Can interviewers effectively rate the likelihood of cases to cooperate?

This paper explores how well interviewers can judge which cases are likely to cooperate with a survey request and which are unlikely. Interviewers in a telephone survey rated the response likelihood of each case after every call they placed on a scale from zero to 100, where zero meant that the case would never cooperate and 100 meant that the case certainly would. Analyses of the ratings reveal that they do correlate with the cooperation rate among the cases, and that the likelihood ratings are influenced in the expected directions by case-and call-level attributes. However, there are also strong interviewer effects. The ratings show promise for use in responsive design procedures.