Table 3:

Interviewer Differences in Prompting Behavior (Based on Ten Recordings Each), Number of Names Elicited, and Mean Time in Eliciting Section.

Interviewer	Average percent of questions prompteda	Percent of q’s using other promptingb	Mean number unique names elicitedc	Mean number of names per questiond	Mean number of minutes in name-eliciting sectione
1	31	10	10.7	3.3	8.5
2	63	3	12.2	3.6	7.3
3	11	2	8.7	2.1	6.5
4	4	36	12.0	3.7	8.9
5	45	33	10.3	3.2	8.9
6	12	3	9.5	2.8	7.6
7	1	12	8.2	2.2	8.2
8	61	9	11.5	3.6	9.3
9	60	53	11.7	3.7	10.0

Notes: Interviewer 10 only conducted two interviews, so was not included in the analysis.

^aWe consider interviewers “prompting” respondents when they mimicked the web administration of the survey by reading the entire prompt and—in particular—telling respondents how many names they could list for a question.

^b“Other prompting language” included instances where the interviewer asked the respondent questions such as, “Is there anyone else?”, “Is that all?”, or “How about your sister?”, or if interviewers told respondents “I have space for x more names.”

^cThe average number of total unique names respondents provided to that interviewer.

^dThe average, across respondents, of the mean number of names provided per question for that interviewer.

^eThe average, across respondents, of the mean number of minutes in the main name-eliciting section for that interviewer. (The mean for web respondents is 9.6, with a higher skew than among interviewers.)