|
Abstract:
|
This paper explores the relationship between proxy measures of burden and survey paradata with survey response for a longitudinal household survey using regression trees. Using data from the Current Population Survey we modeled response propensities over the eight monthly panels of data collection for all households in sample between 2015 and 2017. The paradata included number of attempted contacts and four factors representing concerns expressed by householders (Privacy, Time, Hostility, Gatekeeping). Proxy measures of burden included the following dimensions as described by Bradburn: 1) length (interview mode, number of contact attempts, and number of months in survey); 2) effort (number of adults, self/proxy reporting); 3) stress (income, ethnicity, marital status, and education refusal indicators); 4) frequency (the number of supplemental surveys the participant was previously sampled for); and 5) interest (employment status, disability status, home ownership, presence of children, and education level). These measures were studies in a previous paper using Structural Equation Models. This paper examines the relationship between these five burden dimensions, survey paradata,
|
