2
$\begingroup$

I have a zero-inflated probit model but I know the inflation probability of each observation. So my model is:

$P(y_i = 0|X) = (1-\pi_i) + \pi_i*(1-\Phi(X_i'\beta))$

$P(y_i = 1|X) = \pi_i*\Phi(X_i'\beta)$

and I know what $\pi_i$ is for each observation. Is there a way to estimate the model in R with zeroinfl() or glm()?

Improve this question
$\endgroup$
1
  • $\begingroup$ Are you saying π is a known constant, or π is a known random effect, or π is a known categorical variable (i.e., you have n groups and π is known for each group)? Regardless, I imagine it could be coded-up with optim. I do not know about with glm or zeroinfl(). $\endgroup$
    – Mark Miller
    Commented Oct 18, 2022 at 11:10

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.