Page 1 of 1

help with modelling a causal effect

Posted: Sun Nov 06, 2011 1:21 pm
by gryphones
I am looking for some pointers on possible regression techniques for a problem I'm thinking of doing.

This regression seems overwhelming because I am used to using econometrics as a predictor of multiple X's on a Y that is continuous. Right now I am looking to find a causal relationship between 2 factors that are both discrete, which is new to me.

I am using US census data to try and find the causal effect of someones level of education on divorce.

There are about 14 different possible variables for level of education (educus) and I want to see the effect of these possibilities on marital status (marst), which has 4 possibilities (single/never married, married/in union, divorced/spouse absent, widowed).

I am thinking of using controls for age, employment status, salary, and possibly weight.

Re: help with modelling a causal effect

Posted: Sun Nov 06, 2011 11:11 pm
by startz
Take a look at the material on logit/probit