The Value of the Increasing Effort to Maintain High Response Rates in Telephone Surveys

Declining response rates and coverage in random-digit-dial (RDD) telephone surveys have been observed by many researchers. Several studies have also shown that efforts to increase response rates often do not significantly affect estimates for key outcome variables, but few such studies have been conducted on large scale surveys including a broad range of health services measures. Using the Community Tracking Study Household Survey, a health services survey of roughly 30,000 families per round based on a national RDD sample, we examine the impact on key survey estimates of different simulated levels of effort. Using call history data, we simulate fewer call attempts, fewer refusal conversion attempts, and shorter time periods in the field than were actually pursued, and then re-weight the data according to the simulated outcomes. We then examine the impact of these reduced efforts on weighted estimates, comparing them to the estimates resulting from the complete survey data. These comparisons shed light on whether reducing the level of effort during data collection is likely to affect survey estimates for commonly used health services measures.