0
$\begingroup$

I need to estimate the sample size for a new study with a between-within subject design (4 groups and 2 within factors). In order to do that, I want to perform a power analysis in g-power. I can infer the "possible" effect size of my study from the effect size found in two similar studies already published (I don't have any pilot study do to that unfortunately). The problem is that one study has a within subject design while the other one is a between subject study with two groups. If I am not wrong, the variance, and so the effect size, is different in within vs. between subject design. What can I do to assume a meaningful effect size for my design?

Any idea or suggestion would be really helpful.

Thank you

Claudia

$\endgroup$
  • $\begingroup$ you might want to check out Morris & DeShon (2002). Combining Effect Size Estimates in Meta-Analysis With Repeated Measures and Independent-Groups Designs. Psychological Methods, 7, 105-125. $\endgroup$ – TPM Jan 23 '18 at 21:00
  • $\begingroup$ If you have a linear model, the between/within subject effect is the same. For logistic regression, you can't project the conditional OR, but you can place a lower bound on it by the marginal OR. After that, other repeated measure power calculations apply. You need to know the ICC to get at the intracluster variance. Otherwise you must make some assumptions. Are you using a GEE or mixed model? $\endgroup$ – AdamO Jan 23 '18 at 21:43
  • $\begingroup$ I was trying to use g power for mixed model ANOVA to compute the sample estimation. The study from which I took the effect size used the same type of analysis. $\endgroup$ – claudia Jan 23 '18 at 22:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.