ABSTRACT
Nonresponse is a pervasive problem in applied survey research. No matter how aggressive or ambitious the data collection protocol may be, the reality is that few surveys are able to collect complete data for the entire sample. Even surveys with a legal mandate for sampled individuals to participate, such as the U.S. Decennial Census, are faced with some degree of nonresponse. Common causes for nonresponse include failing to locate or make contact with the sampling unit, the respondent refusing to answer one or more sensitive questions, or the respondent refusing to participate in the survey altogether. Whichever the cause, nonresponse presents a dilemma during the estimation stage.
