Abstract:
|
The dual-frame RDD telephone surveys have been widely used to avoid coverage bias due to cell phone-only populations in many countries. But country-specific characteristics of the cell phone numbering system have led to a variety of drawbacks in conducting telephone surveys. For example, the area code of cell phone numbers in the U.S. has become increasingly unreliable for surveys at the state or local level due to the regional portability of the numbers. In another instance, cell phone numbers in South Korea or some other countries raise a serious concern about RDD sampling and estimation at both national level and sub-national levels because they do not involve area codes useful to select random or stratified samples based on geographies. In this paper, we first present cell phone RDD sample designs to overcome this problem in South Korea. Next, we present several weighting procedures, including post-stratification weighting using respondents' self-reported location and simple weighting to solve the overlap problem in the dual frame. Then, we show the state of data quality at national or provincial levels in the National Adult Tobacco Survey conducted for applying these methods.
|