Closed JohannesWiesner closed 9 months ago
Sorry about the delay @JohannesWiesner ...
It means that due to the correction method applied, your "*" which means between .01 and .05 is actually statistically not significant (ns).
You can see the correction_format
argument of Annotator.configure
for changing this rendering (for example to simply ns
with .configure(correction_format="replace")
. You'll just have to be mindful that the legend will not be accurate anymore (but you see the (ns) in the statistical test result).
But wouldn't it be better if the *
would just completely disappear by default if it doesn't survive the multiple comparisons correction?
Opened a discussion around this suggestion: https://github.com/trevismd/statannotations/discussions/137 (Trying to progressively clean up the issues vs howtos vs why-dont-we ;))
I ran multiple comparison corrections using Benjamini-Hochberg procedure. Some of my tests have a
* (ns)
symbol. Wonder what this means? It does also not appear in the table that is printed to console before running the tests: