Minimizing survey refusal and noncontact rates: do our efforts pay off?

Abstract

"This study investigates the link between the effort undertaken to collect survey data and the nonresponse error on a key survey estimate. For this purpose a threefold analysis was conducted. First, the level of nonresponse error and its composition is charted. Second, it is investigated whether these levels change throughout the fieldwork period. This helps answering the question whether collecting more data implies higher data quality. This type of analysis also provides a possible framework for a dynamic process control during the fieldwork period. A third and final analysis links interviewer efforts (in terms of number of contact attempts) to nonresponse error and its composition. The results show that error due to noncontact is 2.6 times higher than error due to refusal, even though the refusal rate is almost two times higher than the noncontact rate. Also, the results suggest that collecting more data does not necessarily imply higher data quality and that a higher number of contact attempts does not markedly reduce the nonresponse error in absolute terms. The analysis uncovers the underlying process responsible for this latter finding." (author's abstract)