Influence of sponsorship on betting intentions: from linear to deep learning models.

Sports sponsorship by commercial betting providers is an essential source of income for sports entities worldwide. This study examines the formation of betting behavior in sporting events using predictor variables, namely, self-congruence, perceived exposure, subjective norms, goodwill, and involvement, to predict betting intention and actual gambling behavior, comparing linear (PLS-SEM) and nonlinear (artificial neural networks-ANN) methods. Data from 661 Spanish subjects were analyzed. The results of the PLS-SEM model identified subjective norms and self-congruence as the main predictors of betting intention, whereas goodwill showed limited effects. However, the ANN analysis revealed complex patterns not detected by linear methods, where variables such as goodwill acquire greater relative importance when considering complex interactions between variables. The research contributes theoretically by highlighting the need to adopt conceptual frameworks that consider the complex interaction between psychosocial variables, overcoming traditional models that analyze the determinants of gambling behavior in isolation. Additionally, it provides new knowledge about the differentiated and complementary influence of self-congruence, perceived exposure, goodwill, and involvement in predicting gambling behavior, constructs that had not been systematically integrated into a joint predictive model. Methodologically, it contributes by demonstrating that the integration of SEM and ANN approaches improves the robustness and validity of the analysis, offering a more complete view of the factors that influence the intention to gamble. The findings have practical implications for the design of more effective problem gambling prevention strategies and regulations.