|
Abstract:
|
RDD telephone sampling methods have depended on a high coverage of the general population for their use in the survey research industry. While there has always been some level of non-coverage in the sampling frame, the increase of the percentage of the population who exclusive, or nearly exclusively, use cell phones instead of landlines has decreased the coverage of traditional RDD methods to below 80%. While cell phone samples are available, they are not cell phone exclusive or cell phone dominant in nature. Therefore, in order to get a representative sample of the population of interest, the options are costly. This paper proposes a methodology based on propensity score matching that identifies proxies in the traditional RDD sample and weights them accordingly to produce a representative sample at nearly the same cost structure as before.
|
