In a questionnaire study or survey study, how do we avoid bias of selection of a particular group of respondents, so that bias does not affect results of the study?
The obvious possibility would simply be to identify the bias and exclude those respondents/responses from the outset. Alternatively, if your question is something that would likely be better addressed via an experimental/q-experimental design, you could use a counterfactual research design from the survey data, such as propensity score matching. This way, you could tease out the selection bias by comparing only respondents who are similarly situated. In any case, you are not ever required to use all of the available data, but large samples are a luxury in many cases.
If you're talking more about collecting original data, you could try to impose quota sampling to offset the bias; or one of another sampling strategies to offset the prospect for selection bias.
Hope this helps.
Selection bias is avoided by using random sequence generation.
This depends on the project that you are on. Sampling strategies are sometimes quite different. This will rely on the type of respondents that you wish them to participate in your questionnaire. The simplest way is the randomozed saturation sampling for those eho fit in your inclusion criteria which you planned earlier.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire