The studyholism comprehensive model: towards a Bayesian reanalysis

The construct of Studyholism has been defined in the seminal work of Loscalzo and Giannini (2017 1 ) to characterize the obsession toward study in terms of its potential antecedents and outcomes. In a recent work, a path analysis model has been proposed (Loscalzo and Giannini, 2019) to include some antecedents and outcomes of both studyholism and study engagement. The above model was fitted to 1958 Italian college students aged between 18 and 60 years and quite heterogeneous about their year and major of study, as well as concerning the city in which they attended their courses (details in cited works). The above studies bear important implications not only because Studyholism is theoretically defined as an OCD-related disorder, but also for potential insights connected to preventive and clinical practice. For example, in a clinical setting it is important to distinguish between disengaged studyholics and engaged studyholics: although they both have functional impairment this is located in different areas. Factor and path analysis models have been questioned from a statistical point of view: (i) the multivariate normal framework could be a coarse approximation, even after summing related ordinal variables; (ii) statistical tests based on the multivariate normal family of distributions do not closely match the nature of underlying variables. In this work, we took the ordinal nature of most qualitative variables seriously while implementing a Bayesian Structural Equation Model. Model fitting was performed by Markov Chain Monte Carlo simulation. We closed the discussion of results from the Bayesian reanalysis by emphasizing the distinctive achievements as compared to previous work.