Undirected and directed networks of problem gambling: Follow the money trail

Psychometric network analysis offers a granular understanding of problem gambling (PG) by modeling individual symptoms within an interconnected system. Identifying central symptoms can help explain symptom co-occurrence and core symptoms relevant for escalating severity.

We estimated undirected cross-sectional and directed cross-lagged network models based on four representative Norwegian population surveys (2013, 2015, 2019, and 2022; age range 16-74) and assessed network stability, accuracy, and generalizability. Cross-sectional samples ranged from n = 3,232 to 6,034 (47-50% women) and cross-lagged samples included n = 3,043 (2013-2015, 50% women) and n = 2,749 (2019-2022, 47% women). PG symptoms were measured with the Problem Gambling Severity Index alongside self-reported gambling expenditure.

Within undirected networks, financial problems showed the strongest strength centrality and these networks showed strong generalizability, including the 2019 and 2022 surveys per Network Comparison Test (p = .65), edge presence (φ = .79), and edge weight (r = .85) correlations. Within directed networks, borrowing money had the strongest association with symptoms at follow-up (out-strength), while guilt at follow-up was most strongly predicted by other symptoms (in-strength). Generalizability between the 2013-2015 and 2019-2022 networks was inconclusive, with low correlation in edge presence (φ = .08) and high correlation in edge weights (r = .74).

Financial problems and borrowing money emerged as the most central PG symptoms in undirected and directed networks, respectively. Following further investigation, these symptoms may act as informative targets for screening, early intervention, and harm-reduction policy.