Conditional administration of items in the Gambling Harms Measure: Psychometric justification and implications for efficiency.

OBJECTIVE

This study evaluated the psychometric justification for conditionally administering the most severe items within subscales of the Gambling Harms Measure (GHM), a tool assessing gambling-related harms across six domains: financial, psychological, relationship, physical health, work/study, and legal.

METHOD: Data were drawn from the third wave of an ongoing longitudinal study, with 2,849 participants completing the GHM. We computed item endorsement rates, crosstabs comparing each conditional item (most severe item within each subscale) with its preceding item, Cronbach's α with and without conditional items, tests of equivalence, and a paired-samples t test comparing total scores with and without conditional items. A confirmatory factor analysis evaluated potential changes in factorial structure due to conditional administration.

RESULTS

Conditional items were endorsed by 2.8%-4.7% of participants overall, and 1.1%-1.4% participants endorsed these items without endorsing the preceding item. Chi-square tests of independence yielded statistically significant associations across subscales (phi between .49 and .56), indicating moderate-to-strong associations between each conditional item and its preceding item. Cronbach's α for the full scale was .931, decreasing slightly to .927 when using the conditional scoring procedure. Paired t tests showed significant differences between total scores with and without conditional items (Cohen's d = 0.10-.0.12), but equivalence testing supported practical equivalence within 0.1 raw score bounds. Confirmatory factor analysis supported a single-factor model.

CONCLUSIONS

Conditional administration of the most severe items on the GHM preserves psychometric properties while reducing response burden. This method may enhance feasibility in large-scale surveys or screening interviews without compromising reliability or validity. (PsycInfo Database Record (c) 2026 APA, all rights reserved).