Statistical Modelling for Survey Analysis
Richard DaviesCentre for Applied Statistics, Lancaster University
INTRODUCTION
Over the last decade there has been a notable change of emphasis in the statistical analysis of survey data. Increasingly, social science researchers are finding it informative to fit probability models to their data. A wide variety of such models are now used routinely including log linear models, survival models, discrete choice models, latent class models, multi-level models, and many others. These statistical modelling methods contrast strongly in their sophistication with the hypothesis testing and estimation of confidence intervals which once dominated social statistics...