Computing Weights for Student Surveys

• In Event: Data Consideration in Survey Research

Abstract

Representativeness is the essential characteristic of a student sample that allows researchers to generalize results to the population of interest. Weights are needed for sample surveys with less than a 100 percent response rate to make sure that estimates are unbiased and cover the population. Weights are also necessary when students have varying probabilities of selection. How can the results of a survey be weighted to be representative of the population?
This paper demonstrates methods educational researchers can use to compute weights to make estimates of population parameters based on survey responses. Examples will be provided from a large national student survey and from a small fictional survey that may be similar to one an institutional researcher may conduct.

Author

• Peter H. Siegel, RTI International