Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
Error when generating marginal effects from brms model of the form y ~ 0 + intercept + x #113
I am trying to obtain marginal effects using
Here’s a quick example (with mostly unchanged defaults) using the
It seems the error has to do with how the internally generated variable
Actually, it is an oddity in the names assigned by
# In the code: nm = eval(parse(text = "brms:::rename(colnames(X))")) V = vcov(object)[nm, nm, drop = FALSE] # what we get: Browse> nm  "intercept" "tensionL" "tensionM" "tensionH" Browse> vcov(object) intercept tensionM tensionH intercept 7.954110 -7.931893 -7.787033 tensionM -7.931893 15.631325 7.807514 tensionH -7.787033 7.807514 15.885689
So you see that I ask the package to tell me the names and to return the variance-covariance matrix, but the package is inconsistent with itself. That's kinda annoying.
I think I should probably just use the empirical cavariance of the posterior sample anyway, since we realkly don't need this result at all except in a frequestist summary.
I am traveling as I write and will not be able to do anything more with this until late June 2019.