Web Survey Bibliography
Telephone interviewing has gained great importance in empirical social research, as it enables fast and cost eective data-gathering as well as the use of random samples (in contrast to online surveys). Compared to face-to-face interviews, however, the response rates are relatively low. To gain extended understanding of the mechanisms behind unit nonresponse, this study relates response rates to interviewers' voice characteristics. We use contact data generated in a survey with 78 interviewers and 606 respondents (gross sample N=2227) and supplement these data with information interviewers' voices. The survey was conducted during 2007 and 2008 at the University of Mannheim, Germany. To obtain metadata, we recorded the interviewers' voices and analyzed the recordings using praat, a software for acoustic analysis. In a second step, each voice was evaluated on a number of subjective dimensions by 20 raters each (n=156). Using this combined dataset, we study the determinants of interviewer eectiveness with a special focus on objective voice characteristics (e.g. pitch, loudness etc). Additionally, we account for gender, local and foreign accent as well as a variety of subjective voice attributes (friendliness, trustworthiness etc). First results show that (a) objective voice characteristics have greater explanatory power than subjective indicators and (b) a dierentiation by gender is indispensable to obtain valid results. Since subjective indicators are related to the objective measures and interviewer eectiveness only weakly, the further study of objective voice characteristics seems to be a worthwhile opportunity for future research.
Conference homepage (abstract)