April 29, 2016
New Research by Mneimneh and Colleagues Highlights Cross-Cultural Variability in the Effects of Interview Privacy and the Need for Social Conformity on Survey Data Quality
Together with her colleagues from Westat (Roger Tourangeau) and SRC (Beth-Ellen Pennell, Steve Heeringa, and Mike Elliott), SRC Assistant Research Scientist Zeina Mneimneh recently first-authored an original research article in the Journal of Official Statistics entitled 'Cultural Variations in the Effect of Interview Privacy and the Need for Social Conformity on Reporting Sensitive Information.' This article presents work from Zeina's dissertation research.
The objective of this study was to examine the effects of third-party presence in a face-to-face survey interview on reporting patterns for both sensitive survey questions (suicidal behavior and marital ratings) and relatively neutral questions (having high blood pressure, asthma, or arthritis). The authors take an interesting approach in assessing whether or not these effects are moderated by the respondent's need for social conformity and/or characteristics of the country in which the respondent is living. Three country-level characteristics were investigated: individualism, masculinity, and GNI per capita. The authors analyzed data from more than 22,000 interviews completed in nine countries as a part of the cross-national World Mental Health Survey Initiative. The authors used the GLIMMIX procedure in SAS to fit multilevel logistic regression models to the survey responses that accounted for random interviewer effects on the responses.
The authors found that 37% of the interviews analyzed were conducted in the presence of a third person. As expected based on theory and past literature examining bystander effects on survey interviews, the authors found that third-party presence does indeed affect reporting on the sensitive survey questions, and that in the case of suicidal behavior, this effect is moderated by a respondent's need for social conformity. Respondents scoring high on a social conformity scale who had a third party present were less likely to report suicidal behavior than those who had no one else present. For individuals scoring low on need for social conformity, having a third party present significantly increased the odds of reporting suicidal behavior. The country variables were not found to moderate these effects.
Unlike the findings for suicidal behavior, the effects of third-party presence on ratings of marital relationships were not moderated by need for social conformity, but the ratings were still generally higher when a third party was present. This was especially true in countries with middle GNI per capita. No effects were found for the more neutral questions, as expected.
This work has important practical implications for surveys employing face-to-face interviewing. The authors call for more research into the possibility that self-administration of sensitive questions in face-to-face surveys would help to reduce these third-party effects, better training of interviewers on how to achieve and maintain interview privacy (even if these questions are self-administered), and collecting more specific observations about the context of third party presence.