I am running statistics on PCR data, specifically delta CT values. I have 2 forms at 3 doses each as my independent variables (Form A and Form B @ 2, 10, and 20 mg/kg diets each). I am not EXPECTING an interaction effect at all. My primary goal is to look at the form effect between groups (i.e. Form A - 2 mg/kg vs. Form B - 2 mg/kg) and dose effect within groups (i.e. Form A - 2 mg/kg vs. 10 mg/kg vs. 20 mg/kg). Therefore, my main effects are "Form" and "Dose".
With everything said, after running a two-way ANOVA for a few of my genes, most of them do NOT show an interaction effect (p > 0.05). I am not EXPECTING an interaction effect at all, considering my two forms are similar metabolites. HOWEVER, I am getting significant main effects, whether it is one of them or both of them (p
You can do post-hoc tests on the main effects. e.g., for the main effect of dose, you can compare mean(2A,2B) to mean(10A,10B) or to mean(20A,20B), etc., or you can do other sorts of custom contrasts as long as they're justified by the significant main effect.
Thank you Mathew and Stephen for your responses. I really appreciate both your advice.
Mathew, I completely forgot about the Bonferoni correction and will definitely take that into consideration in my thesis.
Stephen, thank you for answering my concern. This is exactly what I wanted to look for. I obviously wouldn't have to do a post-hoc for folate form considering there's only two forms correct?
Thank you in advance gentlemen,
David
Right, if there's a significant main effect for a two-level factor then all you need to do to know the nature of the effect is look at the means.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Modeling
- Analyses of Variance