Testing for Structural and Institutional Correlates of Accuracy in Individual Perceptions of Social Network Structure

• In Event: 373 - Regular Session. Group Processes 3

Abstract

At least since the pioneering studies of Bernard, Killworth, and Sailer (Bernard,Killworth, Sailer, 1984), it has been known that individuals’ perceptions of the social networks are subject to considerable error. Although later work (see e.g., Romney and Faust 1982; Butts, 2003) has suggested that the accuracy of informants in assessing their social environments is greater than implied by earlier studies, it is clear that error levels are often substantial. Since social processes ranging from tie formation and dissolution (Snijders, van de Bunt, Steglich, 2010) to personal power, competition, and influence (Krackhardt, 1990; Burt, 2004) have been hypothesized to depend upon actors’ perceptions of network structure, this gap between what individuals perceive and the actual structure of relations in their environment is of critical theoretical importance. Unfortunately, quantitative studies of this gap have been hindered by a failure to employ methods that estimate both the structure of the network and informant accuracy to account for their joint uncertainty.

We employ a hierarchical Bayesian analysis to address this problem, allowing us to investigate the relationship between where individuals stand in their social environment -- both in network analytic and institutional terms -- and their accuracy in perceiving the interpersonal networks in which they are embedded. As we show, informant accuracy varies with one’s network position. These findings have potentially important implications for both the practical problem of informant selection in settings for which interviewer resources are limited, and for the substantive question of how individuals understand and navigate their social environments.

Author

• Francis Lee, University of California-Irvine