Academic Dishonesty Among Accounting Students: Some Indonesian Evidence

Academic dishonesty among students has been recognised as a major concern in higher education in Indonesia. Accounting research arguably need to give more attention to this issue. This is partly because of the importance of integrity as part of accounting ethics and professionalism. However, little currently known about academic dishonesty among accounting students in Indonesia. We address this issue by surveying 342 accounting students about their perception of academic dishonesty and what motivates such behaviour. Our respondents were from all first, second or third year undergraduate students at one state university in Indonesia. Drawing from Theory of Planned Behaviour, we examine three individual variables - attitude, subjective norms, and perceived behavioural control. In addition, we also examine three situational variables - academic integrity culture, definitional ambiguity, and pressure. Six hypotheses were tested, using a Partial Least Squares-Structural Equation Modelling. The results reveal that 77.5% of respondents admitted committing academic dishonesty. While all individual factors studies have positive significant effect on the intention to commit academic dishonesty, of the three situational factors only pressure and definitional ambiguity have a positive significant effect. Surprisingly, it is found that academic integrity culture does not have a significant effect.


INTRODUCTION
A series of instances of business fraud and corruption have been reported in the last two decades. Examples of such corporate crime included major companies, such as Enron, Tyco International, Citibank, and Satyam Computers (Soltani 2014). In Indonesia itself, it is now common to read reports of corruption happening in government agencies both on the central and local level (Ganie-Rochman and Achwan 2016). There have been various recent 'big corruption scandals' involving politicians, business people, and Indonesia bureaucracy (Prabowo et al. 2017) such as the BLBI case, the Hambalang case, and recently the e-ID graft. One important question about these high profile financial scandals is about the profile of actors involved in such unlawful business practices.
Fraud and corruption are categorised as white-collar crime (ACFE 2014). White-collar crime is defined as "the crimes committed by individuals of high social status during the course of occupations" (Sutherland 1940). This type of offence is highly related with educational level and power. This relationship is supported by the fact that fraud committed by executives and upper-level management, of which 70% of them have a first or postgraduate degree, cause the largest amount of losses (ACFE 2014). In Indonesia, 82 percent of fraudsters are university graduates who serve the local government, city, and province (Maharani 2015) Several studies, such as Nonis and Swift (2001), have found a strong correlation between the frequency of dishonest acts by people when students at university with the tendency towards unethical behaviour as employees in the workplace. Crown and Spiller (1998), Lawson (2004) and Ma (2013) support this relationship. On the other side, there is a common expectation that a university graduate should possess a scholarly personality: the mental power, frame of mind, attitude, and a certain wisdom that belongs to those who have studied at university (Suwardjono 2014). In short, a university graduate is expected to possess ethical sensitivity, as well as ethical judgment, leading to ethical behaviour (Nadelson 2006) Accepting this relationship between education and ethical behaviour (Melé 2005;Floyd et al. 2013;Martinov-Bennie and Mladenovic 2015), business schools in Indonesia have made efforts to strengthen the ethical content of their syllabus. This effort aims to promote awareness of ethics, ethical reasoning ability, and the core principles of ethics that will support students in dealing with the complex environment (AACSB 2004). As the purpose of teaching ethics is to promote the highest level of ethical thinking, this effort is expected to promote an ethical mindset (Fryer 2007)  However, embedding ethics in students is not an easy task as they appear very prone to conduct unethical behaviour such as academic dishonesty (Jensen et al. 2002). Academic dishonesty is an unending problem that has always existed and is common in universities (Thomas 2017). The argument that committing academic dishonesty partly shows a failure in ethics education leads to the extreme view that successful academic dishonesty is a lesson in conducting corruption in the future. This issue is highly neglected in accounting education (Floyd et al. 2013). Moreover, regardless of findings that more than half of business students confess to dishonest practices (Ameen et al. 1996;McCabe et al. 2006), only small number of business school leaders believe that cheating is a problem at their institution (Brown et al. 2010).
The academic dishonesty issue has received considerable critical attention including several studies conducted outside Indonesia and the preliminary survey conducted for this paper. A study carried out by Ameen et al. (1996) among 285 accounting students in four public universities in the USA finds that 56% of respondents admitted dishonesty during exams and written assignments. A meta-study of 46 different studies regarding student cheating in the United States and Canada shows that on average 70% of the students under study acted dishonestly in college (Whitley 1998). More recent studies indicate that up to 86% of college students have been involved in dishonest behaviours in class (McCabe et al. 2006) with a strong indication that dishonesty among students is growing (Simkin and McLeod 2010).
Also, the preliminary survey for this study found that, among 102 students in an undergraduate accounting program at a state university in Indonesia, 74% of the respondents admitted conducting academic dishonesty.
Moreover, according to Adiningrum et al. (2013), there is a discrepancy of understanding on academic dishonesty among staff. Taken together, this paper argues that academic dishonesty is an important issue in Indonesia. Without denying the sensitivity of this issue, the issue needs more attention from accounting scholars.

Research Motivation and Contribution
Interest in teaching business ethics course has been growing globally (Trevino and McCabe 1994). In Indonesia, teaching ethics in business schools is an emerging practice partly to respond to Government (RI 2014) and accreditation body standards (AACSB 2004). However, most of the students tend to perceive academic dishonesty as a common practice in their academic life (McCabe et al. 2002;Smyth and Davis 2004). This paradox drives this study to understand why students are motivated to conduct academic dishonesty.
The criticism increased in the academic community as being partly to blame, due to insufficient attention to addressing instances of academic dishonesty among students. This provides a strong motivation for this study to look closely at the factors that affect academic dishonesty. At the same time try to understand how serious this problem is among accounting students in Indonesia. Ford and Richardson (1994) see two sets of specific factors which possibly affect academic dishonesty among students. Individual factors are personal characteristics, those which are a result of birth and those due to human interaction and development. Situational factors are those that shape and define the situation in which people make decisions. In every decision-making process, these two sets of factors will likely be involved and create a unique interplay in reshaping human behaviour (Ferrell and Gresham 1985;O'Fallon and Butterfield 2005;Craft 2013). This paper discusses academic dishonesty within the context of accounting education, arguing that academic dishonesty by accounting students is a threat to achieving the highest possible ethical behaviour for them as a future accountant. IESBA (2016) stresses that ethical behaviour is a core of integrity while the significance of integrity to the accounting profession is irrefutable. Also, accounting students tend to show lower levels of moral development than non-business students (Armstrong 1987). It means the threat to their integrity like academic misconduct become more prevalent. Moreover, many of them will eventually become professional accountants and business leaders in the future (Guo 2011).
This study adds to and extends the academic dishonesty literature in two ways. First, empirical research on academic dishonesty has mostly been conducted in the context of developed rather than developing countries. This study will extend the boundary of research by investigating this issue in an Indonesian context. It will be able to enrich the literature and bridge knowledge gap in the study of academic dishonesty. Indonesia has different socio-cultural aspects compared to most developed countries. We follow Gray (1988) arguments that accounting systems including accounting education are influenced by culture. Second, this research contributes by attempting to answer for a calling on this research topic (Scrimpshire et al. 2017). Academic dishonesty is known as a widespread phenomenon (Simkin and McLeod 2010), the problem for education institution in Indonesia (Akbar 2008;Adiningrum et al. 2013), and concern for accounting education at the higher institution (Flynn 2003;Ballantine et al. 2014).
The remainder of the article is organised as follows. First, academic dishonesty literature addressing definition is reviewed. Second, a theoretical framework based on personal and situational factors is outlined. The final section provides analysis, and a discussion of the result of the research direction proposed.

Academic Dishonesty
There is no single accepted definition of academic dishonesty (Kibler 1993), and there are many different definitions in the literature (Walker and Holtfreter 2015). While Jensen et al. (2002) identify academic dishonesty specifically as the attempt of students to present the academic work of others as their own, von Dran et al. (2001) define academic dishonesty simply as an intention to behave unethically. Weaver et al. (1991) define academic dishonesty as a violation of institutional rules regarding honesty. Finn and Frone (2004) define it as a breach of regulations and standards needed to complete homework and exams. Staats et al. (2009) see academic dishonesty as a type of deviant behaviour harmful to the development of character, hurtful to others, and endangering the academic integrity of institutions. Academic dishonesty can also be seen in any act or fraudulent effort conducted by students to use illegal or unacceptable means in the production of academic work (Lambert et al. 2003).
This study uses the general classification of Stone et al. (2010). There are many forms of academic dishonesty committed by students that are difficult to observe in a single study. Therefore, this study does not attempt to measure all types of academic dishonesty. Stone et al. (2010) identify eight distinct forms. First, students using other people's work as their own. Second, students cooperating in tasks that should be done individually. Third, students doing homework for someone else. Fourth, students obtaining information about an exam from other students. Fifth, students copying from other students during a test. Sixth, students using an illegal source in completing a task. Seventh, students using an unreliable or inappropriate resource in doing exams. Eighth, students plagiarising by using unreferenced sources from the internet.
Academic dishonesty is a unique unethical behaviour especially in the way students rationalise their cheating behaviour. McCabe (1992) suggests that students use all possible justifications of their cheating action, such as: denial of injury or another adverse outcome, denial of victim, appeal to higher loyalty, and condemnation of condemners. Most using mind block, no understanding of the material, and pointless assignments as the rationalisation.
Academic dishonesty also can be seen as an obstacle to accounting students' moral development. Specifically, in the view that accounting education which should provide an appropriate environment for proper cognitive, moral and (in some cases) faith development in accounting students (Armstrong 1987). However, Armstrong (1987) finds that higher education may not nurture continued moral growth for accounting students. Armstrong (1993) examines a method of teaching ethics and professionalism and its effect on students' moral development. She suggests a course in Ethics and Professionalism can make a difference in students' lives 'by exposing them to ethical theories and principles of professionalism'. She suggests adopting this as a stand-alone course as well as reaffirming moral reasoning through case analysis in existing accounting courses.

Theory of Planned Behaviour
Theory of Planned Behaviour (TPB) is very useful in explaining a complex phenomenon such as ethical/unethical decision-making such as academic dishonesty. This theory has been applied to a broad range of topics from whistleblowing by civil servants (Winardi 2013), decision-making by public accountants (Buchan 2005), software and music piracy by accounting students (Alleyne et al. 2015), to corruption engagement (Othman et al. 2014). Since the introduction of TPB, a range of studies have implemented the theory in various contexts of behaviour (Conner and Armitage 1998;Ajzen 2011).
Individual factors in this study are explained using TPB. TPB focuses on explaining human behaviour (Ajzen 2005). Ajzen (2005) proposes this theory as a refined version of previous theory named the Theory of Reasoned Action (TRA) suggested by Ajzen and Fishbein (1975). The central argument of TPB is that behaviour can be deliberative and planned. Because behaviour cannot be 100 percent under control, perceived behavioural control as a new element in TPB is introduced (Ajzen 1991). According to TPB, there are three forms of beliefs guiding human behaviour. First, behavioural belief is that of possible consequences of the action. Second, normative belief is about other people's normative expectations. Third, control belief is about the existence of factors that could enable or disable performance of the action. Beck and Ajzen (1991) and Stone et al. (2010) show that the intention to cheat and other dishonest acts can be explained by TPB. TPB sees the intention to conduct or not to conduct any action as an important determinant of actual behaviour (Ajzen 2005). This theory assumes that intention will influence behaviour. The intention is an indication of the willingness and effort of the individual to perform a particular action. The general rule that applies is: the stronger the intention to engage in a particular behaviour, the more likely that certain behaviour will occur (Ajzen 1991). Intention becomes a strong proxy for behaviour although intention will not always be translated into actual behaviour (Chandon et al. 2005). Intention is used to explain academic dishonesty practice because the intention is the best predictor of behaviour (Ajzen 2005). Measuring actual academic dishonesty practice is difficult because there is a gap between intention and actual behaviour. McCabe et al. (2012) find the difference in the estimated number of reported instances of academic dishonesty and those actually occurring is due to students not being completely honest in answering questions about their behaviour. According to Hadjar (2017) there is a paradox between negative perception to academic dishonesty and experience in conducting academic dishonesty. This paradox also contributes to the problem in measuring actual dishonest behaviour.
TPB introduces three independent factors in affecting intentions (see Figure 1). First, attitude towards behaviour refers to the extent to which a person has a good rating or better on behaviour. Second, subjective norm relates to the social pressure perceived to perform or not perform a behaviour. Third, perceived behavioural control refers to the ease of the perceived perception to perform the behaviour. In general, the higher attitude and subjective norm and the greater perceived behaviour control, the higher the intention of the individual to perform a behaviour (Ajzen 1991;Beck and Ajzen 1991;Mayhew et al. 2009;Stone et al. 2010;Cronan et al. 2015). TPB is not the only one theory that can be used to explain academic dishonesty. Fitriana and Baridwan (2012) use fraud triangle framework to study academic dishonesty among accounting student in Indonesia. They argue that academic dishonesty behaviour is determined by pressure, opportunity, and rationalisation. Although the result is interesting, we take a different route by selecting a different theoretical framework. Our argument is because academic dishonesty (and fraud as general) is a complex social phenomenon, whose contextual aspects may not be suitable into a particular framework like fraud triangle (Lokanan 2015).

Figure 1 Theory of Planned Behaviour
Source: Ajzen (2005) In this study, TPB is complemented by situational factors. Situational factors used in this study are a culture of academic integrity (Kisamore et al. 2007;Guo 2011), definitional ambiguity (Ellahi et al. 2013), and pressure (Smith and Minhad 2007;Guo 2011;Koh et al. 2011). Individual and situational factors are employed as independent variables. This study examines the relationship between these independent variables on the dependent variable that is the intention to commit academic dishonesty. The complete conceptual framework can be seen in Figure 2.

Attitude toward Academic Dishonesty and Intention to Commit Academic Dishonesty
According to Ajzen (2005), attitudes toward behaviour consist of individual positive or negative evaluations of an object, person, organisation, or event. Attitude consists of beliefs about consequences of the behaviour and an assessment of such behaviour (Alleyne et al. 2013). Students who believe that academic dishonesty practice will produce a good result will have a positive attitude to it. In short, they assess academic dishonesty as a positive action. Thus the intention to commit academic dishonesty will tend to be higher. Conversely, students who believe that the conduct of academic dishonesty will result in an adverse outcome will have a negative attitude towards it. They assess academic dishonesty as a bad behaviour, so their intention to commit academic dishonesty will tend to be lower. Trafimow (1996) finds that attitude is the strongest predictor of behaviour. Stone et al. (2010) find a positive and significant relationship between attitudes toward academic dishonesty and the intention to commit academic dishonesty. These findings are in line with the past research by Beck and Ajzen (1991), Mayhew et al. (2009) and Cronan et al. (2015).
Previous studies find that attitudes toward behaviour are one of the most influential factors for intention to perform academic dishonesty. Majority of the studies agree that attitude has a positive relationship with intention. This study expects that the relationship will remain positive under Indonesia environment. The prevalence of this phenomenon in Indonesia might be partially explained by students' attitude. A study by Agustina and Raharjo (2017) find that students who know that plagiarism is bad, but they still performed it because they think plagiarism as 'a way out' to help them obtained a good mark. Also, Hartanto (2012) stated that students are cheating because they think cheating is a normal act. Therefore, the following hypothesis can be derived: H1: Students with more positive attitude toward academic dishonesty will show higher intention to commit academic dishonesty.

Subjective Norms and Intention to Conduct Academic Dishonesty
Ajzen (2005) defines a subjective norm as the perceived social pressure to perform or not perform a certain behaviour. Subjective norm can also be interpreted as a students' perception that somebody else becomes their reference to think that they should or should not perform the certain behaviour (Ajzen 1991). Students with a particular reference which motivates them to perform academic dishonesty will feel a positive subjective norm. Contrariwise, students who believe their reference will not approve of academic dishonesty will have a negative subjective norm. This helps a student to avoid committing academic dishonesty.
The previous study conducted by Stone et al. (2010) find that subjective norm is a significant predictor of intention to commit academic dishonesty. Similar results are found by Beck and Ajzen (1991), Mayhew et al. (2009) and Cronan et al. (2015). Another study conducted by Ellahi et al. (2013) find that the view from peers provides normative support towards academic dishonesty.
Previous studies outside Indonesia find that subjective norm has a positive relationship with intention (Beck and Ajzen 1991;Mayhew et al. 2009;Stone et al. 2010;Cronan et al. 2015). Another study conducted by Ellahi et al. (2013). This study expects that the relationship will remain positive under Indonesia environment. The argument for this is accounting students have a cohesive and close interaction with their peers. They tend to be more collectivistically-oriented (Teoh et al. 1999). Therefore, their perception also influences by their peers' way of thinking toward academic dishonesty. Hartanto (2012) mentions that pressure from friends could explain the cheating behaviour. Based on the findings outside Indonesia and additional discussion within Indonesian context, this study proposes the following hypothesis: H2: Students with higher subjective norm on academic dishonesty will show greater intention to commit academic dishonesty.

Perceived Behaviour Control and Intention to Conduct Academic Dishonesty
According to Ajzen (2005), perceived behaviour control refers to the individual's awareness of how easy or tough it is to accomplish certain behaviour based on the resources and chances that exist. The more resources and chances that students have in performing academic dishonesty, and the fewer anticipated barriers, the greater the perceived ease felt by the student. The previous study conducted by Stone et al. (2010) find that behavioural control has a positive and significant effect on the intention to commit academic dishonesty. This finding is supported by Beck and Ajzen (1991), Mayhew et al. (2009) and Cronan et al. (2015).
Students are expected to aware of consequences of unethical decision making (Martinov-Bennie and Mladenovic 2015). This issue is addressed in business ethics course which covers a various example of case studies (Baetz and Sharp 2004.). However, academic dishonesty keeps plaguing in Indonesia (Adiningrum 2015). Students think that they can overcome the consequences or think that it does not have any ethical consequences (Agustina and Raharjo 2017). Hartanto (2012) proposes lack of punishment is one factor in cheating behaviour. This factor also possible to affect students' control belief which leads to a decision in committing academic dishonesty.
Drawing from previous studies outside Indonesia and additional discussion within Indonesian context, this study expects that the relationship between perceived behavioural control and intention will remain positive under Indonesia environment. We propose the following hypothesis built on the discussion: H3: Students with higher perceived behavioural control will show greater intention to commit academic dishonesty.

Academic Integrity Culture and Intention to Conduct Academic Dishonesty
Academic integrity culture refers to 'an institution's values regarding promoting academic honesty as well as preventing and punishing academic misconduct' (Kisamore et al. 2007). Academic integrity culture is a key driver of a negative perception toward academic dishonesty (McCabe et al. 2002). Good academic integrity culture leads to lower intention to commit academic dishonesty. Several examples of this culture are faculty member tolerance toward academic dishonesty, penalties for dishonest acts, and code of honour (Kisamore et al. 2007).
Previous research conducted by McCabe et al. (1999) and McCabe et al. (2002) find that good academic integrity culture will result in fewer academic violations. McCabe et al. (1999) and McCabe et al. (2002) find that academic integrity culture was the best predictor of academic dishonesty. Kisamore et al. (2007) find that academic integrity culture lowered the students' perception of the frequency of cheating and suspicion regarding misconduct.
Accounting higher education institutions in Indonesia promote this culture of academic integrity via various instruments (see Minister of National Education Regulation 2010). It can be in the form of the honour code, institutional policies, poster, announcement, etc. Students are brought to a certain value that prohibits academic cheating. Strong culture expected to prevent and overcome academic dishonesty (Siaputra and Santosa 2016). This study expects that the relationship academic integrity culture and intention will remain negative under Indonesia environment. From the above discussion, the following hypothesis can be derived: H4: Higher academic integrity culture will lead to lower intentions to commit academic dishonesty. Smith and Minhad (2007) argue that students will not see academic dishonesty as a bad behaviour when they do not fully understand what constitutes academic dishonesty and what penalties they will face. McCabe and Trevino (1997) found that lack of communication relating to the regulations and policies regarding academic dishonesty will lead to a high number of instances of academic dishonesty.

Definitional Ambiguity and Intention to Conduct Academic Dishonesty
Previous research conducted by Ellahi et al. (2013) has similar results with McCabe and Trevino (1997) and Smith and Minhad (2007). Ellahi et al. (2013) find that definitional ambiguity occurs when students do not have a complete awareness of the code of conduct, or when they do not receive guidance from faculty about academic dishonesty. They find that the definitional ambiguity strongly predicts rationalisation against the dishonest academic behaviour.
Sometimes it is difficult to define what constitutes academic dishonesty. Academic dishonesty is not a unidimensional concept. Thus, it can be seen from different perspectives (Scrimpshire et al. 2017). Without clear rules, students can have a different perception of how certain acts categorise as academic dishonesty (Hartanto 2012). Moreover, students do not feel guilty in committing academic dishonesty because they do not know about it (Sariffuddin et al. 2017). Also, students are found not aware of academic dishonesty (Agustina and Raharjo 2017). Moreover, lack of awareness and understanding of staff are also found (Adiningrum et al. 2013) Drawing from previous studies outside Indonesia and additional discussion within Indonesian context. This study expects that the relationship between definitional ambiguity and intention will remain positive under Indonesia environment. Therefore, the following hypothesis can be derived: H5: Higher definitional ambiguity will cause higher intention to conduct academic dishonesty.

Pressure and Intention to Conduct Academic Dishonesty
Pressure is motivation for dishonesty that may come internally from students themselves or externally from outside (Becker et al. 2006). This study focuses on external pressure which can be in the form of student grades, time, and workload (Ameen et al. 1996;Love and Simmons 1998). Becker et al. (2006) state that a high workload in a limited period would motivate students to perform academic dishonesty. Love and Simmons (1998) find the pressure to be a major determinant of academic dishonesty. Ellahi et al. (2013) conclude that stress encourages the tendency of students to commit academic dishonesty. Koh et al. (2011) find pressure, in the form of deadline pressure, to be a motivator for academic dishonesty. Nevertheless, research conducted by Guo (2011) and Smith and Minhad (2007) conclude that there is no direct relationship between the pressure and academic dishonesty. Given this difference in previous research, this study examines whether higher pressure will lead to the higher the intention to commit academic dishonesty.
Previous studies find that pressure has a positive relationship with intention (Love and Simmons 1998;Becker et al. 2006;Koh et al. 2011;Ellahi et al. 2013). This study expects that the relationship will remain positive under Indonesia environment. Our argument starting from development in Indonesian accounting education regarding learning approach. Student-centred learning approach is one proposed teaching strategy (Santosa and Cintya 2007;Jogiyanto 2009). This approach is followed by a combination of assessment methods to capture student's performance. Students are assessed by using combination any of exams, homework, group-work, presentation, or participation (Minister of National Education Regulation 2005). This situation potentially will create a form of pressure to students when intertwined with other courses which should be learnt by them. The study found that student study load in Indonesia is higher compare to another country (Zubaidah 2015). Thus, students might use their study load as their rationalisation and take a shortcut by committing academic dishonesty. Therefore, the sixth hypothesis in this study is: H6: Higher pressure will cause higher intention to commit academic dishonesty.

RESEARCH METHOD
This study was conducted using a survey method. A survey is a method to collect data from or about people to describe, compare, or explain knowledge, attitudes, and behaviours (Fink 2003). In this study, a survey was conducted using questionnaires. The questionnaire is a set of written questions that have been pre-formulated to record the respondents' answers, usually in alternative answers that have been defined carefully (Sekaran and Bougie 2016).
The respondents in this research were undergraduate accounting students who studied at a state university in Indonesia. It is important to note that the use of only one university may diminish the external validity of this study (Sekaran and Bougie 2016). Data collection instrument was by paper-based questionnaires adopting an existing questionnaire from previous research to measure each variable in this study. The questionnaires were distributed either before or after class with permission from the lecturers and students were asked to complete the questionnaire on the spot. Also, to improve the response rate, the data collection was conducted in mandatory course classes. Respondents were reassured about the privacy of their data as well that of as their personal information.
Data collection took place between April-June 2016 and 352 questionnaire responses were obtained. Ten students did not fill in the questionnaire, so the number of the valid questionnaires was 342, with a response rate of 81.43%.

Preliminary Study
Academic dishonesty is debatably a sensitive topic (Pryor 2004). Our strategy was to assess how sensitive this topic for them. We conducted a preliminary study of 102 students. We asked them a simple question about their experience in committing academic dishonesty. The result shown 75 from 102 admitted their experience in conducting academic dishonesty. We continued our research into the main phase by distributing the questionnaire to them. Table 1 presents the preliminary survey results.

Measurement Model
The evaluation of the measurement model is conducted to assess the convergent and discriminant validity of each indicator (Ghozali and Latan 2015). We used with SmartPLS 3 software to conduct an assessment for the measurement model.

Intention to Conduct Academic Dishonesty
Intention to commit academic dishonesty constructs was measured using eight indicators, asking respondents to answer how likely they would consider performing various types of academic dishonesty. The eight indicators were adopted from Stone et al. (2010) and were measured using a Likert scale from 1 = very unlikely to 5 = very likely. However, we only considered four indicators due to validity issue. The examples of the indicator were: copying from another student during exams, using resources that are not allowed in completing assignments, using resources that are not allowed in completing exams, and performing plagiarism when finishing a written assignment. Higher scores indicate greater intention to commit academic dishonesty. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 2 shows the indicators and outcome measurement model for this variable.

Attitude
Attitudes towards academic dishonesty construct were measured using seven indicators assessing student beliefs about dishonesty, willingness to report dishonesty by other students and helping another student to cheat. The indicators were adopted from Stone et al. (2010). Respondents were asked about the consequences of academic dishonesty by selecting one of the five options using a Likert scale from 1 = strongly disagree to 5= strongly agree. However, we only considered four indicators due to validity issue. The example indicators were: cheating is always wrong, students must proceed to cheat if they know they can get away with the punishment, and I will let other students copy my exam answers if they ask for it. A higher score indicates a more positive attitude towards academic dishonesty. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 3 shows the indicators and outcome measurement model for this variable.

Subjective Norms
Subjective norm constructs were measured using seven indicators assessing student perceived social pressure to perform or not perform academic dishonesty. The indicators were adopted from Stone et al. (2010) and were measured using a point Likert scale. The response formats varied; generally frequency-based options for example from 1 = strongly disagree to 5 = strongly agree. The example indicator was: some of my friends cheated and were not caught. Higher scores indicate a higher subjective norm against academic dishonesty. However, we only considered five indicators due to validity issue. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 3 shows the indicators and outcome measurement model for this variable.

Perceived Behaviour Control
Perceived behavioural control construct was measured using four indicators adopted from Stone et al. (2010). Four indicators designed to assess the ease or difficulty of committing academic dishonesty. Respondents were asked about how easy or difficult it is to commit academic dishonesty by selecting one of the five options using a Likert scale from 1 = strongly disagree to 5 = strongly agree. However, we only considered three indicators due to validity issue. The examples indicator were: if I want to cheat on an assignment, it will be easy for me to do, if I want to cheat on an exam, it will be easy for me to do, in my class, it would be pretty easy for me to cheat, and it is difficult to cheat and not to get caught. A higher score indicates a higher perceived behaviour control against academic dishonesty. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 3 shows the indicators and outcome measurement model for this variable. Respondents were asked about their assessment of academic integrity culture around them by selecting one of the five options using a Likert scale from 1 = very low to 5 = very high. Students' perception of academic integrity culture was measured using seven indicators adopted from Kisamore et al. (2007). However, we only considered three indicators due to validity issue. The example indicators were: academic integrity culture on my campus, campus attention to academic integrity, the degree of punishment for cheating on my campus, the effectiveness of regulation related to academic dishonesty, campus response to academic dishonesty reporting, the degree of tolerance to cheating, and frequency of academic dishonesty reporting on campus. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). The higher score indicates good academic integrity culture. Table 4 shows the indicators and outcome measurement model for this variable.

Definitional Ambiguity
This study measured students' perceptions of definitional ambiguity.
Students' perceptions of definitional ambiguity were measured using four indicators adopted from Ellahi et al. (2013) and were measured using a Likert scale from 1 = strongly disagree to 5 = strongly agree. The indicators were: lecturers do not provide complete information about what plagiarism is, lecturers ignore fraud and plagiarism when they know about it, teaching assistants ignore cheating when checking quiz/homework, there are no policies or regulations that mention academic dishonesty. The higher scores are an indication of higher definitional ambiguity. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 4 shows the indicators and outcome measurement model for this variable.

Pressure
Pressure constructs were measured using three indicators adapted from Smith and Minhad (2007). Respondents were asked about the answers that they think will describe their conditions by selecting one of the five options using a Likert scale from 1 = strongly disagree to 5 = strongly agree. The indicators were: I have a limited time to complete the task, I feel the pressure to accomplish too many tasks within the given time, and I took too many courses for one semester. A higher score indicates a higher pressure. The obtained values for validity and reliability are FL > 0.60; Cronbach's alpha > 0.7; rho_A > 0.70; and AVE > 0.50, therefore meeting the recommended requirements (Field 2013;Ghozali and Latan 2015;Hair et al. 2016). Table 4 shows the indicators and outcome measurement model for this variable. We also tested the discriminant validity for all variables in the model. The square root of the AVE on diagonal lines is greater than the correlation between the constructs in the model. In addition, the value of heterotraitmonotrait (HTMT) was smaller than 0.90. From both results, it can be concluded that all variables meet the discriminant validity (Hair et al. 2016). Table 5 shows the results discriminant validity testing using Fornell-Larcker criterion and HTMT ratio.

Respondent Characteristics
Data collection included respondent characteristics, consisting of gender, cohort, grade point average (GPA), experience in academic dishonesty, and perception of existing control effectiveness to prevent academic dishonesty. Furthermore, respondents with experience in academic dishonesty were asked to answer follow-up questions related to the type of academic dishonesty committed most often, when the first occurrence was, the subjects in which respondents most often commit academic dishonesty and the impact of committing academic dishonesty. General characteristics of respondents are presented in Table 6. Table 6 shows the number of respondents who have committed academic dishonesty as 265 students or 77.5 percent. We asked them further questions to understand more about academic dishonesty behaviour. Table 7 depicts further characteristics of those students who have committed academic dishonesty.
From Table 7 it can be concluded that the type of academic dishonesty mostly committed is cheating, with a percentage of 38.9%. 70.2% of respondents had committed academic dishonesty since the first year of study. 67.9% respondents primarily committed academic dishonesty in a Financial Accounting course. The biggest impact reported by respondents in conducting academic dishonesty is reducing the effort required to perform tasks, with a percentage of 66%.

Structural Model of Academic Dishonesty
The structural model evaluation is intended to assess the quality of the model (Ghozali and Latan 2015). We used bootstrapping process with bias-corrected and accelerated option also with a 5000 resample. The results are presented in Table 8.
This model has the R 2 0.253, which means the level of relationship between variables is 25.3%. Also, adjusted R 2 0.240 indicates that the independent variables in this model can explain 24% change in the dependent variable. The remaining 76% can be explained by other variables which are not included in this study.
The effect size value (f 2 ) of each variable categorized as small with the value from 0.01 to 0.06. The predictive relevance value (Q 2 ) more than 0 means that the model has predictive relevance. The value of variance inflation factor (VIF) in the model for each variable is < 3.3. It can be concluded that there was no collinearity problem. The value of goodness of fit is expressed by the standardized root mean squared residual (SRMR) with value 0.072 < 0.080 which means that the model fits the empirical data (Hair et al. 2016).
We analysed the data using a Partial Least Squares-Structural Equation Modelling. The result is presented in Table 9. As shown in Table 9, the individual factors, ATA, SN, and PBC, significantly and positively affect academic dishonesty intention ATA → IAD β = 0.175, p = 0,000; SN → IAD β = 0.235, p = 0,000; PBC → IAD β = 0.233, p = 0,000 (one-tailed), therefore they fully support H1, H2, and H3. These results are consistent with TPB which postulates attitude, subjective norm and perceived behavioural control as direct antecedents of intentions and important in affecting behaviour. Students who have high ATA, SN, and PBC will be likely to have a high academic dishonesty intention. Furthermore, variable DA and P are also significant and positive for academic dishonesty intention DA → IAD β = 0.124, p = 0.010; P → IAD β = 0.102, p = 0.026 (onetailed), hence they fully support H5 and H6. Students who feel high DA and high P tend to have an intention of performing academic dishonesty.
The results from individual factors analysis support previous studies Beck and Ajzen (1991), Mayhew et al. (2009), Stone et al. (2010) and Cronan et al. (2015) and provide evidence within the Indonesian context. These findings show that these accounting students have a positive attitude toward academic dishonesty, which is worrying from a moral development perspective (Armstrong 1987). They also feel that the practice of academic dishonesty is something common among their peers in higher education. This finding supports previous research by McCabe et al. (2002) and Smyth and Davis (2004). Results also find that accounting students are capable of overcoming existing risk such as academic regulation when they perform academic dishonesty. These results lead to a question about how much attention is given by accounting departments in Indonesia in controlling academic dishonesty. Structures such as code of ethics, information technology, and standard operating procedures are in place but perhaps lack power in affecting students.
The results from situational factors support previous research by McCabe and Trevino (1997), Smith and Minhad (2007), and Ellahi et al. (2013). Most of the students feel an ambiguity exists that makes their perception of academic dishonesty unclear. They know that academic dishonesty is bad, but it is not always wrong when the absence of attention from those charged with governance exists. Pressures faced by students play a major role in affecting intention to commit academic dishonesty. In this case, academic dishonesty is an unintended consequence of high study workload in their study place. Research conducted by Love and Simmons (1998), Becker et al. (2006), Koh et al. (2011), andEllahi et al. (2013) also find that pressure is a major motivator of academic dishonesty. Our finding can be a signal for improvement of assessment design by considering students workload.
Contrary to DA and P, variable AIC did not show a significant effect on intention to conduct academic dishonesty. This finding does not support previous research by McCabe et al. (1999) and McCabe et al. (2002), but is similar to the findings of Kisamore et al. (2007). Accounting students' perceptions of academic integrity culture are not related to the intention of considering misconduct, β = 0.037, p = 0.218. This result contradicts the results of McCabe and others that academic integrity culture is the most important factor in predicting academic misconduct. This study can be related to research (Davis et al. 1992) that shows students' views about academic integrity and their actual behaviour are unconnected. In Table 6, most of the respondents assessed control in their campus as effective or sufficient, but most of them also committed academic dishonesty. Another explanation for this insignificant result is it may be due to the limitation of the study that only observed one university.

CONCLUSION
This study supports the argument that individual and situational factors can increase accounting students' intention to commit academic dishonesty. This study indicates that attitudes, subjective norms, and perceived behavioural control as individual factors significantly and positively affect the intention of accounting students to commit academic dishonesty. Perhaps accounting student attitudes toward academic dishonesty have indeed changed from one of dishonour to "it's fine" ). Furthermore, this study also finds that definitional ambiguity and pressure affect accounting student intention. Surprisingly, this study finds that academic integrity culture does not affect the intention of accounting students to perform acts of academic dishonesty.
Based on these findings, and the understanding of the current educational atmosphere in Indonesia, several recommendations are proposed to reduce the level of academic cheating there. First, this study can be used as a red flag on how policy and strategy on academic dishonesty on any level should be revisited and evaluated. Second, there is a need to educate students to reduce definitional ambiguity and change student's beliefs about academic dishonesty. Third, accounting departments are recommended to make appropriate sanctions for academic dishonesty more visible and have educational and prevention element. Lastly, building in mechanisms that make it easier to detect academic dishonesty is suggested.
These may provide more in-depth understanding of the head of department as an important institutional actor in this issue.
The current study is not without limitations. First, the model does not address possible interaction effect from moderating variables. There is a possibility that a different result may be found. Second, this study only considers academic dishonesty intention without testing students' actual behaviour. Third, this study only sampled accounting students at one university. These limits constrain the ability to draw a general conclusion from this study. Three factors of academic dishonesty are related to situational factors. Therefore, the respondents should represent various universities to capture variation in situational factors. Even though the measure is based on perceptions students, since they study at the similar situation (i.e. one state university), it is possible that they will have invariant perception, and this may create a bias against the hypothesis.
The future studies should broaden the scope of this study by conducting a multicampus investigation in Indonesia (McCabe and Trevino 1997) or a national study (McCabe et al. 2012). That should help to obtain more insight into academic dishonesty phenomena. Subsequent researchers may consider conducting qualitative research about academic cheating among college students, as this kind of research is minimal. Also, as academic dishonesty can be seen as a socially constructed activity the habitus concept of Bourdieu (1977) may be useful as a theoretical lens to understand how academic dishonesty becomes 'a habit' among accounting students in Indonesia. We suggest future research to link academic dishonesty and moral development of accounting students (Armstrong 1987) and to research on how teaching accounting ethics and professionalism can change students' attitude toward academic dishonesty (Armstrong 1993). Lastly, it is the time for accounting researchers to taking academic dishonesty seriously. This problem could be more dangerous than previously considered. Like corruption, perhaps academic dishonesty among accounting student has already become a pervasive and structural problem.