Bifactor Confirmatory Factor Analysis of the Gambling Motives Questionnaire – Financial Among Lottery Loyalty Program Participants: Associations with Gambling Problems

Background: Extensive research has explored the role of gambling motives to understand forces driving gambling behavior to develop appropriate preventions. The Gambling Motives Questionnaire – Financial (GMQ-F) measures four gambling motives (enhancement, social, coping, and financial). These overlapping constructs may be distinct but also represent an overall motivation for gambling. It is unclear the extent to which gambling motivation is a single unitary construct (i.e., a general factor) or a set of relatively distinct motivations (i.e., specific factors). It is also important to ascertain the extent to which motives lead to gambling problems. The current study aims to examine the factor structure of the GMQ-F by testing multiple-factor model configurations and testing the association between latent constructs and Problem Gambling Severity Index (PGSI) scores.

Methods: We utilized data from a lottery loyalty program in a midwestern state. Data from the lottery loyalty participants was supplemented with a survey that included questions such as demographics, GMQ-F, and PGSI. Confirmatory factor analyses with a sample of lottery loyalty program participants (n = 6,785) were performed to test first-order, second-order, and bifactor models. The best-fitting model was then used as independent variables with sociodemographic covariates in a structural equation model (SEM) to test associations with PGSI scores.

Results: The bifactor generated the best-fit indices [χ² ₍₈₈₎ = 3096.026, p < .001, SRMR = .049, RMSEA = .071, CFI = .988, TLI = .983]. The bifactor indices suggested limited multidimensionality with most of the variance explained by overall gambling motives, and limited variances explained by specific motives except financial motives. Omega for the general motive was .983 and Omega for specific motives ranged from .872 to .967. OmegaH for the general motive was .83 and OmegaH for specific motives ranged from .145 to .631. In the SEM model, general motives were associated with a higher PGSI score (b = 1.193). Social motives (b = -0.288) and coping motives (b = 0.533) were associated with PGSI scores.

Conclusions/Implications: Our findings support the value of the GMQ-F as an instrument for measuring gambling motivation. Rather than a set of distinct drivers of gambling, the GMQ-F predominantly reflects the general gambling motives, with less variance explained by specific factors (enhancement, social, and coping). This suggests that models assessing specific gambling motives are at risk of either high collinearity (if all subscale scores are modeled) or confounding general and specific motivation (if a single subscale is used). In considering the potential of the GMQ-F as a unidimensional measure, researchers should consider that the item variances from specific factors (social, enhancement, and coping) are predominantly loading on the general factor. Overall motivation to gamble was associated with problem gambling, and specific gambling motives that may confer added risk (i.e., coping) or lower risk (i.e., social) when adjusting for overall motivation. Other gambling motivations may not influence the risk of problem gambling once one accounts for overall gambling motivation. These findings may be the result of the bifactor latent variable approach and/or the specific characteristics of the lottery player sample.