Using Paradata-Driven Models to Improve Contact Rates in Telephone and Face-to-Face Surveys

This chapter focuses on determining the average best times to call or the sequence of calls that have, on average, the highest contact rates. The telephonic survey uses automated call scheduling software and compares the estimated contact probabilities within a household and finds the window with the highest probability of contact for that household. Beyond telephone surveys, the computerization of face-to-face interviewing should lead us to ask whether this creates the possibility of providing centralized direction in the process of scheduling calls in field surveys. There are large number of factors that influence when and how frequently each case is called for computer-assisted telephone interviewing (CATI) surveys. The chapter concludes with a discussion of important questions for research in this area?including whether large gains in efficiency are possible, are statistical models well suited for this problem, and what should be the goal or ?reward function? of these models.