Hello fellow Researchers!
I used hierarchical regression analysis and the value of effect size (f square) is .18 of model 2 over model 1. What does it imply? Moderate or High?
Thank you all!
f square ≥ 0.02 represents small effect, f square ≥ 0.15 represents medium effect, and f square ≥ 0.35 represents large effect. So in your case .18 will be considered a medium effect.
Squared semi-partial correlation is also used to define effect size in hierarchical regression, it is analogous to eta square where .02, .06, and .14 represents small, medium and large effect sizes, respectively.
Hope this answers your question.
It really implies nothing.
Only subject matter knowledge allows you to interpret effect sizes.
Note that interpreting estimated effect sizes is additionly complicated by the fact that these estimates are subject to uncertainty. Ignoring this may be easy but it eventually gets you nowhere. If you really want to make any practical sense of effect size estimates, they must be sufficiently precise, and it is often extremely hard to get sufficiently precise estimates (it typically requires huge sample sizes).
Article A practical guide to calculating Cohen's f2, a measure of lo...
https://ncss-wpengine.netdna-ssl.com/wp-content/themes/ncss/pdf/Procedures/PASS/Multiple_Regression_using_Effect_Size.pdf
Thank you for sharing this important information with the link. It really helps.
Besides the issue Jochen Wilhelm raises, that you shouldn't think your research context is so unimportant that you can apply verbal labels to effect sizes without consider context (a point also addressed by many about Cohen's verbal phrases), there are additional issues for hierarchical/multilevel/mixed models. See https://www.bristol.ac.uk/media-library/sites/cmm/migrated/documents/variance-partitioning.pdf
Thanks Daniel Wright & Jochen Wilhelm for the critical inputs! Yes without context, effect size as such has no meaning. Thanks!
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication