The Effects of Errors in Paradata on Weighting Class Adjustments: A Simulation Study
Brady Thomas West · 2013
This chapter considers the use of paradata for postsurvey nonresponse adjustments. It presents the results of simulation studies designed to examine the effects of varying levels of error in survey paradata on the effectiveness of postsurvey nonresponse adjustments. The chapter focuses on binary survey process variables as opposed to other forms of paradata like the number of calls made to a sampled case, where setting error parameters for simulation purposes becomes much more complex. The simulation studies presented here aim to examine the link between error-prone paradata and the effectiveness of weighting class adjustments for nonresponse in more detail. The chapter determines whether use of the paradata for nonresponse adjustments would be advisable. If the collected paradata are of high enough quality, future data collections could proceed without the need to purchase the additional information.