OLS vs. Ordered Logistic Regression

[Dave Baik]

Hello,

I have five employee-level variables as below, and I was wondering if I could use the OLS with fixed effect or ordered logistic regression...

Z1 = employee wage

Z2 = employee tenure

Z3 = employee age

Y = employee perception of fairness (ordinal survey item with 5 = very fair... 1 = very unfair)

X = employee job satisfaction (ordinal survey item with 5 = very satisfied...1 = very unsatisfied).

My understanding is that with the two ordinal independent and dependent variables, I perhaps need to use the ordered logistic model. However, I was wondering if using OLS would be incorrect? Furthermore, would it be acceptable to incorporate firm-level fixed effect in the ordered logistic model?

Does STATA allow fixed effects with an ordered logistic model?

Thank you!

[Carlo Lazzaro]

Dave:
1) go -ologit-;
2) check whether -i.whatever- plugged in the right-hand side of your regression equation does the trick.
What above assumes that you are dealing with a cross-sectional dataset.
If you have panel data, see -xtologit- (that allows random effects only, though).

[Dave Baik]

Thank you Carlo!