Are Moral Values Uniquely Divisive?

Download data and study materials from OSF

Principal investigators:

Scott Clifford

Texas A&M University

Email: scottclifford@tamu.edu

Homepage: https://scottaclifford.com/

Jae-Hee Jung

University of Houston

Email: jjung26@central.uh.edu

Homepage: https://jaeheejung.com/


Sample size: 863

Field period: 08/11/2022-03/08/2023

Abstract
Political scientists have long viewed values as a source of structure and constraint in political belief systems and, more recently, as a potential contributor to affective polarization. Some scholars have argued that moral values, in particular, have contributed to disagreement and hostility between partisan groups. Yet, there is little direct and systematic evidence as to which values are moralized. In a pilot study using a convenience sample, we examined 21 values from three different value systems (moral foundations, Schwartz values, and political values) and show that there is meaningful overlap in value moralization between value systems and considerable variation in value moralization within value systems. We replicated this finding on a sample of respondents from the AmeriSpeak panel at NORC and experimentally tested that value moralization amplifies the effect of value disagreement on social polarization. Taken together, our research sheds light on how values may differentially contribute to political and social polarization.
Hypotheses
Value disagreement with another individual will have a greater effect on socially polarized attitudes toward that individual when that value is highly moralized than when it is not.
Experimental Manipulations
Respondents first answered six sets of questions on values that measure whether they agree or disagree with each of the values and the extent to which they moralize their position on each of the values. After that, respondents saw six profiles of hypothetical people. Each hypothetical person was randomly assigned to either agree or disagree with one of the value positions expressed by the respondent in the earlier part of the survey. That is the attribute of interest in our conjoint experiment. In each profile, there was also information on that person’s gender, age, race, partisan identity, career, religion, education, and hobby.
Outcomes
For each profile, respondents answered three questions about attitudes toward the hypothetical person. (1) “How positive or negative are your feelings toward this person?” (2) “How happy or unhappy would you be to have this person as your neighbor?” (3) “Imagine that you need someone to look after your house for a week while you’re out of town. How comfortable would you be having this person look after your house?” Greater values on these variables indicate more favorable or trusting attitudes.
Summary of Results
We regressed each of the outcome variables as a function of whether or not the respondent and the hypothetical person disagree, the respondent’s moralization of their position on the value, and an interaction between the two. There is a significant interaction between our moral conviction variable and value disagreement variable for all three outcome variables (p-values < 0.001). The interaction coefficients are −0.019, −0.026, and −0.025 for the general favorability, neighbor, and house-sitting outcomes, respectively. Moreover, the slopes are steeper for the two more socially interactive outcomes (about neighbor and house-sitting) than for the general favorability outcome, though these differences are not statistically significant (p = 0.061, p = 0.264, respectively).
References
Jung, Jae-Hee and Scott Clifford. 2024. “Varieties of Values: Moral Values Are Uniquely Divisive.” American Political Science Review.