|
Abstract:
|
In recent years, an increasing number of researchers have been able to access survey microdata files. These researchers perform various analyses to summarize the data and to describe relationships in a target population. Many of these researchers use analytic software without having a good understanding of the statistical underpinnings behind the methods being applied. Some of the issues facing the researchers include understanding the survey weights, understanding informative sampling, handling nonresponse, using variances that are model-dependent, incorporating survey design information into the modeling process, understanding the impact of model misspecification, inferring causality, and pooling/combining of survey files. In this paper we review many of these issues, focusing on pooling/combining of survey files.
|
