search

statdivlab / corncob

Count Regression for Correlated Observations with the Beta-binomial
103 stars 22 forks source link

N/A P-Values #138

Closed Desorcy91 closed 2 years ago

Desorcy91 commented 2 years ago

Hello,

When using code :

"Week1corncob$p_fdr[1:23]"

I get several adjusted p-values but also several N/A's. Does this mean that they aren't p<.05 after adjustment? "Week1corncob" is my object. If anyone knows I'd appreciate it.

Thanks,

Katelyn

adw96 commented 2 years ago

Hi Katelyn!

I get several adjusted p-values but also several N/A's. Does this mean that they aren't p<.05 after adjustment?

No, that's not what this means. You will notice that Week1corncob$p[1:23] has NA's in the same place.

Taxa with NA's are "discriminant taxa" -- taxa that are observed in one group only. Unfortunately we can't run a hypothesis test for these taxa, but I certainly wouldn't discard them, because the are clearly strongly associated with one group.

I hope that helps!