Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Mar 1.
Published in final edited form as: Psychol Addict Behav. 2021 Nov 11;37(2):318–330. doi: 10.1037/adb0000793

Big Five Personality Traits and Illicit Drug Use: Specificity in Trait-Drug Associations

Genevieve F Dash 1, Nicholas G Martin 2, Wendy S Slutske 1
PMCID: PMC9091062  NIHMSID: NIHMS1748031  PMID: 34766786

Abstract

Objective.

High neuroticism, low agreeableness, and low conscientiousness are consistent correlates of drug use, though such patterns may be due to common familial influences rather than effects of personality per se. The present study aimed to explore associations of Big Five traits with various forms of drug use independent of confounding familial influences by leveraging differences within twin pairs to identify potentially causal (i.e., within-pair) effects of personality on use.

Method.

980 same-sex twin pairs from the Australian Twin Registry Cohort III (Mage=31.70, 71% female) were interviewed regarding lifetime (mis)use of cannabis, cocaine/crack, prescription and illicit stimulants, prescription and illicit opioids, sedatives, hallucinogens, dissociatives, inhalants, and solvents, and completed a Big Five inventory. Co-twin control analyses predicted use of each drug from all traits simultaneously.

Results.

Individual-level analyses generally showed the expected associations of neuroticism, agreeableness, and conscientiousness with drug use. Familial effects were also somewhat generalized: high neuroticism, high openness to experience, and low agreeableness were associated with use of several drug types. More specificity emerged for within-pair effects. High neuroticism was associated with prescription drug misuse; high extraversion was associated with cocaine/crack and stimulant use; high openness to experience was associated with cannabis use; low agreeableness was associated with cocaine/crack use and illicit opioid use; no within-pair effects emerged for conscientiousness.

Conclusions.

Trait associations common across drugs may be primarily attributable to familial effects. There appears to be more drug-specific influence of personality on use with respect to potentially causal within-pair effects.

Keywords: illicit drug use, prescription misuse, personality, Big Five, twin study

Personality traits have long been posited as a factor underlying substance use and substance use disorder, with high neuroticism, low agreeableness, and low conscientiousness emerging as consistent correlates across studies (Dash et al., 2019; Kotov et al., 2010; Kroencke et al., 2021; Slutske et al., 2005; Sutin et al., 2013; Terracianno et al., 2008). Several explanations for this have been posited. Individuals high in neuroticism are prone to negative affect (Lahey, 2009), and using psychoactive drugs is one method of relieving such experiences of subjective distress (Cooper, 1994; Simons et al., 1998). Individuals low in agreeableness tend to be relatively less concerned with social approval and are more likely to engage in antisocial behaviors, which may manifest in the form of illicit drug use (Costa & McCrae, 1992; Sutin et al., 2013). Individuals low in conscientiousness often display higher levels of impulsivity and lower levels of health-oriented behavior, and may therefore be more likely to engage in potentially hazardous behaviors such as drug use (Raynor & Levine, 2009). There is also evidence that personality domains and drug use have some degree of shared genetic etiology (Agrawal et al., 2004; Littlefield & Sher, 2016; Slutske et al., 2002). This may function such that genes fostering liability to higher neuroticism, lower agreeableness, and/or lower conscientiousness are also those that increase risk for drug use, rather than reflecting a causal effect of the phenotypic expression of certain traits on drug use behaviors.

However, shared genetic influence cannot fully account for associations between personality traits and substance use, and these explanations do not address findings demonstrating some degree of specificity in trait-drug associations (Fehrman et al., 2017; Mahu et al., 2019). Despite genetic overlap between personality and substance use, meaningful sources of potentially differential covariance are left unaccounted for by shared genes: non-shared environmental influences (and error) account for approximately one-third to one-half of the variation in both Big Five traits (Vernon et al., 2008) and illicit drug use (Karkowski et al., 2000; Kendler et al., 2005; Kendler et al., 2000); further, a substantial proportion of the covariance between many trait-drug pairs is attributable to non-shared environmental influence (Agrawal et al., 2004). Such sources of influence may be responsible for deviations from more general patterns of trait-drug association; that is, genetic liability for illicit drug use may operate more generally across drug types (Jang et al., 1995; Kendler et al., 2007), while non-shared environmental liability may be more varied in its influence across drug classes.

Associations between sensation seeking and stimulant use, hopelessness and opioid use, and anxiety sensitivity and tranquilizer use (Mahu et al., 2019) suggest that there is some level of meaningful specificity in the relationship between personality traits and use of particular drugs. Individuals may- consciously or not- select to recreationally use drugs with subjective effects that complement their trait-level cognitive and emotional processes, such that differences in the subjective effects of various drugs foster unique instrumental reasons for use that differentially appeal to individuals of varying personality profiles (Conrod et al., 2000). Consistent with findings that coping motives for substance use are associated with higher trait neuroticism (Chowdhury et al., 2016; Kuntsche et al., 2006), individuals with higher trait neuroticism may be more likely to use substances such as opioids and sedatives, which serve to mitigate both physical and psychological distress (Benotsch et al., 2013; Chinneck et al., 2018; Delić et al., 2017; Fehrman et al., 2017; Mahu et al., 2019). The social facilitation aspect of drugs such as ecstasy, which imbues feelings of social connectedness, empathy, and intimacy, may create particular appeal to individuals high in trait extraversion (Fehrman et al., 2017; ter Bogt et al., 2006; Vreeker et al., 2020). The association of psychedelic drug use (cannabis, hallucinogens) with trait openness to experience (Dash et al., 2019; Erritzoe et al., 2019; LaFrance & Cuttler, 2017) may be explained by the novel perceptual experiences that these drugs tend to facilitate. Individuals high on trait openness to experience are inclined toward seeking new experiences and engaging in introspection (McCrae, 1993; Simons et al., 1998), and expansion motives for use (i.e., expansion of perceptual and cognitive experience) have indeed been found to mediate the relationship between openness to experience and cannabis use (Hawkins, 2012).

Present Study

Although non-twin studies of personality correlates of drug use behaviors are informative, it is also important to address this topic within a design addressing potential genetic and familial confounds inherent in data from unrelated individuals. The present study sought to examine specificity in personality trait-drug use associations by implementing a co-twin control design to explore potential quasi-causal effects of personality traits on different forms of drug use, building on the relatively sparse extant behavior genetic literature in this area (e.g., Agrawal et al., 2004) by including all Big Five traits and wide spectrum of drugs. It was expected that 1) high neuroticism would be quasi-causally related to use of drugs that mitigate physical and/or psychological distress (opioids, sedatives), 2) high extraversion would be quasi-causally related to use of drugs commonly used in social and club settings (cocaine, amphetamine-based stimulants, inhalants [e.g., “whippits”]), and 3) high openness to experience would be quasi-causally related to use of drugs with psychedelic properties (cannabis, hallucinogens). Other probing of personality-drug associations was exploratory.

Methods

Participants and Procedure

Participants were 980 complete same-sex twin pairs of known zygosity (396 monozygotic [MZ] female; 169 MZ male; 299 dizygotic [DZ] female; 116 DZ male) from the Australian Twin Registry Cohort III (Mage=31.70 [SD=2.48], range=27–37 [one twin pair was age 40]; 71% female). It is not standard for Australian demographic surveys to query race and ethnicity; thus, only data on participant ancestry was available. A majority of the sample (84%) reported UK ancestry (Britain, Scotland, Wales), and less than 2% reported indigenous Australian ancestry. See Lynskey et al. (2012) for more information about participants.

Informed consent was obtained from all participants prior to data collection. Participants were surveyed by computer-assisted telephone interview (CATI) in 2005–2009 (participation rate=76%) and a follow-up survey administered via the internet or a mailed paper-and-pencil questionnaire (completion rate=94%). The original data collection was approved by the Washington University and QIMR-Berghofer Institutional Review Boards, and secondary analysis was determined to be exempt by the University of Missouri Institutional Review Board.

Measures

Demographics

Participants were asked to report their biological sex, age, marital status, and educational attainment. Participants were also queried regarding their family’s relative financial stability compared to the average family in the community (“better off,” “about average,” or “worse off”) from when they were ages 6–13.

Big Five Personality Traits

Big Five personality traits were assessed via self-report survey within 2 weeks of the CATI interview using a 74-item version of the NEO-FFI, which has been used in several past studies (e.g., Dash et al., 2019; Davis et al., 2020; Schermer & Martin, 2019). The questionnaire was comprised of the original 60 NEO-FFI items, as well as 14 items pulled from the remaining pool of NEO-PI-R items. These items were selected on the basis that no more than two-thirds were keyed in the same direction, that they were highly correlated with their NEO-PI-R factor scores, and that they adequately represented all scale facets; psychometric evaluation demonstrated that these items improved reliability and factor structure while retaining validity (McCrae & Costa, 2004). Items were scored on a 1 (strongly disagree) to 5 (strongly agree) scale. Scores were generated by computing the item means for each scale. Reliabilities were acceptable for neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness (α=.89, .84, .77, .82, and .85, respectively).

Drug Use

Participants were provided with a respondent booklet containing nine lists of drugs, with each list corresponding to a class of drug (cannabis, cocaine/crack, amphetamine-based stimulants, opioids, sedatives [prescription benzodiazepines], hallucinogens, dissociatives, inhalants, and solvents). Each list contained a comprehensive collection of drugs described by name and by common slang terms, where relevant; the stimulant and opioid classes contained both illicit drugs and prescription drugs with potential for misuse. Participants were asked “Have you ever used any of the items in List [1–9]?” Participants who endorsed use were asked “How many times in your life have you used anything from List [1–9]?” with the lists queried corresponding to those endorsed by the respondent in the previous question. Respondents were also asked which specific drug(s) on the list they had used, or, for lists containing prescription drugs, “used when not prescribed or more than prescribed” (i.e., misuse). As number of lifetime uses was queried as a sum total across all substances within a list, only dichotomous (yes/no) indicators were available for drug subtypes (i.e., prescription and illicit opioids and stimulants).

Analytic Plan

The Co-Twin Control Design

The co-twin control design leverages similarities between twins to create an ex post facto analog to the gold-standard counterfactual experimental design, making it particularly useful for strengthening causal inference in observational data (McGue et al., 2010). Under the classical twin model, MZ co-twins fully share their segregating genes (rg=1.00), and DZ co-twins share, on average, 50% of their segregating genes (rg=.50); co-twins of both zygosities share a common environment (i.e., factors which make twins more similar; rc=1.00); and non-shared environments (i.e., factors which make twins more different plus error) are assumed to be unique to individual co-twins across both zygosities (re=.00). Because MZ co-twins are assumed to be perfectly matched for both genes and common environment, differences between them can be inferred to be the result of non-shared environmental influences. Put simply, the co-twin control model partitions familial (genes, shared environment) and non-shared environmental variance of a predictor (e.g., personality traits) so as to permit examination of the relationship between that predictor and an outcome (e.g., drug use) free of familial confounding. Differences between co-twins (i.e., within-pair differences), who are by nature matched on familial factors, can be inferred to be a potentially causal, or “quasi-causal,” result of non-shared environmental variance in the predictor.

Additionally, the degree of familial confounding present can be ascertained by comparing models comprised of individuals of varying degrees of relatedness. This process aids in determining which observed effects are attributable to familial factors rather than reflecting a potential quasi-causal association between phenotypes: models comprised of unrelated individuals do not control for familial confounding, models comprised of DZ twin pairs partially control for familial confounding, and models comprised of only MZ twin pairs completely control for familial confounding (McGue et al., 2010). Thus, if the magnitude of effect across samples comprised of unrelated individuals, DZ twin pairs, and MZ twin pairs is equivalent, one can infer that there is no familial confounding. Conversely, if the magnitude of the effect is reduced as the genetic relatedness of the subjects increases to the point of elimination of the effect among MZ pairs, one can conclude that there is complete familial confounding. If the magnitude of the effect is reduced as the genetic relatedness of the subjects increases, but the effect is not eliminated among MZ pairs, there is some level of familial confounding in addition to a potential effect of the exposure. When modeled statistically, several other potentially explanatory predictors (e.g., biological sex, age, marital status, educational attainment, socioeconomic status) are also included in the model to more stringently test the quasi-causal effect. A within-pair effect that significantly predicts the outcome of interest even after accounting for the effects of covariates provides further confidence in the potentially causal relationship between the predictor and outcome.

Statistical Power in the Co-Twin Control Design.

Though information on power to detect effects in the co-twin control design is limited, simulations suggest that 1) a sample size of ~1,000 twin pairs is adequate to detect small effects (d=0.10–0.15), 2) a higher MZ:DZ sample size ratio increases power, and 3) a higher proportion of variance attributable to non-shared environmental influences in both predictor and outcome variables increases power (de Moor et al., 2011). The present sample size of 980 pairs approximately meets the recommended threshold; the MZ:DZ ratio was >1:1 (1.36:1), and non-shared environmental variance was significant for all personality and drug use variables (up to 96% of the phenotypic variance). Though not specific to twin data, power calculations for a 1:1 matched case-control design showed that the combined zygosity sample size was sufficient to detect an odds ratio of 1.40 or higher, and that the MZ sample size was sufficient to detect an odds ratio of 1.55 or higher with 90% power (http://sampsize.sourceforge.net/iface/s3.html#cc; Vreeker et al., 2020).

Statistical Analyses

Analyses were conducted using SAS version 9.4 (SAS, Inc., 2014). To examine associations between Big Five traits and drug use while accommodating the clustered nature of the data (i.e., twins nested within pairs), two-level random intercept generalized linear mixed models were run using PROC GLIMMIX. Personality predictor variables were coded to test within-pair effects (i.e., comparison of twin and co-twin’s trait score deviations from their pair average) and between-pair effects (i.e., comparison of twin pair trait score averages across twin pairs) (Slutske et al., 2014). The former provides insight into the association between the predictor and outcome free of familial confounding, including any quasi-causal effects, and the latter provides insight into the aggregate influence of familial factors (genes, shared environment). A negative binomial distribution with a log link function was used to model count variables (number of lifetime uses of cannabis, cocaine/crack, any stimulant, any opioid, sedatives, hallucinogens, dissociatives, inhalants, and solvents), which were positively skewed and overdispersed. A binary distribution with a logit link function was used to model dichotomous variables (any lifetime [mis]use of prescription and illicit opioids and stimulants). Coefficients were exponentiated to produce incidence rate ratios (IRRs) for negative binomial models and odds ratios (ORs) for binary models.

A series of three models were fit for each drug. As variance inflation factors (VIFs) indicated minimal multicollinearity between traits (all VIFs ≤ 1.53), all Big Five traits were included simultaneously in each model so as to estimate the effect of each trait after accounting for the effects of all other traits. This approach was preferred to modeling each trait independently, as it more clearly addresses the question of specificity in associations between traits and drug use by predicting drug use from variance unique to each trait (supplemental models in which traits were modeled independently were also fit). At the first step, models were run at the individual level (“individual-level models”). These models accounted for the clustering of twin pair data so as to approximate data from unrelated individuals. Effects with p>.05 at the individual level were not carried forward. At the second step, co-twin control models were fit in the combined sample of MZ and DZ pairs (“MZ-DZ models”) to examine within-pair and between-pair effects of Big Five traits on drug use while partially controlling for genes and common environment; we opted to use both MZ and DZ pairs at this step so as to optimize power while partially controlling for familial confounding. At the third step, co-twin control models including only MZ pairs were fit (“MZ-only models”) so as to fully control for familial confounding. Sex, age, marital status, educational attainment, and childhood socioeconomic status were included as covariates in all models; zygosity was also included as a covariate in individual-level and MZ-DZ models.

False discovery rate (FDR) adjustment was calculated using the Benjamini-Hochberg procedure (Benjamini & Hochberg, 1995), under which an effect is considered statistically significant if p<(i/m)Q, where i is the p-value rank across all tests (ordered from smallest to largest), m is the total number of tests, and Q is the selected FDR. The FDR was set to .05, resulting in a corrected significance threshold of p<.02. Analyses were not pre-registered. All analyses that were conducted for this study are described here, with the exception of models conducted for the original submission that were not retained in the manuscript revision (output for these analyses is available upon request to the first author).

Results

Sample Characteristics

Descriptive statistics for the Big Five traits are presented in Table 1. Within-pair differences were approximately one-half of a standard deviation among MZ twins, and approximately one-half to two-thirds of a standard deviation among DZ twins. The largest within-pair differences emerged for neuroticism, followed by extraversion and conscientiousness.

Table 1.

Twin pair averages and average magnitudes of discordance for Big Five personality traits

Full Sample Average Twin Pair Average Within-Pair Difference*
N=1,960 MZ (n=1,130) DZ (n=830) MZ (n=1,130) DZ (n=830)
Trait Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Neuroticism 2.59 (0.74) 2.52 (0.68) 2.49 (0.68) 0.34 (0.30) 0.40 (0.34)
Extraversion 3.54 (0.53) 3.44 (0.59) 3.40 (0.63) 0.27 (0.33) 0.32 (0.38)
Openness to Experience 3.29 (0.47) 3.20 (0.56) 3.16 (0.62) 0.22 (0.29) 0.27 (0.32)
Agreeableness 3.79 (0.47) 3.69 (0.60) 3.62 (0.63) 0.24 (0.32) 0.30 (0.38)
Conscientiousness 3.87 (0.54) 3.77 (0.64) 3.70 (0.69) 0.27 (0.35) 0.33 (0.38)
Open in a new tab

Note. MZ=monozygotic; DZ=dizygotic

*

corresponds to the absolute value of the difference between a co-twin’s score and their pair average (raw differences were used for primary analyses).

Average number of lifetime uses of each drug and the prevalence of lifetime use of each drug among MZ and DZ twins are presented in Table 2. Cannabis was the most commonly and most frequently used drug, followed by stimulants, opioids, and hallucinogens. Correlations between study variables and rates of discordance for use of each drug are available in Tables S1 and S2 of the Supplementary Materials, respectively.

Table 2.

Average number of lifetime uses and prevalence of use of each drug type

Drug Use Phenotype MZ (n=1,130) DZ (n=830)
Number of Lifetime Uses Mean (SD) % Any Lifetime Use Mean (SD) % Any Lifetime Use
Cannabis 73.29 (228.19) 64.0% 92.06 (250.90) 67.8%
Cocaine/Crack 2.38 (31.02) 14.1% 2.71 (16.42) 15.8%
Any Stimulant 18.16 (100.49) 28.5% 11.84 (66.47) 29.8%
Any Opioid 5.33 (60.46) 14.5% 8.06 (78.45) 14.2%
Sedatives 3.23 (44.37) 9.9% 2.45 (36.08) 8.9%
Hallucinogens 2.08 (18.09) 12.3% 3.72 (41.23) 17.6%
Dissociatives 0.12 (1.69) 2.7% 1.75 (35.81) 3.3%
Solvents 0.09 (1.04) 2.0% 0.29 (4.20) 1.6%
Inhalants 1.98 (30.96) 9.8% 0.87 (6.94) 9.2%
Any Lifetime (Mis)Use
Prescription Stimulants - 21.4% - 21.8%
Illicit Stimulants - 23.3% - 25.1%
Prescription Opioids - 14.1% - 13.1%
Illicit Opioids - 2.4% - 3.5%
Open in a new tab

Note. MZ=monozygotic; DZ=dizygotic.

Multilevel Models

Although the model-fitting was conducted by drug, the presentation of results is collapsed across personality trait and further organized by model type. A summary of results from individual-level effects is presented in Table 3, a summary of within-pair effects is presented in Table 4, and a summary of between-pair effects is presented in Table 5. Models predicting use of dissociatives and solvents did not converge and were thus omitted from the presentation of results below.

Table 3.

Summary of aggregate effects from individual-level models

Big Five Personality Trait
Drug Use Phenotype Neuroticism Extraversion Openness Agreeableness Conscientiousness
Number of Lifetime Uses IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI)
Cannabis 1.35 (1.08–1.70) 1.31 (0.98–1.75) 3.20 (2.35–4.37) 0.62 (0.45–0.84) 0.67 (0.51–0.88)
Cocaine/Crack 1.45 (0.92–2.26) 2.52 (1.44–4.39) 2.84 (1.55–5.20) 0.24 (0.13–0.42) 0.71 (0.42–1.21)
Any Stimulant 1.21 (0.83–1.74) 1.65 (1.06–2.56) 4.59 (2.87–7.35) 0.33 (0.20–0.54) 0.53 (0.35–0.80)
Any Opioid 1.71 (1.04–2.80) 0.82 (0.44–1.54) 2.86 (1.52–5.40) 0.28 (0.15–0.54) 0.74 (0.41–1.32)
Sedatives 2.99 (1.66–5.38) 1.65 (0.80–3.41) 2.41 (1.14–5.09) 0.37 (0.18–0.77) 0.65 (0.34–1.25)
Hallucinogens 1.44 (0.93–2.22) 0.93 (0.55–1.56) 2.54 (1.39–4.66) 0.62 (0.35–1.09) 1.26 (0.76–2.11)
Inhalants 1.61 (0.96–2.70) 1.14 (0.60–2.19) 3.36 (1.61–6.98) 0.64 (0.31–1.31) 1.38 (0.74–2.55)
Any Lifetime (Mis)Use OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)
Prescription Stimulants 2.72 (1.51–4.91) 2.68 (1.31–5.45) 6.32 (2.97–13.45) 0.34 (0.17–0.71) 0.65 (0.34–1.23)
Illicit Stimulants 1.33 (0.91–1.96) 1.53 (0.96–2.45) 3.95 (2.32–6.69) 0.58 (0.35–0.95) 0.58 (0.38–0.90)
Prescription Opioids 1.48 (1.11–1.98) 1.07 (0.74–1.55) 1.25 (0.85–1.83) 0.56 (0.38–0.83) 0.85 (0.60–1.20)
Illicit Opioids 5.42 (0.40–74.14) 1.62 (0.12–22.05) 11.65 (0.71–192.06) 0.00 (0.00–0.09) 13.48 (0.86–211.69)
Open in a new tab

Note. IRR=incidence rate ratio; OR=odds ratio; CI=confidence interval; bold type indicates significance at FDR-adjusted p-value threshold.

Table 4.

Summary of within-pair effects from co-twin control models

Big Five Personality Trait
Drug Use Phenotype Neuroticism Extraversion Openness Agreeableness Conscientiousness
Number of Lifetime Uses Model IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI)
Cannabis MZDZ 1.38 (1.07–1.78) - 1.84 (1.24–2.72) 0.92 (0.64–1.33) 0.76 (0.55–1.05)
MZ 1.11 (0.81–1.50) - 1.97 (1.22–3.19) 0.86 (0.55–1.34) 0.71 (0.48–1.05)
Cocaine/Crack MZDZ - 2.26 (1.25–4.10) 2.23 (1.13–4.44) 0.21 (0.11–0.39) -
MZ - 1.58 (0.66–3.74) 2.41 (0.96–6.04) 0.33 (0.13–0.83) -
Any Stimulant MZDZ - 2.22 (1.20–4.10) 2.01 (1.01–3.99) 0.19 (0.10–0.37) 0.90 (0.51–1.58)
MZ - 1.22 (0.69–2.14) 1.79 (0.94–3.42) 0.71 (0.38–1.33) 0.59 (0.35–1.01)
Any Opioid MZDZ 1.71 (0.90–3.26) - 1.01 (0.37–2.77) 0.46 (0.19–1.09) -
MZ 2.07 (0.87–4.89) - 0.44 (0.13–1.49) 0.66 (0.19–2.26) -
Sedatives MZDZ 3.88 (2.06–7.31) - 2.00 (0.81–4.91) 0.30 (0.13–0.71) -
MZ 4.60 (2.01–10.54) - 1.19 (0.41–3.51) 0.42 (0.14–1.25) -
Hallucinogens MZDZ - - 1.68 (0.87–3.24) - -
MZ - - 1.21 (0.46–3.19) - -
Inhalants MZDZ - - 2.16 (0.95–4.89) - -
MZ - - 0.79 (0.28–2.21) - -
Any Lifetime (Mis)Use Model OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)
Prescription Stimulants MZDZ 3.16 (1.10–9.06) 2.40 (0.89–6.47) 2.09 (0.72–6.07) 0.43 (0.16–1.15) -
MZ 6.83 (2.24–20.78) 3.62 (0.83–15.83) 3.66 (0.61–12.83) 0.19 (0.04–0.99) -
Illicit Stimulants MZDZ - - 1.59 (0.84–3.03) 0.79 (0.44–1.42) 0.81 (0.49–1.34)
MZ - - 2.95 (0.75–11.65) 0.82 (0.25–2.66) 0.36 (0.13–1.02)
Prescription Opioids MZDZ 1.34 (0.95–1.89) - - 0.62 (0.39–0.97) -
MZ 1.65 (1.02–2.66) - - 0.76 (0.40–1.45) -
Illicit Opioids MZDZ - - - 0.00 (0.00–0.02) -
MZ - - - 0.00 (0.00–0.65) -
Open in a new tab

Note. IRR=incidence rate ratio; OR=odds ratio; CI=confidence interval; MZDZ=model with MZ and DZ twins; MZ=model with MZ twins only; bold type indicates significance at FDR-adjusted p-value threshold; dash (-) indicates that association was not carried forward to co-twin control models.

Table 5.

Summary of between-pair effects from co-twin control models

Big Five Personality Trait
Drug Use Phenotype Neuroticism Extraversion Openness Agreeableness Conscientiousness
Number of Lifetime Uses Model IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI) IRR (95% CI)
Cannabis MZDZ 0.99 (0.73–1.34) - 6.76 (4.41–10.35) 0.32 (0.20–0.50) 0.63 (0.42–0.94)
MZ 1.00 (0.67–1.50) - 6.91 (3.90–12.23) 0.40 (0.22–0.75) 0.39 (0.23–0.65)
Cocaine/Crack MZDZ - 1.28 (0.55–3.01) 5.72 (2.25–14.52) 0.18 (0.07–0.48) -
MZ - 1.17 (0.37–3.69) 5.06 (1.34–19.09) 0.31 (0.08–1.16) -
Any Stimulant MZDZ - 2.26 (0.89–5.68) 6.14 (2.43–15.50) 0.27 (0.10–0.74) 0.26 (0.11–0.62)
MZ - 2.26 (0.84–6.07) 16.59 (6.14–44.81) 0.24 (0.08–0.74) 0.20 (0.07–0.53)
Any Opioid MZDZ 2.08 (1.23–3.53) - 4.34 (2.09–9.04) 0.18 (0.08–0.39) -
MZ 2.15 (l.12–4.12) - 4.56 (1.79–11.60) 0.25 (0.09–0.64) -
Sedatives MZDZ 1.93 (0.91–4.10) - 4.05 (1.34–12.23) 0.53 (0.18–1.56) -
MZ 1.96 (0.74–5.23) - 3.66 (0.88–15.24) 0.51 (0.13–2.00) -
Hallucinogens MZDZ - - 4.19 (1.87–9.37) - -
MZ - - 2.36 (0.69–8.09) - -
Inhalants MZDZ - - 6.38 (2.27–17.90) - -
MZ - - 4.05 (0.96–17.01) - -
Any Lifetime (Mis)Use Model OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)
Prescription Stimulants MZDZ 1.14 (0.57–2.29) 1.25 (0.49–3.14) 20.06 (5.46–73.64) 0.22 (0.08–0.59) -
MZ 1.59 (0.42–6.06) 1.36 (0.25–7.36) 7.02 (1.06–46.27) 0.35 (0.06–2.28) -
Illicit Stimulants MZDZ - - 9.07 (4.40–18.70) 0.39 (0.20–0.78) 0.45 (0.25–0.81)
MZ - - 13.16 (2.35–73.79) 0.31 (0.06–1.62) 0.29 (0.07–1.23)
Prescription Opioids MZDZ 1.70 (1.24–2.31) - - 0.52 (0.34–0.81) -
MZ 1.90 (1.25–2.89) - - 0.78 (0.44–1.38) -
Illicit Opioids MZDZ - - - 0.38 (0.01–11.67) -
MZ - - - 0.19 (0.00–16.18) -
Open in a new tab

Note. IRR=incidence rate ratio; OR=odds ratio; CI=confidence interval; MZDZ=model with MZ and DZ twins; MZ=model with MZ twins only; bold type indicates significance at FDR-adjusted p-value threshold; dash (-) indicates that association was not carried forward to co-twin control models.

Neuroticism

In the individual-level model, high neuroticism was associated with number of lifetime uses of cannabis (IRR=1.35) and sedatives (IRR=2.99), as well as any lifetime misuse of prescription stimulants (OR=2.72) and prescription opioids (OR=1.48); the association with number of lifetime uses of any opioid (IRR=1.71) was nonsignificant after FDR correction. Though the magnitude of effect for any lifetime illicit opioid use was large (OR=5.32), the imprecision of the estimate impeded meaningful interpretation (95% CI=0.40–74.14). In the MZ-DZ model, within-pair effects of high neuroticism emerged for number of lifetime uses of cannabis (IRR=1.38) and sedatives (IRR=3.88); the association with any lifetime prescription stimulant misuse (OR=3.16) was nonsignificant after FDR correction (Figure 1a). Within-pair effects for number of lifetime uses of any opioid (IRR=1.71) and any lifetime prescription opioid misuse (OR=1.34) were nonsignificant alongside comparably robust between-pair effects (IRR=2.08 and OR=1.70, respectively). In the MZ-only model, the magnitude of the within-pair effect of neuroticism on number of lifetime uses of cannabis was reduced by 20% (IRR=1.11, ns), indicating some degree of familial confounding. Interestingly, the within-pair effect sizes for number of lifetime uses of sedatives (IRR=4.60), and any lifetime prescription stimulant (OR=6.83) and prescription opioid misuse (OR=1.65; ns after FDR correction) were increased in the MZ-only model (Figure 1a), though this change in magnitude is difficult to interpret beyond recognition of a potential suppression effect in the MZ-DZ model. Between-pair effects of neuroticism again emerged in notable magnitude for number of lifetime uses of any opioid (IRR=2.15; ns after FDR correction) and any lifetime prescription opioid misuse (OR=1.90).

Figure 1.

Figure 1

Figure 1

Figure 1

Open in a new tab

Incidence rate ratios (IRRs) and odds ratios (ORs) for within-pair effects of neuroticism (a), extraversion (b), openness to experience (c), agreeableness, (d), and conscientiousness (e) in MZ-DZ and MZ-only models

Note. Black marker denotes significance at FDR-adjusted p-value threshold; † indicates binary outcome (any lifetime use) with effect presented as an OR; depiction of CIs is truncated where upper limit > 8 (a, b, c) or > 2 (d, e); please note shifting x-axes.

Extraversion

In the individual-level model, high extraversion was quite robustly associated with number of lifetime uses of cocaine/crack (IRR=2.52), any lifetime prescription stimulant misuse (OR=2.68), and, to a lesser degree, number of lifetime uses of any stimulant (IRR=1.65; ns after FDR correction). In the MZ-DZ model, within-pair effects for number of lifetime uses of cocaine/crack (IRR=2.26) and any stimulant (IRR=2.22) were robust; the magnitude of effect for any lifetime prescription stimulant misuse was notable (OR=2.40), but imprecise and nonsignificant (Figure 1b). Between-pair effects were non-negligible (IRRs=1.28–2.26; OR=1.25), though all were nonsignificant. In the MZ-only model, within-pair effects for number of lifetime uses of cocaine/crack and number of lifetime uses of any stimulant were reduced by 30–45% (IRRs=1.58 and 1.22, respectively) and nonsignificant, suggesting the presence of familial confounding. Though the magnitude of effect for any lifetime prescription stimulant misuse increased by 51% in the MZ-only model (OR=3.62), the imprecision of the estimate (95% CI=0.83–15.83) impeded meaningful interpretation of this effect (Figure 1b). Again, between-pair effects were non-negligible (IRRs=1.17–2.26; OR=1.36), but all nonsignificant.

Openness to Experience

In the individual-level model, high openness to experience showed large magnitudes of effect for all forms of drug use (IRRs=2.41–4.59; ORs=3.95–6.32), with the exception of any lifetime prescription opioid misuse (OR=1.25, ns). The effect for any lifetime illicit opioid use was nonsignificant, though the large magnitude of effect and imprecise estimate (OR=11.65 95% CI [0.71–192.06]) suggest that this may be due to low power. In the MZ-DZ model, within-pair effects only emerged for number of lifetime uses of cannabis (IRR=1.84) and cocaine/crack (IRR=2.23; Figure 1c). Such a pattern, coupled with strong between-pair effects of openness to experience for all drugs (IRRs=4.05–6.76; ORs=9.07–20.06), suggests that familial effects were responsible for most of the associations identified in individual-level models. In the MZ-only model, within-pair effects emerged only for number of lifetime uses of cannabis (IRR=1.97); however, the retained magnitude of the within-pair effect for number of lifetime uses of cocaine/crack (IRR=2.41) suggests that the lack of significance may be due to low power rather than familial confounding (Figure 1c). Between-pair effects of openness to experience emerged for number of lifetime uses of cannabis, cocaine/crack, any stimulant, any opioid, and sedatives (IRRs=4.56–16.59), as well as any lifetime illicit stimulant use (OR=13.16); the effect for any lifetime prescription stimulant misuse was large (OR=7.02), but imprecise and nonsignificant after FDR correction.

Agreeableness

In the individual-level model, low agreeableness was strongly associated with use of most drugs (IRRs=0.24–0.62; ORs=0.00–0.56), except number of lifetime uses of hallucinogens and inhalants (IRRs=0.62–0.64); the association with any lifetime illicit stimulant use was nonsignificant after FDR correction (OR=0.58). In the MZ-DZ model, strong within-pair effects of low agreeableness emerged for number of lifetime uses of cocaine/crack (IRR=0.21), any stimulant (IRR=0.19), and sedatives (IRR=0.30), as well as any lifetime use of illicit opioids (OR=0.00); the effect for any lifetime misuse of prescription opioids (OR=0.62) was nonsignificant after FDR correction (Figure 1d). Strong between-pair effects of low agreeableness also emerged for number of lifetime uses of cannabis, cocaine/crack, any stimulant, and any opioid (IRRs=0.18–0.32), as well as any lifetime (mis)use of prescription stimulants, illicit stimulants, and prescription opioids (ORs=0.22–0.52). In the MZ-only model, within-pair effects for low agreeableness emerged only for number of lifetime uses of cocaine/crack (IRR=0.33); effects for any lifetime (mis)use of prescription stimulants (OR=0.19) and illicit opioids (OR=0.00) were notable, but nonsignificant after FDR correction (Figure 1d). Magnitudes of effect for number of lifetime uses of any stimulant, number of lifetime uses of sedatives, and any lifetime prescription opioid misuse were reduced by 273%, 40%, and 23%, respectively, indicating the presence of familial confounding. Between-pair effects of agreeableness emerged for number of lifetime uses of cannabis (IRR=0.40), any stimulant (IRR=0.24), and any opioid (IRR=0.25).

Conscientiousness

In the individual-level model, low conscientiousness was associated with number of lifetime uses of cannabis (IRR=0.67) and any stimulant (IRR=0.53), as well as any lifetime illicit stimulant use (OR=0.58). In the MZ-DZ model, no within-pair effects emerged, suggesting that the associations identified in the individual-level models could possibly be attributed to familial effects (Figure 1e). Consistent with this, between-pair effects of low conscientiousness emerged for number of lifetime uses of any stimulant (IRR=0.26), as well as any lifetime illicit stimulant use (OR=0.45). A similar pattern of effect was observed in the MZ-only model, wherein no within-pair effects emerged but between-pair effects for number of lifetime uses of cannabis (IRR=0.39) and any stimulant (IRR=0.20) were of notable magnitude (Figure 1e). The between-pair effect for any lifetime illicit stimulant use was stronger in the MZ-only model (OR=0.29), though it was nonsignificant; this may be due to diminished power in the MZ-only model.

Supplemental Models

In addition to predicting drug use from all Big Five traits simultaneously, bivariate models were fit in which trait domains were tested individually. Results of these analyses are available in Tables S3–S5 of the Supplementary Materials. When modeled independently, there appeared to be more within-pair effects of personality on drug use, with less specificity in association across trait-drug pairs. This is likely attributable to overlapping variance across trait domains that is not accounted for when traits are modeled independently. Though VIFs indicated minimal multicollinearity in the simultaneous models, moderate correlations between some traits (up to r=−.50; see Table S1 in the Supplementary Materials) suggest that there is some degree of overlap. As a result, the independent models effectively “double dipped” personality variance, which appears to have resulted in inflated estimates of trait-drug associations.

Discussion

The present study sought to identify specificity in patterns of Big Five trait-illicit drug use associations within a co-twin control framework. By taking such an approach, it was possible to examine the relationships between personality traits and use of specific drugs while controlling for potential familial confounding. Hypotheses were partially supported. Within-pair effects of high neuroticism were associated with sedative use; unexpectedly, strong within-pair effects of neuroticism also emerged for prescription stimulant misuse. Negative affect encompasses many facets of anxiety, which may at least partially explain the relationship between neuroticism and sedative (benzodiazepine) use (Conrod et al., 2000; Vorspan et al., 2015). Past research has identified associations between high neuroticism and prescription stimulant misuse for the purpose of cognitive enhancement (Sattler & Schunck, 2016), suggesting that one explanation for the present results may be that manifestations of neuroticism such as liability to stress and low-self efficacy (Judge et al., 2002) potentiate this method of achieving performance-dependent goals. Of relevance, individuals high in trait neuroticism utilize mental health care services at higher rates (ten Have et al., 2005), and may therefore be more likely to have access to prescription drugs with potential for misuse (e.g., sedatives for anxiety disorders, stimulants for ADHD). It is not uncommon for prescription misuse to occur in the form of taking one’s own prescription in a manner not prescribed (e.g., higher dose, higher frequency, via alternate route of administration) (McLarnon et al., 2011; Votaw et al., 2019), making this a feasible explanation for this pattern of association. Though we expected to identify within-pair effects of neuroticism on prescription opioid misuse, the association was nonsignificant after FDR correction. However, familial effects were identified. High neuroticism is associated with pain experience; these phenotypes are both moderately heritable and share genetic influences (rg=0.18–0.70) (Meng et al., 2020; Nielsen et al., 2012). These overlapping genetic influences may partially explain the relationship between neuroticism and prescription opioid misuse, given the increased risk of prescription opioid misuse among adults with a history of persistent or chronic pain (Groenewald et al., 2019; Smit et al., 2020; Sutin et al., 2019). Hypotheses regarding extraversion were also partially supported: cocaine/crack and stimulant use were associated with within-pair effects of extraversion, consistent with their status as club drugs typically used in social environments. However, the magnitude of these effects was reduced in the MZ-only model, suggesting that these associations may be attributable to familial confounding. While openness to experience did not emerge with unconfounded associations with hallucinogen use as was hypothesized, quasi-causal effects did emerge for cannabis use. Such a pattern indicates that characteristics such as curiosity and introspection play a direct role in cannabis use.

Though agreeableness and conscientiousness are commonly identified as traits negatively associated with substance use (Sutin et al., 2013), their seemingly protective role in this process appeared to be largely attributable to familial factors. Agreeableness was negatively associated with number of lifetime uses of cannabis, cocaine/crack, any stimulant, any opioid, and sedatives, as well as any lifetime (mis) use of prescription stimulants, prescription opioids, and illicit opioids in individual-level models, but within-pair effects only emerged for cocaine/crack, prescription stimulant misuse, and illicit opioid use. Surprisingly, low conscientiousness was only associated with cannabis, any stimulant, and illicit stimulant use in individual-level models. Though the substantive between-pair effects in the co-twin control models are consistent with the notion that commonly-identified conscientiousness-drug use associations are attributable to familial influences rather than causal effects of personality, it is unclear why conscientiousness was only associated with cannabis and stimulant (mis)use in the present sample. This may reflect an idiosyncrasy of the sample, methodological issues (e.g., power), and/or some other factor(s).

The differences across simultaneously and independently estimated models were unexpected. Applying both methods to modeling the Big Five predictors was useful insofar as this approach speaks to both the specificity in trait-drug associations as conceptualized as unique trait variance predicting use of a specific drug, and to the broader associations of complete trait variance with illicit drug use. This may also provide insight into how models of personality and their assessment can introduce noise into studies of associations between trait domains and phenotypic outcomes. While the Big Five trait domains are theoretically orthogonal, this has been repeatedly demonstrated to not be the case in real-world single-respondent Big Five data (Franić et al., 2014); this is, at least in part, attributable to respondent bias and/or common methods variance, which cannot be parsed without taking an often infeasibly resource-intensive multitrait-multimethod approach (Biderman et al., 2011; Biesanz & West, 2004). The results of the present analyses may suggest that the generalized pattern of association of high neuroticism, low agreeableness, and low conscientiousness with drug use identified in past studies may be partially due to respondent bias, method factors, and/or lack of orthogonality of trait constructs in praxis. As neuroticism, agreeableness, and conscientiousness often display the strongest intercorrelations among the Big Five traits (rNA=−.36; rNC=−.43; rAC;=.43; Van der Linden et al., 2010), including in the present sample (rNA=−.35; rNC=−.40; rAC;=.27), this explanation represents one avenue by which the pattern of association of these three traits with drug use is consistently identified when trait domains are modeled independently (i.e., these purportedly independent associations are capturing much of the same variance). As demonstrated in the present study, a strikingly different pattern emerges when instead predicting drug use from variance uniquely captured by each trait domain. Future research may consider addressing this issue using multivariate outcome models that can also account for overlapping variance between correlated drug use outcomes, so as to even more clearly target specificity in trait-drug associations.

Limitations

Several limitations should be noted. First, it is unclear how findings from this Australian sample will generalize to other populations. Drug culture can vary across social groups, cultures, and countries, which may impact who is drawn toward the use of particular drugs and why. This sample was also quite homogenous with respect to ancestry, information on racial and ethnic identity was not collected, and gender identity was not queried. It is thus unclear how such identities may or may not influence the findings presented here, and how these findings may generalize to other groups. Additionally, only dichotomized indexes of prescription and illicit stimulant and opioid (mis)use were available, limiting a more nuanced exploration of these behaviors. Relatedly, rates of dissociative and solvent use were too low to examine within multilevel models. The uncertainty of several of the models, evinced by the wide confidence intervals associated with several of the presented effects, should also be noted. This issue was particularly salient in the MZ-only models, where the restricted sample size diminished power to estimate effects with precision. Coupled with relatively low rates of use for some drugs, this created challenges in gauging the true size of some associations. That is, despite significant associations for some trait-drug pairs, wide confidence intervals limit inference into their likely magnitude. Replication of these effects is needed. Finally, the personality assessment and drug use events were not temporally aligned. Thus, directional inferences must be made with caution and with the stipulation that the personality profiles used to predict drug use were not reported prior to the use event itself. While there is evidence that personality traits remain relatively stable over time, such that fluctuations that may occur beyond childhood are of minimal practical significance (Ferguson, 2010; Roberts & DelVecchio, 2000; Roberts et al., 2006), and limited evidence of personality change following drug use (Kroencke et al., 2021), the temporal ordering of the drug use behavior and the personality assessment must be considered. Despite these limitations, the present study provides novel evidence regarding the nature of personality trait-drug use associations.

Conclusions

Abundant research has investigated the relationship between personality traits and drug use. The present study supports the notion of specificity in trait-drug use associations and further substantiates the value of personality-targeted intervention approaches (Conrod, 2016). One challenge will be to integrate polysubstance use into this picture. As most drug use does not occur in isolation, it will be important for future research to extend the present findings and identify how personality traits relate to more complete patterns of drug use. Several studies have identified distinct illicit drug use typologies, typically encapsulating cannabis use, party drug use (cocaine, stimulants, hallucinogens), prescription misuse (opioids, sedatives), and polydrug use (Dash et al., 2021; Lynskey et al., 2006; Patra et al., 2009). These typologies are differentially associated with substance use outcomes, such as use disorder and related problems. Better understanding of the individual-level risk factors underlying the manifestation of these patterns of drug use could prove invaluable in identifying those liable to high-risk drug use and, therefore, improving early intervention efforts.

Supplementary Material

Graphics

Public Health Significance Statements.

This study indicates that certain personality traits may causally confer risk for use of specific illicit drugs. Commonly identified associations of high neuroticism, low agreeableness, and low conscientiousness with drug use in general may be attributable to familial influences, such as genes and rearing environment, rather than personality per se. Such insight can inform more precise prediction of risk for illicit drug use and targeted prevention efforts.

Acknowledgments

This work was supported by the National Institute on Drug Abuse F31DA054701 (PI: Genevieve F. Dash) and R01DA18267 (PI: Michael T. Lynskey).

The results of the present study have not been previously disseminated at a conference or meeting, nor has any part of this paper been posted to a listserv or shared on a website. The data used in the present study are a subset of a large-scale dataset available to several researchers around the globe, and many published studies have used these data. However, the present study is novel insofar as we have used a unique permutation of predictors, outcomes, and analytic methods to pursue a question previously uninvestigated in these data.

Footnotes

The authors declare no conflict of interest.

References

  1. Agrawal A, Jacobson K, Prescott C, & Kendler KS (2004). A twin study of personality and illicit drug use and abuse/dependence. Twin Research and Human Genetics, 7(1), 72–81. [DOI] [PubMed] [Google Scholar]
  2. Benjamini Y, & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57(1), 289–300. [Google Scholar]
  3. Benotsch EG, Jeffers AJ, Snipes DJ, Martin AM, & Koester S (2013). The five factor model of personality and the non-medical use of prescription drugs: Associations in a young adult sample. Personality and Individual Differences, 55(7), 852–855. [Google Scholar]
  4. Biderman MD, Nguyen NT, Cunningham CJ, & Ghorbani N (2011). The ubiquity of common method variance: The case of the Big Five. Journal of Research in Personality, 45(5), 417–429. [Google Scholar]
  5. Biesanz JC, & West SG (2004). Towards understanding assessments of the Big Five: Multitrait-multimethod analyses of convergent and discriminant validity across measurement occasion and type of observer. Journal of Personality, 72(4), 845–876. [DOI] [PubMed] [Google Scholar]
  6. Chinneck A, Thompson K, Mahu I, Davis-MacNevin P, Dobson K, & Stewart S (2018). Personality and prescription drug use/misuse among first year undergraduates. Addictive Behaviors, 87, 122–130. [DOI] [PubMed] [Google Scholar]
  7. Chowdhury N, Kevorkian S, Sheerin CM, Zvolensky MJ, & Berenz EC (2016). Examination of the association among personality traits, anxiety sensitivity, and cannabis use motives in a community sample. Journal of Psychopathology and Behavioral Assessment, 38(3), 373–380. [Google Scholar]
  8. Conrod PJ (2016). Personality-targeted interventions for substance use and misuse. Current Addiction Reports, 3(4), 426–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Conrod PJ, Pihl RO, Stewart SH, & Dongier M (2000). Validation of a system of classifying female substance abusers on the basis of personality and motivational risk factors for substance abuse. Psychology of Addictive Behaviors, 14(3), 243–256. [DOI] [PubMed] [Google Scholar]
  10. Cooper ML (1994). Motivations for alcohol use among adolescents: Development and validation of a four-factor model. Psychological Assessment, 6(2), 117–128. [Google Scholar]
  11. Costa PT, & McCrae RR (1992). Normal personality assessment in clinical practice: The NEO Personality Inventory. Psychological Assessment, 4(1), 5–13. [Google Scholar]
  12. Dash GF, Martin NG, Agrawal A, Lynskey MT, & Slutske WS (2021). Typologies of illicit drug use in mid-adulthood: A quasi-longitudinal latent class analysis in a community-based sample of twins. Addiction, 116(5), 1101–1112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dash GF, Slutske WS, Martin NG, Statham DJ, Agrawal A, & Lynskey MT (2019). Big Five personality traits and alcohol, nicotine, cannabis, and gambling disorder comorbidity. Psychology of Addictive Behaviors, 33(4), 420–429. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Davis CN, Slutske WS, Piasecki TM, Martin NG, & Lynskey MT (2020). Comparing the potential causal influence of two indicators of early alcohol use on later alcohol use disorder symptoms. Journal of Abnormal Psychology, 129(3), 256–265. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Delić M, Kajdiž K, & Pregelj P (2017). Association of the Five-Factor Model personality traits and opioid addiction treatment outcome. Psychiatria Danubina, 29(3), 289–291. [PubMed] [Google Scholar]
  16. de Moor MHM, de Geus EJC, & Boomsma DI (2011). Statistical power to detect causality in the co-twin control design. Behavior Genetics, 41(6), 900. [Google Scholar]
  17. Erritzoe D, Smith J, Fisher PM, Carhart-Harris R, Frokjaer VG, & Knudsen GM (2019). Recreational use of psychedelics is associated with elevated personality trait openness: Exploration of associations with brain serotonin markers. Journal of Psychopharmacology, 33(9), 1068–1075. [DOI] [PubMed] [Google Scholar]
  18. Ferguson CJ (2010). A meta-analysis of normal and disordered personality across the life span. Journal of Personality and Social Psychology, 98(4), 659–667. [DOI] [PubMed] [Google Scholar]
  19. Fehrman E, Muhammad AK, Mirkes EM, Egan V, & Gorban AN (2017). The five factor model of personality and evaluation of drug consumption risk. In Data Science (pp. 231–242). Springer, Cham. [Google Scholar]
  20. Franić S, Borsboom D, Dolan CV, & Boomsma DI (2014). The big five personality traits: Psychological entities or statistical constructs? Behavior Genetics, 44(6), 591–604. [DOI] [PubMed] [Google Scholar]
  21. Groenewald CB, Law EF, Fisher E, Beals-Erickson SE, & Palermo TM (2019). Associations between adolescent chronic pain and prescription opioid misuse in adulthood. The Journal of Pain, 20(1), 28–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hawkins LW (2012). The impact of personality and affect on college student’s motives for marijuana use [Doctoral dissertation]. Sycamore Scholars. http://scholars.indstate.edu/handle/10484/3905 [Google Scholar]
  23. Jang KL, Livesley WJ, & Vernon PA (1995). Alcohol and drug problems: a multivariate behavioural genetic analysis of co-morbidity. Addiction, 90(9), 1213–1221. [DOI] [PubMed] [Google Scholar]
  24. Judge TA, Erez A, Bono JE, & Thoresen CJ (2002). Are measures of self-esteem, neuroticism, locus of control, and generalized self-efficacy indicators of a common core construct?. Journal of Personality and Social Psychology, 83(3), 693–710. [DOI] [PubMed] [Google Scholar]
  25. Karkowski LM, Prescott CA, & Kendler KS (2000). Multivariate assessment of factors influencing illicit substance use in twins from female-female pairs. American Journal of Medical Genetics, 96(5), 665–670. [PubMed] [Google Scholar]
  26. Kendler K, Gardner C, Jacobson KC, Neale MC, & Prescott CA (2005). Genetic and environmental influences on illicit drug use and tobacco use across birth cohorts. Psychological Medicine, 35(9), 1349–1356. [DOI] [PubMed] [Google Scholar]
  27. Kendler K, Karkowski LM, Neale MC, & Prescott CA (2000). Illicit psychoactive substance use, heavy use, abuse, and dependence in a US population-based sample of male twins. Archives of General Psychiatry, 57(3), 261–269. [DOI] [PubMed] [Google Scholar]
  28. Kendler KS, Myers J, & Prescott CA (2007). Specificity of genetic and environmental risk factors for symptoms of cannabis, cocaine, alcohol, caffeine, and nicotine dependence. Archives of General Psychiatry, 64(11), 1313–1320. [DOI] [PubMed] [Google Scholar]
  29. Kotov R, Gamez W, Schmidt F, & Watson D (2010). Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis. Psychological Bulletin, 136(5), 768–821. [DOI] [PubMed] [Google Scholar]
  30. Kroencke L, Kuper N, Bleidorn W, & Denissen J (2021). How does substance use affect personality development? Disentangling between-and within-person effects. Social Psychological and Personality Science, 12(4), 517–527. [Google Scholar]
  31. Kuntsche E, Knibbe R, Gmel G, & Engels R (2006). Who drinks and why? A review of socio-demographic, personality, and contextual issues behind the drinking motives in young people. Addictive Behaviors, 31(10), 1844–1857. [DOI] [PubMed] [Google Scholar]
  32. LaFrance EM, & Cuttler C (2017). Inspired by Mary Jane? Mechanisms underlying enhanced creativity in cannabis users. Consciousness and Cognition, 56, 68–76. [DOI] [PubMed] [Google Scholar]
  33. Lahey BB (2009). Public health significance of neuroticism. American Psychologist, 64(4), 241–256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Littlefield AK, & Sher KJ (2016). Personality and substance use disorders. In Sher KJ (Ed.), The Oxford Handbook of Substance Use and Substance Use Disorders (pp. 351–378). [Google Scholar]
  35. Lynskey MT, Agrawal A, Bucholz KK, Nelson EC, Madden PA, Todorov AA, Grant JD, Martin NG, & Heath AC (2006). Subtypes of illicit drug users: a latent class analysis of data from an Australian twin sample. Twin Research and Human Genetics, 9(4), 523–530. [DOI] [PubMed] [Google Scholar]
  36. Lynskey MT, Agrawal A, Henders A, Nelson EC, Madden PA, & Martin NG (2012, Oct). An Australian twin study of cannabis and other illicit drug use and misuse, and other psychopathology. Twin Research and Human Genetics, 15(5), 631–641. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Mahu I, Conrod P, Barrett S, Sako A, Swansburg J, Lawrence M, Laroque F, Morin J, Chinneck A, & Nogueira-Arjona R (2019). Specificity of personality relationships to particular forms of concurrent substance use among methadone maintenance therapy clients. Addictive Behaviors, 98, 106056. [DOI] [PubMed] [Google Scholar]
  38. McCrae RR (1993). Openness to experience as a basic dimension of personality. Imagination, Cognition and Personality, 13(1), 39–55. [Google Scholar]
  39. McCrae RR, & Costa PT Jr (2004). A contemplated revision of the NEO Five-Factor Inventory. Personality and Individual Differences, 36(3), 587–596. [Google Scholar]
  40. McGue M, Osler M, & Christensen K (2010). Causal inference and observational research: The utility of twins. Perspectives on Psychological Science, 5(5), 546–556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. McLarnon ME, Monaghan TL, Stewart SH, & Barrett SP (2011). Drug misuse and diversion in adults prescribed anxiolytics and sedatives. Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy, 31(3), 262–272. [DOI] [PubMed] [Google Scholar]
  42. Meng W, Adams MJ, Reel P, Rajendrakumar A, Huang Y, Deary IJ, … & Smith BH (2020). Genetic correlations between pain phenotypes and depression and neuroticism. European Journal of Human Genetics, 28(3), 358–366. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Nielsen CS, Knudsen GP, & Steingrimsdottir OA (2012). Twin studies of pain. Clinical Genetics, 82(4), 331–340. [DOI] [PubMed] [Google Scholar]
  44. Patra J, Fischer B, Maksimowska S, & Rehm J (2009). Profiling poly-substance use typologies in a multi-site cohort of illicit opioid and other drug users in Canada–a latent class analysis. Addiction Research & Theory, 17(2), 168–185. [Google Scholar]
  45. Raynor DA, & Levine H (2009). Associations between the five-factor model of personality and health behaviors among college students. Journal of American College Health, 58(1), 73–82. [DOI] [PubMed] [Google Scholar]
  46. Roberts BW, & DelVecchio WF (2000). The rank-order consistency of personality traits from childhood to old age: a quantitative review of longitudinal studies. Psychological Bulletin, 126(1), 3–25. [DOI] [PubMed] [Google Scholar]
  47. Roberts BW, Walton KE, & Viechtbauer W (2006). Patterns of mean-level change in personality traits across the life course: a meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1–25. [DOI] [PubMed] [Google Scholar]
  48. SAS, Inc. (2014). Base SAS® 9.4 procedures guide: statistical procedures. Cary: SAS Institute Inc. [Google Scholar]
  49. Sattler S, & Schunck R (2016). Associations between the big five personality traits and the non-medical use of prescription drugs for cognitive enhancement. Frontiers in Psychology, 6, 1971. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Schermer JA, & Martin NG (2019). A behavior genetic analysis of personality and loneliness. Journal of Research in Personality, 78, 133–137. [Google Scholar]
  51. Simons J, Correia CJ, Carey KB, & Borsari BE (1998). Validating a five-factor marijuana motives measure: Relations with use, problems, and alcohol motives. Journal of Counseling Psychology, 45(3), 265–273. [Google Scholar]
  52. Slutske WS, Caspi A, Moffitt TE, & Poulton R (2005). Personality and problem gambling: A prospective study of a birth cohort of young adults. Archives of General Psychiatry, 62(7), 769–775. [DOI] [PubMed] [Google Scholar]
  53. Slutske WS, Deutsch AR, Richmond-Rakerd LS, Chernyavskiy P, Statham DJ, & Martin NG (2014). Test of a potential causal influence of earlier age of gambling initiation on gambling involvement and disorder: A multilevel discordant twin design. Psychology of Addictive Behaviors, 28(4), 1177–1189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Slutske WS, Heath AC, Madden PA, Bucholz KK, Statham DJ, & Martin NG (2002). Personality and the genetic risk for alcohol dependence. Journal of Abnormal Psychology, 111(1), 124–133. [PubMed] [Google Scholar]
  55. Smit T, Rogers AH, Garey L, Allan NP, Viana AG, & Zvolensky MJ (2020). Anxiety sensitivity and pain intensity independently predict opioid misuse and dependence in chronic pain patients. Psychiatry Research, 294, 113523. [DOI] [PubMed] [Google Scholar]
  56. Sutin AR, Evans MK, & Zonderman AB (2013). Personality traits and illicit substances: The moderating role of poverty. Drug and Alcohol Dependence, 131(3), 247–251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Sutin AR, Stephan Y, Luchetti M, & Terracciano A (2019). The prospective association between personality traits and persistent pain and opioid medication use. Journal of Psychosomatic Research, 123, 109721. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. ten Have M, Oldehinkel A, Vollebergh W, & Ormel J (2005). Does neuroticism explain variations in care service use for mental health problems in the general population? Social Psychiatry and Psychiatric Epidemiology, 40(6), 425–431. [DOI] [PubMed] [Google Scholar]
  59. ter Bogt TF, Engels RC, & Dubas JS (2006). Party people: Personality and MDMA use of house party visitors. Addictive Behaviors, 31(7), 1240–1244. [DOI] [PubMed] [Google Scholar]
  60. Terracciano A, Löckenhoff CE, Crum RM, Bienvenu OJ, & Costa PT (2008). Five-Factor Model personality profiles of drug users. BMC Psychiatry, 8(1), 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Van der Linden D, te Nijenhuis J, & Bakker AB (2010). The general factor of personality: A meta-analysis of Big Five intercorrelations and a criterion-related validity study. Journal of Research in Personality, 44(3), 315–327. [Google Scholar]
  62. Vernon PA, Martin RA, Schermer JA, & Mackie A (2008). A behavioral genetic investigation of humor styles and their correlations with the Big-5 personality dimensions. Personality and Individual Differences, 44(5), 1116–1125. [Google Scholar]
  63. Vorspan F, Mehtelli W, Dupuy G, Bloch V, & Lépine JP (2015). Anxiety and substance use disorders: co-occurrence and clinical issues. Current Psychiatry Reports, 17(2), 4. [DOI] [PubMed] [Google Scholar]
  64. Votaw VR, Geyer R, Rieselbach MM, & McHugh RK (2019). The epidemiology of benzodiazepine misuse: A systematic review. Drug and Alcohol Dependence, 200, 95–114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Vreeker A, Brunt TM, Treur JL, Willemsen G, Boomsma DI, Verweij KJ, & Vink JM (2020). Comparing ecstasy users and non-users in a population-based and co-twin control design across multiple traits. Addictive Behaviors, 106421. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Graphics
NIHMS1748031-supplement-Graphics.docx (47.4KB, docx)

RESOURCES