EViews.com

Hello!

I need some help please!
My question is, how do I estimate a multinomial logit regression in eviews? I had a look at the estimate equation function, but it only offers binary logit.
Also, are there any additional considerations ( things to note) if one is running multinomial logits on panel/pool data?



Thank you very very much

Ali

Hello!

I need some help please!
My question is, how do I estimate a multinomial logit regression in eviews? I had a look at the estimate equation function, but it only offers binary logit.
Also, are there any additional considerations ( things to note) if one is running multinomial logits on panel/pool data?



Thank you very very much

Ali

To get started, you might search the help files for "multinomial"

There is an example program for using the LogL object to estimate a multinomial logit (Help/Quick Help Reference/Sample Programs & Data/Logl).

As to the panel issue, it really depends on what kinds of structure you want to put on your problem. The sample program above is non-panel aware, and depending on what you wish to assume, there may be changes that you need to make.

many thanks¬

There is an example program for using the LogL object to estimate a multinomial logit (Help/Quick Help Reference/Sample Programs & Data/Logl).

As to the panel issue, it really depends on what kinds of structure you want to put on your problem. The sample program above is non-panel aware, and depending on what you wish to assume, there may be changes that you need to make.

Hi Glenn,

I have a multinomial categorical dependent variable model I want to run for cross sectional data.

I used logistic distribution, uncensored and no truncation under the "censored - censored or truncated data (including tobit)" method option.

Could this be another way of estimating a mutinomial logit?

No, it is not.

No, it is not.

If my dependent variable has values of 0,1,2 and 3 and I use logistic regression under the censored methods but I use left censor 0 and no right censor. No truncation.

Am I converting the dependent variable in 0,1 where 1 can be 1,2 and 3?

Thank you Glenn

The censored regression tools assume that the non-censored responses are continuously observed and that the censoring portions are due to the latent hitting threshold values.