Skip to main content
. 2017 Jan 12;5:e2885. doi: 10.7717/peerj.2885

Table 1. Type I error rates of the anova.traitglm test (site-based bootstrap approach based on the negative-binomial likelihood) and the max r/c test (pmax test based on the Poisson likelihood) for 1,000 simulated negative-binomial data sets using the Gaussian response model and n = m =30.

Value in bold represents the inflated error compared to the nominal level of 0.05.

Test procedure Scenario
Trait random (z and e) Environment random (t and x) Both random (z and x)
anova.traitglm 0.417 0.055 0.045
max r/c 0.047 0.055 0.019