Comparison of proportions with several variables

#1
Hey everyone,

I've been a bit stuck lately on which test to choose or to implement for an assignment. Little bit of context : from a dataframe containing answers of a survey from several hundreds of thousand of people, I needed to ask a scientific question then analyse the data and make inference from it. Since the data range from 1972 to 2012 with a lot of informations on the participant, I choose to analyze the difference of opinion on abortion through the years, and then based on either the social class, highest degree obtaint or religion.

So far so good, but to push it a tiny bit further, I took the two most interesting variables : degree and religion, and I wanted to check if the opinion on abortion would change within a religion (and between religion) based on the highest degree. Basically is a Catholic person with a low degree more likely or not to be against abortion than a Catholic person with a high degree, and then for similar degree is a Protestant more likely or not to be against religion than a Jewish person.


My problem is that at first I thought of using a chi-square test of independance, but I can't really see how to create a contingency table with all 3 variables. I can only think of three options :

  1. A "Religion vs Degree" contingency table containing only the population of either "Yes" or "No" answer regarding Abortion
  2. A "Opinion vs Religion&Degree" contingency table (which means that each column would be like "Protestant Highschool", "Catholic Bachelor", "Jewish Graduate")
  3. A completely different approach that I don't know of
If any of you can help me or give a little hint on the matter, that would be awesome.

Thanks in advance !