search

jasp-stats / jasp-issues

This repository is solely meant for reporting of bugs, feature requests and other issues in JASP.
59 stars 29 forks source link

[Feature Request]: effect sizes for post hoc tests in RM ANOVA factors #1915

Open KilianSander opened 2 years ago

KilianSander commented 2 years ago

Description

As far as I can see JASP's ANOVA module uses the emmeans::eff_size to calculate Cohen's d in post hoc tests. For within-subjects factors this differs from Cohen's dZ which is used in a paired samples t test power analysis. Besides dZ Lakens (2013) mentions some other effect sizes for paired samples, e.g., dav or bias-adjusted Hedges gav. It would be nice to have a choice which effect size is reported for post hoc tests for within-subjects factors.

Purpose

Different effect sizes allow for better comparability between different designs. Power analyses and meta analyses become easier when effect sizes do not need to be converted.

Use-case

No response

Is your feature request related to a problem?

No response

Is your feature request related to a JASP module?

ANOVA

Describe the solution you would like

Add check boxes for effect sizes to report, e.g. "report Cohen's dZ for within-subjects factors", "report Hedges gav", ...

Describe alternatives that you have considered

To get Cohen's d_Z you can run paired t tests in JASP but I suspect this messess up the CI of the resulting effect size as the df in post hoc tests are taken from the ANOVA model.

Additional context

References Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: A practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4, Article 863. https://doi.org/10.3389/fpsyg.2013.00863

tomtomme commented 9 months ago

@KilianSander Sorry, this is still not available in JASP 0.19 beta. But it will not be forgotten. Is there an R-package that implements this, maybe even an update for emmeans since?

KilianSander commented 9 months ago

Thanks for your reply. I'm not aware of an option in emmeans to get a specific effect size. So far, I have calculated the various effect sizes with my own R functions, but effectsize::repeated_measures_d() looks promising.

KilianSander commented 9 months ago

I just realized that repeated_measures_d() is not part of the latest CRAN release (currently 0.8.6), but it is in the development version (0.8.6.6, https://github.com/easystats/effectsize).

tomtomme commented 8 months ago

Same request of providing more effect size options for RM tTest here: https://github.com/jasp-stats/jasp-issues/issues/38

TarandeepKang commented 7 months ago

I would like to second this request and point out that the new feature has now been added to the CRAN version https://easystats.github.io/effectsize/reference/repeated_measures_d.html