Dear All,
As stated above, I want to build a multiple linear regression model based on 4 or 5 independent variable (4 continous and 1 categorical) to predict one dependent variable (continous). Since it is a new approach, I want to do a pilot study first.
I assume that I will use an a-priori power analysis with alpha = 0.05 and power of 95%.
How do I determine the effect size (f2)? Should it be large for the pilot study and medium for the following study? Or there are other approaches?
Hi Rizky Merdietio Boedi ,
the approach depends on your goal. If your pilot study does only want to test methodological aspects, the effect size does not need to change. If you plan to do statistical analyses on the data, you would need to have a very large effect size to reach statistical significance.
Here are two easy to read publications which will hopefully help you with sample size planning and determining the correct effect size:
Article The Art of Planning ‐ How many samples are enough?
Article Significant Significance?
Once you've figured the important points out you can fall back on one of the many free software options for sample size calculation.
Hope I could help!
Tom
Dear Rizky, the most important point is such studies is the avoiding of chance correlations due to the multiplicity of regressors, herewith enclosed a very important paper that explains the issue and I think you should take into consideration...
Hi, Rizky Merdietio Boedi, I recommend that you use G*Power to determine the required sample size. It provides a good estimate of the sample size for multiple regression.You can take a look at the following papers:
Giner-Sorolla, R., Aberson, C. L., Bostyn, D. H., Carpenter, T., Conrique, B. G., Lewis, N. A., & Soderberg, C. (2019). Power to detect what? Considerations for planning and evaluating sample size.
Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using G* Power 3.1: Tests for correlation and regression analyses. Behavior research methods, 41(4), 1149-1160.
"you would need to have a very large effect size to reach statistical significance"
I hope I don't get this wrong, but do you mean large effect size as f2 ≥ 0.35 (large) or f2≥ 0.02 (small)? As setting the f2 to 0.35 gives me a lower number of samples.
2. Thank you Alessandro Giuliani for the recommendation. Do you think that (based on the paper) overextending the sample size also reduces the calculations reliability?
3. Thank you Wang Zheng for the recommendation. I used g*power too to do the analyses. I am just unsure when to use a certain number (in this case f2) to justify the amount of samples I use in a pilot and following studies.
Dear Rizky probability and effect size are two separate problems you need a Great sample size to avoid chance correlation as in the Topliss paper I sent you
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size