2087 individual comments, including 1011 comments on the final item which asked respondents: ‘If you wish to say anything else on this topic, please leave a comment below’. From this sample (n = 2087), 3254 individual coding points were created. The largest code group was ‘Differential Association (Personal/Intrinsic)’ (n = 552) while ‘Other’ (n = 18) was the smallest. Comments coded to ‘No response/irrelevant’ (n = 189) were removed from further analysis, while the remaining ‘Other’ subcategories provided context for students’ perceptions. Table 1 presents the number of references and percentage coverage for each category.

Table 1 reveals student comments associated with diverse criminological theories, which broadly represent the reasons why students did cheat: through learnt behaviours (SLT, DR, and DA) and the self-justification of cheating through excuses (ToN); and why they did not: again, through learning (SLT, DR and DA) and through self-control to remove unwanted consequences of cheating (SCT). The largest proportion of comments were associated with DA (16.96% coverage), followed by SLT (8.91%), ToN ‘Denial of responsibility’ (8.82%) and ToN ‘Denial of the Victim’ (8.21%), respectively. Seemingly, the most common student explanations were those which provided them the learning to cheat (or not) and the justifications to do so. Significantly, the spread reinforces that a singular theory cannot effectively explain why students do, or do not, engage in assignment outsourcing. While many of these factors could have been further divided into opportunity-based perspectives and the weighing-up of risks/rewards, these sub-theories which create the knowledge for how, when, or why to cheat necessitate greater attention.

Why they do

The learning theories accounted for more responses than the other theories, however, many were categorised against positive and negative implications of outsourcing. For example, Social Learning Theory framed suggestions (n = 290) that students did not cheat due to fear of being caught by their institution and the connotations associated with cheating/plagiarism. Conversely, the perception that institutions did not detect cheating, do anything about it or place great importance on managing it was positively attributed to engagement with outsourcing:

Differential Reinforcement was similarly prevalent, coded against 173 comments, for example:

Differential Association was coded to a large proportion of comments (n = 737), with three sub-codes to better align comments to its theoretical principles:

Learnt (n = 27)

This included participants’ descriptions of how they had learnt whether or not to outsource from peers or family, revealing both positive and negative associations with cheating. Some emphasised financial inequalities among students, while others focussed on a more direct, negative influence regarding the commercialisation of higher education:

Pragmatic (n = 158)

Students also reported behaviours learnt situationally where they identified pragmatic reasons/purposes to cheat, demonstrating learning of both the concept and practice of cheating (i.e., when to apply it):

Others’ comments built upon these challenges, where perceived ‘external’ pressures to succeed were reconciled against individuals’ capacity (i.e., wealth), arguably creating fertile ground for outsourcing:

Others expressed disdain for cheating or a strong moral code, alongside an awareness that they may be disadvantaged because they did not cheat:

The DA Personal/Intrinsic (n = 552) comments related to participants’ reasons for not outsourcing, which are presented in the next section (‘Why they don’t’).

Other reasons students gave for outsourcing were coded against Techniques of Neutralisation (ToN) and its five subtypes, as excuses for why they felt it was acceptable to cheat or did not see it as cheating at all. While many of these responses were to questions ‘When is it OK to cheat?’, or ‘Why do you think students use these sites?’, students also offered neutralisations of outsourcing in other items and in the final open-ended question. Generically coded responses for acceptability of cheating (n = 126), included:

The largest number of comments coded to ToN related to ‘Denial of Responsibility’ (n = 287), ‘Denial of the Victim’ (n = 267) and ‘Condemnation of the Condemners’ (n = 231). By contrast, responses were far fewer for ‘Appeal to Higher Loyalties’ (n = 34) and ‘Denial of Injury’ (n = 29).

Denial of Responsibility

Denial of the victim

Condemnation of the condemners

Appeal to higher loyalties

Denial of injury

Why they don’t

As discussed above, many comments represented the converse use of the various learning theories, reasoning why individuals would not outsource. Specifically for Differential Association, the ‘Personal/intrinsic’ subcategory was created to represent most comments associated with honest behaviours, either learnt and/or instilled in them from close relations. Students presented many personal reasons why they would not outsource, seemingly driven by intrinsic motivations for learning, which speak to established moral values and highlight key learning processes:

More comments referred to recognition of their authorship, quality of learning and, specifically, their effort/hard work:

A feature of respondents’ commentary was its bifurcation, extending beyond individual responsibility to recognise external, pragmatic consequences of having unqualified graduates in the community:

Similarly, Differential Reinforcement is useful in framing students’ explanations as to why they would not cheat, where potential benefits were balanced against knowledge of regulatory structures and/or threat of sanction:

Social Learning Theory comments relating to reasons for cheating discouragement included responses such as:

Self-control Theory was coded to 174 comments, focusing on wider implications of getting caught, rather than just a moral dimension against cheating (otherwise coded as DA). These included comments associated with impacts of not learning (including implications of deceit to others) and shame:

Many respondents also mentioned that they had outsourced an assignment but, after obtaining it, chose not to submit it (note: most were double-coded against one or more ‘Other’ category, presented below). This demonstrates that the temptation to outsource can be so strong that some obtain outsourced materials but are then prevented from using them (through internal controls or beliefs).

Other

The’Other’ category comprised comments not connected to any of the criminological theories, usually because they were not contextualised, preventing assessment of whether they were, for or against outsourcing. Some responses also discussed outsourcing matter-of-factly but did not provide any reasoning for the behaviour. The largest group of these comments related to ‘Consequences of cheating’, which was sufficiently disparate to be separated into further sub-categories: ‘Outcomes for cheating’ (n = 50); ‘Legality’ (n = 67); and ‘Student Punishment’ (n = 108).

Outcomes for cheating

Legality

Some comments appeared punitive, using the symbolic weight of the criminal law to express concern:

While others offered a more balanced view:

Student punishment

A further 132 ‘Other’ comments revealed the respondent had never heard of outsourcing and therefore could not comment reliably. Discussion of laziness leading to outsourcing, themselves or (most commonly) by others, was also a common theme throughout the responses. Where these could be rationalised (i.e., other priorities, social life or work, or support from rich parents), comments were coded to DA or DR accordingly. Any comments which were not contextualised were coded to ‘Other – Laziness’ (n = 176).

The final two ‘Other’ categories did not relate to any criminological theory/explanation as to whether the behaviour was un/acceptable, simply citing the obtained assigned as ‘Poor quality’ (n = 26) or seen only as a ‘Learning Resource’ (n = 140). For the latter many were double-coded to DA or ToN, though due to the number of comments it was deemed an important semantic that should be coded individually to determine its frequency (4.30%):

Discussion

Various studies have applied a criminological theoretical frame to explain assignment outsourcing, examine student motivations, opportunity-based approaches, personal and social-cultural institutional causes and prevention techniques (Baird & Clare 2017; Beasley 2014; Clare 2022; Hodgkinson et al. 2016; MacGregor & Stuebs 2012; Nagy & Groves 2021). These studies have established criminology’s value in examining assignment outsourcing; however, as we argued, individual theories have been limited in their scope because they fail to solely account for the complex challenges posed by outsourcing/cheating behaviours. A multifaceted theoretical approach is needed to explain outsourcing as the product of a complex interplay of motivations, decisions, and behaviours.

There is considerable qualitative evidence that university students’ cheating is a response to the assemblage of ‘pressures, stresses, and anxieties associated with being a student’ (Tindall et al. 2021). The criminological literature reveals that both positive and negative affect influence engagement in unethical behaviours, contributing to more favourable attitudes toward cheating (Ruedy et al. 2013; Tindall et al. 2021). To the extent that greater pressure to appear successful and meet common, arguably extrinsic, goals may cause such emotive responses, speaks to the normalisation of cheating as a means to an end (Amigud 2020; Lancaster 2020), where some may see it as morally wrong and others merely as another form of ‘help’. Moreover, the ubiquity of tailored advertisements for outsourcing websites, where the distinction between legitimate academic support services and third-party outsourcing sites is blurry at best (Ellis et al. 2018), exacerbates this pressure, especially in an increasingly isolated online study environment where self-directed learning is lauded. As such, if universities want to prevent cheating by strengthening positive attitudes to academic integrity (i.e., conformist behaviour), then prevention efforts must consider how students learn deviant or conformist strategies in the first instance and the contexts in which they do and do not apply them – as articulately captured by one respondent:

‘This problem is emphatically an institutional one as much as it is a matter of individual transgression.’

Our data demonstrated that explanation of student assignment outsourcing can be found in diverse criminological theories. Wider socio-cultural, and institutional influences can have acute impacts on teaching deviant or conformist behaviours, ultimately shaping whether students will or will not cheat, as students learn dis/honest behaviours from a range of methods, people, and settings. Learning theories were able to account for this through close associations (DA), reinforcement (DR) and broader influences (SLT), where some students stated that they, or others they knew, had ‘acceptable’ reasons to cheat. These could also be categorised as excuses, representing various techniques of neutralisation. In terms of learning, we demonstrated that students experience various influences, from key social institutions (e.g., school) but also from family, friends, and wider social networks, the reach of which need to be addressed given the consequences for behaviour, particularly the influence of peers. Many respondents’ comments also referred to the reinforcement of deviance where institutions, for example, did not take cheating seriously, promote integrity themselves, bother to detect cheating, or apply significant consequences if misconduct was found providing the context for deviance/non-conformist behaviours to proliferate. Knowing students who had cheated and not been caught, seeing tutors ignore witnessed cheating and/or give lenient outcomes led students to believe this was accepted behaviour, and normalised outsourcing as a legitimate method to complete assignments under certain circumstances.

The consequences of these experiences are numerous and far-reaching. Not only can strong messaging on academic integrity, clear misconduct policies and open conversations around expected standards or consequences for cheating be a deterrent, it can also provide the context for students to learn about integrity, conformist behaviours and influence expected morals/norms. In relation to self-control, it is important to note that learning is not limited to delinquent behaviours, as all behaviour can be influenced, where it is the weight, intensity, or frequency of exposure to certain behaviours that lead an individual to offend or adhere to normative values (Akers 1990). While peer influence may not be easily manipulated in HE, the strength of such networks should not be underestimated (Awdry & Ives 2021; O’Rourke et al. 2010). Comments in our dataset showed that peers could influence learning of deviant behaviours (through association and reinforcement) and provide excuses for why it is acceptable to cheat (e.g., needing to cheat to remain competitive in their degree, or that others did it and were awarded grades). Equally then, reinforcing general cultures of academic integrity at institutions may increase the preponderance of honest behaviours, leading students to feel shame if peers thought they were cheating, introducing the strength of SCT. When contemplating the strengthening of institutional policies and practices to address the threat of cheating, it must be acknowledged that students need not be part of deviant social networks to be influenced to cheat, as they see others cheat, hear about it from peers, and/or are taught about cheating by subject tutors and lecturers and/or through generic academic integrity training modules.

As shown in the literature and qualitative data presented here, context is fundamental in students’ decisions to cheat, which emphasises the importance of examining the tertiary setting and how, why, when, and from whom students learn cheating behaviour. Specifically, these findings reinforce the need for universities and the HE sector more widely to acknowledge the situational context of assignment outsourcing and, importantly, the extent of their role and contributions in shaping and facilitating this behaviour. As Clare, (2022) offers, universities need to understand where the type of cheating occurring and where the outsourced assignments are coming from in order to take action and design strategies to address the issue. To date, universities have failed to acknowledge this, conceptualising outsourcing as a problem of students and predatory third-parties, ignoring the various institutional determinants articulated in our respondents’ comments. The multi-theoretical approach employed here innovatively brings together interrogation of the situational or micro-level determinants of outsourcing with a much broader, macro-level perspective of the problem. Taking this philosophical ‘step back’ usefully extends current scholarship beyond reliance on single or dual theoretical explanations and encourages evaluation of the social, political, and structural influences on outsourcing, establishing the utility of an interdisciplinary, multifaceted approach that recognises the diversity of reasons why students do and do not cheat.

Importantly, this process starts long before students arrive at university, but the new educational environment lends itself to new interactions, expectations and influences which must be leveraged. As we have demonstrated, this combined approach can help HE to consider when and how students learn the craft of ‘doing university’, which may include context for students to rationalise and reason why they would or would not engage in cheating. While it was beyond the scope of this paper, logically, considering this macro-level understanding of influences on engagement in outsourcing means that we must now explore suitable prevention and intervention techniques. Criminology can provide the theoretical approach and, crucially, its tools of prevention to strategise and undertake this task, something which has been rare in the educational literature to-date.

Limitations and recommendations

While we made every effort to prevent coding inconsistencies between the authors, this is always a risk with a large dataset. Further, we acknowledge that the coding represents our subjective (albeit collective) understanding of the theories and their definitions. Additionally, one of the questions asked students why they thought other people cheat, with some responses related to their perception of others, rather than self-reported behaviour. Notwithstanding, this is one of the first comprehensive qualitative analyses to consider multiple theories in the consideration of academic cheating. We were limited by our decision to focus on the theories selected and could have created entire codes focussed on situational prevention, which we recognise is needed. As a result, a key recommendation is that further research is completed on this topic and our multi-theoretical approach used to determine effect intervention points and prevention techniques. Situational crime prevention (SCP), if appropriately applied to these theories of deviancy, could provide education with more powerful tools to address the issue of assignment outsourcing and academic dishonesty more broadly. SCP provides the “so what?”, and facilitates transition from the acknowledgement of the need for a multi-theoretical approach, as empirically validated here, to the implementation of practical strategies that shape what can be done about the problem of outsourcing.

Conclusion

The value of our approach is its interdisciplinarity, drawing from—but not limited to—education, economics, criminology, psychology. This allows flexibility in exploring students’ reasons to outsource, without trying to ‘fit’ them to one perspective, evaluated through a qualitative survey methodology. By examining how and when these dishonest behaviours are learnt, we can consider intervention strategies to prevent their development or situational application. The influence of others on cheating behaviours has been strongly evidenced; what this research contributes is the specific theoretical learning implications of this, particularly in the HE context. Our respondents’ comments demonstrated that a simple, linear, and rational ‘if/then’ decision-making process is insufficient. Situational prevention strategies must recognise the interlinked structure and practice of assignment outsourcing and respond accordingly, employing a multifaceted, comprehensive, and arguably cooperative approach. This information should equip HE to consider more effectively what, if anything, we are doing that can influence the learning of honest behaviours, over dishonest ones. As this respondent succinctly articulated: