Can I run a xtologit on a 0.5-paced score as dependent variable?

[Ilka Weichert]

Hi everyone,

I'm working on panel data with firm ID as panelvar and report_period as timevar.

I've tried to find out how to deal with my dependent variable but didn't find anything yet. It is a 0.5-paced score ranging from 0-10 and I wonder what type of regression analysis is suitable for it.
Can I treat it as an ordinal variable and run xtologit right away or do I have to adjust the variable by making the score consist of integer values only?

Thanks a lot for your help!

Best,
Ilka

[Richard Williams]

I don't think it has to be integer coded. But the easiest way to find out is to try it.

It sounds like your variable has 20 possible values? 0, 0.5, 1, 1.5, etc. If those intervals can be regarded as equally spaced then ols regression may be fine.