To Pool or Not to Pool : Accounting for Task Non-Attendance in Subgroup Analysis

Pooling data from different subgroups offers advantages of shrinking standard errors with larger sample sizes and simplifying characterization of the data structure. Testing for poolability requires accounting for differences in response variance or scale among subgroups. This is commonly done by assuming a single scale factor within each subgroup of interest. We use a model of task non-attendance to relax the assumption of a single scale effect across subgroups in DCE data and to compare the preferences of respondents in each group who provide meaningful tradeoff information. Our results suggest that task non-attendance can be a significant problem in the estimation of average preference weights, and by extension in the determination of poolability when conducting subgroup analysis. We propose a relatively simple latent-class/random-parameters Logit (LCRP) model that accommodates task non-attendance across subgroups and directly tests for poolability of responses