After noticing Nate’s recent blog about the idea that people will give more “socially desirable” survey responses to human interviewers, my colleague Jeff Fagan, a criminologist, wrote:
There’s a solid literature showing that people report all sorts of ‘sensitive’ behaviors at higher rates when people respond using any form of electronic response. NIH showed this in both HIV prevention trials on sexual behavior and in research with adolescents and young adults on violence.
Here’s one article (by Turner, Ku, Rogers, Linberg, Pleck, and Sonenstein, from 1998), which has led to several others, including David S. Metzger et al., Randomized Controlled Trial of Audio Computer-assisted Self-Interviewing: Utility and Acceptability in Longitudinal Studies
American Journal of Epidemiology Vol. 152, No. 2 : 99-106 (2000), and Understanding the effects of audio-CASI on self-reports of sensitive behavior, MP Couper, E Singer, R Tourangeau – Public Opinion Quarterly 67: 385-395 (2003).
P.S. To head off some of the comments: Yes, we usually present our own findings on this blog, but I think it can be helpful to use the published literature to illuminate the analyses of Nate and others.