Faculty of Social and Behavioural Sciences
Graduate School of Child Development and Education
Risk Factors for Juvenile Cybercrime:
A Comprehensive Meta-Analytic Review
Research Master Child Development and Education Research Master Thesis
Joyce Standaert (10503609) Supervisor: Inge Wissink January 27th, 2019
Highlights (max. 125 characters)
• This is the first meta-analysis on risk factors for juvenile cybercrime perpetration. • The three most common juvenile cybercrimes were investigated: cyberstalking,
sexting, and hacking.
• A total of 48 articles and 903 effect sizes on (potential) risk factors for cybercrime perpetration were included.
• Prior online and offline deviant behavior seem to be the most important risk factors for cybercrime perpetration.
• Having deviant peers was also identified as an important risk factor for cybercrime perpetration.
Abstract
The past global rise of computer, smartphone, and internet use has been
accompanied with a rise of cyber offenses. For effective (preventive) intervention, it is crucial to determine the risk factors for cyber offenses. The aim of the present meta-analytic review was to identify risk factors for the perpetration of three juvenile cyber offenses: cyberstalking, hacking, and sexting. Forty-eight articles were included from which 903 effect sizes were extracted. Each effect size represented the impact of a (potential) risk factor for cyberstalking (306 effect sizes), sexting (61 effect sizes), or hacking (536 effect sizes). The results showed both similarities and differences between risk factors for the three types of cybercrime. Having deviant peers was positively related to perpetration of all three types of cybercrime, whereas prior online or offline deviant behavior were only positively related to cyberstalking and hacking. Having dark personality traits was identified as a risk factor for cyberstalking and sexting. Further, having low norms and beliefs is a risk factor for both hacking and cyberstalking. Low self-control was only a risk factor for hacking. It was concluded that a substantial number of risk factors can be identified for cybercrime, but that differences exist in risk factors for the different cybercrimes. However, there may be more similarities in risk factors for the cybercrime types, as the included studies did not examine the same variables as risk factors for cyberstalking, sexting, and hacking. Especially research on risk factors for sexting is lacking. Implications for (preventive) intervention are discussed.
Risk Factors for Juvenile Cybercrime: A Comprehensive Meta-Analytic Review
In the past two decades there has been a global rise of computer, smartphone, and internet use. Juveniles often start using a personal smartphone or computer at an early age (Lee, 2018; Madden et al., 2013). These developments have great advantages, but also bring along new challenges, such as how to deal with individuals who show unacceptable behavior or even commit offenses online, and how to prevent such a developmental path. Since new technologies have been developed and have become more innovative, new deviant behavior and new cybercrimes have come forward (Bocij & McFarlane, 2002). Cybercrimes can be divided into cyber-dependent crimes and cyber-enabled crimes (Wall, 2015). Cyber-dependent crimes are dependent upon technology, meaning that the crime would not have existed without the technology. On the other hand, cyber-enabled crimes are traditional crimes that already existed before the cybertechnological
developments, but they can now be performed at a larger scale and in a different form by using cybertechnology (Wall, 2015). In this study, hacking was examined as a form of cyber-dependent crime, whereas cyberstalking and sexting were examined as forms of cyber-enabled criminal behaviors.
The three cybercrimes that are examined in this meta-analysis (cyberstalking, sexting, hacking) are considered to be illegal and prosecutable according to the current Dutch criminal code (Wetboek van Strafrecht, article 285b, 240b, and 138ab). Definitions of cyberstalking, sexting, and hacking may change in reaction to ongoing developments. In this study, the definition of Reyns, Henson, and Fisher (2012) is used, because it is based on the latest developments. Reyns and colleagues (2012) define cyberstalking as
“the repeated pursuit of an individual using electronic or internet-capable devices” (p. 1). Cyberstalking refers to different behaviors, such as sending or posting offensive or false messages, harassing, stealing and using the identity of the victim and acting to be somebody else (Bocij & McFarlane, 2002; Finn, 2004; Sheridan & Grant, 2007). Cyber dating abuse is often used interchangeably with cyberstalking and is commonly measured in the same manner as cyberstalking, although in the context of a dating relationship. Therefore, studies on cyber dating abuse were also included in the present review. Sexting is commonly defined as: “the sending, receiving, or forwarding of sexually explicit messages, images, or photos to others through electronic means, primarily between cellular phones” (Klettke, Halford, & Mellor, 2014, p. 45). As sending and receiving sexts is not illegal in the Netherlands, only the non-consensual creation and dissemination of sexual material without consent is considered as sexting (McGlynn & Rackley, 2017). Finally, hacking was defined as unauthorized trespassing or accessing other computers or networks (e.g., McGuire & Dowling, 2013).
All three cybercrimes can have severe consequences for victims, such as post-traumatic stress disorder, trust issues, depression, anxiety, sleeping problems, and public availability of sensitive information (Bates, 2017; Drebing, Bailer, Anders, Wagner, & Galals, 2014; Furnell & Warren, 1999). Preventing perpetration of cybercrime is
therefore important. Information about risk factors is necessary to design these prevention and treatment programs. First of all, this information could inform us about the necessary content or targets of prevention and intervention programs, and also indicate when, during the lifespan, these programs should be offered. With knowledge on risk factors for cybercrime it is possible to identify risks and needs of cybercrime perpetrators. In these
meta-analyses static as well as dynamic risk factors are identified. Static risk factors are unchangeable and useful in risk assessment, whereas dynamic risk factors are changeable and can be targeted in interventions. Changeable risk factors are also referred to as needs in literature on effective treatment of offenders (e.g., Bonta & Andrews, 2007). If
treatment programs are offered in line with the static and dynamic risk factors that are present in offenders, it can be expected that recidivism rates in perpetrators are reduced.
The peak age for committing general cybercrime has been found to fall between 18 and 30 years (United Nations Office on Drugs and Crime, 2013). The peak ages for cyberstalking, hacking, and sexting specifically seem to be at a somewhat earlier age than has been found for general cybercrime. Police records for hacking, for example, showed that juveniles between 12 and 25 were most likely to commit a hacking offense (Ruiter & Bernaards, 2013). Concerning sexting behavior two reviews and a meta-analysis showed that older teenagers and young adults (16-24 years old) were most likely to be involved in sexting (Lounsbury, Mitchell, & Finkelhor, 2011), with a peak around 18 years old (Madigan, Anh, Rash, van Ouytsel, & Temple, 2018). Another study focusing on
cybercrime file data (2006-2011) indicated that youth under 18 committed more hacking crimes than adults, and a comparable amount of child pornography crimes, such as forwarding sexts (Zebel, de Vries, Giebels, Kuttschreuter, & Stol, 2013). Further, it was found that in convicted youth online threatening behavior, spreading pornographic material, and hacking were the cybercrimes perpetrated most (Oosterwijk & Fischer, 2017; Zebel et al., 2013). Following these findings, it is important to meta-analytically study the risk factors for juvenile (12-23 years old) cybercrime perpetration.
Even though rates of cyber deviant behavior are rising, research is unequally divided over the different types of cyber deviant behavior. There are numerous studies on the least severe forms of cyber deviant behavior, such as cyberbullying, illegal
downloading, internet addiction, and gaming (e.g. Chen, Ho, & Lwin, 2017; Kuss & Griffiths, 2012). However, less is known about more severe cyber deviant behaviors, such as cyberstalking, sexting, and hacking. Moreover, most research has focused on characteristics of, and consequences for, victims, but not on characteristics of juvenile perpetrators and antecedents of juvenile cybercrime. Thus, even though it is needed, a meta-analysis of the risk factors for juvenile cyberstalking, sexting and hacking has not been conducted before. So far, most meta-analyses and reviews on juvenile crime risk factors focused on risk factors for traditional crimes. It is unknown, though, whether these risk factors are also relevant for the explanation of cybercrime perpetration. If this is the case, antecedents of the perpetration of cybercrime are similar to antecedents of the perpetration of traditional crime. However, if unique risk factors are present in
cybercrime perpetrators, this would mean that ‘traditional’ prevention and intervention programs might need to be adapted. An important question is what variables can be identified as important static or dynamic risk factors for cybercrime. In the past, many researchers have produced much knowledge on risk factors for traditional crime forms. Based on this body of knowledge, Andrews and Bonta (2010) identified the ‘big four’ or most important risk and protective factors for criminal behavior: ‘history of antisocial behavior’, ‘antisocial personality pattern’, ‘antisocial cognition’, and ‘antisocial associates’.
This history of antisocial behavior risk domain includes prior offenses or prior antisocial activities. Juveniles might have been arrested in the past or committed multiple preceding offenses (Andrews & Bonta, 2010). Not much is known about recidivism rates among cyberstalkers, sexters, and hackers. It was found that for offline stalking,
recidivism rates were around 50% (Rosenfeld, 2003). It would therefore be interesting to learn whether researchers have examined variables such as prior offenses or prior
antisocial activities and what their findings were.
Several personality traits are included in the antisocial personality pattern risk domain. First, the general theory of crime argues that a person’s level of self-control explains criminal behavior (Gottfredson & Hirschi, 1990). Lack of self-control originates in childhood through parenting. Parents’ attachment to children leads to protective and controlling behavior, which then leads to development of self-control in children. When parents are emotionally unavailable to their children, children are at risk of developing low self-control. Low self-control is expressed through impulsiveness, higher levels of risk-taking, not overseeing the consequences of one’s actions and insensitivity. When a person experiences low self-control, (cyber)crime can be perceived as an easy and instant satisfaction (Gottfredson & Hirschi, 1990). The general theory of crime has proven to be important for perpetration of offline crime (Pratt & Cullen, 2000). However, research also shows that low self-control could be a predictor for perpetration of cybercriminal
behavior (e.g. Donner, Marcum, Jennings, Higgins, & Banfield, 2014; Holt, Bossler, & May, 2012; Moon, McCluskey, & McCluskey, 2010).
Next to the general theory of crime that explains low self-control, personality traits can be described following the Big Five personality characteristics: extraversion,
agreeableness, conscientiousness, neuroticism, and openness to experience (Digman, 1990). There is some preliminary evidence that introversion, neuroticism, and less openness to experience are risk factors for cybercriminal behavior (Rogers, Smoak, & Liu, 2006). Finally, some studies on traditional and cybercrime also examine dark personality traits, consisting of Machiavellianism, narcissism, psychopathy, and sadism (Paulhus, 2014). However, there are mixed findings on the significance of these traits in conducting cybercriminal behavior (Duncan & March, 2019; Withers, 2019).
The third risk domain, antisocial cognitions, comprise “attitudes, values, beliefs, rationalizations, and a personal identity that is favorable to crime” (Andrews & Bonta, 2010, p.59). An example of a risk factor in this risk domain is a lower level of moral decision making, which is often measured by the moral decision-making scale (MDKS; Rogers et al., 2006). The MDKS consists of social moral values (i.e. guidelines for social conduct), internal moral values (i.e. beliefs about right and wrong), and hedonistic moral values (i.e. desire to have pleasure in life) (Zezulka & Seigfried-Spellar, 2016). For different cybercriminal behaviors it has indeed been found that perpetrators show relatively low internal moral values (Seigfried, Lovely, & Rogers, 2008; Zezulka & Seigfried-Spellar, 2016) and low social moral values (Rogers et al., 2006).
The final risk dimension encompasses that juveniles have peers that are involved in antisocial or criminal activities (Andrews & Bonta, 2010). This risk domain leans on Akers’ (1998) social learning theory, which consists of four components: differential association, definitions (attitude towards committing crime), imitation, and
learning theory to cybercrimes were inconclusive (e.g. Hinduja & Ingram, 2009; Phillips, 2015).
The present study
The present review has two different aims. First, we aimed to synthesize primary research to identify risk factors for perpetration of cyberstalking, sexting, and hacking by juveniles. This was done by categorizing variables that were examined as risk factors in primary research into several general risk domains. A separate meta-analysis was conducted for each of these risk domains to estimate risk domain effects. Second, we aimed to examine whether risk domain effects were moderated by gender, as boys seem to commit more cybercrimes than girls (Hutchings & Chua, 2016), implying that there may be differences in risk factors between offending boys and girls. We also examined ethnicity as a moderator as it was found, for instance, that attitudes towards sexual behavior (Ahrold & Meston, 2008) and the acceptance of hacking behavior (Kshetri, 2013) differ between ethnic groups. Finally, educational stage was tested as a moderator, as it could inform what risk factors should be targeted given the educational setting of a perpetrator.
Method Literature Search
To search for relevant studies, four electronic databases were searched: ERIC, PsycINFO, Web of Science, and Google Scholar. The search was concentrated on three categories: (1) age (mean age of respondents between 12 and 23 years), (2) cybercrime (cyberstalking, sexting, hacking), and (3) study type (quantitative studies). The search procedure (see Appendix A for details) was performed until May, 2019. Further, the
reference lists of all included studies were scanned to identify additional studies that may have been missed in the electronic search. Finally, all included studies were entered in Google Scholar to determine whether additional studies could be identified by the ‘cited by’ function.
Eligibility Criteria
Prior to the literature search, inclusion and exclusion criteria concerning study characteristics were established. First, the mean age of the participants had to be between 12 and 23 years. Second, only studies on risk factors for perpetration were considered and not for victimization. Third, solely articles reporting on risk factors for cyberstalking (including cyber dating abuse, digital dating abuse, electronic intrusion), hacking, and sexting (including forwarding images or videos without consent, image-based sexual abuse, non-consensual dissemination, coercive sexting, revenge porn, and sextortion) were included. Not included were articles on other cybercrimes, such as identity theft, illegal downloading, and online scams. Fourth, only studies that reported bivariate statistics were included. Therefore, only studies that reported correlations, t-tests, chi-squared tests, bivariate odds ratios, and mean and standard deviations were included. Multivariate statistics were excluded, as it is not possible to calculate standard errors and variance values for multivariate statistics (Lipsey & Wilson, 2001). Fifth, studies had to be written in English or Dutch. Sixth, there was no restriction on the year of publication, due to the novelty of the field. Finally, for the same reason, not only published studies in peer-reviewed journals were included, but also dissertations, government reports and master theses.
Using the electronic databases, a total of 2.126 articles was found (175 from ERIC, 808 from PsycINFO, and 1.143 from Web of Science). The Google Scholar search yielded one additional article. After removing duplicates 1.626 articles were left. These articles were screened on title and abstract based on the inclusion- and exclusion criteria. Two reviewers blindly screened the 154 articles in Rayyan. The two reviewers had an agreement of 96.1%.
Based on abstract and title 173 articles were included. Of these articles the full text was read to decide on eligibility. Initially, studies focusing on cyber aggression, cyber harassment, trolling, online hate, and online insults were included, but this strategy was later changed to include solely studies on cyberstalking, since the above mentioned cyber deviant behaviors could often not be defined as severe and were thus excluded. Further reasons to exclude articles are reported in the flow diagram (Figure 1). Finally, 48 articles were included in the meta-analysis. Of these, 24 studies reported on
cyberstalking, 15 on sexting, and 10 studies on hacking. For an overview of the included studies and their study characteristics, see Appendix B.
Data Coding
The 48 studies were coded in SPSS. Several coding categories were described in a SPSS file that was created for these meta-analyses. Separate SPSS files were set up for cyberstalking, sexting, and hacking. Ten percent of studies were double coded by two reviewers, which showed an interrater agreement of 99,9.%. All SPSS files contained the same variables: variables on study level, variables on sample level and variables on effect size level (Lipsey & Wilson, 2001). Many different variables were coded, in order to be
able to use them as possible moderators during later moderator analyses. A more elaborate description of all coded variables can be found in Appendix C.
In order to analyze the data, risk domains were created in which each of the risk factors was placed. A risk domain is a category in which risk factors are placed that are similar (e.g. parental education and parents’ income represent SES) or have an underlying common factor (e.g. Machiavellianism, psychopathy and narcissism are all considered to be dark personality traits). Thirty-five risk domains were identified for cyberstalking perpetration, 12 for sexting perpetration, and 22 for hacking perpetration. For each of the risk domains an average effect was calculated. For an overview of all risk domains and the risk factors they contain, see Appendix D.
Not all studies reported effect sizes in Pearson’s r (between a specific risk factor and cybercrime perpetration). Other statistics that were reported were t-tests, chi-squared tests, and bivariate odds-ratios. Formulas from Lipsey and Wilson (2001) and Lenhard and Lenhard (2016) were used to transform these statistics to Pearson’s r. One study reported Kendall’s rank correlation, which was transformed to r with Walkers (2003) formula. Sometimes the exact statistic of non-significant results was not reported. To prevent assigning a zero value, authors were mailed whenever possible. However, for nine of the cyberstalking and six of the sexting effect sizes it was not possible to retrieve a true value and a value of zero was assigned. While this approach is conservative and leads to an underestimation of the real effect size (Lipsey & Wilson, 2001), it was the best approach at hand as to not having to exclude these effect sizes. For hacking such a null assignment was not necessary, since all non-significant results were reported.
or female) or coded as a protective factor instead of a risk factor (e.g. high self-control). In those cases, the effect sizes were inverted.
Statistical Analyses
Before performing the main analyses, all continuous variables were mean centered, while for each dichotomous variable a dummy-variable was created (e.g. published and not-published). Next, the z-scores in each risk domain were checked for outliers. An effect size was considered to be an outlier when the z-value was above 3.29 SD or below -3.29 SD (Tabachnik & Fidell, 2013). No outliers were found in the risk domains and therefore no adjustments had to be made. Moreover, all Pearson’s r correlations were transformed into Fisher’s z-scores, because z-scores have a normal sampling distribution (Silver & Dunlap, 1987). For interpretability purposes Fisher’s z was transformed back to Pearson’s r correlation after analysis. This process leads to less bias than averaging correlations without transformations (Silver & Dunlap, 1987). Finally, the standard error and variance were calculated (Lipsey & Wilson, 2001).
This meta-analysis uses a three-level random effects model. This model allows the use of multiple effect sizes per study. Typically, a meta-analysis cannot contain variables that are dependent upon each other, which is the case if multiple variables from the same study are used (Lipsey & Wilson, 2001). The three-level random effects model controls for this dependency. Since multiple effect sizes per study are considered, three levels of variance were modelled (Assink & Wibbelink, 2016). Level 1 is the sampling variance. Level 2 is the variance between effect sizes from the same study (within study variance). Level 3 is the variance between studies. When significant level 2 or level 3 variance was found for a risk domain, moderator analyses were conducted to establish
whether moderators could explain this variance. Several sample characteristics were used as moderators: the percentage of males in the sample, the percentage of respondents with an ethnic majority background in the sample, educational stage, and for cyberstalking the moderator ‘subtype of cyberstalking’ was added. Analyses were conducted in R Studio with the rma.mv function of the metaphor-package (Viechtbauer, 2010). A separate analysis was conducted for each of the risk domains. The manual of Assink and Wibbelink (2016) was used to conduct the analyses.
Publication Bias
A common problem in meta-analyses is the ‘file drawer problem’ (Rosenthal, 1979). This refers to the problem that it is difficult to find all existing research. Studies with non-significant results are often not accepted by journals, and not publicly available. Whether or not publication bias may be a problem in a meta-analysis can be inspected by creating funnel plots. The funnel plot shows the potential risk for publication bias (Duval & Tweedie, 2000a). A funnel plot looks at the standard error in combination with
Fisher’s z values. When there is evidence for publication bias, the funnel plot shows an asymmetric pattern, which can be resolved with a trim-and-fill analysis (Duval & Tweedie, 2000a). Due to the many small domains in this study, a trim-and-fill analysis was not performed, as this analysis is aimed at restoring the symmetry of the funnel by imputing missing studies and with the many small domains symmetry cannot be expected in the first place. Even though with visual inspection it seems like most of the risk
domains are symmetric, we cannot conclude there is no publication bias. Funnel plots are displayed for all cyberstalking (Appendix E), sexting (Appendix F), and hacking
Results Cyberstalking
Regarding cyberstalking, N = 24 articles and k = 29 independent samples were included. The articles were published between 2010 and 2019. Studies were conducted in the United States (k = 18), Canada (k = 4), the United Kingdom (k = 1), the Netherlands (k = 1), Turkey (k = 1), Portugal (k = 1), Belgium (k = 1), Australia (k = 1), and Spain (k = 1). In total, 306 effect sizes were extracted from the manuscripts, with an average of 11.3 per independent sample. The total sample consisted of N = 20,368 juveniles.
The overall relation between all studied risk factors and cyberstalking was significant, r = .174, p <.001. All 306 studied risk factors were categorized and brought back to 35 risk domains (see Table 1). Fifteen of these risk domains were significantly related to cyberstalking perpetration. Concerning the interpretation of the strength of the mean correlations, Cohen’s (1988) criteria label correlations equal or greater than .10, .30, and .50 as respectively small, medium, and large correlations. Large relations with cyberstalking perpetration were found for previous cybercrime perpetration (r = .572) and previous cyberstalking victimization (r = .545). Moderate relations were found between previous offline violence perpetration (r = .395), previous offline victimization (r = .313), and having deviant peers (r = .300) and cyberstalking perpetration. Small relations with cyberstalking perpetration were found for having dark personality traits (r = .200), substance abuse (r = .159), mental health problems (r = .159), attachment problems (r = .146), high computer preoccupation (r = .137), length of relationship (r = .133), and negative gender norms and beliefs (r = .121). Finally, very small relations with
cyberstalking perpetration were found for risk behavior (r = .096), sexual risk behavior (r = .096), and a lack of social support (r = .084).
All cyberstalking risk domains were checked for heterogeneity of within-study variance (level 2) and between-study variance (level 3). If level 2 or level 3 variance was significant, moderator analyses were performed to find variables that could explain this variance. Moderator analyses are reported in Table 2. Eighteen domains showed
significant level 2 or level 3 variance. Since each risk domain needed to contain five or more independent samples in order to be able to conduct meaningful moderator analyses, moderator analyses were conducted for only ten risk domains. Five significant
moderating effects were found in three risk domains. The results showed that both the percentage of respondents with an ethnic majority background, F(1, 8) = 17.402, p = .003, and the specific subtype of cyberstalking, F(1, 10) = 16.022, p = .003, were significant moderators for the risk domain being male. The results indicated that the relation between being male and cyberstalking perpetration increased when the
percentage of juveniles with an ethnic majority background in the sample increased (b = .239). The relation between being male and cyberstalking perpetration also increased when the study specifically focused on cyberstalking (mean r = .095) compared to (the more specific) cyber dating abuse (mean r = -.077), indicating that relatively more females perpetrated cyber dating abuse (compared to cyberstalking). Second, the percentage of males in the sample was found to be a significant moderator for the risk domain high computer preoccupation, F(1, 14) = 6.754, p = .021, b = -313. The results indicated that the relation between high computer preoccupation and cyberstalking became smaller when the percentage of males in the sample increased. Third, both the
percentage respondents with an ethnic majority background, F(1, 7) = 7.152, p = .032, b = .190, and educational stage, F(1, 7) = 6.194, p = .003, were significant moderators for the risk domain mental health problems. The relation between mental health problems and cyberstalking perpetration increased when the percentage respondents with an ethnic majority background in the sample increased. Furthermore, for juveniles in university the relation between mental health problems and cyberstalking perpetration (mean r = .202) was stronger than for juveniles in middle and high school (mean r = .071).
Sexting
Concerning sexting, N = 15 articles were included with k = 19 independent samples. The articles were published between 2012 and 2019. Studies were conducted in the United States (N = 9), Europe (N = 6), South Korea (N = 1), Australia (N = 1),
Botswana (N = 1), and Canada (N = 1). Sixty-one effect sizes were extracted from the manuscripts, resulting in an average of 3.2 effect sizes per independent sample. The total sample consisted of N = 16,816 juveniles.
The overall relation between all studied sexting risk factors and sexting was significant, r = .106, p < .001. All specifically studied risk factors were categorized and brought back to twelve risk domains (see Table 3). Of these twelve risk domains, only two appeared to be significant. That is, there was a small relation between dark
personality traits and sexting perpetration (r = .148), and a very small relation between being male and sexting perpetration (r = .071). For several effects, no risk domain could be created, because there was just one risk factor. One of these effects is peer pressure, which showed a significant medium effect of .490. Furthermore, a significant small effect
was found for positive attitudes on sexting. These results should be interpreted with caution though.
All sexting risk domains were checked for heterogeneity of within-study variance (level 2) and between-study variance (level 3) as well. Three domains showed significant level 2 or level 3 variance, but since just one risk domain contained more than five independent samples, a moderator analysis could only be conducted for this risk domain (i.e. being male). There appeared to be a moderating effect of educational stage on the relation between being male and sexting perpetration, F(1, 16) = 5.562, p = .031. Additional results indicated that being male was related more strongly to sexting perpetration for university students (mean r = .209) than for middle or high school students (mean r = .051) (see Table 4).
Hacking
Finally, for risk factors related to hacking perpetration, N = 10 articles were included with k = 11 independent samples. Manuscripts were published between 2006 and 2018. Studies were conducted in the United States (N = 6), South Korea (N = 1), China (N = 1), Canada (N = 1), Australia (N = 1), and one study was conducted in 31 different countries. A total of 536 effect sizes were extracted from the manuscripts, with an average of 48.7 effect sizes per study. The total sample was N = 72,218 juveniles.
The overall relation between all studied risk factors and hacking perpetration was significant, r = .073, p = .014. In Table 5 results for all 21 hacking risk domains are shown. Seven risk domains were significantly related to hacking perpetration. A
moderate relation between having deviant peers and hacking perpetration was found (r = .335). Besides, small relations were found with hacking perpetration and prior online
deviant behavior (r = .299), low moral beliefs (r = .233), low self-control (r = .127), and prior offline deviant behavior (r = .119). Finally, two very small relations were found between hacking perpetration and low school preoccupation (r = -.027) and high computer preoccupation (r = .062).
Homogeneity analyses of within-study variance (level 2) and between-study variance (level 3) indicated that eleven risk domains had significant level 2 or level 3 variances. However, since risk domains had to be based on at least five independent samples in order to be able to conduct meaningful moderator analyses, moderator
analyses were conducted for only seven risk domains (see Table 6). The results indicated a moderating effect of educational stage for the risk factor being male, F(1, 19) = 6.398, p = .020. Additional results indicated that the relation between being male and hacking was stronger for middle or high school students (mean r = .109) than for university students (mean r = -.074). This result also indicates that females are relatively more likely to perpetrate hacking in university than in middle and high school. Moreover, there was a moderating effect of educational stage for low self-control, F(1, 82) = 13.793, p < .001. Results indicated that the relation between low self-control and hacking perpetration was stronger for students in middle or high school (mean r = .235) than for university students (mean r = .064). Furthermore, the percentage of males had a moderating effect on
computer preoccupation, F(1, 87) = 9.401, p = .003. Results in Table 6 indicate that the relation between high computer preoccupation and hacking perpetration became less strong as the percentage of males in the sample increased (b = -.239). Finally, both the percentage of males in the sample, F(1, 113) = 16.064, p <.001, and the educational stage, F(1, 113) = 6.139, p = .015, were moderators for the risk factor computer skills,
and additional results indicated that the relation between computer skills and hacking became less strong as the percentage of males in the sample increased (b = -.286). Finally, the relation between high computer skills and hacking perpetration was only present for juveniles in middle or high school students (mean r = .114) and not for university students (mean r = .002).
Discussion
The present review is the first to provide a three-level meta-analytic overview of risk factors for juvenile perpetration of three different types of cybercrimes:
cyberstalking, sexting, and hacking. The first aim of this study was to estimate the effect of different risk domains for cyberstalking, sexting, and hacking perpetrated by juveniles between the age of 12 and 23. For each form of cybercrime, different risk domains have been studied, given the variables that were tested as risk factors in the available primary research: 35 risk domains for cyberstalking, 12 risk domains for sexting, and 21 risk domains for hacking. The second aim of this study was to examine whether the overall effect of risk domains was moderated by gender, ethnicity, and educational status.
Overall Effect of Risk Domains
For cyberstalking, 15 significant risk domains were found. Overall, it appears that juveniles are at heightened risk of cyberstalking perpetration when they have committed prior online or offline crimes, or when they have been victims of online or offline crimes themselves. Also, when juveniles show attachment problems, have longer romantic relationships, and have more deviant peers the risk of cyberstalking perpetration increases. Finally, other important risk factors are using drugs, having mental health problems, having negative beliefs about the other gender, spending considerable time on
computers or smartphones, and having dark personality traits. Regarding sexting, only two significant risk domains were found. Dark personality traits had a small impact, and being male had a very small impact. Further, a medium effect was found for the relation between peer pressure and sexting perpetration, however, this was only based on one study. Concerning hacking, seven risk domains were significant. It seems that having deviant peers, low moral beliefs, a history of having committed online or offline crimes, and low self-control are risk factors for hacking.
Interestingly, prior perpetration and victimization of online and offline crime show medium and large relations with both cyberstalking and hacking perpetration. Prior research has identified a ‘bully-victim cycle’, where persons who are a victim of bullying also become a bully themselves (Aleem, 2016). There is evidence that this is even more true for cyberbullying than for traditional bullying (e.g. Li, 2007; Mishna, Khoury-Kassabri, Gadalla, & Daciuk, 2012). This behavior can be explained by the social
learning theory, where children copy the behavior of the bullies, become more aggressive and show more disruptive behavior themselves (Aleem, 2016; Aker, 1998). For both hacking and cyberstalking, a moderate impact of having deviant peers was found. Being affiliated with peers that are also involved in cybercriminal activities may lead to more acceptance of committing cybercrimes, even when the juvenile perpetrator knows the consequences for victims (Bossler & Burruss, 2012; Gordon, 2000; Holt, Burruss, & Bossler , 2010). Hence a possible explanation for the relations between both hacking perpetration and low moral beliefs, and cyberstalking perpetration and negative gender norms and beliefs.
Several non-significant risk domains are interesting as well. Being male was only a significant risk factor for sexting perpetration, whereas literature generally sees males as perpetrators of cybercrimes and females as victims (Hutchings & Chua, 2016). Further, we expected that low self-control would be an important risk factor for cybercriminal behavior, but this was only true for hacking and not for cyberstalking or sexting. A possible explanation is that hacking is a thrill-seeking offense, whereas cyberstalking and sexting are more relational offenses. In prior research it has been found that thrill seeking is an important moderator of the relation between self-control and crime. Juveniles who are low in self-control and show thrill seeking behavior, are more likely to commit a crime than juveniles with low self-control and little thrill-seeking behavior (Burt & Simons, 2013). Lastly, it is often presumed that hackers have limited social skills, but strong computer skills (Barber, 2001). This assumption was not evidenced by the present findings, since no significant relations with hacking perpetration was found for autistic traits, computer skills, and personality traits (such as self-centeredness, introversion, and agreeableness) was found. However, this may be explained by the fact that the studies included in this review may have mainly captured the so-called ‘scriptkiddies’, who are teenagers with limited computer knowledge trying – and often succeeding – in hacking by using online tutorials (Barber, 2001).
When comparing the currently identified risk factors for the different forms of cybercrime to the risk factors for offline crimes, there are some noteworthy findings. First, there seems to be some overlap in risk factors, as prior involvement in crimes, having dark personality traits, low norms and beliefs, and having deviant peers are risk factors for both cybercrime and offline crime. On the other hand, differences in risk
factors have also been found, as personality traits (Big Five) and self-control could not be identified as risk factors for cybercrime. However, it should be noted that primary
research has mainly tested the more ‘traditional’ risk factors for cybercrime and little other – possibly cybercrime specific – risk factors.
It is also interesting to look at the overlap of risk factors across the different types of cybercrime to determine whether the same risk factors are associated with different types of cybercrime, or whether each cybercrime type has unique risk factors. As
discussed above, some risk factors seem to pose a risk for crime in general, such as prior deviant (online and offline) behavior, having deviant peers, low moral beliefs, and dark personality traits. Only one risk factor was identified that seems to be specific for
cybercrime, which is a high computer preoccupation. The results for low self-control and being male seem to be inconclusive, meaning that these variables pose risk factors for some cybercrime types, but not for other cybercrime types. It is also noteworthy that the included studies did not examine the same variables as risk factors for cyberstalking, sexting, and hacking, which makes it more difficult to make statements about the overlap between the different cybercrimes. Therefore, more research is needed to study overlap in risk factors across different types of cybercrimes.
Moderator Effects
For risk domains with sufficient data, moderator analyses were performed. First, an interesting finding is that educational stage was found to be the most common significant moderator. Considering hacking, three risk domains were moderated by educational stage. For being male, having low self-control, and having good computer skills, it was found that being in middle and high school constitutes a greater risk for
hacking perpetration. A smaller impact of these risk domains was found in university students. The risk factor being male was also moderated by educational stage sexting. Males committed significantly more sexting offenses in university than in middle/high school. Finally, the relation between high computer preoccupation and both cyberstalking and hacking perpetration was moderated by the percentage of males in the sample: when the percentage of males in the sample increased, the relation between high computer preoccupation and cyberstalking and hacking decreased. This result indicates that high computer preoccupation is in particular a risk factor for hacking perpetration by girls. This provides evidence for the gender paradox (Hoeve, Vogelvang, Wong, & Kruithof, 2012). The gender paradox assumes that girls in general commit less crimes or show les behavioral problems. However, when girls commit crimes or show behavioral problems, they are more severe than male problem behavior.
Limitations
This study has several limitations. First, several risk domains did not consist of many effect sizes and moderator analyses were sometimes based on a low number of studies and effect sizes. The field of cybercrime is relatively new, and therefore relatively little primary research on risk factors for cybercrime perpetration is yet available. This study should therefore be seen as a first exploratory overview of risk factors for three types of cybercrime committed by juveniles. There have not been prior reviews or meta-analyses on risk factors for juvenile cybercrimes, although a review on risk factors for cyberbullying is available (Chen et al., 2017). It is therefore recommended that this review is updated when the body of primary research has increased to see whether current results can be replicated and/or should be adjusted. It may be very well possible that
‘new’ risk factors for cybercrime can be identified in future research. Until now, mainly traditional risk factors have been studied, whereas it may be possible that there are more specific or unique risk factors for cybercrime. Further, the quality of the available
primary studies was sometimes rather low, for example because a clear description of the instruments used for measuring cybercrime perpetration was lacking. Second, it should be stressed that the reported relations are correlational and not causal. It is therefore not possible to interpret the identified risk factors as predictors for cyberdelinquent behavior. Instead, the studied factors are only correlates of cyberdelinquent behavior. Nevertheless, these factors could be important in prevention and treatment efforts, and future
longitudinal research may perhaps confirm these correlates as true predictors of
cybercrime. Third, because some risk domains were rather small, it was not possible to review publication bias for each risk domain in a reliable manner or resolve publication bias with a trim-and-fill analysis (Duval & Tweedie, 2000b).
Implications
The results from this review have implications for clinical practice. The results could especially contribute to strengthening prevention and intervention programs that are aimed at reducing (the risk of) cybercriminal behavior. First, it seems important to monitor juveniles that previously have been a perpetrator or victim of online or offline crimes. In a review for traditional crimes it was found that the majority of the studies supported the victim-offender overlap (Jennings, Piquero, & Reingle, 2012). Studies looking into the victim-offender overlap for traditional crime specifically for juveniles found small to medium correlations (Barnes & Beaver, 2012; Beckley et al., 2017; Posick, 2013). Since the present review found medium and large effects for prior online
victimization and perpetration, it seems that the victim-offender overlap for online crime is at least as large as, or possibly even larger than, for traditional crime. Further, juveniles are highly influenced by their (deviant) peers. Juveniles might be less likely to imitate the behavior of deviant peers when the reward for committing cybercrime is reduced (Clarke, 1997). Another aspect of prevention programs that might be important is reducing the time spent online, since high computer preoccupation is an important risk factor. Furthermore, the criminal justice system should attempt to increase chances of getting caught for cybercrime. Research shows that victims often do not officially report victimization of cybercrime to the police, as they feel that the perpetrator will not be caught and/or that the police will not put sufficient effort into catching perpetrators (Centraal Bureau voor de Statistiek, 2019).
Oosterwijk and Fischer (2017) wrote a review of interventions for juvenile perpetrators of cybercrime, including interventions for cyber aggression, sexting, and hacking. Regarding cyberstalking, interventions seem to focus mainly on female victims and male perpetrators (Halder, 2015; King, 2008), whereas no gender effect was found for juvenile perpetrators in the current study. Interventions should therefore not only focus on male, but also on female perpetrators.
Regarding sexting, the available interventions solely focus on victims and prevention programs only focus on stopping the sending of sexts (Döring, 2014;
Oosterwijk & Fischer, 2017). Until now, no intervention has taken the role of perpetrators into account who force a person to make sexts or distribute sexts of other persons
(Oosterwijk & Fischer, 2017). Recently, a new intervention specifically for sexting among 12-17 year olds started in the Netherlands, called ‘respect online’ (Jonker & Van
Diessen, 2017). The aims of this intervention are to teach rules for safe and respectful online behavior, recognizing peer pressure, and offer support for parents. This study showed evidence for the importance of peer pressure. Still, relatively little is known about the perpetrators of sexting. Therefore, more research is necessary to integrate risk factors with intervention and prevention programs.
Most interventions are designed for hacking perpetrators (Oosterwijk & Fischer, 2017). Some interventions focus on warning juveniles when they are about to commit a hacking offense, whereas others focus on teaching societal values. The present study found that in particular juveniles with low moral beliefs are involved in hacking. It might be that they have other moral beliefs in ‘hacking ethics’ than in ‘societal ethics’. They are often punished legally and still think they did nothing wrong (Kao, Fu-Yuan Huang, & Wang, 2009). Juveniles may need to learn the differences between right and wrong behavior in the context of hacking. Changing the ethical code of a hacker might have great potential in preventing recidivism of cybercriminal behavior. An intervention that was recently launched in the Netherlands, Hack_Right, applies exactly this strategy of changing the ethical code of a hacker (Halt, 2018). This may be a promising approach, but a proper evaluation of this intervention is necessary to determine its effectiveness. .
Conclusion
The present study identified different risk factors for three forms of cybercriminal behaviors: cyberstalking, sexting, and hacking. Overall, the largest impact was found for prior (online and offline) crime perpetration and victimization. A risk factor that seems to be specifically important for cybercrime is a high computer preoccupation. The results also showed that the impact of several risk domains is moderated by educational stage.
Some risk domains are more important for juveniles attending middle or high school than for juveniles attending university. This review presents a first overview of risk factors for cybercrime, and more primary research is necessary to obtain a better grasp of the
variables that can be designated as true risk factors for juvenile cybercrime perpetration.
Declarations of Interest
None.
Acknowledgements
We thank drs. Janneke P. C. Staaks for her help with the literature search.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
References
References marked with an asterisk indicate studies included in the meta-analysis. Ahrold, T. K., & Meston, C. M. (2010). Ethnic differences in sexual attitudes of U.S.
college students: Gender, acculturation, and religiosity factors. Archives of Sexual
Behavior, 39, 190–202. doi:10.1007/s10508-008-9406-1
Aleem, S. (2016). Bullying behaviour among school students: A review. Indian Journal
of Health and Wellbeing, 7, 976–981. Retrieved from
https://web.a.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=1&sid=4ff4ba81-04e8-453d-bc8f-0fd80c3eff40%40sessionmgr4007
Akers, R. (1998). Social learning and social structure: A general theory of crime and
deviance. Boston, MA: Northeastern University Press.
Andrews, D., & Bonta, J. (Eds.). (2010). The Psychology of Criminal Conduct (5th ed.).
New Providence, NJ: Matthew Bender & Company, Inc, LexisNexis Group. Assink, M., & Wibbelink, C. J. M. (2016). Fitting three-level meta-analytic models in R:
A step-by-step tutorial. The Quantitative Methods for Psychology, 12, 154–174. doi:10.20982/tqmp.12.3
Barber, R. (2001). Hackers profiled – Who are they and what are their motivations?
Computer Fraud & Security, 2, 14–17. doi:10.1016/S1361-3723(01)02017-6
Barnes, J. C., & Beaver, K. M. (2012). Extending research on the victim-offender overlap: Evidence from a genetically informative analysis. Journal of
Bates, S. (2017). Revenge porn and mental health: A qualitative analysis of the mental health effects of revenge porn on female survivors. Feminist Criminology, 12, 22– 42. doi:10.1177/1557085116654565
Beckley, A. L., Caspi, A., Arsenault, L., Barnes, J. C., Fisher, H. L., Harrington, H., . . . Moffitt, T. E. (2017). The developmental nature of the victim-offender overlap.
Journal of Developmental and Life-Course Criminology, 4, 24–49.
doi:10.1007/s40865-017-0068-3
Bocij, P., & McFarlane, L. (2002). Online harassment: Towards a definition of cyberstalking. Prison Service Journal, 139, 31–38.
Bonta, J., & Andrews, D. A. (2007). Risk-need-responsivity model for offender assessment and rehabilitation. Rehabilitation, 6, 1-22. Retrieved from
https://www.publicsafety.gc.ca/cnt/rsrcs/pblctns/rsk-nd-rspnsvty/rsk-nd-rspnsvty-eng.pdf
Bossler, M. A., & Burruss, G. W. (2012). The general theory of crime and computer hacking: Low self-control hacking? In Cyber Crime: Concepts, Methodologies,
Tools and Applications (pp. 1499-1527). Hershey, PA: IGI Global.
*Brewer, R., Cale, J., Goldsmith, A., & Holt, T. (2018). Young people, the internet, and emerging pathways into criminality: A study of Australian adolescents.
International Journal of Cyber Criminology, 12, 115–132.
doi:10.5281/zenodo.1467853
*Bui, N. H., & Pasalich, D. S. (2018). Insecure attachment, maladaptive personality traits, and the perpetration of in-person and cyber psychological abuse. Journal of
Burt, C. H., & Simons, R. L. (2013). Self-control, thrill seeking, and crime: Motivation matters. Criminal Justice and Behavior, 40, 1326–1348.
doi:10.1177/0093854813485575
Centraal Bureau voor de Statistiek. (2019). Digitale Veiligheid en Criminaliteit 2018. Retrieved from https://www.cbs.nl/nl-nl/publicatie/2019/29/digitale-veiligheid-criminaliteit-2018
Chen, L., Ho, S. S., Lwin, M. O. (2017). A meta-analysis of factors predicting
cyberbullying perpetration and victimization: From the social cognitive and media effects approach. New Media & Society, 19, 1994–1213.
doi:10.1177/1461444816634037
*Clancy, E. M., Klettke, B., & Hallford, D. J. (2019). The dark side of sexting – Factors predicting the dissemination of sexts. Computers in Human Behavior, 92, 266– 272. doi:10.1016/j.chb.2018.11.023
Clarke, R. V. (1997). Situational Crime Prevention. Succesful Case Studies (2nd ed.).
Guilderland, NY: Harrow and Heston.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum.
*Daskaluk, S. (2016). Cyber dating abuse: How coercive control and attitudes about
dating aggression affect health and relationship quality (Master’s thesis,
University of Windsor, Windsor, Canada). Retrieved from https://scholar.uwindsor.ca/etd/5809/
Digman, J. M. (1990). Personality structure: Emergence of the five-factor model. Annual
*Donner, C. M., Marcum, C. D., Jennings, W. G., Higgins, G. E., & Banfield, J. (2014). Low self-control and cybercrime: Exploring the utility of the general theory of crime beyond digital piracy. Computers in Human Behavior, 34, 165–172. doi:10.1016/j.chb.2014.01.040
Döring, N. (2014). Consensual sexting among adolescents: Risk prevention through abstinence education or safer sexting? Cyberpsychology: Journal of Psychosocial
Research on Cyberspace, 8. doi:10.5817/CP2014-1-9
*Doucette, H., Collibee, C., Hood, E., Stone, D. I. G., DeJesus, B., & Rizzo, C. J. (2018). Perpetration of electronic intrusiveness among adolescent females: Associations with in-person dating violence. Journal of Interpersonal Violence, 1–21.
doi:10.1177/0886260518815725
Dreßing, H., Bailer, J., Anders, A., Wagner, H., & Gallas, C. (2014). Cyberstalking in a large sample of social network users: Prevalence, characteristics, and impact upon victims. Cyberspsychology, Behavior, and Social Networking, 17, 61–67.
doi:10.1089/cyber.2012.0231
Duncan, Z., & March, E. (2019). Using Tinder to start a fire: Predicting antisocial use of Tinder with gender and the Dark Tetrad. Personality and Individual Differences,
145, 9–14. doi:10.1016/j.paid.2019.03.014
*Duerksen, K. N., & Woodin, E. M. (2019). Technological intimate partner violence: Exploring technology-related perpetration factors and overlap with in-person intimate partner violence. Computers in Human Behavior, 98, 223–231. doi:10.1016/j.chb.2019.05.001
Duval, S., & Tweedie, R. (2000a). Trim and fill: A simple funnel plot based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56, 455– 463. Retrieved from http://www.biometrics.tibs.org
Duval, S., & Tweedie, R. (2000b). A nonparametric ‘trim and fill’ method of accounting for publication bias in meta-analysis. Journal of the American Statistical
Association, 95, 89–98. doi:10.1080/01621459.2000.10473905
*Ewijk, N. van (2018). Explaining variables of adolescents’ cyberstalking behavior:
Internet addiction, gender-differences and educational level (Unpublished
master’s thesis). University of Amsterdam, Amsterdam, The Netherlands. Finn, J. (2004). A survey of online harassment at a university campus. Journal of
Interpersonal Violence, 19, 468–483. doi:10.1177/0886260503262083
Furnell, S. M., & Warren, M. J. (1999). Computer hacking and cyber terrorism: The real threats in the new millennium. Computers & Security, 18, 28–34.
doi:10.1016/S0167-4048998006-6
Gordon, S. (2000, September). Virus writers: The end of the innocence? Paper presented at the 10th Annual Virus Bulletin Conference, Orlando, FL. Retrieved from
https://pdfs.semanticscholar.org/c0f9/db4fdb5945eace504eff659b5ed7ddf43e37.p df
Gottfredson, M., & Hirschi, T. (1990). A General Theory of Crime. Stanford, CA: Stanford University Press.
Halder, D. (2015). Cyber stalking victimisation of women: Evaluating the effectiveness of current laws in India from restorative justice and therapeutic jurisprudential perspectives. Temida, 103–130. doi:10.2298/TEM1504103H
Hinduja, S., & Ingram, J. R. (2009). Social learning theory and music piracy: The differential role of online and offline peer influences. Criminal Justice Studies,
22, 405–420. doi:10.1080/14786010903358125
*Holt, T. J., Bossler, A. M., & May, D. C. (2012). Low self-control, deviant peer associations, and juvenile cyberdeviance. American Journal of Criminal Justice,
37, 378–395. doi:10.1007/s1203-011-9117-3
Holt, T. J., Burruss, G. W., & Bossler, A. M. (2010). Social learning and cyber-deviance: Examining the importance of a full social learning model in the virtual world.
Journal of Crime and Justice, 33, 31–61. doi:10.1080/0735648X.2010.9721287
*Hu, Q., Xu, Z., & Yayla, A. A. (2013). Why college students commit computer hacks:
Insights from a cross culture analysis. Paper presented at the 17th Pacific Asia
Conference on Information Systems, Jeju, South Korea. Retrieved from
https://pdfs.semanticscholar.org/7d13/27d548c05300e6337a04f2f8ad66a1fcc75a. pdf?_ga=2.220175953.959096644.1579539297-1663215328.1570790980 Halt. (2018, December 18th). Hack_Right: jonge hackers weer op het rechte pad.
Retrieved from https://www.halt.nl/actueel/hack_right-jonge-hackers-weer-op-het-rechte-pad/
Hutchings, A., & Chua, Y. T. (2016). Gendering cybercrime. In T. J. Holt (ed.),
Cybercrime through an interdisciplinary lens (pp. 167–188). Oxon: Routledge.
Jennings, W. G., Piquero, A. R., & Reingle, J. M. (2012). On the overlap between victimization and offending: A review of the literature. Aggression and Violent
Jonker, M., & Diessen, C. van. (2017). Toeleidingshandleiding halt-interventie sexting:
Respect online: Een interventie voor jongeren die lichte vormen van seksueel grensoverschrijdend gedrag hebben vertoond. Retrieved from
https://www.rutgers.nl/sites/rutgersnl/files/PDF/Halt-interventie%20sexting%20Toeleidingshandleiding.pdf
Kao, D. Y., Fu-Yuan Hang, F., & Wang, S. J. (2009). Persistence and desistance: Examining the impact of re-integrative shaming to ethics in Taiwan juvenile hackers. Computer Law & Security Review, 25, 464–476.
doi:10.1016/j.clsr.2009.05.009
King, M. S. (2008). Restorative justice, therapeutic jurisprudence and the rise of emotionally intelligent justice. Melbourne University Law Review, 32, 1096– 1126. Retrieved from http://www.mulr.com.au/issues/32_3/32_3_10.pdf *Kircaburun, K., Jonason, P. K., & Griffiths, M. D. (2018). The Dark Tetrad traits and
problematic social media use: The mediating role of cyberbullying and cyberstalking. Personality and Individual Differences, 135, 264–269. doi:10.1016/j.paid.2018.07.034
Klettke, B., Hallford, D. J., & Mellor, D. J. (2014). Sexting prevalence and correlates: A systematic literature review. Clinical Psychology Review, 34, 44–53.
doi:10.1016/j.cpr.2013.10.007
Kshetri, N. (2013). Cybercrime and cyber-security issues associated with China: Some economic and institutional considerations. Electronic Commerce Research, 13, 41–69. doi:10.1007/s10660-013-9105-4
Kuss, D. J., & Griffiths, M. D. (2012). Internet gaming addiction: A systematic review of empirical research. International Journal of Mental Health and Addiction, 10, 278–296. doi:10.1007/s11469-011-9318-5
*Lee, B. H. (2018). Explaining cyber deviance among school-aged youth. Child
Indicators Research, 11, 563–584. doi:10.1007/s12187-017-9450-2
*Lee, C., Moak, S., & Walker, J. T. (2016). Effects of self-control, social control, and social learning on sexting behavior among South Korean youths. Youth & Society,
48, 242–264. doi:10.1177/0044118X13490762
Lenhard, W., & Lenhard, A. (2016). Calculation of Effect Sizes. Dettelbach, Germany: Psychometrica. doi:10.13140/RG.2.1.3478.4245
Li, Q. (2007). New bottle but old wine: A research of cyberbullying in schools.
Computers in Human Behavior, 23, 1777–1791. doi:10.1016/j.chb.2005.10.005
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: SAGE publications.
Lounsbury, K., Mitchell, K. J., & Finkelhor, D. (2011). The true prevalence of ‘sexting’. Durham, NH: Crimes against Children Research Center.
Madden, M., Lenhart, A., Cortesi, S., Gasser, U., Duggan, M., Smith, A., & Beaton, M. (2013). Teens, social media, and privacy. In Pew Internet & American Life
Project. Retrieved from
www.pewinternet.org/2013/05/21/teens-social-media-and-privacy/
Madigan, S., Ly, A., Rash, C. L., van Ouytsel, J., Temple, J. R. (2018). Prevalence of multiple forms of sexting behaviour among youth: A systematic review and meta-analysis. JAMA paediatrics, 172, 327–335. doi:10.1001/jamapediatrics.2017.5314
*Makgale, O. L., & Plattner, I. E. (2017). Sexting and risky sexual behaviours among undergraduate students in Botswana: An exploratory study. Cyberpsychology:
Journal of Psychosocial Research on Cyberspace, 11, article 1.
doi:10.5817/CP2017-2-1
*Marcum, C. D., Higgins, G. E., & Nicholson, J. (2017). I’m watching you: Cyberstalking behaviors of university students in romantic relationships.
American Journal of Criminal Justice, 42, 373–388.
doi:10.1007/s12103-016-9358-2
*Marcum, C. D., Higgins, G. E., & Poff, B. (2016). Exploratory investigation on
theoretical predictors of the electronic leash. Computers in Human Behavior, 61, 213–218. doi:10.1016/j.chb.2016.03.010
McGlynn, C., & Rackley, E. (2017). Image-based sexual abuse. Oxford Journal of Legal
Studies, 37, 534–561. doi:10.1093/ojls/gqw033
McGuire, M., & Dowling, S. (2013). Cybercrime: A review of the evidence (Research Report 75). Chapter 1: cyber-dependent crimes. Retrieved from
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attac hment_data/file/246751/horr75-chap1.pdf
*Melander, L. A. (2010). Explaining college partner violence in the digital age: An
instrumental design mixed methods study (Doctoral dissertation, University of
Nebraska, Lincoln, United States). Retrieved from
https://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1000&context=sociol ogydiss
*Ménard, K. S., & Pincus, A. L. (2012). Predicting overt and cyber stalking perpetration by male and female college students. Journal of Interpersonal Violence, 2, 2183– 2207. doi:10.1177/0886260511432144
Mishna, F., Khoury-Kassabri, M., Gadalla, T., & Daciuk, J. (2012). Risk factors for involvement in cyber bullying: Victims, bullies and bully-victims. Children and
Youth Services Review, 34, 63–70. doi:10.1016/j.childyouth.2011.08.032
Moon, B., McCluskey, J. D., & McCluskey, C. P. (2010). A general theory of crime and computer crime: An empirical test. Journal of Criminal Justice, 38, 767–772. doi:10.1016/j.jcrimjus.2010.05.003
*Morelli, M., Bianchi, D., Baiocco, R., Pezzuti, L., & Chirumbolo, A. (2016). Not-allowed sharing of sexts and dating violence from the perpetrator’s perspective: The moderation role of sexism. Computers in Human Behavior, 56, 163–169. doi:10.1016/j.chb.2015.11.047
*Murray, D. L. (2014). A survey of practices and perceptions of students in one catholic
high school on the use of the internet in relation to safety, cyberbullying, and sexting (Doctoral dissertation, University of San Francisco, San Francisco, United
States). Retrieved from https://repository.usfca.edu/diss/89/
*Norman, J. (2017). Implications of parenting behaviours and adolescent attachment for
understanding adolescent sexting (Doctoral dissertation, University of Windsor,
Ontario, Canada). Retrieved from https://scholar.uwindsor.ca/cgi/viewcontent. cgi?article=8287&context=etd
*Novo, F., Pereira, F., & Matos, M. (2014). Cyber-aggression among Portuguese adolescents: A study on perpetration, victim offender overlap and parental
supervision. International Journal of Cyber Criminology, 8, 94–110. Retrieved from https://www.cybercrimejournal.com
Oosterwijk, K., & Fischer, T. F. C. (2017). Interventies Jeugdige Daders Cybercrime. Den Haag, The Netherlands: WODC.
*Ouytsel, J. van., Ponnet, K., & Walrave, M. (2017). Cyber dating abuse: Investigating digital monitoring behaviors among adolescents from a social learning
perspective. Journal of Interpersonal Violence, 1–22. doi:10.1177/0886260517719538
*Patchin, J. W., & Hinduja, S. (2018). Sextortion among adolescents: Results from a national survey of U. S. youth. Sexual Abuse, 28, 1–25.
doi:10.1177/1079063218800469
*Peskin, M. F., Markham, C. M., Shegog, R., Temple, J. R., Baumler, E. R., Addy, R. C., … Emery, S. T. (2017). Prevalence and correlates of the perpetration of cyber dating abuse among early adolescents. Journal of Youth and Adolescence, 46, 358–375. doi:10.1007/s10964-016-0568-1
*Phillips, E. (2015). Empirical assessment of lifestyle-routine activity and social learning
theory on cybercrime offending (Master’s thesis, Bridgewater State University,
Bridgewater, United States). Retrieved from
https://vc.bridgew.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&htt psredir=1&article=1024&context=theses
Pratt, T. C., & Cullen, F. T. (2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38, 931–964. doi: 10.1111/j.1745-9125.2000.tb00911.x
*Preddy, T. M. (2015). Assessment and investigation of electronic aggression in the
romantic relationships of emerging adults (Doctoral dissertation, University of
Tennessee, Knoxville, United States). Retrieved from https://trace.tennessee.edu/ cgi/viewcontent.cgi?article=4846&context=utk_graddiss
Paulhus, D. L. (2014). Toward a taxonomy of dark personalities. Current Directions in
Psychological Science, 23, 421–426. doi:10.1177/0963721414547737
Posick, C. (2013). The overlap between offending and victimization among adolescents: Results from the second international self-report delinquency study. Journal of
Contemporary Criminal Justice, 29, 106–124. doi:10.1177/1043986212471250
*Reed, L. A., Tolman, R. M., & Ward, L. M. (2016). Snooping and sexting: Digital media as a context for dating aggression and abuse among college students.
Violence Against Women, 22, 1556–1576. doi:10.1177/1077801216630143
*Reed, L. A., Tolman, R. M., Ward, L. M., & Safyer, P. (2016). Keeping tabs:
Attachment anxiety and electronic intrusion in high school dating relationships.
Computers in Human Behavior, 58, 259–268. doi:10.1016/j.chb.2015.12.019
*Reed, L. A., Ward, L. M., Tolman, R. M., Lippman, J. R., & Seabrook, R. C. (2018). The association between stereotypical gender and dating beliefs and digital dating abuse perpetration in adolescent dating relationships. Journal of Interpersonal
Violence, 1–25. doi:10.1177/0886260518801933
*Reyns, B. W., Henson, B., Fisher, B. S. (2012). Stalking in the twilight zone: Extent of cyberstalking victimization and offending among college students. Deviant
*Rogers, M., Seigfried, K., & Tidke, K. (2006). Self-reported computer criminal behaviour: A psychological analysis. Digital Investigation, 3, 116–120. doi:10.1016/j.diin.2006.06.002
*Rogers, M., Smoak, N. D., & Liu, J. (2006). Self-reported deviant computer behaviour: A big-5, moral choice, and manipulative exploitive behaviour analysis. Deviant
Behavior, 27, 245–268. doi:10.1080/01639620600605333
Rosenfeld, B. (2003). Recidivism in stalking and obsessional harassment. Law and
Human Behavior, 27, 251–265. doi:10.1023/A:1023479706822
Rosenthal, R. (1979). The file drawer problem and tolerance for null results.
Psychological Bulletin, 86, 638–641.
http://dx.doi.org/10.1037/0033-2909.86.3.638.
Ruiter, S., & Bernaards, F. (2013). Are crackers different from other criminals? A
comparison based on Dutch suspect registrations [Verschillen crackers van andere criminelen? Een vergelijking op basis van Nederlandse verdachtenregistraties].
Tijdschrift voor Crimonologie, 55, 342–359.
*Schell, F. (2018). The role of friends in cyber dating abuse: An examination of attitudes,
normative beliefs and reinforcement behaviours (Honours thesis, King’s
University College, London, Canada). Retrieved from https://ir.lib.uwo.ca/psychK_uht/72
*Schnurr, M. P., Mahatmya, D., & Basche, R. A. (2013). The role of dominance, cyber aggression perpetration, and gender on emerging adults’ perpetration of intimate partner violence. Psychology of Violence, 3, 70–83. doi:10.1037/a0030601
