Hello,
It would be nice to add the eta square and omega square effect size in the GAMLj3 Custom Contrast Tests table.
Also, I was wondering if an additional improvement in this table would be to include the Welch t' test for heterogeneous variances (this is the default tests for comparing two means in the core package of R). In that situation, a new measure named Hedges' gs* effect size that was suggested by Delacre et al. seems to be an interesting choice:
https://thenewstatistics.com/itns/2021/ ... e-unequal/
Best regards,
Michael
Suggestion of improvement for GAMLj3: add effect sizes in the Custom Contrast Tests table
- mcfanda@gmail.com
- Posts: 530
- Joined: Thu Mar 23, 2017 9:24 pm
Re: Suggestion of improvement for GAMLj3: add effect sizes in the Custom Contrast Tests table
Eta square for contrasts can be done (in future versions). For the Welch's, there is already the possibility to select "robust standard errors", which work quite well for heterogeneous variance. For newer effect sizes, we wait to see if they are reliable and useful, then we implement them.
- mcfanda@gmail.com
- Posts: 530
- Joined: Thu Mar 23, 2017 9:24 pm
Re: Suggestion of improvement for GAMLj3: add effect sizes in the Custom Contrast Tests table
Version.3.4.0 (coming out tomorrow) computes variance effect size for custom contrasts.
Re: Suggestion of improvement for GAMLj3: add effect sizes in the Custom Contrast Tests table
This is great, thank you very much.