Did many regression analysis in the past, but zero-inflated poisson regression taking my too much time! Looking ways to understand to interpret the result obtained from STATA.
What you get from a ZIP are IRRs. Try to interpret an IRR. Have you seen the following page? Also Joseph M. Hilbe had an addendum to his book Negative Binomial Regression 2007 explaining this, but I could not find it online now maybe due to his loss in March 12.
http://stats.idre.ucla.edu/stata/output/zero-inflated-poisson-regression/
thank you. the link is great. It works for me.
In addition, I take an opportunity to ask you more- the dependent variable has only three data options, 0, 1, and 2 = the count of the chronic diseases/individual. I did Poison, but which is not a best fit model in which 53% is zero. Should I go for zero-inflated or Hurdle? please advice me.
Yes try those. A negative binomial model could also help a lot.
Additionally you have to check weather there is a random/cluster effect or no
Why not a ordinal ordinal logistic regression ? You only have only three counts.
I recommended you to use either of the Count data regression model based on the comparison criteria like AIC, BIC & LR measures.
More, I invite you to read this pdf (link: https://www.diva-portal.org/smash/get/diva2:729147/FULLTEXT01.pdf%20[2020-04-16].
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression