Chapter 18: Calculating and Using Weights

A basic idea behind survey research is that it is hard to talk to every member of a group, but researchers can talk to a few people (i.e., sample) and their voices can speak for the population as a whole (de Leeuw et al., 2008). Furthermore, researchers do not have to actually talk to the few chosen representatives of the group, they can send them a self-administered survey as a means to capture opinions, attitudes, beliefs, emotions, or self-report of a variety of behaviors. However, as soon as these simple ideas are put into motion, many important questions emerge almost immediately. This chapter will focus on just one of these questions: How can the results of a survey be weighted in order to be representative of the population? Representativeness is the essential characteristic of a sample that enables generalizations back to the population. However, perfectly representative samples are almost impossible to obtain. The methods addressed in this chapter are critically important for educational researchers who are attempting to make precise estimates of population parameters from sample statistics that are based on survey responses.