categorical DV fixed effect analysis

[Bart Boersma]

hello,

I have dependent variable (self-assessed health) with a range of excellent (1) to poor (5). I wish to perform a fixed effect analysis (as I have panel data, that is not random), but none of the models seem to suffice. Does anyone know a good probit analysis for fixed effects?

kind regards
Bart

[Carlo Lazzaro]

Bart:
since you have an ordered dependent variable, you may want to take a look at -xtologit-, although it allows random effect only.
Conversely, you may want to consider a pooled -ologit- with statndard errors clustered on -panelid-.
Anyway, -probit- does not seemt to be the right tool for your data analysis.

[Bart Boersma]

Carlo:
thank you for your fast reply, I have quick follow up question: how do i Intepret the cut1 cut2 cut3 cut4 in my model.
In the example of rologit.pdf they have the example of cars and their repair states: values “Poor”, “Fair”, “Average”, “Good”, and “Excellent.
in this example they say that cut1 is poor, as poor=1 fair=2 and so on, does stata do this automatically, so in my case cut1 would be excellent (1) and no cut would be poor (5).

[Carlo Lazzaro]

Bart:
your feeling is correct.

[Bart Boersma]

thank you!