logistic regression with multigroup nominal and ordinal independent variables

[Magnus Ploug]

Hi Stata experts

I am doing multiple regression i Stata followed by logistic regression. My dep.var is binary. Among my variables are both ordinal and nominal variables (called O and N). When the results of the logistic regression is showed it looks something like

mi estimate,or: logit dep.var i.O i.N

ordinal variable subgroup (O)
- 2 _____OR 0.5
- 3 _____OR 0.3
- 4 _____OR 0.3

nominal variable subgroup (N)
- 2 _____OR 1.2
- 3 _____OR 1.7

Is it correct to interpret this OR as beeing between group 2 and 1 and between 3 and 1 (and 4 and 1) for each variable?
What if I would like the result between group 3 and 2 (4 and 2 ect. )displayed for each variable? Can that be done

Best regards
Magnus

[Weiwen Ng]

Magnus,

Your interpretation is correct. Stata chooses the lowest coded value as the reference category by default. You can alter that:

mi estimate,or: logit dep.var ib2.O ib2.N

[Magnus Ploug]

Thank you very much! That works perfect for my needs.

I will try to use the functions better next time.

However when trying the dataex I get the error code

logit jernstatus i.Gender ib3.ASA

. dataex
input statement exceeds linesize limit. Try specifying fewer variables

But I will try to work it out for my next posting
Thanks again!

Magnus