Social Capital and Social Cohesion. Ascertaining Social Cleavages from Everyday Interaction in a Latin American Context

Recent debates on social capital have disputed about the association between bonding and bridging social capital. Putnam (2007) conjectured that they would be complementary, as long as personal networks were large enough. Research by DiPrete and colleagues (2011) provided evidence against the compatibility of both types of social capital. Recently, Abascal & Baldassarri (2015) argued that these hypotheses are contingent on compositional effects affecting diversity. This debate has strong implications for the characteristics of social cohesion and social trust at the macro-social level because, following Granovetter's (1973) seminal argument, bonding social capital would be associated with fragile social structures characterized by lack of solidarity and distrust.

I will examine Chileans' personal networks to establish to what extent their social connections cut across the social boundaries of class and territory. I will argue that Chilean social structure stems from mostly homophilous social networks garnering kin and close friends. At a macro-social level this composition yields social and geographic segregation, reducing the opportunities to access distant social circles. Additionally, regarding the normative side of social capital, segregation fosters distrust among socially diverse groups and amplifies perceptions of conflict.

Data consist of a positional network generator of 13 occupations used in the first wave of Chile's Longitudinal Social Survey (ELSOC), a 2016 nation-wide survey. The survey also includes indicators of social trust and perception of conflict. Three indicators of social capital can be generated from these data: diversity, volume and quality (highest status). As a first step in the analysis, social capital indicators are regressed (multilevel) on selected demographic, territorial and ideological covariates. In a second step, personal networks indicators, considering their composition, along with individual and territorial data are used to predict individual levels of trust and perception of conflict . Finally, I run comparable analyses for kin and non-kin contacts.