Nonresponse Adjustments for a Telephone Follow-up to a National In-Person Survey
Hüseyin Göksel, David R. Judkins, and William D. Mosher
The National Survey of Family Growth (NSFG), Cycle IV Baseline, was based on in-person interviews in 1988 with 8,450 women selected from several cycles of the National Health Interview Survey (NHIS). Nonresponse adjustments for Cycle IV Baseline were described in an earlier article. In 1990, a telephone reinterview, the CATI Phase, was conducted with about 5,700 of the same women who were interviewed in 1988. Some of the NHIS variables that predicted nonresponse to the Baseline also predicted nonresponse to the CATI Phase, but several Baseline variables were considerably better predictors. Mobility was the most powerful predictor, but race, Hispanic origin, education, and other socioeconomic variables and several variables specific to the subject matter of NSFG were also closely associated with nonresponse. The article describes how the determinants of nonresponse were modeled, how the sample weights were adjusted for nonresponse, and presents the results of an evaluation.
Response propensity; Automatic Interaction Detection (AID).