Interpretation of betta coefficient table significance

[cyklee]

Hi Amy,

Firstly, quick easy fix to report. In the vignette diversity-hypothesis-testing.Rmd

The values 171 and 0.475 were hardcoded and do not match the output above.

My main question is related to the vignette section, where it said: "A hypothesis test for a change in richness would not be rejected at any reasonable cut-off (p = > 0.05, essentially)" - could you elaborate on what that means? What is the null hypothesis here? I've read the betta function vignette, DivNet/issues/34, #66 (both of which explained interpreting the estimates excellently), but not much was said about the coefficient table p-values.

Thank you for your help and for breakaway and DivNet, both of which have grown so much since I last checked it out!

Kevin

[ailurophilia]

Hi Kevin,

Thanks for bringing this to our attention! The null hypothesis we are testing is a null of no difference in mean diversity comparing soil samples at day 0 and day 82 (so equivalently, a null that mean soil diversity at day 82 minus mean soil diversity at day 0 is equal to zero).

Best, David

[cyklee]

Hi David,

Thank you for your reply. That makes sense. Just to be clear, would the p-value for that (i.e. significant difference in means between the two days) be represented by the second row (i.e. 0.412)? If so, what does the p-value of the intercept (Day 0, 0.000 here) represent? I am asking because apparently every group in my comparison is significant except the intercept 🙃.

Best wishes, Kevin

[ailurophilia]

Hi Kevin,

The intercept in this model is the mean on day 0, so the p-value in the first row is for a test of a null hypothesis that mean diversity on day 0 is 0. In general, the p-values in each row of $table in betta output are test the null hypothesis that the parameter whose estimate is summarized in that row is equal to zero. Exactly what interpretation these p-values have depend on how you have parametrized your model.

All best, David

[cyklee]

Hi David,

Thank you, I think that thoroughly explained how to interpret the result! I hope others will find this conversation useful as well. I will close this issue now.

Many thanks, Kevin