I am a graduate student of Statistics in Korea.
I have a question so I can write.
I am studying Zeroinflated NB2 Modeling using R.
What I want is the overdispersion parameter Std.error of alpha.
In R packages (pscl)
Variance = mu + mu ^ 2 / alpha, but the form I want is
Variance = mu + mu ^ 2 * alpha type.
So when I look at the summary (zeroinfl (y ~ x | z)) through the pscl package, I get an alpha value, and if I take a reciprocal, I get the alpha value I want. But
The value of Std.error of alpha should not be taken as a reciprocal.
This is the problem.
One Std.error of alpha value can be obtained using the alpha value obtained earlier
However, the modeling I study is n = 200, and it has 200,000 data with group = 1000.
So, for every 200, one summary appears. That is,
I can get the alpha value, but I wonder how to get the value of Std.error (NB2 Ver.) In each summary.