I am using this code (below) to use FDR to correct p values because I ran three exploratory growth curve models. How many p values do you need to correct? Do I just correct all of the p values for the one model in which I had significant findings, that significant path across the three models, or something else? Because of the kind of growth curve model I used, there would be 12 p values per model
p.vect
In False Discovery Rate (FDR) correction, you typically correct all p-values being tested simultaneously.
You are considering the p-values for the terms (coefficients, parameters) of the model ? Usually, people don't employ p-value correction for these. Although when I was in graduate school --- oh, some decades ago --- we were taught to make these corrections, for example in a factorial anova.
You might take a step back and consider why you would make this kind of correction, and if you do or don't want to make this kind of correction.
Sal Mangiafico a reviewer has asked to see corrected p-values since these were exploratory models
I'm confused about the number of p-values you have for three models and 12 p-values. ... But I would say, adjust the p-values within each model. ... There aren't really rules about this. In reality, p-value adjustment is at the judgement of the analyst. But if the reviewer requires it...
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs