Abstract:
|
The National Health Interview Survey (NHIS) recently underwent its first major questionnaire redesign since 1997, with the goal of reducing respondent burden by shortening the questionnaire and streamlining instrument flow. The redesigned questionnaire was first fielded in quarter 4 of 2018 during a bridge period and continued into 2019 and beyond. Using 2018-19 NHIS data, we evaluate the extent to which the redesigned questionnaire (versus the previous questionnaire) minimizes the variability of responses across interviewers, and the extent to which interviewer effects vary by question and interviewer characteristics. Since interviewer effects are often difficult to estimate in face-to-face surveys due to non-interpenetrated designs, we estimate multilevel models that control for respondent, interviewer, and county characteristics to help separate interviewer from area effects. Preliminary analyses of 86 variables reveal smaller interviewer effects, on average, with the redesigned questionnaire although the size of effects and differences by questionnaire design vary by content (e.g., health status, sociodemographics).
|