Assessing the Use of a Pre-Field Screening Service to Identify Nonworking Cellphone Numbers in Ohio

Abstract

In recent years, most RDD telephone surveys have introduced a cellphone frame to augment the landline sample to improve coverage and reduce variance. Due to a number of challenges it can be more expensive to complete cellphone interviews and often, for this reason, the proportion of the total sample allocated to the cell phone frame has been kept artificially small to reduce costs. As more people shift away from using landlines it becomes increasingly important to increase the allocation of RDD sample to the cellphone frame. We investigate the effectiveness of the Cell-Wins service provided by Marketing Systems Group which flags cellphone numbers that are actively being used by a person at the time the flag is applied. If accurate, this flag could reduce data collection costs by identifying nonworking numbers prior to data collection. This paper presents an assessment of this flag in the State of Ohio by which we (1) assess the under coverage rate if identified inactive telephone numbers were excluded from the sampling population and (2) determine the cost efficiency associated with excluding telephone numbers identified as inactive. We found that the accuracy of the Cell-Wins flag was high for telephone numbers identified as inactive and only introduced an under coverage rate of 2.4%. Furthermore, we determined that utilizing the Cell-Wins flag was cost efficient decreasing data collection costs a net 12%.