Depending on the Dark Triad: exploring relationships between malign personality traits, substance and process addictions

ABSTRACT

Background and objective

The present study aimed to contribute to the understanding of addiction formation and maintenance by exploring the relationships between Dark Triad (DT) personality trait levels and addiction tendencies to “substances” and “processes.” Incorporating an adapted “Addiction as Excessive Appetites” model within a biopsychosocial framework, relationships between DT trait levels, alcohol, and exercise addictions were examined.

Methods

Two hundred and eighty-nine adults living in the UK (192 women, 97 men) completed an online questionnaire measuring DT trait levels and indicative addiction scores to alcohol (substance) and exercise (process). These were interrogated using correlation analysis, group mean comparisons and multivariate analysis of variance (MANOVA).

Results

Significant positive correlations were found between psychopathic, grandiose narcissistic and Machiavellian trait levels with addiction scores to alcohol and exercise. MANOVA demonstrated significant main effects of alcohol and exercise with respect to DT trait levels. In most cases, those in the higher addiction risk groups showed higher psychopathic and Machiavellian trait scores.

Conclusions

Given the positive associations between DT trait levels and addiction and elevated DT levels in highest risk groups, the data was consistent with the proposed model. Machiavellianism trait levels were more strongly associated with process and substance addictions than found in previous studies.

KEYWORDS:

• Dark
• Triad
• alcohol
• exercise
• addiction

Introduction

First named by Paulhus and Williams (2002), the Dark Triad (DT) is a postulated concept incorporating three universal traits at the sub-clinical level (psychopathy, narcissism, and Machiavellianism). It has been used to explain a broad range of human behaviors including happiness (Egan et al., 2014), friendship selection (Lyons, 2019) and entrepreneurial motives (Hmieleski & Lerner, 2016). However, one area that has been relatively unexplored is the relationship between the DT and addiction. Whilst a recent literature review found evidence for positive associations between psychopathy and addictions to substance and non-substances, it concluded associations between Machiavellianism and addiction were “unclear” (Jauk & Dieterich, 2019). Analogously, addictions to non-substance processes or behaviors are less widely explored and are rarely acknowledged as contributing toward clinical disorders (Rennert et al., 2014; Zendle & Bowden-Jones, 2019). Overall, few, if any, studies have examined the role of “all” DT levels as potential psychological determinants of addiction formation and attachment whilst also distinguishing between substance and process addiction types. This paper explores the potential interdependence of these concepts within an adapted addiction as excessive appetites psychological model (Orford, 2001a).

Whilst it is perhaps intuitive to suspect that addictions to a “bad for you” poison (alcohol) and a “good for you” endeavor (exercise) have little in common, a closer examination reveals that when both activities are taken to excessive or addicted levels, they are more closely related than preconceptions predict. Six core component symptoms and consequences of exercise addiction that have been identified as salience, mood modification, tolerance, withdrawal symptoms, personal confiict, and relapse (Griffiths, 2005), also feature in alcohol addiction (Alcohol Change, 2018; NHS, 2018). Both exercise and alcohol cause the release of dopamine, endorphins, and serotonin (Fischer & Ullsperger, 2017; Renzoni, 2020), commonly referred to as the pleasure principle (Powledge, 1999; Shaffer & Corliss, 2017). The addictive qualities of these responses are so strong they have been referred to as the “addiction pathway” by neuropsychologists (Powledge, 1999). This process demonstrates that distinctions between “substance” and “process” addictions are inherently complex, given that even a process addiction (such as exercise) has substance (neurotransmitters) elements.

Biopsychosocial and adapted addiction as excessive appetites model

The biopsychosocial model is a conceptual amalgamation of various risk factors that can contribute to addictive behaviors. Spheres of psychological, biological, or social factors demonstrate a range of possible influences which can increase the risk of addiction (See, Figure 1)

Figure 1. Biopsychosocial risk of addiction model, with Dark Triad positioning. Adapted from the biopsychosocial model of substance use problems (The Centre for Addiction and Mental Health (CAMH), 2020).

. This becomes particularly great in overlapping areas, such as genetic predisposition (biological) being present with post-traumatic stress disorder (psychological). Although the biopsychosocial model presents a useful snapshot of DT positioning within overall risk spheres, to understand how they may translate to addictive behaviors it is necessary to examine dynamic psychological processes also. The addiction as excessive appetites model (Orford, 2001b) provides such an explanation. Addiction, it is proposed, flows through a series of stages beginning prior to taking up the activity through to attachment and commitment. Orford stated that any activity can become addictive, of which alcohol is a special case. As it is hypothesized that DT personality trait levels are an element of personality, and that Orford’s (2001b) model demonstrates personality influences all stages of addiction, the present study proposes that DT trait levels will also influence addiction stages. Adjusting the model to incorporate the potential influence of the DT and the psychological-biological affective response (pleasure principle) provides a hypothesized better fit to explain addiction progression for both alcohol and exercise simultaneously (see, Figure 2).

Figure 2. Adapted addiction as excessive appetites model.

The adjusted features reflect an appreciation of the broad range of interdependent biopsychosocial factors that affect the decision to take up or cease a behavior at every stage. In phases one to four, the addiction flows are still likely to be highly similar for substance and behavioral addictions, as the driving factors and the reward systems in operation are the same mechanisms. Therefore, DT traits will affect all but physical dependency phase for alcohol and all stages for exercise addiction.

The present study tackles two research questions:

1. Are component elements of the DT differentially associated with increased risk of addiction tendency?

In concurrence with existing evidence outlined by Jauk and Dieterich (2019), it was predicted that significant positive associations between each addiction type and some DT traits would be found, but that the strength of associations would differ depending on each trait. If found to be the case, then DT traits could be declared as being differentially associated with increased risk of addiction tendency.

• (2) Does the type of addiction mediate the relationship with the DT traits?

By examining DT associations with alcohol and exercise addictions the current study provides an opportunity to test the appropriateness of the adapted addiction as excessive appetites model within the biopsychosocial framework. If the pleasure principle and model are accurate reflections of the etiology of both addiction forms, then same directional associations between DT traits and them will be observed. Therefore, it was hypothesized that addiction type would not mediate the relationship with DT traits. Instead, where one DT trait was associated with alcohol addiction then this would be mirrored with same directional associations with the same trait and exercise addiction. This hypothesis contradicts current evidence where positive associations between narcissism and non-substance addictive behavior and no associations between narcissism and substance use disorder have been identified (Jauk & Dieterich, 2019).

Materials and methods

A combination of opportunity sampling and access through gatekeepers was used to recruit 289 participating adults who each completed an online self-report survey. This composed of 192 (66%) women and 97 (34%) men. Ages ranged from 18 to 79 years old (M = 39.8). Gatekeepers involved in drug and alcohol recovery and sporting organizations ensured participation from individuals in the higher exercise and alcohol addiction behaviors categories.

Qualtrics was used to deploy the Dark Triad Index 2 (DTI-2) (Lawson et al., In prep), the Alcohol Use Disorders Identification Test (AUDIT; Babor et al., 2001) and Exercise Addiction Inventory (EAI; Griffiths et al., 2005). These provided indicative scores for the six dependant variables for each participant: DT personality trait levels (x4) and measures of their current addictive behaviors.

The DTI-2 (Lawson et al., In prep) contained 60 items; 20 (across 4 subscales) on psychopathy and Machiavellianism and 10 (across 2 subscales) each for positive (grandiose) narcissism and negative (vulnerable) narcissism. Participants indicated a “level of agreement” to each statement on a 6-point Likert scale ranging from strongly disagree (0 point) to strongly agree (5 points). DTI-2 items were randomized, and four items were reverse scored to reduce potential order effects. No such randomization was put in place for the alcohol or exercise addiction scales to comply with the conventions of the application of these scales.

The AUDIT (Babor et al., 2001) examined alcohol addiction risk and is widely used to screen for dependency and alcohol use disorders (Lavoie, 2016). It contains 10 items, split between hazardous use, dependence symptoms and harmful consumption domains. The sum total of all items indicates addiction risk. Scores between 0 and 7 represent lower risk, 8–15 increasing risk, 16–19 higher risk, and 20+ possible dependence. AUDIT has been found to be robust and accurate (Allen, Litten, Fertig & Babor, 1997; Foxcroft et al., 2015). Assessments demonstrate strong Cronbach’s alpha scores (α > .80), indicating high internal consistency (Allen et al., 1997). Furthermore, the scale is described as “outstanding” and “excellent” in identifying dependency (Lundin et al., 2015).

The EAI contains Likert-style items assessing the six component symptoms and consequences of addiction. Each are scored between 1 and 5, the sum of which provides the addiction risk score. Scores range from 6 to 30, with those above 23 indicating potential dependence. The EAI has been found to have high internal reliability (Cronbach’s α = .84; Terry et al., 2004), high content, concurrent and construct validities (Lichtenstein & Jensen, 2016; Terry et al., 2004).

Pilot and live study respondents were briefed on the nature of the research, and that this had received approval from Psychology Research Ethics Committee, Oxford Brookes University (1920/137). Due to an elevated risk of potential demand characteristics and response bias associated with research on ’bad’ personality traits and stigma associated with addiction, it was necessary to be slightly vague about the true nature of the research. Respondents believed they were taking part in a “personality and free-time activities” survey, rather than a DT and addictions analysis. A post-survey debrief explained the nature and reasons for the deception and further consent was obtained from participants.

Results

Descriptive statistics

DT scores ranged from 13 to 72 (psychopathy), 15 to 74 (Machiavellianism), 0 to 46 (grandiose narcissism) and 5 to 49 (vulnerable narcissism). Histograms and Q–Q plots demonstrated no obvious violations of normality for psychopathy and Machiavellianism, supported by non-significant Shapiro–Wilk statistics, low skew, and kurtosis. Significant Shapiro–Wilk statistics for narcissism (pPOSITIVE NARCISSISM 0<.05, pNEGATIVE NARCISSISM 0<.01) suggested violations of normality. Histograms and Q–Q plots demonstrated imperfect normal distributions in leptokurtic (positive narcissism) and platykurtic (negative narcissism) deviations. However, small skew and kurtosis values were observed. Data skew divided by the standard error was −1.71 (positive narcissism) and 0.93 (negative narcissism), lower than the recommended acceptable cutoff point of 2, identified by Navarro and Foxcroft (2019).

AUDIT produced continuous data ranging from 0 to 40. More than half (n = 158, 55%) were in the lowest risk (0–7) category, approximately a third (n. = 94, 33%) scored as increasing risk (8–15), and fewer were categorized as higher risk (n = 17, 6%) (16–19) or possibly dependent (n = 20, 7%) (20+). Large skew (1.51, SE = 0.14) and kurtosis (3.22, SE = 0.29) were observed. A highly significant Shapiro–Wilk statistic (p < 0.001) and deviations of residuals in Q–Q plot data demonstrated normality had been violated.

The EAI produced continuous data with a range of 24 (minimum = 6, maximum = 30). Potential exercise addicted scores were identified in 36 (12%) respondents.

Some deviations from normal was observed in histogram, and Shapiro–Wilk data (p < 0.05). However, small skew and kurtosis values and Q–Q plot residuals did not indicate strong deviations.

Dark Triad and addiction associations

Substantive and significant positive correlations were observed between DT trait levels and AUDIT scores, and DT trait and EAI scores. This was true for all combinations of relationships apart vulnerable narcissism and EAI (See, Table 1).

Table 1. Dark Triad and addiction associations.

Dark Triad and addiction relationship	Spearman’s Rho (rs)	Significance (p)	Coefficient of determination (rs^2)
Psychopathy and AUDIT	.37	p < 0.001	0.14
Machiavellianism and AUDIT	.29	p < 0.001	0.08
Grandiose narcissism and AUDIT	.13	p < 0.05	0.02
Vulnerable narcissism and AUDIT	.14	p < 0.05	0.02
Psychopathy and EAI	.18	p < 0.01	0.03
Machiavellianism and EAI	.15	p < 0.01	0.02
Grandiose narcissism and EAI	.16	p < 0.01	0.03
Vulnerable narcissism and EAI	.05	p = ns	<0.01

Addiction type and Dark Triad scores

Where substantive and significant correlations between AUDIT and DT trait scores were identified, this was mirrored in same directional correlations between EAI and the same DT trait. For example, correlations between Machiavellianism and AUDIT (r_s = .29, p < 0.001, N = 289) compared with Machiavellianism and EAI (r_s = .15, p < 0.01, N = 289).

Inferential statistics

AUDIT and EAI variable scores were used to create two new independent variables AUDIT Risk and EAI Risk, each with two levels: High/Other, using cutoff point recommendations outlined by Public Health England (2017) and Terry et al. (2004). Thirty-seven respondents scored 16 or higher AUDIT points and 36 individuals scored greater than 23 EAI points. Mean DT scores in High and Other risk category levels were calculated for comparative analysis.

A two-way factorial Multivariate Analysis of Variance (MANOVA) was conducted to explore for any group differences (High v Other AUDIT and High v Other EAI) on the DT trait levels. Significant main effects were produced for both alcohol and exercise addiction types against DT trait scores (Pillais’ TraceAUDIT V = .076, F(4, 282) = 5.830, p < 0.001 and Pillais’ TraceEAI V = .042, F(4, 282) = 3.089, p < 0.05, with DT mean scores higher in High-risk than Other-risk level respondents. The multivariate effect sizes were estimated to be .076 (AUDIT) and .042 (EAI) meaning that 7.6% of DT trait score variance was accounted for by the AUDIT High versus Other variable levels and 4.2% by the EAI High versus Other variable levels. No significant interaction effects occurred.

Univariate testing revealed significant effects of alcohol risk on psychopathy and Machiavellianism (p < 0.001 for both), and exercise risk on psychopathy and Machiavellianism (p  <.02 and p < 0.01, respectively). Scores were all greater in the High than in the Other group levels. A significant effect was produced between EAI risk and positive narcissism (p < 0.05), with higher positive narcissism scores present in the higher level (see, Figure 3). Insignificant effects were produced for the remaining univariate combinations of addiction risk and narcissism.

Figure 3. Dark Triad scores across addiction type and risk.

Discussion

Are component elements of the Dark Triad differentially associated with increased risk of addiction tendency?

It was predicted that positive and significant associations between levels of Dark Triad (DT) traits and addiction tendencies to alcohol and exercise would be found. Further, that these associations would differ depending on the trait in question. Overall, the results strongly supported the predictions. Weak to moderate positive correlations between each of the DT traits and AUDIT and EAI scores were found (with only exercise and vulnerable narcissism bucking the trend). When split into High/Other dichotomous risk category levels, MANOVA analysis confirmed that those in the High-risk groups displayed significantly higher psychopathy and Machiavellianism trait scores than in the Other group. However, univariate tests revealed the picture was not so clear for either form of narcissism. Positive narcissism was significantly associated with increased risk of exercise addiction, but the remaining narcissism/addiction combinations were not significant.

Comparing the High against Other AUDIT levels demonstrated higher DT scores in the High group levels with the biggest mean differences occurring in psychopathy, closely followed by Machiavellianism. For exercise, mean difference DT traits scores for those in the High-risk compared with Other level were similar for psychopathy and Machiavellianism. The difference in positive narcissism scores, whilst still substantive and significant, was smaller. Therefore, as predicted, the traits were not behaving as a unified entity and instead their relationships with the two addictions differed. Combining all the evidence led to the conclusion that elements of the DT were associated with addiction tendency in a structurally similar way but that the specific associations differed depending on the DT trait in question. Therefore, DT traits were found to be differentially associated with increased risk of addiction tendency.

Does the type of addiction mediate the relationship with the DT traits?

All significant correlations between AUDIT and DT traits were positive, as were those between the EAI and Dark Triad traits (apart from exercise and negative narcissism).

When examining the High-risk and Other groups, MANOVA results demonstrated that these categories had significant main effects on DT scores for both alcohol and exercise addictions. Mean DT scores were significantly higher in the High-risk compared with Other group level.

Clear and consistent patterns of resultant DT trait scores regardless of addiction type, led to the conclusion that addiction type did not significantly mediate the relationship with any of the DT traits. This was in keeping with the predictions made via the adapted addiction as excessive appetites model.

Critique of the present study

The evidence demonstrated elements of concurrence and contradiction with the preexisting consensus summarized by Jauk and Dieterich (2019)(See, Figure 4)

Figure 4. Dark Triad and addictions results comparison.

. Positive associations between psychopathy and both addiction types are reported in both studies. However, notable differences include clear associations with both alcohol and exercise addictions and Machiavellianism and a positive association between vulnerable narcissism and alcohol in the present study, neither of which are identified by Jauk and Dieterich (2019).

The discrepancies between previous research and results may be due to a wide range of possibilities. It could be that the results reflect genuine patterns of behavior not previously found by researchers or that these are due to the specific characteristics of the sample, which may not be representative of the wider population. The sampling strategy utilized to capture sufficient respondents from extreme ends of alcohol and exercise addiction, for example, may have skewed the results.

This paper has provided new information on Machiavellianism and addiction and demonstrated clear similarities between DT trait scores across addiction types and addiction risk level. However, whether process addictions are real and should be recognized as such is a subjective issue and evidence on this was mixed. Addiction indicator score distributions demonstrated differences between alcohol and exercise addictions. However, DT-level changes from the Other to High risk category levels showed tremendous similarities. However, this pattern is not proof that all process addictions would demonstrate the same relational pattern as substance ones because only one substance and one process was investigated. It would be useful in future research to explore more combinations of process and substance addictions to discover whether the patterns of behaviors uncovered by this research are specific to alcohol and exercise exclusively or fit a more general pattern between other process and substance addictions. It may also be interesting to examine whether DT trait levels demonstrate greater or lesser connectedness with other specific process or substance addictions.

Acknowledgments

Grateful acknowledgments go to Kirsty Walter, Oxford Brookes University, who assisted with the formatting of the online survey using Qualtrics and nine gatekeepers from alcohol and drug recovery groups and sporting organizations who assisted in recruitment of participants.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

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