My regression model has binary variables for the dependent variable, so I have to use logit regression approach for the analysis. This means that my model is nonlinear, the question is after running the logit model, I checked for autocorrelation and heteroskedasticity problem. some of my model suffer from one or both of these problem. So, xtlogit model (my data is panel) is applied to see whether fixed or random is more efficient. Now, how to tackle the abovementioned problems when Hausman test result decided random or fixed I have to apply? when I apply this command in Stata :
xtlogit D.V IDV, re vce (robust)
This command doesn't work nor with cluster option. Only for PA, Population Average which produces semirobust for standard error in regression but I need fixed and random estimation not PA
So, any advice for this issue please and a great thanks for you in advance.
In Google searching, nothing clear in this regard
Actually logistic regression models are linear because they are linear in the regression coefficients. The tests and fixups are the same as for any linear model.
See: https://b-ok.cc/book/999818/40e7c5
Now tell us about the panel part i.e. your data and your models. Best, David Booth
a lot of thanks to you David Eugene Booth for your reply. I attached a file described the issue that I face. please check it and give me your solution.
Hi. See if this book doesn't help you a bit. Best, D. Booth
https://b-ok.cc/book/3588849/f27729
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression