Using logistic or other method?

[Andre Thunberg]

Hi!
I use STATA 17. I have a dataset with 826 observations and hundreds of variables.
My outcome is completing referral, which is defined as yes=1, no=0
I want to know association between various variables; binary like "presence of danger sign" yes/no, continuous like "distance" (km), and categorical like way of transportation (car/bus/walking/taxi etc), age group or wealth quintile.

I tried like this:
logistic complete_yn transport_yn i.assetcat_final5 r_severe i.refertransport1 after12 distance i.age_grp

First, I have a question why not all the answers to i.refertransport1 was shown. There are six possible answers, but only 6 was shown. Same for assetcat_final5 which is wealth quintile and should have 5 answers, but only show 4.

Second, is there a better way to do the analysis you think?

I apologize if this is not the right place to ask questions like this. Otherwise I would really appreciate your input.

Thank you!

[George Ford]

logit gives you coefficients, and margins is required to interpret.
logistic gives you odds ratios, which no one knows how to interpret.
reg might be useful, if the DV has a mean away from zero (the coefs will be nearly identical to the marginal effects on logit).

make sure your categorical answers are evenly spaced and try simple models. missing values or linear correlation with the categories and other variables could cause some to not know. and, if you have 5 categories, only 4 coefficients are available (dummy trap).

[Rich Goldstein]

if I understand your question, please read https://en.wikipedia.org/wiki/Dummy_...le_(statistics) on the dummy variable trap