The multitude of gambling activities corresponds to heterogeneous ways of analyzing behavioral outcomes and may partially underlie the lack of replication in gambling research. The current study incorporated complementary analyses to provide an approach to investigate associations with multi-dimensional gambling data that we demonstrated in a discovery/community sample of 2,116 twins (54.86% female; Mage=24.90) and a replication/clinical sample of 619 siblings (30.37% female; Mage=28.00). Our proposed approach was twofold. First, we used confirmatory factor analyses (CFAs) to derive a general gambling frequency measure across eight gambling activities and to test the common liability hypothesis. Second, we conducted latent class analyses (LCAs) to derive gambling frequency subtypes and investigated their theoretical correspondence with the Pathways Model. Our study identified robust support for the common liability hypothesis of gambling, suggesting a shared mechanism for multiple gambling behaviors – including activities controversially defined as “gambling.” Using LCAs, we identified four novel gambling frequency subtypes with analogous behavioral profiles and correlates across samples and some resemblance with the Pathways Model. The subtype with the highest rates of problem gambling had a frequent appetite for particular gambling activities and demonstrated externalizing psychopathology comparable to the “antisocial impulsivist” pathway. Using co-twin control models, we determined that risk-taking, sensation seeking and antisocial personality disorder predicted gambling frequency above and beyond shared genetic and environmental factors. In sum, we illustrated the utility of multi-dimensional statistical techniques for disentangling the structure and typology of different gambling behaviors and discussed our results in context of the psychometric, empirical and theoretical implications.
