Exploration of the antecedents of digital piracy through a structural equation model

Computers & Education 78 (2014) 294e305

Contents lists available at ScienceDirect

Computers & Education journal homepage: www.elsevier.com/locate/compedu

Exploration of the antecedents of digital piracy through a structural equation model Yavuz Akbulut* Anadolu University, Faculty of Education, Turkey

a r t i c l e i n f o

a b s t r a c t

Article history: Received 4 April 2014 Received in revised form 20 June 2014 Accepted 26 June 2014 Available online 3 July 2014

The prevalence of unauthorized downloading and duplication has been a serious ethical and financial threat. The current study explored the antecedents of digital piracy attitudes and intentions in a country with a high piracy rate. A structural equation model was proposed to investigate the interrelationships among the proposed antecedents and digital piracy intentions. The model was tested with 268 high school students, 610 undergraduate students and 406 adults. Latent variables of interest were derived from the recent literature on piracy, which were facilitating conditions, optimism bias, previous experiences, prosecution risk, current habits, attitudes towards digital piracy and behavioral intention to conduct piracy. The model revealed a positive relationship between facilitating conditions and optimism bias. Optimism bias and prior experiences had a positive relationship with current digital piracy habits, whereas prosecution risk had a decreasing influence on these habits. Relationships among piracy habits, attitudes and intentions were significant as well. The structural equation model revealed acceptable and consistent fit values across three samples. Findings maintain that further research may resort to the assumptions of the Theory of Planned Behavior and the Theory of Reasoned Action to explain the nature of behavioral intentions towards digital piracy and eliminate undesirable piracy acts. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Computer-mediated communication Humanecomputer interface Lifelong learning Pedagogical issues Public spaces and computing

1. Introduction Rapid developments in online connectivity and digital compression technologies provided novel opportunities for interaction and diffusion of information. On the contrary, such advances have increased the unauthorized duplication and use of digital products. Referred to as piracy, this misconduct included behaviors such as copying digital materials, illegal installation, Internet piracy and loading a singleuser program to multiple machines (Prasad & Mahajan, 2003). Even though the digital goods have high development costs, duplication is almost free of charge and does not influence the performance of the product in the original computer. Thus, software piracy and illegal file downloading, increases the concerns for both intellectual property rights and lost sales (Bhattacharjee, Gopal, & Sanders, 2003). For instance, the Business Software Alliance has recently investigated the volume and value of unlicensed software used in personal computers in 2011 (BSA, 2012). An extensive survey with 14.700 respondents, which represented 82 percent of the global PC market, indicated that 57 percent of the world's PC users admitted software piracy. In addition, the commercial consequences of this shadow market increased from $58.8 billion in 2010 to $63.4 billion in 2011. PC shipments to emerging economies tend to increase this volume, because developing countries are the places where piracy rates are the highest. As an illustration, the piracy rates in Turkey (i.e., 62%) are quite higher than the worldwide average (i.e., 42%), and it almost doubles the western European average (i.e., 33%). Accordingly, it is quite relevant to investigate the antecedents of software piracy attitudes and intentions in such a problematic setting. Turkey is listed among Middle Eastern and African countries while reporting the piracy rates (BSA, 2012). The report reveals that the piracy rates in Turkey decreased from 65% in 2007 to 62% in 2011. On the other hand, the commercial value of unlicensed software in Turkey increased from $365 million to $526 million in the same time span probably due to the growth of software market in similar emerging

* Department of Computer Education & Instructional Technology, Faculty of Education, Anadolu University, Yunusemre Campus, 26470, Eskisehir, Turkey. Tel.: þ90 (222) 335 0580x3465; fax: þ90 (222) 335 0573. E-mail addresses: [email protected], [email protected]. URL: http://yavuzakbulut.home.anadolu.edu.tr. http://dx.doi.org/10.1016/j.compedu.2014.06.016 0360-1315/© 2014 Elsevier Ltd. All rights reserved.

Y. Akbulut / Computers & Education 78 (2014) 294e305

295

economies. Frequent pirates in such emerging economies tend to “install nearly four times as many programs of all sorts per new PC as do frequent pirates in mature markets” (BSA, 2012, p. 2). Furthermore, frequent pirates are more than twice as likely to live in a developing economy as they are to live in a mature one. Some of the pirating instances in emerging economies like Turkey can be attributed to the lack of a general understanding about which ways of acquiring software are legal and which are not (BSA, 2012). The lack of such awareness can be exemplified with the fact that even computer technology students do not take compulsory courses related to computer ethics in Turkey (Namlu & Odabasi, 2007). A recent empirical work cites the lack of educational precautions and maintains that the lack of policies regarding unethical computer use worsens the problem (Beycioglu, 2009). For example, stores in such emerging economies may sometimes be stocked with illegal copies of original software, which increases the undeliberate piracy rates among end users as well (BSA, 2012). Unethical or unhealthy digital behaviors of Turkish users have been investigated through several empirical studies. However, a content analysis on ScienceDirect reveals that relevant studies are generally conducted in undergraduate settings with pre-service teachers. These
, Duran, & Fraser, 2012), works have addressed the extent and predictors of unethical computer or Internet use (Beycioglu, 2009; S¸endag unhealthy online behaviors (Aricak, 2009), and psychosocial variables which predict problematic computer and Internet use (Ceyhan, 2008; Ceyhan & Ceyhan, 2008; Çuhadar, 2012; Tekinarslan, 2008). Very few studies have investigated contexts other than formal educational institutions, and proposed empirically supported hypotheses to investigate the behavioral antecedents of digital piracy. Limited number of studies in such contexts have addressed IT professionals in particular, and listed digital piracy antecedents as gender, age and experience (e.g., Mishra, Akman, & Yazici, 2006). Follow-up works with the same target population revealed further interrelationships among the code of ethics, attitudes towards unethical software use, training on computer ethics, and awareness of license conditions in government and private sectors (Akman & Mishra, 2009). In short, the current study aimed to investigate the antecedents of digital piracy through proposing a structural equation model, which is expected to be consistent both in and outside formal educational contexts. In this regard, facilitating conditions with regard to digital piracy, optimism bias, prior piracy experiences, perceived prosecution risk, current piracy habits and attitudes towards piracy were used to predict the future piracy intentions. The next section summarizes the theoretical framework, which is followed by research hypotheses and the rationale for proposing current latent variables. 2. Theoretical framework Most of the antecedents investigated in the current context were borrowed from a recent structural equation model (Nandedkar & Midha, 2012). In the study, researchers confirmed the proposed model with 219 university students. Through resorting to the arguments of the Theory of Reasoned Action (i.e., TRA, Fishbein & Ajzen, 1975), vital components in a music piracy framework were regarded as attitude and behavioral intention. That is, attitude towards music piracy was regarded as the major determinant of participants' piracy intentions. In addition to interrelationships among piracy habits, attitudes and behavioral intentions; the model highlighted the influence of perceived risks on eliminating piracy (Tan, 2002). That is, increasing perceived risk was expected to let individuals avoid digital piracy (Chiou, Huang, & Lee, 2005). Furthermore, it was hypothesized that facilitating conditions were positively related to music piracy attitudes (Limayem, Khalifa, & Chin, 2004), which could not be retained statistically. The authentic contribution of the study was the introduction of optimism bias into the model, which attenuated the relationship between perceived risks and piracy attitudes. In essence, individuals having an optimism bias (i.e., unrealistic optimism) perceive their likelihood of experiencing negative events less than other individuals around them. Such people tended to engage in digital piracy because they perceived themselves to be at a lower risk than other individuals around them. The model of Nandedkar and Midha (2012) is illustrated in Fig. 1. Many researchers have explained the reasons of digital piracy through resorting to different theoretical frameworks. For instance, Cronan and Al-Rafee (2008) used the Theory of Planned Behavior (i.e., TPB, Ajzen, 1991) as a framework to determine the predictors of digital piracy. Their factor structure explained 71 percent of the piracy intention, which sheltered attitude, past piracy behaviors, perceived behavior control, and moral obligation (i.e., feeling of guilt). Likewise, Taylor, Ishida, and Wallace (2009) resorted to the modified version of Perugini and Bagozzi's (2001) Model of Goal Directed Behavior, which was based on the TPB. The proposed structure was validated for both movie and music piracy. The model highlighted the importance of attitudes towards digital piracy, frequency of past piracy behaviors along with motivations and intentions underlying digital piracy. Theoretical frameworks challenging the arguments of the TPB have emerged as well. For instance, Jacobs, Heuvelman, Tan, and Peters (2012) reviewed several frameworks including the aforementioned ones, and came out with a refined model, which is based on a recent application of the Social Cognitive Theory (Bandura, 1986). The idea was that descriptive and prescriptive norms influenced deficient selfregulation, but did not have a direct influence on behavioral intentions (LaRose & Kim, 2007). This idea was in contrast to the TPB and consistent with the social cognitive theory. It was noted that e in addition to common antecedents e less conscious influences were on stage

Fig. 1. Illustration of the previous model by Nandedkar and Midha (2012).

296

Y. Akbulut / Computers & Education 78 (2014) 294e305

such as deficient self-regulation. Jacobs et al. (2012) focused on this idea and proposed their own model to explain the behaviors of illegal movie downloaders. Important factors in the model were the drive to view different and new movies, the social environment and attitudes towards the behavior, and the degree to which downloading has embedded itself in individuals' daily routine. Even though the explained variance values are relatively small in these emerging models, they are promising as they try to relieve the tension between producers and consumers through building a consensus between the two, and through attenuating the blame on end-users (Jacobs et al., 2012). The next section discusses individual hypotheses of the current study through resorting to relevant theoretical rationales and sample empirical works. 3. Hypotheses 3.1. Facilitating conditions Facilitating conditions are defined as factors in users' environment that facilitate the act of piracy (Nandedkar & Midha, 2012). Several studies revealed that the ease of engaging in digital piracy increased individuals' likelihood of engaging in such behaviors (Gupta, Gould, & Pola, 2004; Limayem et al., 2004). These findings have a theoretical explanation as well. The extension of the TRA by Ajzen and Fishbein (1985) predicts that individuals are more likely to engage in an activity that is easier to do as opposed to the ones that are more difficult. Thus, individuals' PC experiences may be considered as a personal facilitating condition. Accordingly, several studies revealed that one's perceived computer experience and competency predicted piracy behaviors (D'Astous, Colbert, & Montpetit, 2005; Gan & Koh, 2006). The pilot study of the current model proposal revealed a singularity problem between computer competencies and environmental facilitating conditions. Thus, the environmental conditions have been considered in particular, and the study was conducted at a research context where computer competencies were homogeneous across samples. 3.2. Optimism bias The next variable of interest was the optimism bias, whose influence on piracy was acknowledged by Nandedkar and Midha (2012). When individuals are given a chance to compare themselves to an average person, they tend to conclude that their futures are relatively better, and bad things are less likely to occur in their lives than that of an average person (Heine, 1993). That is, they show an unrealistic level of optimism about future life events (Weinstein, 1980), particularly about the negative ones (Gouveia & Clarke, 2001). Within the context of piracy, individuals tend to pirate unauthorized digital products because of their optimism bias, which leads them to regard the risk of confronting with negative consequences to be lower than others (Nandedkar & Midha, 2012). Both optimism bias and facilitating conditions are regarded as significant predictors of digital piracy. The current study posits that there can be a relationship between the two within the context of digital piracy. That is, environmental input may explain a certain amount of variation in optimism bias. A supporting argument appeared in the field of behavior genetics. That is, measurement of optimism, mental and self-rated health in 3053 twin individuals revealed that genetic factors explained less than half of the variation in optimism whereas the remainder being due to environmental influences (Mosing, Zietsch, Shekar, Wright, & Martin, 2009). A study on computer anxiety in Turkey also revealed that optimistic individuals tended to have less computer anxiety whereas pessimistic ones had high levels of anxiety (Ceyhan, 2006). Besides, differences between optimists and pessimists were consistent across different types of computer anxiety (i.e., affective anxiety, damaging anxiety, learning anxiety). Thus, the relationship between optimism and facilitating conditions may be sustained in further settings. Based on above arguments, the following research hypotheses evolve: H1. Facilitating conditions will be positively related to optimism bias. H2. Optimism bias will be positively related to users' piracy habits.

3.3. Previous experiences Studies on personality and social psychology revealed that if individuals engage in a specific behavior frequently, this makes it more likely that they will repeat that behavior in future (Aarts & Dijksterhuis, 2000). In essence, if a behavior is habitual, behavioral responses are more likely to be activated automatically. This idea has been theoretically justified through the self-perception theory of Bem (1967), which regards behavior as something individuals repeat over time. Another rationale comes from Beck and Ajzen (1991) who maintained that past behaviors are significant determinants of current unethical behaviors. Such a perspective has been tested successfully in the context of consumers' digital piracy behaviors (Cronan & Al-Rafee, 2008; Gupta et al., 2004; Taylor et al., 2009). Thus, H3. Prior piracy experiences will be positively related to users' current piracy habits.

3.4. Perceived risks The issue-risk-judgment (IRJ) model proposed by Tan (2002) integrated previous interrelating theories and hypotheses on ethical decision making with respect to software piracy. The model posits that the higher the magnitude of the negative consequence, the lower will be individuals' piracy acts. These perceived risks are categorized as financial, social, performance and prosecution risks. The pilot implementation of the current study revealed significant correlations among these four variables where the prosecution risk suppressed the others. Thus, it was decided to include prosecution risk in the current model. This decision was theoretically justified through a previous research in the same context, which maintained that the existence of computer ethic rules and regulations was negatively and significantly correlated with the reported tolerance towards illegal software use. On the other hand, the effect of other general ethical rules was not

Y. Akbulut / Computers & Education 78 (2014) 294e305

297

significant enough to explain the illegal use (Akman & Mishra, 2009). Thus, prosecution risk may be the prominent perceived risk to decrease digital piracy in Turkey, whereas other risks are not that dominant. Thus, H4. The perceived prosecution risk will be negatively related to user's attitudes towards digital piracy.

3.5. Habit, attitude, and intention Both the TRA and TPB maintained that behavioral intention is influenced by attitude. That is, favorable reactions toward a specific behavior are regarded as indicators of developing attitudes, which are generally correlated with behavior (Ajzen & Fishbein, 1980). In addition, aforementioned studies revealed positive relationships between habits and attitudes, and habits and intentions (Aarts & Dijksterhuis, 2000; Bem, 1967; Beck & Ajzen, 1991; Gupta et al., 2004; Nandedkar & Midha, 2012). Such relationships have been empirically confirmed in terms of music piracy. For instance, the intention to traffic music online depended on individuals' attitudes towards music piracy (D'Astous et al., 2005). In addition, individuals with a positive attitude toward piracy were found more likely to use pirated software in a recent empirical work (Gupta et al., 2004). Finally, the relationships among habit, attitude and intention were retained by Nandedkar and Midha (2012) as well. Thus, the following hypotheses evolve: H5. Current piracy habits will be positively related to attitudes towards piracy. H6. Current piracy habits will be positively related to behavioral intention. H7. Attitudes towards piracy will be positively related to behavioral intention.

4. Material and methods 4.1. Sample Demographic profile of the participants with regard to age, gender and computer use is summarized in Table 1. Of 1284 usable responses, 920 were males (71.65%) and 364 (28.35%) were females. Mean age of respondents were 22.9 with a standard deviation of 6.3, which represented a relatively young sample. However, this was not regarded as a confounding issue because the literature revealed that younger people were more likely to be involved in piracy (BSA, 2012; Gupta et al., 2004; Mishra et al., 2006; Rob & Waldfogel, 2006). Participants' perceived PC competencies did not vary across the samples whereas their daily Internet use did. That is, high school students' daily PC and Internet use was less than that of undergraduate students and adults. 4.2. Measures Nandedkar and Midha (2012) resorted to different scales regarding the current latent variables and proposed potential items to measure music piracy attitude, intention, habit, optimism bias, facilitating conditions and perceived risks. Items sheltered by each variable were generated in other studies. That is, items for attitude and intention towards piracy (Chiou et al., 2005), facilitating conditions (Limayem et al., 2004), optimism bias (Heine, 1993), and perceived risks (Tan, 2002) were borrowed from other studies. The current study resorted to the measures of Nandedkar and Midha (2012) after authors' written permission, adapted them to the current context, and added further items to measure previous digital piracy experiences. Items were translated and back-translated by two educational technology scholars. Two researchers of educational technology, an expert on value education and a guidance and counseling scholar reviewed the instrument in order to fine-tune the items and instructions in a way that the original content could be preserved and the legibility for the target group could be facilitated. A pilot study was conducted with 32 bilingual undergraduate students biweekly in which Turkish and English forms were administered successively. The implementation revealed Pearson r values of 0.75 or higher between the factors of each form. A second pilot implementation was realized with 559 undergraduate students who were conveniently reached from different departments. Even though all factors had internal consistency coefficients of 0.80 or higher, there were several non-adaptive or complex items with high error covariance values. Through an expert panel, complex items were examined and repetitive ones were eliminated. The data after these modifications revealed better fit indices (Table 2). Furthermore, the pilot implementation revealed that the most significant perceived risk was the prosecution risk; whereas social, financial or performance risks were secondary. Because all

Table 1 Profile of the participants. High school (n: 268)

Age PC competency/10 Internet competency/10 Daily PC use (Hours) Daily Internet use (Hours) Gender Female Male

Undergraduate (n: 610)

Adults (n: 406)

Mean

SD

Mean

SD

Mean

SD

17.15 7.51 7.46 3.68 3.53

2.47 2.65 2.70 1.89 1.89

21.63 7.32 7.42 4.28 4.07

2.85 2.43 2.45 1.86 1.87

28.60 6.98 7.16 4.26 4.00

7.25 2.57 2.59 2.06 2.04

n

%

n

%

n

%

86 182

32.09 67.91

181 429

29.67 70.33

97 309

23.89 76.11

298

Y. Akbulut / Computers & Education 78 (2014) 294e305

Table 2 CFA results of the pilot implementation. Index

Acceptable

First draft

Modified

Evaluation rationale

c2

N/A N/A 0  c2/df  3 0  RMSEA  0.08 0  SRMR  0.08 0.90  NFI  1.00 0.90  NNFI  1.00 0.90  CFI  1.00 0.90  GFI  1.00

2450.66 629 3.9a 0.066 0.092a 0.94 0.95 0.96 0.84a

1559.61 524 2.98 0.060 0.068 0.95 0.96 0.97 0.86a

N/A N/A Kline (2005), Sumer (2000) Hooper, Coughlan, and Mullen (2008) Brown (2006) Thompson (2004) Tabachnick and Fidell (2001) Tabachnick and Fidell (2001) Schumacker and Lomax (1996)

df

c2/df RMSEA SRMR NFI NNFI CFI GFI a

Beyond acceptable values.

perceived risks were very highly correlated (p < .001), the most prominent one (i.e., prosecution risk) was addressed in the current study in order to cope with multicollinearity and singularity problems (Pallant, 2007). Confirmation of the final scale across different samples was realized in the current study. The final form included 29 Likert items whose ratings ranged from 1 (strongly disagree) to 5 (strongly agree) except for the construct of optimism bias. The items in the construct of optimism bias addressed the likelihood of unwanted events on a 5-point scale ranging from 1 (much below average) to 5 (much above average). As suggested in previous successful implementations (Heine, 1993; Nandedkar & Midha, 2012), optimism bias score was calculated through subtracting 3 (same as an average person) from an individual's score. Thus, optimism bias can take values between 2 and þ2 and the signs (þ/) show the direction of the bias (Nandedkar & Midha, 2012). Variables and internal consistency coefficients, standardized coefficients of each item in the measurement models and error variances are summarized in Table 4. As illustrated in the table, each variable had sufficient number of items (i.e., at least 3) to sustain a robust measure (Tabachnick & Fidell, 2001), reflected ideal internal consistency coefficients (Huck, 2012) and acceptable factor loadings (Worthington & Whittaker, 2006). 4.3. Procedure The data collection tool was administered online. Through the permission of the manufacturers, links to the data collection tool were embedded in a popular social networking application, which had more than 500,000 active Turkish users per month. The links had specific prompts to invite high school students, undergraduate students and adults. Application members were invited to respond to the measures voluntarily. They were also free to withdraw from the study any time they wanted. These precautions may have reduced the response rate, but they were likely to increase the likelihood of obtaining genuine responses. It was believed that online administration could help the researcher to retrieve robust and more focused data (Baltar & Brunet, 2012), since some participants in formal school settings may not be active computer users. In other words, as adults were not in formal educational settings, a more relevant platform with regard to the computer use habits can be accessed through online administration. In order to reduce the social desirability response bias, the term ‘piracy’ was not explicitly used in scale instructions. The data collection was continued till sufficient number of respondents were accessed in each dataset. For instance, after reviewing different sources regarding the sample size in confirmatory factor analyses, Worthington and Whittaker (2006) recommended a minimum of five cases per parameter to be estimated whereas the 10:1 ratio was considered optimal. Thus, the data collection procedure continued till a minimum of five cases per item were accessed in each group. After the data were collected, bad or missing data were eliminated through checking the validation questions. This was deliberately done since deceptive self-presentation in online survey administrations could have a contaminating influence on responses (Castro, 2013; Chesney & Penny, 2013). The procedure helped the researcher to eliminate several faking participants such as 100-year-old high school students or 11-year-old civil engineers. In addition, participants who answered all questions with the same pattern were excluded. The elimination led the researcher to ignore 38 participants from university students (5.86%), 14 participants from high school students (4.96%), and 21 participants from adults (4.92%). Since the participation was voluntary, the ratio of the bad data was not high. After the three datasets were prepared, internal consistency coefficients and correlations were calculated through IBM SPSS Statistics 21. Furthermore, the measurement and structural models were tested through LISREL 9.01 across samples.

Table 3 Evaluation of the measurement model across samples. Index

Acceptable

High school

Undergraduate

Adults

Evaluation rationale

c2

N/A N/A 0  c2/df  3 0  RMSEA  0.08 0  SRMR  0.08 0.90  NFI  1.00 0.90  NNFI  1.00 0.90  CFI  1.00 0.90  GFI  1.00

748.29 356 2.102 0.064 0.070 0.94 0.96 0.97 0.85a

1184.61 356 3.328a 0.062 0.073 0.96 0.97 0.97 0.88a

1114.61 356 3.131a 0.072 0.077 0.94 0.95 0.96 0.84a

N/A N/A Kline (2005), Sumer (2000) Hooper et al. (2008) Brown (2006) Thompson (2004) Tabachnick and Fidell (2001) Tabachnick and Fidell (2001) Schumacker and Lomax (1996)

df

c /df 2

RMSEA SRMR NFI NNFI CFI GFI a

Beyond acceptable values due to sample size and high number of parameters.

Y. Akbulut / Computers & Education 78 (2014) 294e305

299

Table 4 Factors, items and the summary of the measurement model across samples. Variable

Attitude

Intention (I will …)

Habit

Optimism bias (I'm likely to …)

Facilitating conditions

Prosecution risk

Prior experiences

Sample

High school (n: 268)

Undergraduate (n: 610)

Adults (n: 406)

Items

Loading

Error

Loading

Error

Loading

Error

Downloading/duplicating unauthorized software is good … pleasant … useful … attractive Alpha … duplicate them without hesitation. … download them without hesitation. … duplicate them for family and friends. … download them for family and friends. … encourage family and friends to download them. Alpha Downloading/duplicating them is automatic for me. I do not even think twice before downloading/duplicating them. The number of products I downloaded/duplicated is high. I have recently downloaded/duplicated them. Alpha … contract cancer. … have an unhappy family life. … have an unhappy working life. … drop out of school/quit work. … have an early heart attack. … become senile with old age. Alpha inappropriate copyright measures lack of awareness campaigns knowing how to access unauthorized software products Alpha If I download them, I will be caught for the infringement of law. If I duplicate them, I will be caught for the infringement of law. If I use them, I will be caught for the infringement of law. Alpha I used to download/duplicate unauthorized products. I used to share them on the Web. I used to download/duplicate not-genuine products. I used to share not-genuine products on the Web. Alpha

0.82 0.79 0.73 0.69 0.84 0.75 0.85 0.82 0.8 0.68 0.88 0.77 0.85

0.33 0.37 0.46 0.53

0.88 0.87 0.79 0.68 0.88 0.75 0.86 0.84 0.89 0.72 0.91 0.81 0.81

0.22 0.25 0.38 0.54

0.85 0.88 0.73 0.51 0.83 0.85 0.83 0.86 0.82 0.75 0.91 0.8 0.84

0.27 0.23 0.46 0.74

0.8 0.73 0.86 0.67 0.72 0.79 0.56 0.72 0.74 0.85 0.85 0.86 0.41 0.73 0.81

0.36 0.47

0.37 0.52

0.25

0.78 0.75 0.87 0.69 0.62 0.7 0.6 0.77 0.76 0.85 0.88 0.91 0.26 0.66 0.91

0.39 0.44

0.34

0.79 0.69 0.86 0.71 0.73 0.7 0.57 0.74 0.67 0.84 0.81 0.91 0.28 0.66 0.87

0.87

0.24

0.86

0.25

0.92

0.15

0.88 0.89 0.75 0.82 0.81 0.8 0.87

0.22

0.92 0.91 0.81 0.88 0.88 0.84 0.91

0.15

0.93 0.94 0.81 0.89 0.85 0.83 0.91

0.14

0.44 0.28 0.33 0.36 0.53 0.41 0.27

0.56 0.48 0.37 0.68 0.48 0.45 0.28 0.26 0.83

0.43 0.33 0.34 0.36

0.44 0.27 0.29 0.21 0.47 0.34 0.35

0.49 0.47 0.51 0.67 0.45 0.56 0.34 0.17 0.92

0.34 0.23 0.23 0.3

0.28 0.32 0.26 0.33 0.43 0.37 0.3

0.52 0.62 0.51 0.64 0.4 0.43 0.23 0.17 0.93 0.17

0.34 0.21 0.28 0.31

5. Results 5.1. Measurement model The evaluation of the current measurement model across samples is provided in Table 3. In addition, item statistics are provided in Table 4. The total variance explained by the complete measurement model was 59.29% for the high school group, 61.99% for undergraduates and 62.47% for adults. In addition, explained variance values for individual sub-scales ranged from 57% to 86%. Almost all fit indices were within the acceptable limits to confirm the factor structure. The ratio of df to chi square was beyond acceptable limits in two large samples (i.e., undergraduates and adults). However, since the chi-square statistic is sensitive to sample size and tends to reject models with large samples, € reskog & So € rbom, 1993). Besides, the current cutoff point for c2/df this was not considered a serious deviation from an acceptable model (Jo was determined as three, even though there were recommendations in the literature to accept values below five (Sumer, 2000; Wheaton, Muthen, Alwin, & Summers, 1977). The GFI is also sensitive to the magnitude of samples and number of parameters. The index has become less popular in recent years (Steiger, 2007) and there have been recommendations not to use the GFI at all (Sharma, Mukherjee, Kumar, & Dillon, 2005). Aside from these, other indices were acceptable. For instance, even if the stringer upper limit for the RMSEA was considered as 0.07 (Steiger, 2007), most values in the measurement and structural models met this criteria. Besides, most of the fit statistics (CFI, NFI, NNFI) were above the cut-off value of a perfect fit (i.e., 95; Tabachnick & Fidell, 2001; Thompson, 2004). 5.2. Structural model Correlations among the variables of interest are provided in Table 5. Similar relationships among the variables were observed across the groups. Proposing unique structural models for each sample might be possible through resorting to these relationships. However, a model to embrace all three samples was sought for. The model is illustrated in Figs. 2e4. The proposed structural equation model repeated the same pattern with regard to fit indices as illustrated in Table 6. That is, c2/df ratio was slightly problematic in two large samples. Regardless, most fit indices were plausible and all proposed paths among the latent variables had statistically significant t values. Summary of individual hypotheses are provided for each sample as follows:

300

Y. Akbulut / Computers & Education 78 (2014) 294e305

Table 5 Correlations among the latent variables. Variable

Sample

Intention

Habit

Optimism

Facilitating conditions

Prosecution risk

Prior experiences

Attitude

HS U A HS U A HS U A HS U A HS U A HS U A

0.653*** 0.681*** 0.657*** e

0.568*** 0.600*** 0.612*** 0.752*** 0.790*** 0.813*** e

0.275*** 0.179*** 0.057 0.300*** 0.139*** 0.102* 0.326*** 0.178*** 0.199*** e

0.217*** 0.039 0.134** 0.214*** 0.076 0.138** 0.261*** 0.123*** 0.184*** 0.264*** 0.335*** 0.353*** e

0.024 0.170*** 0.078 0.171** 0.179*** 0.214*** 0.206*** 0.158*** 0.187*** 0.120* 0.228*** 0.213*** 0.419*** 0.431*** 0.419*** e

0.457*** 0.488*** 0.406*** 0.563*** 0.594*** 0.581*** 0.545*** 0.586*** 0.570*** 0.349*** 0.180*** 0.173*** 0.103 0.018 0.112* 0.080 0.179*** 0.132**

Intention

Habit

Optimism

Facilitating conditions

Prosecution risk

*p < .05; **p < .01; ***p < .001; HS: High school students; U: Undergraduates; A: Adults.

H1. The effect of facilitating conditions on optimism was significant for high school students (b ¼ 0.29; t ¼ 4.01; p < .001), undergraduate students (b ¼ 0.35; t ¼ 7.24; p < .001), and adults (b ¼ 0.35; t ¼ 6.11; p < .001). H2. The effect of optimism bias on habit was significant for high school students (b ¼ 0.21; t ¼ 3.52; p < .001), undergraduate students (b ¼ 0.11; t ¼ 3.00; p < .01), and adults (b ¼ 0.13; t ¼ 2.88; p < .01). H3. The effect of prior experiences on habit was significant for high school students (b ¼ 0.59; t ¼ 8.77; p < .001), undergraduate students (b ¼ 0.65; t ¼ 14.80; p < .001), and adults (b ¼ 0.61; t ¼ 11.55; p < .001). H4. The negative effect of prosecution risk on habit was significant for high school students (b ¼ 0.19; t ¼ 3.29; p < .001), undergraduate students (b ¼ 0.11; t ¼ 2.85; p < .01), and adults (b ¼ 0.17; t ¼ 3.76; p < .001). H5. The effect of habit on attitude was significant for high school students (b ¼ 0.64; t ¼ 8.83; p < .001), undergraduate students (b ¼ 0.68; t ¼ 15.55; p < .001), and adults (b ¼ 0.66; t ¼ 11.99; p < .001). H6. The effect of habit on intention was significant for high school students (b ¼ 0.64; t ¼ 8.23; p < .001), undergraduate students (b ¼ 0.73; t ¼ 14.24; p < .001), and adults (b ¼ 0.80; t ¼ 13.70; p < .001). H7. The effect of attitude on intention was significant for high school students (b ¼ 0.31; t ¼ 4.78; p < .001), undergraduate students (b ¼ 0.22; t ¼ 5.68; p < .001), and adults (b ¼ 0.18; t ¼ 3.96; p < .001). In brief, seven hypotheses were retained across three samples. Details of the structural equations evolving from the current analyses are provided in Table 7. In essence, the contribution of the current model to predict digital piracy intention range from 77 percent to 86 percent across different age groups. Therefore, the model looked acceptable and consistent across three samples.

5.3. Alternative models The original model retained by Nandedkar and Midha (2012) could not be retained with the current samples (RMSEA> 0.083; c2/ df > 4.82; CFI & IFI < 0.90; SRMR > 0.13). However, as the fit values were somewhat close to acceptable limits, further administrations after modification indices may retain the original model in Turkey. Alternative models that had acceptable fit values across samples are illustrated in Figs. 5e7. However, none of them were robust enough to accommodate all current variables that were determined through the review of contemporary literature. These models are provided here

Fig. 2. Results of the structural equation model for high school students.

Y. Akbulut / Computers & Education 78 (2014) 294e305

301

Fig. 3. Results of the structural equation model for university students.

so that future scholars may adopt the one that is most relevant to their unique research and implementation contexts. In Fig. 5, prior experiences and prosecution risk simultaneously predicted digital piracy intention through the full mediation of attitude (RMSEA < 0.066; c2/ df < 3.18; CFI & IFI > 0.90; SRMR < 0.089). In Fig. 6, optimism bias is added to the previous equation, which slightly reduces the fit values (RMSEA < 0.067; c2/df < 3.21; CFI & IFI > 0.89; SRMR < 0.094). Finally, Fig. 7 provides a more complicated model where optimism bias, prior experiences and prosecution risk predicted attitude, intention and habit (RMSEA < 0.063; c2/df < 2.96; CFI & IFI > 0.91; SRMR < 0.087). The model was statistically strong, but facilitating conditions could not be integrated anywhere in the model. 6. Discussion and conclusive remarks 6.1. Discussion Findings revealed a measurement model and a structural model to predict digital piracy intentions, which were consistent across three samples. The 29-item measure sheltered seven latent variables: Facilitating conditions, optimism bias, previous experiences, prosecution risk, current habits, attitudes towards piracy and behavioral intention to conduct piracy in future. The structural equation model revealed the influence of facilitating conditions on optimism. Then, optimism bias, prior piracy experiences and prosecution risk predicted current piracy habits, which further influenced both behavioral attitude and intention regarding digital piracy. The positive relationship between attitude and intention was retained as well. Hence, the current variables which were determined in accordance with the TRA and TPB were robust enough to explain behavioral intentions towards digital piracy. There were several similarities and differences between the current model and the model proposed by Nandedkar and Midha (2012). Both models underlined the importance of the relationships among piracy habits, attitudes and intentions. In addition, both models embraced optimism bias and perceived risks to explain digital piracy. One of the difference of the current model is the particular influence of the prosecution risk on eradicating piracy habits. Unfortunately, rather than the impact of general ethical rules, strict regulations are found more plausible to eliminate digital piracy in Turkey, which was previously suggested in the literature (Akman & Mishra, 2009). This is an unfortunate implication, which means that ethical conduct can be sustained in the current context through strict precautions rather than the self-conscience of individuals or peer-pressure. However, such a speculation needs to be re-tested through developing alternative measures and administering them in larger contexts to see the proposed influence of financial, social and performance risks on ethical decision making (Tan, 2002). The next contribution of the current model was the predictive role of facilitating conditions and optimism bias on piracy habits. Nandedkar and Midha (2012) already proposed a path between facilitating conditions and attitude, but their data could not retain this hypothesis. The current work retained a path between facilitating conditions and optimism bias, which further predicted piracy habits. That is, the ease of engaging in digital piracy was related with individuals' piracy habits (Gupta et al., 2004; Limayem et al., 2004). In Turkey, even strict precautions may empower facilitating conditions. For instance, knowing how to access unauthorized digital products is a common facilitating condition, which increases digital piracy (D'Astous et al., 2005; Gan & Koh, 2006). Recent strict precautions adopted by the Turkish government such as blocking Twitter and YouTube somewhat triggered such facilitating conditions. More specifically, Turkish users learned how to change their DNS settings or to set up anonymous VPN services (i.e., virtual private network) to access forbidden sites. These

Fig. 4. Results of the structural equation model for adults.

302

Y. Akbulut / Computers & Education 78 (2014) 294e305

Table 6 Evaluation of the structural model across samples. Index

Acceptable

High school

Undergraduate

Adults

Evaluation rationale

c2

N/A N/A 0  c2/df  3 0  RMSEA  0.08 0  SRMR  0.10 0.90  NFI  1.00 0.90  NNFI  1.00 0.90  CFI  1.00 0.90  GFI  1.00

812.75 367 2.215 0.067 0.108 0.93 0.96 0.96 0.83a

1247.47 367 3.399a 0.063 0.088 0.96 0.97 0.97 0.87a

1157.05 367 3.153a 0.073 0.086 0.94 0.95 0.96 0.83a

N/A N/A Kline (2005), Sumer (2000) Hooper et al. (2008) Kline (2005) Thompson (2004) Tabachnick and Fidell (2001) Tabachnick and Fidell (2001) Schumacker and Lomax (1996)

df

c /df 2

RMSEA SRMR NFI NNFI CFI GFI a

Beyond acceptable values due to sample size and high number of parameters.

are high-level technical skills, which can take considerable time to teach in our undergraduate courses, but strict precautions led most people to learn such skills incidentally. Finally, forbidden things are more attractive as they are intriguing. Thus, value-related precautions such as awareness raising campaigns should be assumed for effective results rather than strict prohibitions and prosecutions, which can trigger piracy further. The final contribution of the current model was the involvement of prior experiences in predicting current piracy habits. Individuals who used to engage in specific behaviors are more likely to repeat that behavior in future (Aarts & Dijksterhuis, 2000), which was retained in terms of the digital piracy in international studies (Cronan & Al-Rafee, 2008; Gupta et al., 2004; Taylor et al., 2009). The current study retained this hypothesis in Turkey. Unfortunately, the lack of education and awareness campaigns on computer ethics (Namlu & Odabasi, 2007), the lack of proper precautions (Beycioglu, 2009) and the abundance of facilitating conditions (BSA, 2012) may convert current digital piracy instances into habitual behaviors in future, which requires immediate action. 6.2. Practical implications Reducing piracy rates is a difficult endeavor particularly because of the increasing demand for software. Still, large scale confirmation of current structural models may provide practitioners with valuable suggestions. Revisiting the suggestions of Nandedkar and Midha (2012) is quite plausible considering the similarity of their model with the current model with regard to the relationships among habits, attitudes and intentions. First of all, they call for orientation programs to enhance individuals' thinking and behavior. Such an implication is quite relevant in the current context regarding the lack of relevant education on computer ethics in Turkey (Namlu & Odabasi, 2007) and the interrelationships among ethical codes, attitudes, training and awareness (Akman & Mishra, 2009). Second, making digital products available through inexpensive rates may reduce the digital piracy. Such a practical precaution is consistent with the theoretical framework of Jacobs et al. (2012), since it can attenuate the blame on end-users and relieve the tension between manufacturers and consumers through building a peaceful consensus. In addition to above practical suggestions, unique implications emerge from the current structural model. First of all, facilitating conditions contributed to current piracy habits. In this regard, policy makers need to revisit current copyright measures and regulations, and organize awareness campaigns to reduce piracy rates. Second, prosecution risk was the only variable which was negatively related to piracy habits. This finding should not be used by policy makers as an excuse to impose further penalties. Rather, the focus on value education and awareness campaigns should be empowered so that individuals refrain from misconduct through their own will rather than an external pressure. Within the scope of the current theoretical framework, several alternative equations were tested and illustrated. Statistical analyses suggested that it was possible to propose unique models for each sample through slight modifications in the current paths. However, proposing a model which was applicable to different populations was found more plausible. Such a consistency may guide policy makers while determining the standards for coping with digital piracy. While adapting this consistent structure to other contexts, the assumptions of the TPB and the TRA can be used to explain the nature of behavioral intentions towards digital piracy. 6.3. Limitations and recommendations for further research One of the limitations of the study is the high proportion of male participants in the sample. A completely different pattern could be observed in female-dominated settings. In this regard, replicating the study may reveal different results in a context where the gender Table 7 Structural equations for each sample and variable. Variable

Sample

Equation

Attitude

High school University Adults High school University Adults High school University Adults High school University Adults

0.638 0.676 0.662 0.314 0.225 0.178 0.211 0.114 0.134 0.290 0.346 0.355

Intention

Habit

Optimism

           

Habit Habit Habit Attitude þ 0.641  Habit Attitude þ 0.734  Habit Attitude þ 0.797  Habit Optim  0.190  Prosec þ 0.592  Prior Optim  0.105  Prosec þ 0.648  Prior Optim  0.168  Prosec þ 0.613  Prior Facilitating conditions Facilitating conditions Facilitating conditions

Error

R2

0.593 0.543 0.562 0.235 0.188 0.145 0.558 0.543 0.558 0.916 0.88 0.874

0.407 0.457 0.438 0.765 0.812 0.855 0.442 0.457 0.442 0.084 0.12 0.126

Y. Akbulut / Computers & Education 78 (2014) 294e305

303

Fig. 5. Alternative model 1.

distribution represented a different pattern. However, the current sample's being young and male-dominated can be regarded as an advantage as well. That is, recent reports indicate that unethical behaviors are more common among young and male populations (BSA, 2012; Lau & Yuen, 2014). Thus, the current sample may be considered more relevant and more representative of the world's current shadow market with regard to piracy. The statistical consistency of the current model across three different samples should be considered with caution. That is, the current model may not be that consistent in different settings other than online platforms. More specifically, the current research context was a unique social networking site where users could be too homogeneous with regard to critical variables examined in the current study, and with regard to critical covariates such as PC use habits. This comment is valid for the proposed alternative models as well. These models were not considered plausible, because none of them could accommodate current research variables simultaneously. Changing the research context and integrating further constructs into the model may empower their validity in further research. Further researchers should also consider the limited coverage of current factors and items. Even though each factor had sufficient number of items and plausible fit indices to measure the desired construct, they were abbreviated forms of the original measures implemented in the literature. Besides, perceived risk was evaluated only in terms of the prosecution risk, since other perceived risks (e.g., social, financial, performance) was not considered significant in the pilot implementation. Particularly the social risk factor can be investigated in further research to investigate the influence of the social proximity to pirates on unethical downloading and duplicating behaviors (Tan, 2002). Thus, developing further items for the current constructs and creating robust indicators to measure other types of perceived risks can be a plausible step to fine-tune the current proposal. The contribution of the model for each latent variable varied slightly across the samples. In addition, even though the whole model looked acceptable across different age groups consistently, it seems that the robustness of the model is influenced by age. That is, the explained variance of the structural models increased in accordance with age (i.e., high school students: 77%, undergraduates: 81%, adults:

Fig. 6. Alternative model 2.

304

Y. Akbulut / Computers & Education 78 (2014) 294e305

Fig. 7. Alternative model 3.

86%). Thus, age appears to be a significant predictor of piracy intention as suggested in the literature (Mishra et al., 2006). In this regard, further works may need to include age as a covariate of the dependent variable. An interesting addition to the model can be realized through scrutinizing on further contextual variables that predict piracy. Inclusion of the perceived risk and facilitating conditions contributed to this purpose to some extent; however, the items were too general and far from addressing the unique organizational culture, which may either facilitate or impede piracy. Indeed, previous studies showed that ethical decision making varied according to unique scenarios or group dynamics (Akman & Mishra, 2009; Haines & Leonard, 2007; Leonard & Haines, 2007). Therefore, single issue approaches similar to the current study may be inadequate for describing ethical decision-making. In this regard, in addition to the social context mentioned above, the organizational environment should be taken into account in future works where different scenarios of piracy and different organizational interactions were investigated. Further individual variables can be included in the model. For instance, individuals' current technology competencies may be a facilitator of piracy. However, all PC and Internet use variables addressed in the background questionnaire correlated positively with facilitating conditions. Thus, these variables were excluded since they created singularity problems. Another alternative is to enhance the model through the inclusion of tangible and encouraging factors. That is, tangible outcomes of being unethical may be quite effective in increasing piracy. Recent work retained this idea and maintained that promotions may lead individuals to be more likely to act unethically than preventions (Gino & Margolis, 2011). Finally, the low level of anticipated guilt toward illegal downloading has been reported as a predictor of piracy in recent research (Cronan & Al-Rafee, 2008; Wang & McClung, 2012), which can be integrated into future models. In short, finetuning current scales and variables may be helpful to understand the influence of such individual and contextual variations on digital piracy.

References Aarts, H., & Dijksterhuis, A. (2000). Habits as knowledge structures: automaticity in goal-directed behavior. Journal of Personality and Social Psychology, 78(1), 53e63. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice Hall. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179e211. Ajzen, I., & Fishbein, M. (1985). From intentions to actions: a theory of planned behavior. In J. Hugland, & J. Beckman (Eds.), Action-control: From cognitions to behavior (pp. 11e39). Heidelberg: Springer. Akman, I., & Mishra, A. (2009). Ethical behavior issues in software use: an analysis of public and private sectors. Computers in Human Behavior, 25, 1251e1257. Aricak, O. T. (2009). Psychiatric symptomatology as a predictor of cyberbullying among university students. Eurasian Journal of Educational Research, 34, 167e184. Baltar, F., & Brunet, I. (2012). Social research 2.0: virtual snowball sampling method using Facebook. Internet Research, 22(1), 57e74. Bandura, A. (1986). Social foundations of thoughts and action. Englewood Cliffs, NJ: PrenticeeHall. Beck, L., & Ajzen, I. (1991). Predicting dishonest actions using the theory of planned behavior. Journal of Research in Personality, 25(3), 285e301. Bem, D. J. (1967). Self-perception: an alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74(3), 183e200. Beycioglu, K. (2009). A cyberphilosophical issue in education: unethical computer using behavior e the case of prospective teachers. Computers & Education, 53(2), 201e208. Bhattacharjee, S., Gopal, R. D., & Sanders, L. (2003). Digital music and online sharing: software piracy 2.0? Communications of the ACM, 46(7), 107e111. Brown, T. A. (2006). Confirmatory factor analysis for applied research. NY: Guilford Press. Business Software Alliance (BSA). (2012). Shadow market: 2011 BSA global software piracy study (9th ed.) Retrieved May 16, 2014 from http://globalstudy.bsa.org/2011/ downloads/study_pdf/2011_BSA_Piracy_Study-Standard.pdf. Castro, R. (2013). Inconsistent respondents and sensitive questions. Field Methods, 25(3), 283e298. Ceyhan, E. (2006). Computer anxiety of teacher trainees in the framework of personality variables. Computers in Human Behavior, 22(2), 207e220. Ceyhan, A. A. (2008). Predictors of problematic Internet use on Turkish university students. CyberPsychology & Behavior, 11(3), 363e366. Ceyhan, A. A., & Ceyhan, E. (2008). Loneliness, depression and computer self-efficacy as predictors of problematic internet use. CyberPsychology & Behavior, 11(6), 699e701. Chesney, T., & Penny, K. (2013). The impact of repeated lying on survey results. JanuaryeMarch 2013 SAGE Open, 1e9. Chiou, J. S., Huang, C. Y., & Lee, H. H. (2005). The antecedents of music piracy attitude and intentions. Journal of Business Ethics, 57(2), 161e174. Cronan, T. P., & Al-Rafee, S. (2008). Factors that influence the intention to pirate software and media. Journal of Business Ethics, 78(4), 527e545. Çuhadar, C. (2012). Exploration of problematic Internet use and social interaction anxiety among Turkish pre-service teachers. Computers & Education, 59(2), 173e181. D'Astous, A., Colbert, F., & Montpetit, D. (2005). Music Piracy on the web e how effective are anti-piracy arguments? Evidence from theory of planned behavior. Journal of Consumer Policy, 28, 289e310. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory and research. Boston: Addison Wesley. Gan, L. L., & Koh, H. C. (2006). An empirical study of software piracy among tertiary institutions in Singapore. Information & Management, 43, 640e649. Gino, F., & Margolis, J. D. (2011). Bringing ethics into focus: how regulatory focus and risk preferences influence (Un)ethical behavior. Organizational Behavior and Human Decision Processes, 115, 145e156. Gouveia, S. O., & Clarke, V. (2001). Optimistic bias for negative and positive events. Health Education, 101(5), 228e234.

Y. Akbulut / Computers & Education 78 (2014) 294e305

305

Gupta, P. B., Gould, S. J., & Pola, B. (2004). To pirate or not to pirate: a comparative study of the ethical versus other influences on the consumer's software acquisition-mode decision. Journal of Business Ethics, 55(3), 255e274. Haines, R., & Leonard, L. N. K. (2007). Situational influences on ethical decision-making in an IT context. Information & Management, 44, 313e320. Heine, S. J. (1993). Cultural variation in unrealistic optimism. Unpublished master's thesis. Vancouver, B.C., Canada: The University of British Columbia. Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural equation modelling: guidelines for determining model fit. The Electronic Journal of Business Research Methods, 6(1), 53e60. Huck, S. W. (2012). Reading statistics and research (6th ed.). Boston: Pearson. Jacobs, R. S., Heuvelman, A., Tan, M., & Peters, O. (2012). Digital movie piracy: a perspective on downloading behavior through social cognitive theory. Computers in Human Behavior, 28, 958e967. €reskog, K., & So €rbom, D. (1993). LISREL 8: Structural equation modeling with the SIMPLIS command language. Chicago, IL: Scientific Software International Inc. Jo Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). NY: Guilford Press. LaRose, R., & Kim, J. (2007). Share, steal, or buy? A social cognitive perspective of music downloading. CyberPsychology & Behavior, 10(2), 267e277. Lau, W. W. F., & Yuen, A. H. K. (2014). Internet ethics of adolescents: understanding demographic differences. Computers & Education, 72, 378e385. Leonard, L. N. K., & Haines, R. (2007). Computer-mediated group influence on ethical behavior. Computers in Human Behavior, 23, 2302e2320. Limayem, M., Khalifa, M., & Chin, W. W. (2004). Factors motivating software piracy: a longitudinal study. IEEE Transactions on Engineering Management, 51(4), 414e425. Mishra, A., Akman, I., & Yazici, A. (2006). Software piracy among IT professionals in organizations. International Journal of Information Management, 26(5), 401e413. Mosing, M. A., Zietsch, B. P., Shekar, S. N., Wright, M. J., & Martin, N. G. (2009). Genetic and environmental influences on optimism and its relationship to mental and self-rated health: a study of aging twins. Behavior Genetics, 39(6), 597e604. Namlu, A. G., & Odabasi, H. F. (2007). Unethical computer using behavior scale: a study of reliability and validity on Turkish university students. Computers & Education, 48, 205e215. Nandedkar, A., & Midha, V. (2012). It won't happen to me: an assessment of optimism bias in music piracy. Computers in Human Behavior, 28, 41e48. Pallant, J. (2007). SPSS survival manual: A step by step guide to data analysis using SPSS for windows (3rd ed.). Maidenhead: McGraweHill Education. Perugini, M., & Bagozzi, R. P. (2001). The role of desires and anticipated emotions in goal-directed behaviours: broadening and deepening the theory of planned behaviour. British Journal of Social Psychology, 40, 79e98. Prasad, A., & Mahajan, V. (2003). How many pirates should a software firm tolerate? an analysis of piracy protection on the diffusion of software. International Journal of Research in Marketing, 20, 337e353. Rob, R., & Waldfogel, J. (2006). Piracy on the high C's: music downloading, sales displacement, and social welfare in a sample of college students. Journal of Law and Economics, 49, 29e62. Schumacker, R. E., & Lomax, R. G. (1996). A beginner's guide to structural equation modeling. NJ: Lawrence Erlbaum Associates. Sharma, S., Mukherjee, S., Kumar, A., & Dillon, W. R. (2005). A simulation study to investigate the use of cutoff values for assessing model fit in covariance structure models. Journal of Business Research, 58(1), 935e943. Steiger, J. H. (2007). Understanding the limitations of global fit assessment in structural equation modeling. Personality and Individual Differences, 42(5), 893e898. Sumer, N. (2000). Yapısal esitlik modelleri [Structural equation models]. Turkish Psychological Articles, 3(6), 49e74.
, S., Duran, M., & Fraser, M. R. (2012). Surveying the extent of involvement in online academic dishonesty (e-dishonesty) related practices among university students S¸endag and the rationale students provide: one university's experience. Computers in Human Behavior, 28(3), 849e860. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). MA: Allyn & Bacon. Tan, B. (2002). Understanding consumer ethical decision making with respect to purchase of pirated software. Journal of Consumer Marketing, 19(2), 96e111. Taylor, S. A., Ishida, C., & Wallace, D. W. (2009). Intention to engage in digital piracy: a conceptual model and empirical test. Journal of Service Research, 11(3), 246e262. Tekinarslan, E. (2008). Computer anxiety: a cross-cultural comparative study of Dutch and Turkish university students. Computers in Human Behavior, 24(4), 1572e1584. Thompson, B. (2004). Exploratory and confirmatory factor analysis: Understanding concepts and applications. Washington: American Psychological Association. Wang, X., & McClung, S. R. (2012). The immorality of illegal downloading: the role of anticipated guilt and general emotions. Computers in Human Behavior, 28, 153e159. Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and Social Psychology, 39(5), 806e820. Wheaton, B., Muthen, B., Alwin, D. F., & Summers, G. (1977). Assessing reliability and stability in panel models. Sociological Methodology, 8(1), 84e136. Worthington, R. L., & Whittaker, T. A. (2006). Scale development research: a content analysis and recommendations for best practices. The Counseling Psychologist, 34, 806e838.