Hi, Everyone I have a statistical issue.
I divided all my samples into type and age groups:
1. Young A;
2. Old A;
3. Young B;
4. Old B.
I need to compare A vs B under young and aged groups, separately. At the same time, I need to compare young vs old under A and B groups, respectively.
Based on my statistical knowledge, it seems that I have more than two groups so ANOVA should be my test. However, all my groups could be regarded as two independent and unconnected groups and in each group, I compare two sub-groups (because I never compare Old A vs Young B or Young A vs Old B). So Actually I should use independent t-test or Mann Whitney U test.
However, when I look into similar literature, they prefer ANOVA, which really confuses me.
The more important question: why did you divide age in the first place??
Dear Shan Lu,
Assuming your dependent variable is quantitative and the assumptions are reasonable, I think that you can find out more about your data using a 2 x 2 factorial ANOVA. Having age as one factor and A/B as the other and running a factorial ANOVA, you will have the opportunity to find out if there is an interaction between age and A/B as well as the main effects (old vs. young; A vs. B) if there is no interaction and the simple effects (these same comparisons under each level of the other) if there is an interaction.
Good luck with your study!
Jim McLean
ANOVA is the way to go.
I personally prefer using linear mixed effect (LME) model and estimated marginal means to contrast between group differences. It is difficult to adjust for covariates with ANOVA, but with LME you can. This is the paper that I share to my students: Article Categorical Data Analysis: Away from ANOVAs (transformation ...
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA