Shouldn't the number of observations that gets reported by logLik(model) or gets used in AICc(model) or BIC(model) be the product of the number of rows and the number of columns of model$y?
Currently, the nobs function in aicc.gllvm.R returns this product but other functions including inf.criteria in gllvm.auxiliary.R and logLik in logLik.gllvm.R only use the number of rows.
Indeed. I have corrected this some time back when I implemented AICc, yet forgot to address the same issue in BIC and logLik, apparently. Thanks for noting this, it has now been addressed in my PR #51.
Shouldn't the number of observations that gets reported by
logLik(model)
or gets used inAICc(model)
orBIC(model)
be the product of the number of rows and the number of columns ofmodel$y
?Currently, the
nobs
function in aicc.gllvm.R returns this product but other functions includinginf.criteria
in gllvm.auxiliary.R andlogLik
in logLik.gllvm.R only use the number of rows.