Ben Vogel (Dept. of Mathematical Sciences, MSU)

11/21/2024  3:10pm

Abstract:  A brief historical introduction to network scale-up models, followed by a discussion of two new proposed advances in Bayesian Network scale-up models. Specifically, we propose to increase the accuracy and reliability of hard-to-reach population size estimates from Network Scale-up Methods (NSUM) by extending the model in Laga et al. (2023a). We incorporate correlated respondent social network size with reported subpopulation contacts and compensate for data censoring in Aggregate Relational Data. Correlations are directly estimable from NSUM survey data and simulations demonstrate that subpopulation size estimates are biased when data correlations are excluded. Two data sets from McCarty et al. (2001) and Salganik et al. (2011) are analyzed, providing both hard-to-reach population size estimates and novel sociological information about the structure of social networks in these communities.