search

jamovi / jamovi

jamovi - open software to bridge the gap between researcher and statistician
https://www.jamovi.org
Other
562 stars 61 forks source link

Omega squared (ω2) is wrong. #504

Open Freeopenstat opened 6 years ago

Freeopenstat commented 6 years ago

Hi, When I do the exact same analysis, a repeated measures ANOVA in Jamovi and in JASP I get the exact same results in everything except for ω2. Here it is a big difference. Either Jamovi or JASP is wrong, or both.

Thanks for a great program! JAMOVI Omega 2.pdf

raviselker commented 6 years ago

So it looks like we are using the regular calculation of omega squared, while JASP is doing a generalized version of omega squared. I think we should probably use the generalized version as well, because the regular one is not completely appropriate for repeated measures designs. I'll fix this quickly and the change will be added to the next release. Thanks for reporting!

jonathon-love commented 6 years ago

where we at with this @raviselker ?

jonathon-love commented 6 years ago

@raviselker?

raviselker commented 6 years ago

If you can find me a good reference of how it's calculated for RM and mixed ANOVA's I'll implement it. I found it difficult to find consistent sources of information.

ericcfields commented 5 years ago

@raviselker @jonathon-love

I'm not sure about calculation of generalized omega squared, but I believe at least omega squared and partial omega squared is relatively straightforward for repeated measures designs:

Kirk, R. E. (2013). Experimental Design (4th ed.). Thousands Oaks, CA: SAGE Publications.

Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: A practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4(863).

It would be good to add this for repeated measures, because currently the only option is eta squared, which is known to be biased. However, one issue may be that I don't think omega squared is calculated the same way for fixed and random factors, both of which would be possible in the RM ANOVA menu.