ANCOM Giving Different Results For Same Numbers

[Stephanieorch (Stephanie Orchanian)]

I'm running ANCOM comparing days in a time series to the initial day. I get that at day 6 lactobacillus is significantly different but day 7 it's not, even though the numbers are identical and the W score is extremely high. I've attached the ANCOM files for both timepoints. Why is it significant in one and not the other?

ancom_day_7.csv (6.7 KB)
day_7_percent-abundances.csv (9.0 KB)
day_6_percent-abundances.csv (8.3 KB)
ancom_day_6.csv (6.1 KB)

[mortonjt (Jamie Morton)]

Just followed up with @Stephanieorch offline. The consensus was that these statistics are relative. The threshold for chosing a W statistic is not fixed, but rather dependent on the particular dataset of interest.