ABSTRACT
What do you do when, despite your best efforts, some people cannot be reached or refuse to participate in the survey? This chapter presents methods that can be used to attempt to reduce bias resulting from nonresponse. In weighting class methods, the weights of respondents in a weighting class are increased to compensate for nonrespondents in the class. Calibration, which was introduced in Chapter 4, adjusts the weights of respondents using information about the population from an independent source. Imputation methods create a “complete” data set by using a statistical model to fill in values for data missing because of nonresponse. All survey-takers should report response rates, describe the steps taken to adjust for nonresponse, and assess nonresponse bias that may be present after those steps are taken.
