Computing Sampling Weights in Large-scale Assessments in Education

Abstract

Sampling weights are a reflection of sampling design; they allow us to draw valid conclusions about population features from sample data. This paper explains the fundamentals of computing sampling weights for large-scale assessments in educational research. The relationship between the nature of complex samples and best practices in developing a set of weights to enable computation of unbiased population estimates is described. Effects of sampling weights on estimates are shown, as well as potential consequences of not using weights when analysing data from complex samples. Illustrative examples are provided in order to make it easy to understand the rationale behind the mathematical foundations.