Description
Interrogating a customer, member or address list before sampling so coverage error is known, and fixing a rule for who in a household or firm should answer.
Coverage error is often the largest error in self-administered surveys, yet the list is usually accepted as given. Dillman's procedure asks five questions of any frame: does it contain everyone in the population, does it contain people who are not, how and when is it updated, does it contain duplicates relative to the sampling unit, and what other fields does it hold that could support follow-up or nonresponse analysis. Each question is grounded in a real failure: a health-plan list missing members who opted out, an alumni list padded with donors and spouses, children of separated parents listed twice. The second half settles who answers within a selected unit, recommending the most-recent-birthday instruction or naming the person most responsible for the topic, since Kish tables are too complex for mail. The rigour lies in deciding, before fieldwork, whether residual coverage error is acceptable or the survey should be redesigned.
Attributes
- Discipline
- Research Design & Rigor
- Loss type
- Siloed
- Stage
- Discovery, Definition
- Data
- Analytical / expert
- Effort
- Days
Steps
- Find the list keeper and learn how, from what sources and for what purpose the list was compiled.
- Check whether every member of the target population is on the list; obtain missing units or estimate the coverage consequence.
- Identify units outside the population and plan to delete or screen them out.
- Learn how and when the list is updated, and test a sample of contact details if in doubt.
- Find exact and near duplicates relative to the sampling unit and correct selection probabilities.
- Note extra fields (tenure, plan, order history, alternative contacts) for stratification, follow-up and nonresponse analysis.
- Fix a within-unit selection rule (most recent birthday, or the person most responsible for the decision).
- Judge whether remaining coverage error is acceptable, or redesign or drop the survey.
Sources
- Mail and Internet Surveys (Dillman, 2007)
- Quantitative Social Research Methods (Singh, 2007)
Also known as
- Sample list evaluation
- Frame coverage assessment
- Most recent birthday method
- Respondent selection
Categories
- Research Operations
- Research Design & Rigor