Computing predicted values and confidence intervals from ologit model

[Bruno Hoepers]

Hi all,

I have a dataset with information on employees from a firm. The dependent variable is “Prom”, or the promotion to different managerial positions inside the firm. It’s an ordinal variable and it assumes four values: 0 (no promotion), 1 (promotion to position A), 2 (promotion to position B), 3 (promotion to position C).

The model follows below:


ologit Prom revClass_reversed mba masters phd foreign_language, vce(robust)


The independent variable of interest is “revClass_reversed”, or their ranking inside the firm based on performance. It’s reverse-coded so that higher values mean a higher ranking (from 0 to 119).

The goal: to compute predicted values and confidence intervals for the four promotion categories.

I’m having problems doing it. Could someone help?

Thank you.

[Fei Wang]

After running ologit, you may run margins for predicted probabilities of each category, and run margins, dydx(revClass_reversed) for the partial effect of revClass_reversed on the probability of each category.

[Bruno Hoepers]

Thank you Fei.