Examining the Association Between Interviewer and Respondent Speaking Pace in Telephone Interviews

Telephone interviewers are typically trained to speak at a pace of two words-per-second to enhance respondent cognitive processing. Although interviewer speaking pace varies across different question characteristics such as question length and complexity, the pace at which respondents answer questions in a telephone survey and whether pace varies by question characteristics has received scant attention. Furthermore, although there is a longstanding hypothesis that the speed at which interviewers ask questions influences the speed of respondent replies and that this in turn influences the quality of answers provided by respondents, few empirical studies directly examine the relationship between interviewer speaking pace and respondent speaking pace. This thesis examines the association between question-level interviewer and respondent speaking pace among the first two conversational turns in telephone interviews. Given lack of replication of how question characteristics are associated with the pace of interviewer question administration in previous research, I start by examining whether question linguistic and cognitive complexity, question sensitivity, and the position of the question in the interview are related to the pace of interviewer question administration. I additionally examine whether question linguistic and cognitive complexity, question sensitivity, respondent familiarity, and the position of the question in the interview are related to the pace of respondent initial replies to questions. Finally, I examine whether interviewer speaking pace predicts respondent speaking pace and if this relationship is moderated by question complexity. Using behavior coded transcripts from the Work and Leisure Today 2 Survey (AAPOR RR3=7.1%), I find that on average, interviewers speak at a pace of 3.15 (95% CI=3.136, 3.154) words-per-second and respondents reply at a pace of 1.33 (95% CI=1.319, 1.335) words-per-second. Interviewers ask linguistically complex questions at both a slower and a faster pace (depending on the indicator for question linguistic complexity), and respondents reply to linguistically complex questions faster than to questions that are not as linguistically complex. No other question characteristics are associated with interviewer or respondent speaking pace. Furthermore, interviewer question-asking pace is a significant positive predictor for respondent pace (b=0.13, p=0.006). The relationship between interviewer and respondent speaking pace is significantly moderated by question linguistic complexity. Advisor: Kristen Olso