Fmax
While working on the final project, I violated the assumption of homogeneity. My cell sizes are discrepant by a ratio of 5:1. Since it is more discrepant than the 4:1 ratio, I know my Fmax value needs to be lower than 10 but since it is not quite to the 9:1 ratio, do I have to lower the critical value to less than or equal to 3? My Fmax value is 3.6 and I am thinking that ANOVA should be robust to this violation since my cell sizes are not necessarily close to the 9:1 ratio discrepancy, but I just wanted to check.
posted by Anonymous @ 8:57 AM
3 Comments:
At 1:40 PM,
T.J. said…
Let me make sure I understand: the official rule is to go to the 3 ratio with your cells, but since your cells are not really close to 9:1 ratio, than you are wondering if it's okay that you are a bit above the ratio?
It's a good question. It helps to remember that the Fmax ratios and cutoff points are somewhat arbitrary. (Why not 10:1 or 8:1?) That said, you wouldn't want to push them too far.
You should be fine, so long as you document what you did, and argue for why you can bend the general Fmax rules a bit here. Good thoughts.
At 2:17 PM,
Anonymous said…
Thanks! Now would my argument for this be in my "results" section of the project or under the section where we explain why we chose our specific analysis?
At 8:04 AM,
Mari Clements said…
Normally, this would be in a Preliminary Analyses subsection of the Results section, although it could arguably also go in the Method section.
As you are not writing a Method section, either the Results section or the explanation of analysis section would be OK.
Post a Comment
<< Home