The objective of this article is to test conflicting hypothesis predicting the causal mechanism linking civil society associations engagement, political participation and political efficacy. The works by Toqueville, Almond & Verba, and Putnam gave the most important contribution to the socialization hypothesis, claiming the importance of non-political associations for increasing political participation, by enhancing members’ civic skills and political efficacy. According to this approach civil society associations should work as ‘schools of democracy’. Yet, empirical analyses showed mixed support to this theory and authors recently began to propose an alternative selection model hypothesis: no causal link exists between the two types of participation, rather citizens with higher political efficacy are induced both to join social associations and to participate in political action. Since the The Civic Culture research in the early 60s, the Italian context, characterized by low social participation, low political efficacy, but high political participation has been a paradox in light of the neo-Tocquevillian theories, most of all considering that association membership has grown in the last 50 years and that political participation has decreased, while political efficacy has remained low. Using a three-wave panel data of Italian National Election Studies, causal models are tested through structural equation modelling (SEM) in order to explore this paradox and to shed light on the direction of causality relating these three variables in Italy.
