Hey everyone,
a
I am trying to wrap my head around the difference between a regular orderd logit and a fixed or random effect one. I vaguely understand the difference between a fixed and random effect ordered logit, but i do not understand how a regular ordered logit is different from the two. In addition to this is it statistically correct to include both the results from a normal, fixed and random effect logit in an assignment or if you use one of them the others should not be used. So to summarize to me fixed effects only takes into account variable that vary over time and removes time-invariant characteristics and random effects are used when you think the differences between entities have some influence on your dependent variable but the variation is random and not correlated with the predictor and error term. How do i test whether i can use a random effect model when using logit? For normal regression a hausman test can be done but this cannot be computed for logit regressions. I am using panel data, with an ordinal dependent variable.

Thank you for taking the time to read this!

Kind regards,

Max Verheijen