The Fixed Effects Ordered Logit Model with Cross-Sectional Data

[simon]

Hi Everyone,

My question is whether I can run fixed effects ordered logit model with cross-sectional data.

I'm working on an immigration issue. I have two time points, 2003 and 2013. My dependent variable is how many immigrants should we let in a country with 5 categorical choices: Increased a lot, Increased a little, Remain the same as it is, Reduced a little, Reduced a lot. And I would like to see "industry" and "location" fixed effects.

My questions are

1) whether I can run fixed effect ordered logit model with cross-sectional data;

2) if I can, would it be an appropriate approach to create a sereis of dummy variables of year, indstry, and location.

Many thanks in advance!

Simon

[Carlo Lazzaro]

Simon:
welcome to the list.
Since you have panel data, you should consider -xtlogit-, that allows -re- specification only (please note that when Stata does not allow a seemingly straightforward statistical procedure, has wise background reasons to behave that way).
Over and above your query, there's no need to bother yourself with dummies creation, when -fvvarlist- can do it for you.

[Richard Williams]

If you feel comfortable with dichotomizing your dependent variable you could run an xtlogit, fe model.

[simon]

Many thanks for your kind answer/ suggestion.

Would you mind if I ask one more question?

Unfortunately, my data is not panel data, but repeated cross-sectional data. Would it be still valid to run "xtlogit" model?

Kind regards,

Smion

[Carlo Lazzaro]

Simon:
what do you mean by repeated cross-sectional data?
The sample from which the data waves were collected is not composed of the same units? Or what else?

[simon]

Carlos:

Precisely! The sample was collected annually by three-stage probability sampling method, but a unit, (here, an interviewed individual) was not same every year..

Kind regards,

Simon

[Carlo Lazzaro]

Simon:
so you have a survey data structure.
I would take a look at -svy- prefix, that allows -ologit- (please, see -help svy_estimation-).