Psychological Pathways to Gambling-Related Harm: Age, Personality, and Implications for Persuasive Health Communication

1. Introduction

Gambling-related harm is increasingly treated as a public health issue rather than simply a matter of participation frequency (Abbott, 2020). Risk taking also changes across adulthood and is closely tied to individual differences in impulsivity and sensation seeking (Rolison et al., 2014; Steinberg et al., 2008). Impulsivity reflects a tendency to act with insufficient forethought, particularly when rewards are immediately available (Moeller et al., 2001), whereas sensation seeking reflects the pursuit of novel, intense, or stimulating experiences. The two tendencies overlap, but they are not interchangeable: one centers on self-regulatory control and the other on the attraction of stimulation. This distinction matters for gambling because the same level of gambling participation may carry very different consequences for people with different psychological profiles.

Personality provides a second layer of explanation. Big Five traits and Dark Triad traits have been associated with risk taking across domains, including gambling (Crysel et al., 2013; Nicholson et al., 2005). Conscientiousness is generally linked to planning and behavioral restraint, while openness is linked to novelty and exploration; Dark Triad traits capture distinct tendencies involving grandiosity, strategic manipulation, and callous or impulsive behavior (McCrae & Costa, 1987; Paulhus & Williams, 2002). These traits may therefore shape whether gambling is approached as an occasional activity, a source of stimulation, or a behavior that becomes difficult to regulate (Onyedire et al., 2021; Refaie et al., 2024). Examining age and personality together can clarify why some individuals experience greater gambling-related harm.

This research also has a communication component. Contemporary gambling is embedded in digital platforms, sports media, sponsorships, and promotional environments that can normalize betting and frame it as entertaining, skill based, or socially integrated (Lopez-Gonzalez et al., 2018; Thomas et al., 2018). Health communication intended to reduce gambling-related harm may therefore benefit from identifying the psychological mechanisms that make different audiences more vulnerable to particular persuasive cues. The present study does not test persuasive messages directly; rather, it examines age and personality related correlates that can inform subsequent message design and intervention research.

1.1. Age and Gambling-Related Harm

Age is consistently associated with risk behavior, but gambling patterns are not uniform across outcomes. Younger adults often report stronger gambling motives and greater involvement in newer gambling formats, while older adults may gamble frequently for different reasons (Aragay et al., 2021; Clarke, 2008). Earlier gambling onset and gambling during young adulthood are also associated with later gambling-related problems and co-occurring risk behavior (Emond et al., 2022; Pérez, 2024; Quinn et al., 2023). Across international studies, younger age is associated with greater problem gambling risk (Dellosa & Browne, 2024). These findings suggest that age may be more informative for gambling-related harm than for simple participation frequency.

Digital access further complicates age differences. Internet-based gambling has expanded the opportunity to gamble privately, rapidly, and across settings (Gainsbury et al., 2015). Although younger adults have often been associated with online gambling, recent work shows substantial overlap between online and land based in person gambling participation, including a sizeable mixed-mode group (Hing et al., 2022). It follows that the present study treats problem severity and preferred gambling medium as distinct outcomes.

H1. 

Younger age will be associated with greater problem gambling severity.

H2. 

Age will be associated with preferred gambling medium, with younger adults more likely to report online/app-based gambling.

1.2. Personality, Self-Regulation, and Gambling

The Dark Triad includes narcissism, Machiavellianism, and psychopathy: overlapping but distinct subclinical traits found in the general population (Paulhus & Williams, 2002). Narcissism is characterized by grandiosity, entitlement, and a desire for admiration and has been linked to risky choice (Leder et al., 2020; Yang et al., 2018). Machiavellianism is more strategic and calculated, whereas psychopathy is more consistently associated with impulsive and sensation-seeking tendencies (Malesza & Ostaszewski, 2016; van Dongen et al., 2022). These distinctions make the Dark Triad potentially relevant to different pathways into gambling-related risk rather than to a single generalized ‘dark’ profile.

Big Five traits may operate through more conventional self-regulatory pathways. Conscientiousness reflects organization, planning, and restraint and is typically associated with lower impulsivity (Lange et al., 2017). Openness reflects interest in novelty and experience, characteristics that may overlap with sensation seeking (Abu Raya et al., 2023; de Vries et al., 2009). Meta-analytic and large sample work has also linked Big Five traits to problem gambling, although the pattern of specific traits is not fully consistent across studies (Dash et al., 2019; Dudfield et al., 2023; von der Heiden & Egloff, 2021). Rather than assuming that one personality configuration explains gambling risk, the present study examines how multiple traits relate to impulsivity, sensation seeking, gambling behavior, and problem severity.

H3a. 

Personality traits will predict impulsivity.

H3b. 

Personality traits will predict sensation seeking.

H4. 

Personality traits will predict gambling behavior and problem gambling severity.

1.3. Gambling Attitudes and Sports Fanship

Gambling can also be embedded in broader attitudes and social identities. Sports fanship changes across the lifespan and can shape how people experience sports-related media (Gantz & Lewis, 2023). Betting may enhance aspects of the viewing experience for some consumers, suggesting that gambling behavior can become intertwined with entertainment and fandom (Lee et al., 2023; Lee et al., 2025). This creates an additional route through which personality might relate to gambling, not only through harm or frequency, but also through satisfaction and sports fanship.

H5. 

Personality traits will predict gambling satisfaction and sports fanship.

This study examines whether age and personality distinguish gambling-related harm from broader gambling engagement and whether impulsivity, sensation seeking, and fanship help explain these patterns. The applied goal is to identify psychologically meaningful audience differences that can be tested in future persuasive health communication interventions.

2. Materials and Methods

2.1. Study Design and Participants

This cross-sectional study used an online survey to examine associations among age, Big Five and Dark Triad personality traits, impulsivity, sensation seeking, gambling behavior, problem gambling severity, gambling satisfaction, and sports fanship. Participants were recruited through Prolific and completed the survey in Qualtrics.

A total of 86 adults completed the survey. Three respondents were excluded because they completed the questionnaire in less than 388 seconds (approximately 6.5 min), which was less than half of the average completion time of 839 seconds and was treated as an indicator of insufficient attention. The final analytic sample was N = 83. Eligibility required participants to be at least 18 years old, fluent in English, and able to provide informed consent online. Participants ranged in age from 18 to 78 years (M = 48.02, SD = 16.73); 54.2% identified as female and 45.8% as male.

2.2. Procedure

Data collection occurred from Summer 2025 through Fall 2025. After reviewing the IRB approved consent information and agreeing to participate, respondents completed the survey online. Survey items were randomized within Qualtrics where appropriate to reduce order effects. The survey took approximately 5-10 min, and participants were compensated through Prolific after completion (via departmental funds of the second author).

2.3. Measures

2.3.1. Big Five Personality

Big Five personality traits were measured with the 10-item Ten-Item Personality Inventory (TIPI) (Gosling et al., 2003). Participants rated how well paired descriptors applied to them on a five-point scale from 1 (strongly disagree) to 5 (strongly agree). Two items represented each of the five traits: openness, conscientiousness, extraversion, agreeableness, and emotional stability.

2.3.2. Dark Triad Traits

Dark Triad traits were assessed with the 12-item Dirty Dozen (Jonason & Webster, 2010). Four items assessed each trait: narcissism, Machiavellianism, and psychopathy. Participants responded on a five-point scale from 1 (not well at all) to 5 (extremely well).

2.3.3. Gambling Outcomes and Attitudes

General gambling frequency was assessed with a single item asking how often participants gambled across contexts (e.g., casino, online, sportsbook), with responses ranging from never to always (7 days per week). Participants also indicated their most commonly used betting medium, with response options including online/app, casino, or other.

Problem gambling was assessed with the nine-item Problem Gambling Severity Index (PGSI) (Cooper & Marmurek, 2023; Ferris & Wynne, 2001). Items referred to the previous 12 months and used response options ranging from never to almost always. Higher scores indicated greater problem gambling severity.

Gambling satisfaction was assessed with three items measuring satisfaction and regret related to prior gambling choices, adapted from Lee and Choi (2017). Sports fanship was assessed with two items measuring enjoyment of watching sports, adapted from Wann and Branscombe (1993). Both measures used five-point response scales.

2.3.4. Impulsivity and Sensation Seeking

Impulsivity was assessed with the 20-item short UPPS-P Impulsive Behavior Scale (Cyders et al., 2014). Sensation seeking was measured with the eight item Brief Sensation Seeking Scale (BSSS-8) (Hoyle et al., 2002). Participants responded to both measures using five-point agreement scales.

2.4. Power Analysis

An a priori power analysis conducted in G*Power 3.1 (Faul et al., 2009) indicated that a minimum of 77 participants was required to detect a large effect (f² = 0.35) with 95% power at α = 0.05 in a multiple regression with nine predictors; a single-predictor model required 40 participants. The full sample therefore met the planned threshold for detecting large effects in the full sample regression models.

2.5. Statistical Analysis

Analyses were conducted in R version 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). Bivariate associations were examined descriptively, and ordinary least squares regression was used for the reported age and personality models. The significance criterion was α = 0.05. The primary analyses tested H1-H5. Exploratory post hoc models then examined two indirect effects: conscientiousness → impulsivity → problem gambling and age → narcissism → problem gambling, and one interaction model testing narcissism × openness as a predictor of sensation seeking. Because these models were exploratory and the data were cross-sectional, indirect effects are interpreted as statistical pathways rather than necessarily causal mechanisms.

During the statistical analysis, ChatGPT (OpenAI; GPT-5.6 Sol) was used to assist with debugging and refining R code. The authors reviewed and verified all code and statistical output and take full responsibility for the analyses reported here.

3. Results

3.1. Participant Characteristics

Participant characteristics are summarized in Table 1.

3.2. Reliability

Across all 12 Dirty Dozen items, internal consistency was high (α = 0.93, 95% CI [0.91, 0.95]). The BSSS-8 (α = 0.86, 95% CI [0.83, 0.90]), short UPPS-P (α = 0.87, 95% CI [0.84, 0.90]), and PGSI (α = 0.95, 95% CI [0.94, 0.96]) also demonstrated good to excellent internal consistency. For descriptive completeness, α across the 10 TIPI items was 0.68 (95% CI [0.59, 0.75]); however, the TIPI is designed to yield five two-item trait scores rather than a single unidimensional personality score, so this coefficient should not be interpreted as the reliability of one overall personality construct.

3.3. H1: Age and Problem Gambling Severity

A simple linear regression tested whether age predicted problem gambling among participants who reported gambling. The model was significant, F(1, 49) = 4.22, p = 0.045, R² = 0.08 (f² = 0.09). Age was a significant negative predictor of problem gambling, b = −0.02, SE = 0.01, t(49) = −2.05, p = 0.045. Thus, younger participants reported greater problem gambling severity, supporting H1.

3.4. H2: Age and Preferred Gambling Medium

The reported simple regression examining age as a predictor of betting medium was not significant, F(1, 48) = 1.65, p = 0.205, R² = 0.03 (f² = 0.03). Age was not a significant predictor, b = −0.005, SE = 0.004, t(48) = −1.29, p = 0.205. H2 was not supported.

3.5. H3a and H3b: Personality, Impulsivity, and Sensation Seeking

The personality model predicting impulsivity was significant, F(8, 42) = 3.80, p = 0.002, R² = 0.42, adjusted R² = 0.31. Conscientiousness was the only significant individual predictor, b = −0.36, SE = 0.15, t = −2.45, p = 0.019, with higher conscientiousness associated with lower impulsivity. H3a was supported at the omnibus model level.

The personality model predicting sensation seeking was also significant, F(8, 42) = 3.43, p = 0.004, R² = 0.40, adjusted R² = 0.28. Openness showed a positive trend but did not reach the conventional significance threshold, b = 0.39, SE = 0.19, t = 1.99, p = 0.053; no other individual trait was significant. H3b was supported at the omnibus model level, although evidence for specific individual traits was limited.

3.6. H4: Personality and Gambling Outcomes

Personality traits did not significantly predict general gambling behavior, F(8, 73) = 1.06, p = 0.40, R² = 0.10; no individual predictor was significant (all ps > 0.18). In contrast, the reported model predicting problem gambling among participants who gambled was significant, F(8, 41) = 4.87, p < 0.001, R² = 0.52. Narcissism was a significant positive predictor, b = 0.47, p = 0.018, and Machiavellianism showed a positive trend, b = 0.42, p = 0.065. No other predictors were significant (all ps > 0.11). H4 was therefore partially supported.

3.7. H5: Personality, Gambling Satisfaction, and Sports Fanship

The model predicting gambling satisfaction was not significant, F(8, 40) = 1.13, p = 0.37, R² = 0.18, and no individual predictor was significant (all ps > 0.14). The model predicting sports fanship was significant, F(8, 41) = 2.37, p = 0.034, R² = 0.32, although no individual personality trait reached statistical significance. H5 therefore received mixed support: personality did not predict gambling satisfaction, whereas the sports-fanship model was significant at the omnibus level.

3.8. Exploratory Indirect-Effect Analyses

3.8.1. Conscientiousness → Impulsivity → Problem Gambling

Given the negative association between conscientiousness and impulsivity and prior work linking self-regulation to risky behavior (Canale et al., 2017; Costa & McCrae, 1992), an exploratory indirect-effect model tested whether impulsivity statistically linked conscientiousness to problem gambling. The indirect effect was significant, b = −0.365, 95% CI [−0.885, −0.061], p = 0.011. Higher conscientiousness was associated with lower impulsivity, which in turn was associated with lower problem gambling severity.

The direct effect of conscientiousness on problem gambling after accounting for impulsivity was not significant, b = 0.302, 95% CI [−0.231, 1.005], p = 0.248, and the total effect was not significant, b = −0.063, 95% CI [−0.594, 0.554], p = 0.840. The proportion mediated estimate was unstable and not interpretable (5.82, p = 0.834).

3.8.2. Age → Narcissism → Problem Gambling

Because age predicted problem gambling and narcissism was positively associated with problem gambling in the personality model, a second exploratory indirect-effect analysis tested age → narcissism → problem gambling. Narcissism is also associated with risky decision-making in prior work (Crysel et al., 2013; Yang et al., 2018). The indirect effect was significant, b = −0.014, 95% CI [−0.027, −0.002], p = 0.024. Increasing age was associated with lower narcissism, which was in turn associated with lower problem gambling severity.

The direct effect of age after accounting for narcissism was not significant, b = −0.007, 95% CI [−0.024, 0.009], p = 0.377, whereas the total effect was significant, b = −0.021, 95% CI [−0.042, −0.001], p = 0.040. The reported proportion mediated was 0.665 (p = 0.043). This result shows a statistically consistent indirect association.

3.9. Exploratory Moderation: Narcissism × Openness → Sensation Seeking

An exploratory moderation model tested whether openness altered the association between narcissism and sensation seeking. The interaction was not significant, b = 0.014, p = 0.934. However, significant positive main effects were observed for narcissism (b = 0.278, p = 0.005) and openness (b = 0.470, p = 0.011), consistent with literature linking these traits to risk-taking, novelty, and exploration (Abu Raya et al., 2023; Crysel et al., 2013; de Vries et al., 2009). The overall model was significant, R² = 0.322, p < 0.001. Thus, narcissism and openness were independently associated with sensation seeking in this model, but openness did not moderate the association between narcissism and sensation seeking.

4. Discussion

The present study examined whether age and personality help distinguish gambling participation from gambling-related harm. The clearest pattern was that younger age predicted greater problem gambling severity, while age did not predict preferred betting medium. Personality also mattered selectively rather than uniformly: conscientiousness was associated with lower impulsivity, openness showed a positive trend with sensation seeking in the primary personality model, and narcissism predicted greater problem gambling severity among participants who gambled. Personality did not reliably predict general gambling behavior or gambling satisfaction, and the significant sports-fanship model did not identify a significant individual trait. These findings point to a more specific account of gambling risk in which age and personality are more informative for vulnerability and self-regulation than for gambling participation alone.

4.1. Age and Gambling-Related Harm

The negative association between age and problem gambling is consistent with evidence that gambling-related problems are more common among younger adults than older adults (Dellosa & Browne, 2024). Developmental differences in impulsivity and reward sensitivity may contribute to this pattern (Leshem, 2016; Steinberg et al., 2008), but the contemporary gambling environment also matters. Mobile access, rapid betting opportunities, and the integration of gambling into sports and digital media increase the number of contexts in which gambling cues can occur (Gainsbury et al., 2015; Thomas et al., 2018). The current data therefore support distinguishing age-related vulnerability to harm from perhaps popularized simple assumptions about who uses a particular gambling medium.

H2 was not supported: age did not significantly predict preferred betting medium. This null result is compatible with evidence that online and in person gambling participation frequently overlaps (Hing et al., 2022). For health communication research and practice, this suggests caution about treating online gambling as a youth only issue. Age tailored communication may still be useful when addressing vulnerability to gambling risks; digital harm reduction messages may need broad reach across adult age groups.

4.2. Personality, Impulsivity, and Sensation Seeking

Conscientiousness was the clearest personality correlate of impulsivity. This finding is consistent with the conceptualization of conscientiousness as planning, restraint, and self-control and with prior evidence linking personality to impulsive behavior (Lange et al., 2017; Mao et al., 2018). In practical terms, the result shifts attention from broad personality labels to a more specific self-regulatory mechanism. A health message designed for people prone to acting quickly may be more useful if it supports concrete precommitment, such as setting time or spending limits before gambling versus simply presenting general information about risks.

Openness showed a positive (but non-significant) trend in the primary sensation-seeking model, while openness was significant in the exploratory moderation model. Conceptually, openness and sensation seeking share an orientation toward novelty and exploration (Abu Raya et al., 2023; de Vries et al., 2009), but the present sample does not establish that openness is a stable predictor of gambling-related risks. Future work should test whether novelty based message strategies, alternative sources of stimulation, or message framing around exploration affect gambling intentions for people higher in openness versus other types of personalities.

4.3. Narcissism and Problem Gambling

Narcissism was the only individual personality trait that significantly predicted problem gambling severity in the reported gambling-specific model. This is consistent with research linking narcissism to risky decision making and gambling disorder related processes (Rogier & Velotti, 2018; Yang et al., 2018). Machiavellianism approached significance but did not meet the conventional threshold, which is consistent with the possibility that a more calculated risk orientation does not map onto gambling in the same way as narcissistic risk-taking (Jones, 2013). Broadly, Dark Triad traits should not be treated as interchangeable predictors when studying gambling or considering audiences in health message strategies.

For persuasive health communication, narcissism may be relevant because gambling promotions can emphasize competition, skill, confidence, and status. Prior work has identified illusion of control processes as important in gambling (Lopez-Gonzalez et al., 2018). A useful next step would be to experimentally test whether messages that correct overconfidence, make chance more salient, or provide personalized probability and loss feedback reduce risky intentions among individuals higher in narcissism.

4.4. Gambling Attitudes, Sports Fanship, and Media Context

Personality did not significantly predict gambling satisfaction, and no individual trait significantly predicted sports fanship, even though the fanship model was significant overall. These findings should caution others in attributing fandom to particular personality profiles. The sports context remains relevant because betting can change the viewing experience and become integrated with entertainment motivations (Lee et al., 2023; Lee et al., 2025). From a health communication perspective, responsible gambling messages placed within sports media environments may reach consumers at moments when betting is socially and emotionally salient (like during a major sporting event like the World Cup or Super Bowl). Whether those messages should be tailored by personality is an important question for future research.

4.5. Exploratory Psychological Pathways

The first exploratory indirect-effect model was consistent with a self-regulatory pathway: higher conscientiousness was associated with lower impulsivity, which was associated with lower problem gambling severity. The total effect of conscientiousness on problem gambling was not significant. The indirect effect therefore should not be seen as strong evidence that conscientiousness protects against gambling problems through impulsivity. Instead, it identifies a pattern worth testing prospectively with temporal separation between trait measurement, self-regulatory behavior, and gambling outcomes (e.g., in a controlled experimental or field-based study design).

The second exploratory model indicated an indirect association from age to problem gambling through narcissism. Younger participants reported higher narcissism in the model, and higher narcissism was associated with greater problem gambling severity. This finding suggests that age differences in gambling-related harm may partly coincide with personality differences rather than operating as a purely demographic age based effect.

The narcissism × openness interaction was not significant. Instead, narcissism and openness showed additive main effects on sensation seeking in that exploratory model. This distinction is useful in that the data do not support the idea that openness amplifies narcissistic risk tendencies. Future studies should test these traits independently before building more complex interaction models.

4.6. Implications for Persuasive Health Communication

The relevance of these findings to health communication lies in audience segmentation by psychological mechanism rather than by demographic category alone. The data suggest at least three candidate mechanisms for message research: age linked vulnerability to gambling harm, self-regulatory difficulty reflected in impulsivity, and narcissism related risk tendencies. These mechanisms could be paired with various message features; for example, cumulative loss information and long-term consequence framing for younger adults (e.g., loss of thousands of dollars in a year), precommitment and pause prompts for people high in impulsivity, and probability or illusion of control corrections for people high in narcissism (e.g., underscoring actual gambling odds over time for users).

At the same time, tailoring carries risks if it stigmatizes audiences or implies that personality itself is pathological. The more defensible target is the decision process: overconfidence, impulsive action, misperception of odds, or normalization of betting. Messages and platform features (e.g., gambling applications) should therefore be evaluated for effects on perceived risk, willingness to gamble, betting intentions, and actual behavior, rather than assumed to work because they appear theoretically matched to a trait.

4.7. Future Research

Future research should connect these individual differences to models that explicitly distinguish planned from reactive risk behavior. The Prototype Willingness Model has recently been applied to sports betting (Lambuth et al., 2026) and is particularly relevant because it separates reasoned intentions from situational willingness. Impulsivity may be more closely related to reactive willingness, whereas personality and fanship may shape the social prototypes and norms that make betting seem attractive or typical. Longitudinal and experimental designs could test these possibilities directly.

Message effects studies are also needed. A factorial experiment could cross message strategy (e.g., probability correction, social norm correction, self-regulation prompt) with audience characteristics measured before exposure. This would allow researchers to determine whether personality informed communication improves risk comprehension or reduces gambling willingness beyond a one size fits all message. Behavioral outcomes, repeated exposure, and real or simulated betting decisions would provide stronger evidence than self-reported attitudes alone.

4.8. Limitations

In regard to limitations, the study was cross-sectional; temporal ordering cannot be established, and the exploratory indirect-effect analyses should not be interpreted as causal mediation. The final sample was rather modest (N = 83), and gambling specific models were based on approximately 49-51 participants. The study also used self-report data from an English-speaking Prolific sample and did not use a clinical or population representative sampling frame. Generalizability to adolescents, clinical populations, other countries, or people with severe gambling disorder is therefore limited. Several constructs were measured with brief scales or single items (e.g., TIPI was intentionally brief, gambling frequency was measured with one item, etc.). The study also did not manipulate media messages or health communication interventions; future research should examine specific message designs and tailoring.

5. Conclusions

Age and personality were more informative for gambling-related vulnerability than for gambling participation in general. Younger age predicted greater problem gambling severity, conscientiousness was associated with lower impulsivity, and narcissism predicted greater problem gambling among participants who gambled. Exploratory analyses further identified indirect statistical pathways involving impulsivity and narcissism, while the narcissism × openness interaction was not significant. These findings support a mechanism focused approach to future gambling health communication: rather than assuming that one message will work for all audiences, research should test whether self-regulation, overconfidence, and age-related vulnerability alter responses to risk messages. Larger longitudinal and experimental studies are needed to guide targeted interventions.