Audit of the baseline survey of residential buildings in site II
An assessment of the completeness and reliability of data from a survey of the physical characteristics of residential buildings, the properties where they stood, and the immediate neighborhoods of a sample of 4,443 residential properties in St. Joseph County, Indiana (Site II of the Housing Assistance Supply Experiment). Outcomes of survey attempts were accounted for, and tests were conducted for nonresponse bias. Analysis weights were computed to minimize such bias. Checks were also made for response errors or inconsistencies and for anomalies in the survey instrument or field procedures. Less than 1 percent of the applicable responses were unusable, and only one serious form of nonresponse bias was found--in questions about interior public areas of large apartment buildings. Responses to descriptive questions were found reliable. The lesser reproducibility of responses to evaluative questions seems to reflect random response error and only slightly reduces the reliability of parameter estimates.