I have collected data from 80 subjects. They were divided in 5 groups and performed a rating task of audio files (once as a baseline - same for all groups, second time in experimental condition - different for each group, one group was a control group- kept on rating in the same condition as the baseline).
I have then fitted a lmm model using lmer() from lmerTest (which is using lme4 but also providing the p-values for anova). My fixed effects are: Group (5 levels), Condition (2 levels), GenderS (gender of the speaker), GenderP (gender of the participant) and the interaction between all of those. Participant and Item (audio file) were added as random effects.
Each participant rated 100 audio files (40 in the baseline condition and 60 in the experimental condition) giving a total of 8000 data points (1600 in each group).
I have then run anova() (and Anova() from car package) on the model and got significant Group:Condition interaction (p
A few quick suggestions:
Post hoc is focused/more powerful.
ANOVA is omnibus/global.
Hope this helps,
Matt
Thank you for your suggestions. I actually did try emmeans as well (I though that these two are now actually connected). The thing is I originally wanted to look only at the group:codition interaction but then (from looking at the data) I realized that both, gender of the speaker and participant, may be important.
I though that looking at this step-wise - see that there is no difference if we just consider "group:condition" and look further would be the easiest way to go but now I am stuck. I think what Matt says makes sense too but I am not sure what would be a good way to conclude here.
I have found these two articles very useful and I will leave these here since it may help someone else. To be honest I am still not sure how to approach this problem best when writing but your comments helped me a lot too.
Article Evaluating significance in linear mixed-effects models in R
Article Relationship between Omnibus and Post-hoc Tests: An Investig...
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA