Study of Income Nonresponse in a Survey of Japanese Elderly

A large amount of economic research has been done evaluating the characteristics of individuals who refuse to participate in a survey, unit nonrespondents, and the effects these missing individuals have on the survey sample’s respresentiveness of the target population (cf. Hill and Willis 2001; Horowitz and Manski 1998). While the nature of unit nonresponse has been thoroughly examined, the study of individuals who refuse to answer specific questions within a survey, item nonrespondents, has been largely neglected. This negligence is surprising because the loss of information and sample representativeness that results from item nonresponse can be just as significant as the loss of information resulting from unit nonresponse. Moreover, past studies have found that survey questions that ask for a respondent’s personal income are particularly prone to item nonresponse (cf. Gronhaug et al. 1988 or Sheth et al. 1980). Since income is one of the most commonly used variables in economic research, the identification of a nonresponse bias from item nonresponse to an income measure has important implications for past and future economic research. This paper aims to identify significant predictors of item nonresponse to a question about personal income in a sample of elderly Japanese using a logistic regression framework. In thi