2 Way ANOVA or multiple t tests with correction?

#1
Hi all,

I am trying to do some data analysis, and I'm not sure whether to use a 2 way ANOVA or multiple t tests with Bonferroni correction.

I am looking at the effect of a chemical on bacterial binding to different polymers. So for each polymer, I want to know if there is a difference in binding between the treatment group and the control group.

I initially wanted to do a 2 way ANOVA with multiple comparisons because I have two independent variables (polymer and treatment group) and one dependent variable (bacterial binding). However, would multiple t tests with correction be more appropriate if I am only comparing each polymer +/- treatment, and not comparing the polymers to each other?

I really hope a 2 way ANOVA with multiple comparisons is ok for this data, because I have spent a LONG time making all my figures using the ANOVA data.... :D

Thanks in advance for you input!!!
 
Last edited:

noetsi

Fortran must die
#2
Why would you not compare the impact of the polymer on the DV? I can't see any advantage if that is an IV.

In general ANOVA is recommended to multiple t test, although I have not seen anyone do what you are.
 
#3
Thanks for your response!

I do want to look at the impact of the polymer, but I only want to compare each polymer to itself +/- treatment, not to the other polymers. So the experimental design is "for each polymer, is there a difference in binding +/- treatment?" I am not looking to see if the binding differs between polymers.
 

noetsi

Fortran must die
#4
It seems then that you are in practice running one predictor (treatment)within each "group" (the specific polymer). That is one regression for each polymer. In that case ANOVA and single t test should generate identical results - its your choice which you use.