Careful When You Click? How the Dark Triad of Personality Can Influence the Likelihood of Online Crime Victimization

Abstract

Cybercrime is a growing problem, with increasing numbers of people reporting they have been a victim. However, the literature has tended to focus on the characteristics of the perpetrator and has often neglected to examine how the individual differences of victims may have an impact. This paper investigates how the Dark Triad – Machiavellianism, narcissism, and psychopathy – may increase the chances of being a victim of online crime. To do this, the Cyber Routine Activities Theory was applied, which suggests victimization is a result of two things: 1) a user’s routine online activity which may bring them into contact with nefarious others and/or makes them an attractive target, and 2) the lack of a “capable guardian” who can defend against such nefarious others. Using an online survey (N = 328), we measured Internet users’ Dark Triad traits, along with their engagement in various online activities and the preventative measures used against potential criminals. Findings demonstrated that narcissism and psychopathy increased the likelihood of victimization, but Machiavellianism did not. These relationships were moderated by gender. However, contrary to other work using the Cyber-RAT, preventative measures (e.g. knowledge of computers, presence of anti-virus programs) did not seem to impact on the likelihood of victimization. The challenges of using these findings to reduce cybercrime and future work are then discussed.

CLASSIFICATIONS:

• personality
• human factors
• psychology and law
• social interaction

Introduction

An Overview of Cybercrime Prevalence

Cybercrime is “an umbrella term to describe different online threats such as malware, scams and hacking” (Martens et al., 2019; p139). The goal for many of these crimes is to obtain money from the victim. Individuals can be tricked into providing log-in details or directly contribute money to fraudulent causes; computers can also be held hostage using viruses, and individuals can be extorted through blackmail and threats of releasing personal information/images. As well as this, some cybercrimes have no goal other than to disrupt the online experiences of others. Many computer viruses are intended only to be an annoyance, rather than to accomplish a goal. In the UK last year, there were over a million incidents of “computer misuse” and 700,000 incidents of computer virus infection (Elkin, 2019).

In this paper, we examine an under-explored aspect of cybercrime; namely, how individual differences may make a person more or less likely to become a victim of these acts. In particular, we postulate that the Dark Triad—Machiavellianism, narcissism, and psychopathy - may be instrumental in increasing the likelihood of victimization, by influencing the types of behavior those high in these traits engage in online. Research often focuses on the characteristics of the perpetrator, but investigating the characteristics of the victim may be more beneficial as these individuals have motivation to change so they are no longer at risk.

Explaining Cybercrime Victimisation

One key model in explaining crime victimization is the Lifestyle Exposure Theory (LET, Hindelang et al., 1978), which suggests that individuals become vulnerable to crime through the activities they engage in during their daily life. For example, an individual who engages in careless behavior such as leaving their house unlocked when going out, or who regularly takes walks through crime-rife areas is more likely to be a victim. This model was then further expanded into the Routine Activities Theory of victimization (RAT, Cohen & Felson, 2016). Here, three components are purported as necessary for victimization to occur: a motivated offender, the lack of a capable guardian, and a suitable target. The latter component here maps onto the LET lifestyle activities idea, arguing that the behaviors the individual engages in are what makes them “suitable” or not. Since its inception, the RAT has been successfully applied to explain a variety of crimes such as fraud of the elderly (DeLiema, 2018), sexual harassment (Clodfelter et al., 2010), and violent crime (Podaná, 2017).

As well as offline crime, the RAT has also been applied to online crime, successfully explaining general victimization (Leukfeldt & Yar, 2016), and also other deviant online behavior such as online aggression (Melander & Hughes, 2018). However, there has been some debate amongst authors as to how to conceptualize the different aspects of the model to provide the greatest explanatory power. Some researchers (e.g. Bossler & Holt, 2009; Marcum, 2008) consider the “motivated offender” component of the model as behavior by a potential victim that may bring them into contact with an offender, rather than whether an offender is “there”. So, illegally downloading a film or browsing a torrent site may be considered for this component. Other researchers (e.g. Kalia & Aleem, 2017) consider this kind of behavior more as making an individual a “suitable target”; that by engaging in these activities they make themselves attractive to nefarious online operatives. Ngo and Paternoster (2011) has suggested that suitability/attractiveness of a target can be influenced by how careful individual is online. Do they click links that are sent to them by unfamiliar address, or do they enter their personal details on sites with little care for security? Other demographic aspects have also been included in the “suitability” component such as gender, age, and socio-economic background (Bossler et al., 2012).

The “capable guardian” aspect of the RAT has also prompted some definitional debates (e.g. Reyns et al., 2016; Vakhitova, Reynald, & Townsley, 2016). Most researchers include the installation of security software such as antivirus as a “digital” guardian that prevents access from offenders (Williams, 2015). But, this may depend on the crime being examined. For example, crime in which children are commonly victimized (such as online abuse or cyberbullying) may also include physical guardianship by the child’s parents, who monitor their online activity (Navarro & Jasinski, 2012). Furthermore, the user’s computer competency has also been included in some work as a form of “personal” digital guardianship; the idea being that an individual with greater IT competency would be less likely to fall victim to crime (Bossler et al., 2012)

To resolve these issues when translating RAT components to an online context, in this paper we utilize Choi’s (2008; Choi & Lee, 2017) cyber-RAT model which considers only online lifestyle and digital guardianship as factors in victimization. The former subsumes conventional online behavior such as buying goods; more “deviant” online behavior such as obtaining illegal software; and general caution when interacting online, such as examining links before clicking on them. The latter refers to both physical software that may be a barrier to crime, such as anti-virus programs, and also knowledge of computer/Internet matters.

Surprisingly, although the extant literature is replete with examinations of personality and crime perpetration, there is still a lack of diversity in has examining victimization, particularly in an online context. Almost all research in this area has focus on trait self-control - demonstrating that low levels of this trait can lead to greater victimization (Louderback & Antonaccio, 2021; although see Bossler & Holt, 2010 for a contrasting argument). Thus, there is considerable scope to broaden our understanding of how personality traits outside of self-control may influence the likelihood of becoming a victim of online crime. In this paper, we investigate a particular set of traits that have been shown to exert considerable power over online behaviors in the past—the Dark Triad.

An Overview of Dark Triad

The Dark Triad of personality (Paulhus & Williams, 2002) consists of three traits: Machiavellianism, describing a tendency for plotting and subterfuge; narcissism, describing a tendency toward high self-regard; and psychopathy, describing a tendency for low empathy and impulse control (Paulhus, 2014). These three traits have all been implicated in various studies which have examined socially questionable behaviors. For example, all three traits positively correlate with enjoyment of others misfortune (Porter et al., 2014) and of bullying others (van Geel et al., 2017). More specifically, those who score high on Machiavellianism often engage in deceit when dealing with others (Baughman et al., 2014). Those high in psychopathy tend to display less compassion and empathy to others overall (Lee & Gibbons, 2017), and narcissists usually perceive others incompetent compared to themselves (Kong, 2015). There is also evidence that directly links the dark triad to behavior online. Those high in Machiavellianism often make dishonest posts on social media in order to manipulate others (Abell & Brewer, 2014), and those high in psychopathy will leave unpleasant comments for others on Facebook (Lopes & Yu, 2017). Narcissism has received considerable attention with regards to online interactions. Narcissists will often attempt to elicit responses from other users, and react angrily when the desired reaction is not forthcoming (Errasti et al., 2017). They also, unsurprisingly, tend to craft their social media profiles ensure what they post is as socially desirable as possible (Kapidzic, 2013).

Given the robust link between the Dark Triad and socially transgressive behavior, and in engaging in undesirable behavior online, it seems highly likely there will be an association between the traits and deviant online behavior (e.g. visiting piracy sites, downloading hacked software). Moreover, the cyber-RAT suggests a link between online lifestyle and online crime victimization. Therefore, we hypothesize a link between the Dark Triad and victimization, mediated by the engagement of deviant online behavior. As narcissism, psychopathy, and Machiavellianism tend to increase, so too does deviant online behavior, and therefore likelihood of online crime victimization (H1, H2, and H3 respectively).

As well as this, both narcissism and psychopathy are linked with a propensity to engage in risky behavior online, which may make someone an attractive target for a criminal (Malesza & Ostaszewski, 2016). Narcissists tend to suffer from over-confidence and thus are unlikely to come to harm from risky behavior (Buelow & Brunell, 2014). Those high in psychopathy tend to underestimate the risk present in situations (Hosker-Field et al., 2016), leading to more risky actions (Yao et al., 2019). Therefore, we expect a link between narcissism (H4), psychopathy (H5), and cybercrime victimisation, mediated by engaging in online behavior that makes one an attractive target for victimization (e.g. clicking links in email from unknown senders, sharing log in passwords).

The picture of Machiavellianism is less clear in the literature. There are links with the trait and riskier behavior (Rim, 1966), and Sakalaki et al. (2007) suggest that those high in Machiavellianism may be careless when there is monetary gain in a situation. However, other researchers have suggested Machiavellianism may predict more distrusting, less risky behavior (Ináncsi et al., 2018). Thus, we tentatively predict a link between Machiavellianism and cybercrime victimization mediated by attractiveness as a target (H6) but do not specify the direction.

We can also consider the relationship between the Dark Triad and “normal” Internet use. Surprisingly, little research has examined this—most has focused on pathological Internet use such as “trolling” others (e.g. March, 2019). Therefore, we can only offer exploratory hypotheses for this link. As we have suggested the Dark Triad is linked with other types of web browsing (i.e. deviant activities), we tentatively suggest the same positive link for narcissism (H7), psychopathy (H8) and Machiavellianism (H9) with cybercrime victimization, mediated by conventional Internet use.

The second component of the cyber-RAT examines the influence of a “capable guardian” in victimization, manifested as the use of anti-crime software (such as anti-virus) and knowledge of Internet/computer related matters. No literature to date has examined the link between individual differences and computer skills or software use, making hypotheses on this aspect difficult. Instead, we offer an exploratory research question this: how do individual differences predict the use of digital guardianship? (RQ1).

The Dark Triad, Internet Use, and Gender

When considering crime victimization, females tend to have a greater fear of becoming victim (Chataway & Hart, 2018) and are more likely to take self-protective action to avoid it than males (Custers et al., 2017). However, this is often driven by concerns over the physical aspects of crime, such as sexual harassment or assault (Donnelly & Calogero, 2018). How then does this translate to online crime?

The extant literature is replete with examples of how males and females utilize the Internet differently. Amongst habitual Internet users, females are more likely to communicate with strangers online (Ang, 2017). Men are more likely to view sexually explicit material when lonely or lacking a committed relationship compared to women (Weber et al., 2018), and men are more likely to be problematic Internet users, whereas females are more likely to use their smartphone to access (Lee et al., 2018).

There is also evidence of differences in expressions of the Dark Triad between genders (Jonason & Davis, 2018). Overall, females tend to demonstrate narcissistic tendency both in public and when in private, whereas males tend to focus more on overt observable narcissism (Barnett & Sharp, 2017). Differential patterns have also been found for psychopathy. Overall, females tend to show lower rates of psychopathy than males (Vogel & Lancel, 2016) (although it should be noted, much of the research on this trait including that cited here uses a clinical population, whereas the Dark Triad traditional focuses on sub-clinical individuals). Nevertheless, reasonable assumption of a gender difference can be made, even in a typically-developed population. Although there is a dearth of research examining Machiavellianism and gender, the extant literature does support the idea of a difference. Szabó and Jones (2019) found the tendency to plan ahead related to levels of Machiavellianism and the gender of the planner. They found that for males, Machiavellianism tended to facilitate planning ahead; by contrast amongst females it seemed to hinder that tendency.

Therefore, it is likely that there will be differences in online behaviors which may be driven by differences in the Dark Triad according to the individual’s gender. That is, the relationship between narcissism, psychopathy and Machiavellianism on deviant online behavior, normal online behavior, and attractiveness as a target will be moderated by an individual’s gender (H10).

Summary and Hypotheses

In this paper, we offer an investigation into how personality traits may influence the likelihood of online crime victimization. Using the Cyber-RAT model as a basis, we examine how the Dark Triad of personality may influence normal and deviant online activities, and also how it may impact on care taken online which may alter attractiveness as a target. We also examine the role of technological barriers as a “guardian” on victimization. Overall, we hypotheses that the Dark Triad will influence online behavior, which in turn will lead to greater crime victimization.

Method

Participants

Sample size was determined using G Power. A moderate effect size (f² = .15) with power of .8 and an α of .05 was used. Six main predictor measures were used, meaning a minimum sample of 98 participants was required.

Participants were recruited from the crowd-sourcing site www.prolific.ac in return for payment of .80p (around £1). Overall, 350 participants began the survey. Of these, 22 did not finish substantial portions of the survey and were removed from the sample, leaving 328 completed surveys. The average age of participants was 26.32 years (SD = 9.66, minimum = 16, maximum = 75). In the sample, 120 participants reported their gender as male, 206 as female, 1 as non-binary, and 1 did not record their gender.

Materials and Design

This study used an online survey design, with the link distributed via Qualtrics. Participants responded to several scales regarding the Dark Triad traits, their online browsing behavior, and the online crimes they had been a victim of.

Dark Triad

Narcissism was measured using the NPI-13 (Gentile et al., 2013; example item: “I know I am a good person because everyone keeps telling me so”). Machiavellianism was measured using the 20-item Mach IV scale (Christie & Geis, 2013; example item: “it is wise to flatter important people”), and psychopathy was measured using the appropriate component from the Short Dark Triad (Jones & Paulhus, 2014; example item: “people who mess with me always regret it”). All items were measured on a 5-point Likert scale with a higher score indicating a greater presence of that trait.

Online Behaviour and Crime Victimisation

Measures of deviant online behavior were based on items from Bossler and Holt (2009), with some additions from the researchers, giving 17 items in total (e.g. “how often have you pirated a copy of a film in the last 6 months”). Participants suitability of a target was based on items from Ngo and Paternoster (2011); giving eight items in total (e.g. how often have you opened unfamiliar attachments in the last six months”). Both were answered on a 5-point Likert scale from “never” to “all the time” with a higher score indicating greater frequency.

Measures of normal online behavior were created by focus group. An online survey was provided on the crowd-sourcing website Prolific (prolific.co) recruited 50 individuals from the UK (24 male, mean age = 41.21, SD = 5.36) and asked for three activities that they performed daily online. Removing duplicates and unsuitable answers, this produced 13 unique online behavior items (e.g. “how often do you shop on Amazon?”) answered on a 5-point Likert scale from “not at all” to “very often” with a higher score indicating greater frequency.

Digital Guardianship

Two items were used, adapted from items in Bossler and Holt (2009) which asked participants “how skilled are you in using computers for everyday tasks? (e.g. web browsing, word processing)” and “how skilled are you in using computers for complex tasks (e.g. diagnosing and fixing problems)” on a 5-point Likert scale from “not at all” to “fairly” with a higher score indicating greater competence. Participants were also asked to answer “yes” or “no” whether they used additional software to protect their computer from intrusion or harm.

Crime Victimization

To create our pool of crimes, we examined the most reported cybercrimes that had occurred in the previous year as indicated by the National Crime Agency (Panlogic, 2020). This indicated the most common were hacking of users’ accounts (e.g. social media, email), attempts to “phish” a user by providing a malicious or compromising link to click on, distribution of malevolent software, and Distributed Denial of Service (DDoS) attacks. The latter is pertains more to organisations’ websites rather than individual users, so was discarded from the list of potential crimes for this study.

Examination of other pertinent cybercrime sources (e.g. Irwin, 2021) indicated that online fraud when buying good was also extremely common. Accordingly, we summarized the main types of crime as information compromise (e.g. when an individual’s account was compromised or their personal information stolen), catching a virus (wherein the purpose is simply to damage or cause destruction to an user’s device), losing money (through purchasing fake goods or non-delivery, or having money stolen), and software compromise (the installation of something malevolent such as a botnet or ransomware). Whilst not an exhaustive list, it was felt these categories well represented some of the most pertinent cybercrimes.

For each of these categories, we generate specific instances of a crime that exemplified them. For parsimony, the smallest number of crimes was used that covered each category. In total nine crimes participants were used in total. Participants indicated how often they had ever been a victim of these crimes: never, 1-2, 2-3, 4-5, or more than 5 times. Following data collection, a measurement model was constructed using SPSS AMOS to test our categorization. This showed excellent goodness-of-fit for the data (χ² = 41.19, df = 21, p = .005; CFI = .96, RMSEA = .05, SRMR = .04)—see Figure 1.

Figure 1. Measurement model for types of online crime. Notes. All visible pathways a significant at p < .05.

Procedure

Participants were recruited via Prolific and presented with the Qualtrics link. The initial page explained some background to the study, but without giving explicit details of the hypotheses. No mention of “personality” or “the Dark Triad” was given anywhere; participants were simply told they would be asked questions about themselves, and whether they had been a victim of online crime. They were reminded that their answers were anonymous, and that they could withdraw (by closing the browser window) at any time.

If participants were happy to proceed, they were required to tick boxes indicating they had read the instructions and understood what the survey would ask. Doing so started the survey. Participants were first presented with the demographic measures. They then moved on to the Dark Triad scales, the online activities scales, and then the crime victimization measures in that order. The order of presentation for the Dark Triad and online activities scales was randomized uniquely for each participant.

Upon finishing, participants were thanked and told to email the researcher if they would like a full debrief on the study (none did). They were then diverted back to the Prolific site for compensation.

Results

Data Preparation

All negatively worded items on the Dark Triad scales were reverse coded. Each scale was then averaged to produce a single Machiavellianism (α = .76), psychopathy (α = .77), and narcissism (α = .85) score for each participant.

Normal online activities (RA), deviant online activities (DA), and attractiveness as a target (AT) did not measure a psychological construct, but rather the frequency of engaging in particular behaviors. Therefore, items in each measure were summed (rather than averaged) to give a better indicator of participants’ engagement with each type of behavior. Participants responses within each of the four crime categories were collapsed into a single score—see Table 1.

Table 1. Descriptives for Main Dependent Variables.

	M	SD
Dark Triad traits
Narcissism	3.62	.94
Machiavellianism	3.70	.63
Psychopathy	2.81	.96
Online lifestyle
Normal online activities	35.72	6.02
Deviant online activities	20.95	8.91
Attractiveness as a target	11.76	4.18
Digital guardianship
Skilled at every-day tasks	4.43	78
Skilled at advanced tasks	3.24	1.25
Crime
Information compromise	.36	.57
Catching a virus	.24	.60
Losing money	.45	.53
Software compromise	.26	.51

Notes. Crime figures are given by responses “how often have you been a victim of this crime?” with 1 = never, 2 = 1-2 times, 3 = 2-3 times, 4 = 4-5 times, and 5 = more than 5 times. Digital guardianship also asked participants “do you have additional software to guard against intrusion on your computer?”; 215 participants answered “yes”.

Data Analysis

An initial structural equation model (SEM) was created in SPSS AMOS to test the relationships between the main variables. The Dark Triad were visualized as observed variables, whilst online lifestyle was considered a latent variable containing normal and deviant online behaviors, along with attractiveness as a target. Guardianship was a second latent variable, represented by simple and complex computer skills, and whether participants used other software to protect their devices. The aim with this model was to create a parsimonious conceptualization of the relationships between our key variables.

Goodness-of-fit indices indicated a moderate-to-poor fit (χ² = 156.94, df = 57, p<.001; CFI = = .894, RMSEA = .073, SRMR = .074). Examination of the standardized regression weights indicated that the guardianship latent variable did not contribute to the crime victimization latent variables (β = -0.058, SE = .053, p = .415). There was also little contribution from the Dark Triad traits to the guardianship latent variable (for Machiavellianism β = .003, SE = .073, p = .973; for narcissism β = .0138, SE = .049, p = .108; for psychopathy β = -0.036, SE = .055, p = .709). However, the contribution from the lifestyle latent variable was significant for crime (β = .746, SE = .012, p<.001), as were the contribution from two of the three Dark Triad traits to the lifestyle variable (Machiavellianism β = .092, SE = .331, p = .164; for narcissism β = .177, SE = .220, p = .007; for psychopathy β = 440, SE = .254, p<.001)

A second model was therefore conducted which did not include the guardianship latent variable. This had a much improved goodness of fit (χ² = 72.89, df = 31, p<.001; CFI = .95, RMSEA = .06, SRMR = .046). A chi-square difference test produced a value the critical value for the appropriate degrees of freedom; this the second model without guardianship is considered the “better” model. Thus, the response to RQ1 suggests that individual differences do not contribute significantly to digital guardianship choices—see Figure 2.

Figure 2. The Dark Triad’s relationship with online lifestyle and online crime victimization.

Notes. Figures indicate standardized effects. * = p<.05. ** = p<.001.

To examine these relationships in more detail, further models were created, with all three Traits as exogenous variables, crime as a latent endogenous variable and a single lifestyle aspect as a mediator. Bias-corrected confidence intervals were calculated using 1000 bootstrap samples of the data. An effect is said to be statistically significant if zero is not within the confidence interval produced. Three such models were produced, one for each lifestyle aspect.

Only psychopathy showed a significant indirect effect to crime victimization via deviant online behavior, supporting H3, but not H1 and H2. For attractiveness as a target, all three Dark Triad traits predicted an increase in this variable, and accordingly an increase in victimization, supporting H4, H5, and H6. For normal online activities, narcissism and psychopathy predicted an indirect effect to increased victimization, but Machiavellianism did not, supporting H7, H8, but not H9.

Finally, the same analysis was performed above, for each gender separately. This demonstrated some differences in the indirect effects for males and females, supporting H10. Most notably, no indirect effects were found for the Dark Triad via normal online activities on crime victimization for females. Furthermore, psychopathy showed consistent effects for males across all three lifestyle aspects, but only for an indirect effect via deviant behavior for females—see Table 2.

Table 2. Indirect Effect Pathways for Dark Triad Traits on Crime Victimization via Online Lifestyle Components.

		Confidence interval
Dark Triad trait	Mediator	Overall	Males	Females
Machiavellianism	Normal online activities	β = −0.009, SE = .014,	β  = −™0.038, SE = .031,	β = .018, SE = .024,
		[−0.036, .006]	[−0.120, −0.002]	[−0.001, .087]
Narcissism		β = .034 SE = .020	β = .033 SE = .030	β = .021 SE = .026
		[.003, .065]	[-0.003, .099]	[−0.005, .106]
Psychopathy		β = .039 SE = .024	β = .071 SE = .051	β = .003 SE = .021
		[.012, .098]	[.009, .159]	[−0.016, .062]
Machiavellianism	Deviant online activities	β = .019, SE = .027	β = −0.007, SE = .046,	β = .073, SE = .038,
		[−0.027, .062]	[−0.095, .068]	[.013, .136]
Narcissism		β = .029 SE = .040	β = .029 SE = .058	β = 0.00 SE = .037
		[−0.021, .107]	[-0.057, .125]	[-0.049, .072]
Psychopathy		β = .186 SE = .056	β = .256 SE = .096	β = .077 SE = .038
		[.094, .287]	[.122, .429]	[.031, .164]
Machiavellianism	Attractiveness as a target	β = .060, SE = .036,	β = .051, SE = .065,	β = .080, SE = .041
		[.015, .129]	[−0.046, .183]	[.023, .160]
Narcissism		β = .095 SE = .039	β = .012 SE = .075	β = .101 SE = .046
		[.036, .165]	[−0.118, .120]	[.039, .184]
Psychopathy		β = .187 SE = .052	β = .440 SE = .070	β = .038 SE = .045
		[.093, .281]	[.314, .549]	[−0.037, .109]

Notes. Figures in bold indicate a significant indirect effect. β = standardized beta weights, SE = standard error. Figures in square brackets indicate 95% confidence intervals.

A summary of all hypotheses and whether they were supported can be found in Table 3, and correlations between all variables can be found in Table 4.

Table 3. Summary of Hypotheses.

Hypotheses	Predictor	Mediator	Outcome
H1	Narcissism	Deviant online behavior	Not supported
H2	Psychopathy		Supported
H3	Machiavellianism		Not supported
H4	Narcissism	Attractiveness as a target	Supported
H5	Psychopathy		Supported
H6	Machiavellianism		Supported
H7	Narcissism	Normal online behavior	Supported
H8	Psychopathy		Supported
H9	Machiavellianism		Not supported

Note. For all hypotheses, crime victimization was the outcome variable. H10 was a moderation hypotheses relating to gender and so is excluded from this table.

Table 4. Correlations between the Dark Triad, Lifestyle Activities, Computer Skills, and Online Crime Victimization.

Variable	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.
1.Narcissism
2.Machiavellianism	.31**
3. Psychopathy	.54**	.56**
4. Normal activities	.23**	.10	.23**
5. Deviant activities	.26**	.26**	.41**	.27**
6. Attractiveness as a target	.36**	.32**	.46**	.35**	.39**
7. Skill in everyday tasks	.05	−.03	−.03	.18**	−.03	−.03
8. Skill in complex tasks	.14*	.10	.12*	.32**	.25**	.19**	.39**
9. Information compromise	.13*	.15**	.19**	.10	.26**	.45**	−.04	.07
10. Caught a virus	.13*	.09	.17**	.10	.31**	.31**	−.16**	.04	.40**
11. Lost money	.23**	.13*	.27**	.20**	.33**	.36**	.09	.11	.37**	.20**
12. Software compromise	.16**	.23**	.26**	.147**	.37**	.36**	−.02	.20**	.42**	.44**	.42**

Notes. * = p < .05. ** = p < .001.

Discussion

In this paper, we examined how individual differences may impact on online behavior, which in turn may change the likelihood of crime victimization. Using the Cyber-RAT model as a conceptual basis, we postulated that the Dark Triad of personality would impact on online lifestyle, which in turn would increase crime victimization. We also suggested a moderation of this relationship by gender and investigated the “guardianship” aspect of the RAT model.

Overall, there was strong support for the idea that personality impacts online crime victimization. The Dark Triad was broadly predictive of different kinds of online behavior, which in turn led to a greater likelihood of victimization. Looking at our initial model collapsed across gender, narcissism and psychopathy appear to be significant predictors of online lifestyle overall, whereas Machiavellianism is not. This is perhaps not surprising given that those high in Machiavellianism are supposed to be adept at “reading” social situations and avoid risk (McCutcheon, 2003). By contrast those high in narcissism and psychopathy are likely to demonstrate the lack of care and enjoyment of risk that makes increases the likelihood of victimization.

However, it is apparent that these relationships are greatly diverse across gender, and as such an “overall” picture for Internet users may not be fitting. Instead, the gender of the Internet user has to be considered when trying to predict victimization. As has been found in other papers (e.g. Hartung et al., 2022) intensity of Dark Triad traits were not equal across males and females, with the former scoring higher on both narcissism and Machiavellianism. Males also had a greater online presence for normal and deviant internet activities, and were more attractive as a target. This did not seem to translate into higher crime victimization though, with only software compromise showing a significant difference across gender. When looking at gender within the predictive models, Machiavellianism become a significant predictor in a number of models, and narcissism became non-significant. Psychopathy was perhaps the most consistent predictor across gender. This finding is consistent with previous work which suggests a more difference in trait expression males and females (Rahafar et al., 2017). Moreover, males usually engage in more socially transgressive online behavior (Donner, 2016) and in more risky behavior overall (Byrnes et al., 1999). Although we can say overall that the Dark Triad may influence crime victimization, fully extricating differences in expression across gender will need to be examined in future studies.

We also did not find support for the “guardianship” component in the Cyber-RAT model for explaining crime victimization. In fact, previous work has shown mixed support for this component in an online context, with some researchers also finding no impact of guardianship in crime (Marcum et al., 2010). One reason for this is that quantifying “guardianship” may be challenging. Many computers come with virus-protection pre-installed, and the public understanding of the risks of viruses and how to avoid them is fairly high (Seybert & Lööf, 2010). In the original conceptualization of the RAT, guardianship referred to physically interceding between the criminal and victim. In an online context, researchers have maintained this idea when looking at parental involvement with children’s Internet browsing, to avoid issues with cyberbullying (Navarro & Jasinski, 2012). When applied to adults, its importance may be lessened. Nevertheless, given its prominence in the Cyber-RAT it is worth examining this aspect in closer detail in future studies to properly pin down its impact.

Measurement Issues in the Current Work

A challenge when measuring as diverse concept as “internet use” is that creating a complete list of all possible behaviors is challenging if not impossible. In this paper, we have used previous work on these topics to provide sufficiently rigorous measure of how people behave online but, of course, these are not exhaustive. New technologies and the ever-developing sophistication of the Internet means that these measures may become outdated over time. Researchers should be vigilant to this matter and closely examine whether any changes are required when performing research in future.

The measure we have used for the Dark Triad are validated and robust; however, other scales are available. In particular, the NPI-13 gives an overall measure of narcissism but can also be broken down into three sub-scales. Grandiose/exhibitionist narcissism relates to self-aggrandizement and showing off, leadership/authority narcissism to demonstrating power over others, and entitlement/exploitative narcissism to manipulating others to get one’s way. Each of these are quite different and can alter behavior in different ways (e.g. Kajonius & Björkman, 2020). Psychopathy too can be separated into several sub-scales depending on the measure used. The Self-Report Psychopathy Scale version 3 (SRP-III) is perhaps the most well-known (Gordts et al., 2017), detailing four sub-scales of antisocial behavior, impulsive thrill-seeking, interpersonal manipulation, and cold affect. Thus, we may have more diverse measures available to look at the Dark Triad in more detail. It should be noted that this level of granularity may compromise the brevity of a questionnaire; most tools to measure psychopathy for example are well over 100 items.

Finally, work on maladaptive traits has suggested the Triad should be extended to a Dark Tetrad, which adds sadism to the collective (Međedović & Petrović, 2015). Work has already shown that this trait may offer a unique predictive path for attitudes and behavior (e.g. Tsoukas & March, 2018). Furthermore trait sadism has also been identified as a factor in some social media behaviors such as trolling others on Facebook (Craker & March, 2016). As the Dark Tetrad has not yet penetrated the psychological literature to the same extent as the Dark Triad, it is difficult to say how sadism may relate to online crime victimization. Nevertheless, it may still be worth including a measure of sadism in other work.

Future Work on Cybercrime

This work is one of the first to demonstrate that certain personality traits may impact on the likelihood of becoming a victim of online crime, through the type of online activities an individual engages in. This further expands our understanding of victimization and the RAT model by showing how individual differences are an important component in predicting crime victimization. With this identified, we may then ask—how can this knowledge help reduce victimization?

Ironically, the specific traits being examined here—i.e. the Dark Triad—may make this problematic. Narcissists are known for perceiving others as less intelligent and capable than themselves (Krizan & Bushman, 2011). Those high in psychopathy too tend to view others as “lesser” (Morrison & Gilbert, 2001), and Machiavellianism is associated with retaining less feedback and finding it less useful (Malachowski et al., 2013). Thus, if we identify those high in the Dark Triad and attempt to warn them of their increased susceptibility to crime, those individuals are less likely to listen and act on this advice. Little research has examined how to increase advice taking in those high in Dark Triad traits, however evidence suggest narcissist do not listen to advice from others because they believe others are incompetent (Kausel et al., 2015). Efforts to increase perceptions of competence in others may help to encourage narcissists to listen to advice.

Another barrier to reducing victimization may be the lack of instant gratification. If factors such as pirating movies online increase the likelihood of victimization, then to reduce this would require the individual to stop taking part in such activities. The Dark Triad has been associated with an inability to delay gratification (Baughman et al., 2016) and it may be that the instant gratification of acts such as pirating movies outweighs the long-term gratification, which in this case would be not becoming victim to cybercrime. Perhaps here the solution may be to first improve self-control in turn affecting the ability to delay gratification.

In this paper, we have only examined the occurrence of crime, but of course there are other important metrics of victimization to be considered—namely money and time. Crime victims can lose a considerable amount of both if they are targets for criminals. It may be interesting therefore to examine measures of loss in future work. Money can be quantified fairly easily if it has been stolen from an account; time however may be more challenging as there is little clear record of how long it took an individual to “solve” a problem. The spread of responses to these measures may also be considerable; victimization could result in only a small inconvenience of time lost if a simple password reset is required, but there could also be a cost of several days if a virus means a new hard drive must be reformatted and programs reinstalled. As these metrics are the ultimate “cost” of most online crime, their inclusion of a model looking at victimization is very important in subsequent studies.

It should also be noted that the Cyber-RAT does not necessarily offer a complete view of crime victimization. Other researchers have noted further influential variables; for example Yaghoobi et al. (2016) examined the role of adult attachment styles in crime victimization likelihood. Securely attached individuals were less likely to be victims of cybercrime compared with those with an anxious attachment style. Although the RAT has provided an illuminating picture in this work, clearly, there are other aspects of personality that may be important in predicting victimization.

Finally, we have attempted to cover some of the most widespread crimes in this paper, based on prevalence reports from published data. However, it would be very difficult—perhaps impossible—to cover all instances of cybercrimes that may occur. Furthermore, the nature of online crime is constantly evolving as new technologies emerge that allow different ways of interacting with—and exploiting- others. For example, a high prevalence of revenge porn has been evidenced (Marganski & Melander, 2018; Walker & Sleath, 2017), wherein sexualized images of an individual are made available to others without the model’s consent as a form of retaliation concurrent with the increase in phone camera technology and internet availability. We have also seen in our work how the application of the “guardian” component can be difficult in a cybercrime context. Any model of victimization may need to be malleable to take account these new crimes as they emerge.

Conclusions

This paper has empirically demonstrated how personality traits may impact on cybercrime victimization, by dint of their influence on online activities outlined by Cyber Routine Activities Theory. It has extended previous work by highlighting other individual differences important in cybercrime; it has also further highlighted some of the difficulties of using the traditional “guardianship” component of the RAT in modern crime. In future work, we suggest using additional measures of victimization—namely, money and time lost through crime—and to considering the flexibility of any psychological model to take into account new crimes generated by technological advancement.

Ethical Approval

Informed consent was obtained from all participants in this study. This study met with the British Psychological Society ethical guidelines for human participants, and with the ethical standards of the host institution.

Author Notes

Chris Stiff, Keele University, obtained his degree and PhD from the University of Southampton in 2005. After working at post-doc positions in Bristol and Nottingham University, he began a lectureship at Keele in 2009. Chris is now a senior lecturer in the School of Psychology.

Meike Reeves currently works as a Psychological Wellbeing Practitioner for Cornwall partnership NHS trust. She graduated with a degree in Psychology from the University of Southampton and went on to graduate from Keele University with a Masters degree in Cognitive Psychology.

Disclosure Statement

The authors declare there are no conflicts of interest.

Data Availability Statement

Data for this study is available from the authors on request.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.