Sample Design Effects for Social Surveys: Recent Research and Current Issues

Abstract

Design effects are useful measures of the impact of a sample design on the precision of survey-based estimates. However, they are often misunderstood. Research in recent years has provided some clarity regarding definitions and some guidance regarding practical applications, including both issues relating to the use of predicted design effects to inform sample design and issues relating to the use of estimated design effects to inform inference. An overview of the contributions of this research is provided. However, several issues remain unresolved and survey researchers often use approximations and simplifications that may or may not be reasonable. Important areas for future research are suggested.