Consumer Cellular Database: More Efficient, but at What Cost?
Caroline Scruggs
RTI International
Marcus Berzofsky
RTI International
Thomas Duffy
RTI International
Bo Lu
Division of Biostatistics, The Ohio State University
Timothy Sahr
Ohio Colleges of Medicine Government Resource Center
As telephone surveys increase the proportion of respondents who come from the cellphone frame, the question of how to accurately and efficiently target respondents in small geographic areas has become a major issue for survey methodologists. In this paper, we evaluate MSG's Consumer Cellular Database to determine: (1) the accuracy level of the address information on the database (2) if those obtained through the database are different in terms of demographic characteristics and key outcome variables compared to those on the RDD cellphone sample and (3) if the database yields respondents at a lower cost. The Consumer Cellular Database uses multiple publicly available sources to construct household and person-level files. Using the 2017 OMAS, our evaluation selected a stratified random sample of cellphone numbers using the Rate Center Plus method to minimize the classification error associated with using rate centers as proxies for counties followed by further stratification by whether the number was on the database. We assess efficiency through yield rates and call attempts. We evaluate differences by comparing demographic characteristics and key outcome variables.