Message-ID: <CAPxEaERx_yNqkM8nznVFOGqVTSGNfD+0FvVswG3J+0CbSOC0fg@mail.gmail.com>
Date: 2020-11-05T17:48:39Z
From: Gian Maria Niccolò Benucci
Subject: interaction effect in capscale, how to calculate
Hello sig-r-eco community,
I would love to have some insights on my interpretation of capscale() and
how to assess the variance explained by the interaction effect.
Imagine a model like this: Var1 + Var2 + Var1:Var2
> RsquareAdj(capscale(otu_table ~ Var1 + Var2 + Var1:Var2, metadata,
distance = "bray"))$adj.r.squared
[1] 0.281792
Then I can obtain the variance of the main factors
> RsquareAdj(capscale(otu_table ~ Var1 , metadata, distance =
"bray"))$adj.r.squared
[1] 0.1270805
> RsquareAdj(capscale(otu_table ~ Var2, metadata, distance =
"bray"))$adj.r.squared
[1] 0.09308548
Then, is this the right way to calculate the Adj.R2 for the interaction?
> RsquareAdj(capscale(otu_table~ Var1:Var2 + Condition(Var1) +
Condition(Var2), metadata, distance = "bray"))$adj.r.squared
[1] 0.05174793
However, if I sum the variances altogether I do not get the variance
explained by the full model
0.09308548 + 0.1270805 + 0.05174793 = 0.2719139
I looked online but I could not find any decent explanation of this.
Thank you for your help,
--
Nico
*----- Do not print this email unless you really need to. *
*Save paper and protect the environment! -----*
[[alternative HTML version deleted]]