regression analyses…Clinton = Loyalty & Obama = Fairness

Jon Haidt, the most public advocate of Moral Foundation Theory wondered….

very interesting. I wonder if regression analysis would give cleaner results here. It does seem that clinton supporters always show the more liberal profile. What would happen if you compute a difference score between liking for clinton and liking for obama, and then use that as the DV to be predicted in a regression equation that includes politics and all 5 foundation scores. Which foundations predict the difference when politics is included in the equation?

Results are below….

Clinton Obama 5 Foundations Regression

and controlling for sex and education (both of which replicate results that pollsters have found, meaning that women prefer Clinton and more educated people prefer Obama)…

Clinton Obama 5 Foundations Regression with Sex & Education

In both cases, the fairness foundation is the strongest predictor of preference for Obama and the effect remains significant controlling for politics, age, and gender. Loyalty is the strongest predictor of preference for Clinton. I ran these results separately for conservatives and liberals and these effects are driven largely by liberals.

As I have mentioned before, the Loyalty/Clinton result is fairly explainable. It’s a good confirmation that our scale, the MFQ, is indeed tapping individual differences in endorsement of loyalty as a virtue. Loyal liberals should be somewhat favorable to Clinton, who has been an icon of Democratic politics for years.

The Fairness/Obama result is perhaps less clearly interpretable, but I might suggest that Obama supporters tend to be people who support the underdog and might not want 2 families (Bush/Clinton) dominating the White House as they have for 20+ years. Or possibly Obama supporters are those liberals whose sense of fairness has been offended by some of the scandals of the Clinton years

However, my personal theory would be that Obama supporters are actually more liberal than Clinton supporters in terms of liberal values (Fairness being the most liberal of values), but identify less with the term liberal, hence the result that we see when controlling for self reported liberalism.



Also read...