An overview of RDD sampling and available options
Realistically, the term “RDD sample” is used more generically than one might like. Academics immediately focus on a pure probability or EPSEM approach, but there are still options to consider and each company or survey research center has its own preferences.
At the other end of the spectrum, some people even attach the term RDD to listed samples, forgetting the basic premise that the RD portion of RDD stands for “random digit.” Perhaps a better term for those samples might be to forget the RDD label and simply call them “randomly selected listed samples.”
Each study has a specific purpose and intended use, a carefully designed target respondent, a degree of desired statistical accuracy, a timeline which largely determines how and when the sample will be administered, and – certainly from an entrepreneurial viewpoint somewhere in the work flow matrix – a potential profit or loss opportunity.
The trade-offs to consider are myriad and may substantially affect the cost of both the sample and the collection of the data, and the degree to which the sample is truly random. For instance:
- Do you want to include or exclude dedicated cell phone blocks or prefixes?
- Because cell phones must be dialed “by hand” or risk substantial TCPA penalties, does the research design include or exclude mixed-mode working blocks?
- The “industry standard” typically specifies that valid working blocks are excluded from the sampling frame unless there are at least three known households in that block. What level of inclusion do you wish to specify?
- In general, how do you want to define the geographic area to be covered? ZIP? County? State? Census division? DMA®? CBSA? Etc.
- What types of intentional exclusions from the sampling frame are desired, if any? Low income areas? Retirement communities? High Hispanic zips? Urban, suburban or rural areas?
- And, so forth.
Click here for a full list of options for RDD samples.
|
