Typology-Based Prevention for Adolescent Online Gambling: Latent Profile Evidence from a Nationally Representative Korean Sample

Background and Purpose: Adolescent online gambling represents a growing public health crisis, yet existing prevention frameworks predominantly focus on reducing risk exposure while neglecting the independent role of protective capacity. Moreover, most measurement approaches treat gambling-related cognitions and behavioral tendencies as a single or binary construct, limiting the precision of risk identification. Grounded in the Theory of Planned Behavior and Social Cognitive Theory, this study pursued three aims: (1) to empirically re-examine the factor structure of gambling-related cognitive and behavioral tendencies among Korean adolescents, (2) to identify latent profiles based on the derived factor structure, and (3) to examine how these profiles differ in their associations with online gambling experience—with particular attention to the independent protective role of gambling refusal self-efficacy.

Methods: Data were drawn from the 2024 Korean Youth Gambling Survey conducted by the Korea Problem Gambling Agency (N = 13,368), a nationally representative sample collected via stratified cluster sampling. Principal component factor analysis with promax rotation was applied to 14 gambling-related items originally categorized as gambling perceptions and attitudes, yielding a three-factor solution. Latent Profile Analysis (LPA) was then conducted using the three derived factor scores as continuous indicators. Associations between latent profiles and online gambling experience were estimated using weighted logistic regression, adjusting for sex, age, school level, monthly allowance, and prevention education participation.

Results: Factor analysis revealed a three-factor structure comprising “gambling outcome expectancies”, “gambling exposure vulnerability”, and “gambling refusal self-efficacy”, with cumulative explained variance of 70.93%—empirically validating a more differentiated measurement framework than previously applied. LPA identified four profiles: Low-Risk High-Efficacy (61.5%), Low-Risk Low-Efficacy (10.5%), Moderate-Risk Low-Efficacy (23.4%), and High-Risk Vulnerable (4.5%). Relative to the Low-Risk High-Efficacy group, the High-Risk Vulnerable group showed markedly elevated odds of online gambling (aOR = 12.09, 95% CI [6.52, 22.43]). Critically, the Low-Risk Low-Efficacy group—adolescents with low risk exposure but insufficient refusal self-efficacy—showed significantly elevated odds (aOR = 1.74), demonstrating that compromised protective capacity alone is sufficient to elevate online gambling risk, independent of risk factor level. Furthermore, prevention education participation showed a paradoxical positive association with gambling in several subtypes, suggesting current single-session lecture-based programs may inadvertently stimulate curiosity rather than reduce risk.

Conclusions and Implications: These findings carry direct implications for social work practice and youth gambling prevention policy. First, the three-factor structure provides practitioners with a more actionable assessment framework—distinguishing between adolescents who need cognitive reframing, those requiring environmental boundary-setting, and those needing refusal skill development. Second, the Low-Risk Low-Efficacy finding reframes prevention as a dual-track social work mandate: building protective capacity must be treated as a primary intervention target, not merely a supplement to risk reduction. Within multi-tiered systems of support, profile-based screening enables differentiated intervention—from peer-led normative education for low-risk youth to intensive wraparound case management for the High-Risk Vulnerable group. The paradoxical education effect further calls on social workers to dismantle lecture-based prevention in favor of structured, multi-session, skill-based programs embedding refusal training, cognitive distortion correction, and peer norm restructuring.