Attrition and Response Effects in the Dutch National Mobility Panel. Methods and Applications

Contains fulltext : 140472.pdf (publisher's version ) (Closed access)In this chapter, we study the effects of under-reporting and attrition on inferences from the Dutch National Mobility Panel. Using refreshment samples, we assess the biases in the panel. We also investigate potential sources of bias, and distinguish between response effects and selective attrition. We use a random effects stochastic censoring model and exploratory methods to estimate the size of the attrition bias. We indicate some limitations on the type of inferences from the panel, and discuss possibilities to minimize attrition bias by choosing an appropriate sample design