Conference Paper 19 Sinapis - National Symposium of Probability and Statistics, 26-30 July 2010, São Pedro, Brazil
Misspecification effects in the analysis of longitudinal survey data
Misspecification effects (meffs) measure the inflation of the sampling variance of an estimator as a result of the use of complex sampling schemes. Many longitudinal social survey designs employ multi-stage sampling, leading to some clustering of the sample and to meffs greater than one. For a model for panel data we consider methods for estimating parameters which allow for complex schemes. An empirical study using longitudinal data from the British Household Panel Survey is conducted, and a simulation study is performed.