If you have a frame or list of all possible respondents in that specific community, you may use a probabilistic sample and your inference would be over all the community. With a Convenience sampling you have not statistic evidences on the specific community, because it only produce inference over the sample. You may do a probabilistic sample of all the online respondents asking them complementary information. Then the target of your inference would be the people that responded to your website. This population may be the same that that of the specific community or not.
Related to Guillermo's point about the inference not being about the frame or list about about the people who responded, Andrew Gelman has a good blog discussion here (http://andrewgelman.com/2014/08/06/president-american-association-buggy-whip-manufacturers-takes-strong-stand-internal-combustion-engine-argues-called-automobile-little-grounding-theory/). Basically he was criticizing someone who claimed that there is statistical grounding in probability samples. Gelman points out that some of these may only have a 10% response rate, and therefore there really is not an established statistical literature that describes who to use these estimate.
Convenience sampling can be extremely valuable for hard-to-reach (although electronically connected) populations. Under certain assumptions, convenience samples can also be used for model-based inference
https://www.rand.org/content/dam/rand/pubs/monograph_reports/MR1480/MR1480.ch4.pdf
I would use stratified sampling with a question which divides the answers by country, for example. Stratified sampling improves precision theoretically and is a technique important for the statistical analysis.
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics