C2.5 Considerations of Unobserved Heterogeneity in Internet-Based Survey Data

Wednesday, March 14, 2012: 11:50 AM
Regency
Derek Smolenski, PhD, Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, MN

Sampling of study participants via the Internet assumes that individuals recruited through the same source websites constitute a homogeneous study population.  The data, however, can reflect multiple distributions of risk behavior and exposure data that are combined to reflect a single distribution given the absence of directly-observed grouping variables.  Contemporary analytic techniques, such as latent class analysis (LCA), provide an opportunity to model participant heterogeneity which can improve analytic precision and validity.