The Association between the Five-factor Model of Personality and Problem Gambling: a Meta-analysis

This meta-analysis examined the associations between five-factor personality model traits and problem gambling. To be eligible for inclusion in the meta-analysis, studies had to provide effect size data that quantified the magnitude of the association between all five personality traits and problem gambling. Studies also had to use psychometrically sound measures. The meta-analysis included 20 separate samples from 19 studies and 32,222 total participants. The results showed that problem gambling was significantly correlated with the five-factor model of personality. The strongest personality correlate of problem gambling was neuroticism r = .31, p = < 0.001, 95% CI [0.17, 0.44], followed by conscientiousness r = − .28, p = < 0.001, 95% CI [-0.38,-0.17] ), agreeableness r = − .22, p = < 0.001, 95% CI [-0.34, − 0.10], openness r = − .17, p = < 0.001, 95% CI [-0.22,-0.12], and extraversion r = − .11, p = .024, 95% CI [-0.20,-0.01]. These results suggest problem gamblers tend to share a common personality profile – one that could provide clues as to the most effective ways to prevent and to treat problem gambling.

Problem gambling, also termed gambling disorder and pathological gambling, is a behavioral addiction characterized by persistent gambling behavior despite significant negative consequences that can include financial hardship, legal problems, relationship and occupational dysfunction, and significant emotional distress (Blanco & Bernadi, 2014); Brunborg et al., 2016). The diagnostic criteria for gambling disorder in the Diagnostic and Statistical Manual of Mental Disorders − 5 (DSM-5; American Psychiatric Association, 2013) includes additional elements, such as restlessness or irritability when reducing or attempting to stop gambling, a preoccupation with gambling, and a tendency for gambling to occur when feeling distressed. disorders of various types (Malouff et al., 2005), smoking (Malouff et al., 2006), excessive drinking (Malouff et al., 2007), and the dramatic and emotional cluster (cluster B) of personality disorders (Samuel & Widiger, 2008). In each meta-analysis, there were significant associations between multiple five-factor traits and the psychological problem. For instance, Malouff et al. (2005) found that all of the five-factor traits except openness were related to symptoms of clinical disorders. Samuel and Widiger found significant associations between all of the five traits except openness and multiple personality disorders.
The relationship between personality and problem gambling is not yet clear. Where the overall pattern of findings among related studies is not clear, a meta-analysis can be useful, so we set out to complete a meta-analysis of the association between the five-factor model of personality and problem gambling. We focused on studies using the five-factor model because we wanted to use an empirically supported model and because studies using the model provide an opportunity to assess each five-factor trait against the others in the same sample of participants. Because there have been numerous studies of the five-factor model and problem gambling, it is clear that researchers consider the relationship important. What is missing is an aggregation of the findings in a meta-analysis.

Aims of the meta-analysis
The purpose of the present meta-analysis was to examine the association between the fivefactor model of personality and problem gambling. We hypothesized that high neuroticism, low conscientiousness, and low agreeableness would be associated with problem gambling, because these personality traits have been found to be associated with other types of addictive behavior involving alcohol, cannabis, tobacco, and Internet gaming (Dash & Slutske, 2019;Malouff et al., 2005;Müller et al., 2014).

Eligibility criteria
Studies had to meet three criteria for inclusion in the meta-analysis: (1) The related report had to include effect sizes for the association between each of the five-factor personality traits and problem gambling, (2) the report had to state the number of participants, and (3) the study had to use psychometrically sound measures.

Search strategy
A protocol for this meta-analysis was published in the International Prospective Register of Systematic Reviews, registration number: CRD42021237773, in February 2021. In March 2021, two researchers systematically searched the following databases: EBSCO, EBSCO Open Dissertations, Google Scholar, ProQuest, and PubMed. Keywords used were five factor or big five, and gambl*. No date or language parameters were set for the electronic search. To reduce the search results in the Proquest database from 6283 to 112, we added quotation marks to the search terms "five-factor" and "big 5." In August 2021, we repeated the electronic database search and included a date parameter set to 2021-2022 to find any recently published studies. We also examined reference lists of included articles and emailed corresponding authors of included studies requesting unpublished data. No unpublished studies were found.

Data extraction and coding
One researcher manually extracted data from the included studies and recorded it on an electronic spreadsheet. Data extracted to calculate the effect sizes included correlations, and independent group means and standard deviations of problem gambling and healthy control groups. Coded descriptive data included: (1) study authors and publication date, (2) number of participants, (3) mean age, (4) percentage female, (5) five-factor model of personality measure used, (6) problem gambling measure used, (7) study design (correlational or between groups), (8) evidence of validity and reliability of the measures used, and (8) sample type. Then a second researcher checked data entries, and a third researcher independently coded entries needed to calculate effect sizes. Inter-rater reliability between the first two coders and the independent coder was 93%. Consensus between coders resolved all disagreements.

Data analysis
Analyses were performed using the Comprehensive Meta-Analysis software (CMA Version 3.3.070; Borenstein et al., 2014). A composite score was computed for studies reporting multiple outcomes for a single trait based on the same participants. Effect sizes were calculated using a meta-analysis analogue of Pearson's r. For the 13 studies that reported means and standard deviations for groups of problem gamblers and others, Hedge's g (Hedges, 1982) was calculated and then converted to r.
We used a random-effects model because it recognizes within-study and between-study variance and assumes that the true effect size differs among studies (Borenstein et al., 2009). To measure heterogeneity, both the I² statistic and Cochran's Q were calculated. The I 2 statistic quantifies the level of heterogeneity (Higgins et al., 2003). I 2 is the proportion of variance across studies that is due to true effects rather than sampling error. The Cochran's Q statistic was computed to examine whether all studies in the present meta-analysis assessed the same effect (Higgins et al., 2003).

Quality assessment
Assessment of study quality involved evaluating the validity and reliability of all measures used in studies included in this meta-analysis.

Study selection
Following the removal of 159 duplicates, 239 records were retained for screening. Of the 239 records, 20 studies seemed to meet the inclusion criteria. Out of these 20 studies, one was removed because group-assignment in a treatment study was treated as a covariate in the key reported results. Hence, 19 studies were included in the meta-analysis. One study had two independent samples, leading to a total of 20 samples to analyze. Figure 1 presents a PRISMA Flow Diagram (Moher et al., 2009) containing information about the study selection process.  Table 1 shows key study characteristics. The meta-analysis included 20 samples, with a total of 32,222 participants and produced 100 effect sizes (20 effect sizes for each of the five personality factors). The most common problem gambling scale used was the SOGS, and 12 studies used a version of the NEO to measure the five-factor model of personality. The data file is at https://rune.une.edu.au/web/handle/1959.11/31788.

Quality Assessment
All measures used in studies included in the present meta-analysis demonstrated reliability (see Table 2). All measures used in the present meta-analysis also had evidence of validity.
Cochran's Q statistic was significant across all five personality factors, indicating heterogeneity and supporting the use of a random effects model. Table 3 presents meta-analytical summary statistics for the association between the five-factor model of personality and gam-bling for all 20 samples. Figures 2, 3, 4, 5 and 6 illustrate the analyses of each individual personality factor and its association with problem gambling for the 20 samples included in the meta-analysis.  Table 1 Descriptive information about studies included in the meta-analysis

Publication bias
Funnel plots for all five personality traits show a symmetric distribution. Duval and Tweedie's (2000) trim and fill method did not suggest trimming any studies. These results suggest an absence of small-study effects. Table 5 shows the classic fail-safe N and Orwin's fail-safe N analyses for each personality factor.

Discussion
The present meta-analysis provided a statistical review of the association between the fivefactor model of personality and problem gambling. The findings from the 20 samples supported the hypothesis that gambling disorder was significantly associated with higher scores on neuroticism, and lower scores on conscientiousness and agreeableness. The results also showed problem gambling was significantly associated with lower scores on both openness and extraversion. Cohen (1988) suggested that r be interpreted as a small effect when r = .10, a medium effect when r = .30, and a large effect when r = .50. The effect size for neuroticism, 0.31, was a medium effect, and the effect size for conscientiousness, − 0.28, was nearly medium. The other effect sizes were small.
Because the findings are correlational, they do not provide evidence that the traits cause problem gambling. However, the findings are consistent with possible causes of problem gambling. The implications of the findings vary from trait to trait, as described below.
Individuals scoring high on neuroticism tend to be worrisome, anxious, self-conscious, and depressed. Hence, some individuals may use gambling to escape these negative feelings, at least for a short while (Mackinnon et al., 2016).  Low conscientiousness involves apathy, impulsivity and a disregard of social norms. Impulsivity could play a factor in problem gambling by its focus on the extreme short-term over the longer term (Ioannidis et al., 2019). Low agreeableness is characterized by a tendency to be unsocial, inconsiderate, and competitive. Disagreeable behavior may be an antecedent to relationship and occupational dysfunction, consequences that are characteristic of gambling disorder (Widinghoff et al., 2019). The competitive element of this trait could lead individuals to continue gambling despite losses (Parke et al., 2004).
Individuals low in openness tend to avoid change, to be closed-minded, and to prefer routine. The change-avoidant characteristic could contribute to persistent gambling by keeping a person repeating the behavior that is causing problems (Myrseth et al., 2009). The relationship between extraversion and problem gambling was the lowest in magnitude among the five personality factors. Low extraversion involves low engagement with others and is typically associated with maladaptive emotion regulation strategies (Baranczuk, 2019).    Table 3 Summary statistics for the meta-analysis of the association between pathological gambling and the five-factor model of personality We used the sub-sample of non-comorbid pathological gamblers found in Table 3 of the study of Mann et al. (2017)  of other social sources of excitement, and because of low mood that can be briefly improved by the excitement of gambling.
The meta-analysis showed that all the five-factor personality traits are related to problem gambling in specific ways. However, those same, seemingly undesirable traits might have adaptive value in certain situations. For instance, low agreeableness might help a person avoid being swindled by a new romantic partner.
The personality characteristics associated with problem gambling tend to be associated with other addictive disorders, including alcohol, cannabis, and nicotine use disorders (Dash & Slutske, 2019;Malouff et al., 2005;Müller et al., 2014). Studies have shown that the same personality profile is associated with various psychological problems (Malouff et al., 2005) and with the dramatic and emotional cluster (cluster B) of personality disorders (Samuel & Widiger, 2008;Quilty et al., 2021). It is therefore not surprising that problem gamblers are highly comorbid with nicotine dependence, substance use disorders, mood disorders, anxiety disorders (Lorains et al., 2011), and cluster B personality disorders, particularly  (Brown et al., 2015). It could be that personality factors help push a person toward addictive behavior.
The main limitations of the present meta-analysis are that the findings (a) are correlational, (b) are entirely based on self-report, (c) combine problems relating to various types of gambling, and (d) are based on mainly English-speaking participants. The correlational findings do not show the direction of the causal relationship between personality and problem gambling. Personality may cause problem gambling, problem gambling may lead to certain personality traits, the relationship may be bidirectional, or some third variable, such as specific genes, may lead to both certain personality traits and problem gambling. Selfreport measures rely on a person's insight and honesty, making them vulnerable to biases. Individuals problematically engaging in different types of gambling activities, e.g., betting on horse races and playing slot machines, may differ in important ways. Individuals who are problem gamblers in different cultures might show a different pattern of personality. It is unknown whether the findings of this meta-analysis could be generalised to every type of gambling and every culture.
The present meta-analysis has advantages over the results of any single study in that the meta-analysis included results from different sets of researchers examining individuals in different countries, used different measures, and analyzed a very large overall group of participants. Aggregating findings across many different studies helps increase the generalizability of findings.   If problem gambling results from attempts to reduce the negative affect of neuroticism, implementing treatment strategies that reduce negative affect may prove helpful in managing problem gambling. Clinicians could devise treatment plans to focus on identifying and implementing ways to improve an individual's overall affective state. Additionally, clinicians may need to consider the possible influence of personality on treatment processes. The personality profile associated with gambling disorder, including low conscientiousness and low agreeableness, may make it challenging to successfully treat individuals for problem gambling. A person with the personality of low conscientiousness and low agreeableness may not consistently attend appointments or undertake therapeutic assignments. Clinicians may need to make special efforts to overcome these client tendencies. In this regard, Ramos-Grille et al. (2014) found problem gamblers with low scores on conscientiousness had higher rates of treatment failure and relapse.
Clinicians who help problem gamblers could consider personality-focused strategies that have shown success with other addictive disorders; for instance, the Preventure Programme delivers brief interventions targeting personality risk factors associated with substance abuse. The interventions include psychoeducation, motivational enhancement therapy, and cognitive behavioral therapy (Edalati & Conrod, 2019).
Future research on personality and problem gambling could explore whether the findings of the meta-analysis apply to problems with specific types of gambling and apply in cultures not examined so far. Studies could examine whether personality-focused preventive efforts and treatments are effective. Studies could also examine whether different types of treatment for problem gambling change specific personality traits.

Funding Open Access funding enabled and organized by CAUL and its Member Institutions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.