Pilot study plays a major role in research. What is the appropriate size for a pilot study?How should one determine the sample size for a pilot study especially when the actual sample size is more than 350?
I would be very grateful if you could help.
The question you have to ask: "What is the purpose of the pilot study?" If you want to have an initial estimation for an effect size for example, pilot studies are not such a good idea in the first place. This is because pilot studies have small sample sizes and small sample sizes yield very unstable estimations for effect sizes like r (Schönbrodt & Perugini, 2013). Therefore, it may be the case that your estimation for an ES is much too high and you decide to have a smaller sample size for your main study and will miss the true effect, since the main study is underpowered for the true ES. Or, your ES estimation is too small (or even in the opposite direction...), which will lead to very large sample sizes for the main study (which will cost lots of ressources) or even that a main study will not be conducted, since the ES was in the wrong direction.
So, in my opinion, pilot studies make only sense to test the whole setting, if the experiment and apparatus work fine and to get qualitative feedback from the participants, to be able to adjust the setting, if necessary. For this purpose your sample size does not need to be very huge.
How do you know that your main study will have 350 participants? How did you determine it?
Please look at the following below links which may help you in your analysis:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4876429/pdf/10.1177_0962280215588241.pdf
https://ncss-wpengine.netdna-ssl.com/wp-content/themes/ncss/pdf/Procedures/PASS/Pilot_Study_Sample_Size_Rules_of_Thumb.pdf
Thanks
Evans -
Rainer starts and ends with very good questions:
1) "What is the purpose of the pilot study?"
and
2) "How do you know that your main study will have 350 participants? How did you determine it?"
In between Rainer notes that using a pilot study to determine a sample size may easily under- or overestimate your needs, and of course that can vary by variable, say questions on a survey. However, in W.G. Cochran, Sampling Techniques, 3rd ed, 1977, a classic, a pilot study is what he suggests if you have no better knowledge of your population. An adequate sample size depends on the methodology, and what sample size it would take with given standard deviation to obtain a desired standard error. (The situation under a model-based approach - prediction - is analogous, but perhaps more complex. Also, power considerations under hypothesis testing also depend on standard deviation.)
But you seem to already know what sample size you need with your data and methodology to obtain the accuracy you desire. How is that? If you used an online calculator, beware that that is just looking at one "question," and those calculators generally assume simple random sampling with no finite population correction (fpc) factor, that your question is yes/no (for proportions), and without knowledge of the standard deviation, it assumes the worst case (not even possible with continuous data). Generally, even if you are looking at proportions, this will overestimate sample size needs.
A pilot study might also be used to test questions for clarity, or to test the logistics of data collection.
So, there are various things you may be doing, or wanting to do, which are not clear here.
Cheers - Jim
It is actually a survey of auditors but would want to conduct a pilot study to check the reliability of my research instrument.
You may read the attached document. It will be of help to you.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size