Analysis of the Survey Response Behavior: An experience from a pilot survey of the health and living status of the 50s and beyond in Japan

Exploiting a survey of aged population implemented in Tokyo, we examine the targeted individual's decision to respond the survey. The sampling of potential respondents is based on the resident registry compiled by the local governments that carries all targeted individual's information on sex, age and exact street address. We matched this data with the land price of the street address and the survey administrative information that records interviewer's information. Our empirical findings reveal that whether targeted individual responds the survey or not depends on age, gender and land price. Most significantly the decision critically depends on interviewers' unobserved heterogeneity. We speculate the interviewer's effort to obtain responses crucially determine whether the targeted individual responds to the survey. Given the random assignment of interviewers to the targeted individuals, we argue that interviewers' heterogeneity can be used as an excluded variable for the Heckman sample selection correction.