What is the minimum sample size for a pilot study in a mix method research?
Start with around 20–30 participants for the quantitative part and 5–15 for the qualitative part.
Piloting what? Qualitative or quantitative? Both are very different, and the type of analysis, length of the instrument, and type of sample all play a part in determining the specifics. For example, piloting a questionnaire with 8 questions for a sample of 30 is very different than planning on exploratory factor analysis with 40 questions and a large, diverse sample.
Thanks David, i mean for Quantitative only. Length of the instrument is based on 110 items, to be asked from a known population.
110 items are a great quantity. Has the questionnaire been used before and found reliable & valid with other populations? If completely new, then you need to collect enough data to verify the questions do what they purport to do and make sense. Conducting more complex analysis, such as CFA or SEM would require a large sample. Again, if that is not your purpose, then you can a priori determine sample size based on your research questions. 10-40 are common, and you have to determine if you’re conducting cognitive interviews as well. The rules of thumb only make sense depending on your purpose; there can be many other factors.
Hertzog, M. A. (2008). Considerations in determining sample size for pilot studies. Research in nursing & health, 31(2), 180-191.
There is a difference between pilot tests and pre-tests. Pilot tests are usually rather small and assess things like question wording. Pre-tests are larger and assess whether things like scales function as planned.
David L Morgan Questionnaire went through content validation process where feedback was collected through subject/domain experts- i believe this is what you mean by pre tests?
I would consider that a pilot test. A pre-test examines how well a questionnaire works in the intended sample -- for example, whether a set of scale items has sufficient reliability.
People define pretests and pilot tests in multiple ways. To me, a pretest is a test of the questionnaire, and a pilot test is a test of the larger data collection process. Thus, to me a pilot test is larger. I wouldn't trust content experts to know how people will actually respond to a survey; you really need to test how the questions work in practice. For federal studies of the public, OMB approval is required for any data collection of more than 9 people, so pretests are often limited to 9 people. In general, that is enough to provide a lot of useful information. In my experience, I generally stop learning something new after the first 5 responses. But it does depend on what you want to test. Also, be sure to get diversity among those in the pretest. Sometimes different groups will approach the same question differently, and getting responses from 9 people who are all the same may not tell you much.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling