Running head: MORAL FOUNDATIONS AND IDEOLOGY
Tracing the threads:
How five moral concerns (especially Purity) help explain culture war attitudes
September 5th, 2011
Word count: 11,287
Commentators have noted that the issue stands taken by each side of the American “culture war” lack conceptual consistency and can even seem contradictory. We sought to predict and understand the psychological underpinnings of culture war attitudes using Moral Foundations Theory. Over two studies, 24,739 participants, and 20 such issues (e.g. abortion, immigration, same-sex marriage), we found that endorsement of five different moral foundations predicted judgments about these issues over and above ideology, age, sex, religious attendance, and political interest. Our results suggest that dispositional tendencies, particularly a person’s moral intuitions, may underlie, motivate, and unite ideological positions across a variety of issues and offer new insights into the multiple “moral threads” connecting disparate political positions.
Word count: 116
Keywords: moral intuitions, characteristic adaptations, ideology, political psychology, moral values
Tracing the threads:
How five moral concerns (especially Purity) help explain culture war attitudes
Imagine two Americans, Libby and Connie. Libby believes abortion should be legal and supports tight restrictions on gun purchases, while Connie believes that abortion is tantamount to murder and that any restrictions on gun purchases violate the Second Amendment of the U.S. Constitution. Which one of these two people is more likely to favor capital punishment?
Most Americans know intuitively that the answer is Connie, because of her conservative stance on abortion and gun control. But what makes these positions hang together? Why is Connie for the death penalty if she’s pro-life? Why does Libby believe in individual freedom in the case of abortion, but not in the case of gun purchases?
One possibility is that there is no unifying principle, other than the fact that the two major political parties in the U.S. have staked out opposing positions on these issues. Perhaps people simply know what position the political “team” they support has taken, and they adopt a menu of such positions even when some of them entail internal contradictions (Converse, 1964). Cohen (2003), for example, found that people were more favorably disposed to a policy position if they believed it was proposed by their own political party than by the opposing one, even when the policy content was kept identical.
However, many political scientists and psychologists have argued against the notion that people’s issue positions passively track their liberal-conservative “team” preference. Instead, these researchers suggest that people are psychologically prepared (by their genes, childhood experiences, personality characteristics, positions in society, etc.) to adopt some policy positions more easily than others (see, for example, Carney, Jost, Gosling, & Potter, 2008; Duckitt, 2001). Such scholars search for coherence among the issues that divide liberals and conservatives by examining their fit with a variety of cognitive structures (e.g., Lakoff, 1996), epistemological orientations (Hunter, 1991), or personality traits, existential needs, and motivated cognitions (Jost, Glaser, Kruglanski, & Sulloway, 2003). We agree with the thrust of this work and suggest that moral intuitions are one powerful and largely unexplored psychological mechanism that underlies ideology in general and issue positions in particular. In this article we present a novel method for conceptualizing and measuring the moral factors that predispose people to accept some political positions more readily than others. We describe five moral “foundations” and show that endorsement of these foundations predicts people’s attitudes on a wide range of culture war issues, above and beyond their demographic characteristics – including, importantly, ideological self-placement. We argue that these findings add depth and richness to our understanding, by revealing the multiple (and often unexpected) moral concerns that seem at play for a given issue, while also offering a relatively comprehensive approach to the study of a broad range of political attitudes.
2. Previous Conceptualizations of the Liberal-Conservative Divide
In his 1991 book Culture Wars, James Hunter noted a realignment occurring in American politics, in which divisions within major religious denominations were becoming more important, while differences across denominations were shrinking. Orthodox Jews, conservative Catholics, and evangelical Protestants sometimes found more in common with each other than with their more liberal co-religionists, particularly on issues related to sex, gender, family life, and what was widely perceived to be the debasement of popular culture.
Hunter proposed that underlying this new division was a fundamental disagreement over the nature of moral authority. On one side of this “culture war” were the “orthodox,” who believed that moral truths existed independently of human preferences, and were grounded in “an external, definable, and transcendent authority” (Hunter, 1991, p. 44). On the other side were the “progressives,” who saw moral truths not as fixed but as works in progress, which had to be reinterpreted by each generation for its own time. Once an individual took a position on the nature of moral authority (whether because of childhood socialization or innate temperament – Hunter took no position), that person would be “prepared” to adopt one side or the other on most of the culture war issues. Issues that pitted a traditional, Bible-based, or standard-affirming position vs. a modern, liberationist, or relativist position were especially prone to becoming battlegrounds in the culture war.1
A second attempt to explain the coherence among partisan positions came from George Lakoff (1996) in his book Moral Politics. Lakoff proposed that Americans generally construe the nation as a family, with government as a parent, but they disagree on the cognitive model of the family that they prefer. Conservatives are those who think of the ideal family as being headed by a “strict father,” and liberals are those more prone to idealize families headed by a “nurturant parent.” When applied to politics, conservative positions cohere because they tend to be those that impose strict discipline and “tough love” for the children’s own good in a world full of danger and competition. Liberal positions cohere because they are often attempts to provide the resources and freedom that individuals need to develop their talents in a world that is relatively safe and cooperative.
Most recently, John Jost and his colleagues have argued that basic personality traits prepare some individuals to become conservative, others to become liberal. In a comprehensive meta-analysis of the psychological correlates of conservatism, Jost et al. (2003) found that conservatives (compared to liberals) have higher needs for order, structure, and closure; they are lower on tolerance of ambiguity, integrative complexity, and openness to experience, and they score higher on measures of death anxiety and fear of threats to the stability of the social system.
Jost et al. (2003) propose that logic or coherence is to be found not in the issues themselves, but in two overarching habits of minds predisposed to conservatism: first and foremost, a resistance to or dislike of change, and secondly, a tendency to accept (or even prefer) social inequality. These are the two psychological “threads” that stitch together seemingly unrelated stances on political issues.
3. Beyond change and inequality: Moral Foundations Theory
Hunter (1991), Lakoff (1996), and Jost et al. (2003) begin from different points, yet converge on the idea that coherence in culture war attitudes can be traced largely to disparate affinities toward change versus stability, and to the related tension between hierarchy (which generally supports stability) and equality (which often mandates change). Differential comfort with change and the desire to adhere to the dictates of tradition and traditional authority certainly captures one central aspect of the liberal-conservative dynamic, but recent research suggests that the psychological divide between liberals and conservatives is more multi-dimensional in nature.
We undertook the present inquiry to determine whether Moral Foundations Theory (MFT, Haidt & Graham, 2007; Haidt & Joseph, 2004) might reveal the influence of moral motives beyond those related to change and inequality.
Moral Foundations Theory (based in part on the work of Fiske, 1991, and Shweder, Much, Mahapatra, & Park, 1997), argues that human groups construct moral virtues, meanings, and institutions in variable ways by relying, to varying degrees, on five innate psychological systems. Each system (akin to the five kinds of taste receptors on the tongue) produces fast, automatic gut-reactions of like and dislike when certain patterns are perceived in the social world, which in turn guide moral judgments of right and wrong. The foundations were identified by a simultaneous review of the literature on morality in anthropology and evolutionary psychology (Haidt & Joseph, 2004). The five systems, or moral foundations, are harm/care, fairness/reciprocity, ingroup/loyalty, authority/respect, and purity/sanctity. In terms of previous research on ideology, fairness/reciprocity relates to a concern for inequality and authority/respect relates to a preference for stability versus change (Graham, Nosek, Haidt, Iyer, Koleva, & Ditto, 2009).
Briefly, the harm/care foundation leads us to disapprove of acts and individuals that cause pain and suffering and to approve of those who prevent or alleviate harm. Some associated virtues are compassion and kindness. The fairness/reciprocity foundation makes us sensitive to issues of equality and justice and leads us to frown upon acts and people that violate these principles. The ingroup/loyalty foundation is based on our sense of obligation and attachment to groups (e.g. our family, company, team, church, or country), leading us to approve of those who contribute to the group’s well-being and cohesion, and who maintain their distance from outsiders. Relevant virtues are loyalty and patriotism. The authority/respect foundation is based on our tendency to create hierarchically structured societies of dominance and subordination. This foundation includes approval of individuals who fulfill the duties associated with their position on the social ladder, for example by showing good leadership and wisdom, or respect and obedience. Lastly, the purity/sanctity foundation is based on the uniquely human emotion of disgust in response not just to biological contaminants (e.g. feces or rotten food), but also to various social elicitors like spiritual and physical corruption, or the inability to control one’s base impulses (see Rozin, Haidt, & McCauley, 2008, on disgust). Some virtues associated with this foundation are chastity, temperance, and spirituality.
The five moral foundations are posited to be universally present (see Haidt & Kesebir, 2010, on the evolutionary processes that may have shaped the five foundations), but morality is a complex and culturally variable construction. Different societies build different moralities, and they do so in part by resting their moral virtues, claims, and institutions to varying degrees on the five moral foundations, much as the world’s strikingly different cuisines all rely on the same five basic tastes. Furthermore, subcultures within the same society may also elaborate and emphasize the five foundations to differing degrees. Some preliminary findings suggest that gender, socioeconomic class, and ethnic background are all associated with differential endorsement of foundation-related concerns (Haidt, Koller, & Dias, 1993; Koleva, Graham, Ditto, Haidt, & Iyer, 2008).
Most importantly for the current purposes, several studies have now found that political liberals and conservatives show a clear and consistent pattern of differences in the moral weight they place on the various foundations (Graham, Haidt, & Nosek, 2009; Haidt & Graham, 2007; McAdams, Albaugh, Farber, Daniels, Logan, & Olson, 2008; van Leeuwen & Park, 2009). Specifically, liberals rate considerations of harm and fairness as significantly more important moral factors than ingroup, authority, or purity. To liberals, acts are perceived as immoral primarily to the extent that they harm others or treat people unfairly. Social conservatives, in contrast, show a more even distribution of reliance upon all five foundations2. Liberals and conservatives often disagree about what is harmful and what is unfair, but the most striking political differences involve the ingroup, authority, and purity foundations.
In short, MFT represents a broader attempt to identify the moral concerns that motivate culture war positions. It captures the emphasis of past approaches on differential sensitivity to stability and change (reflected in the authority/respect foundation), and (dis)comfort with inequality (reflected in the fairness/reciprocity foundation), while also pointing to three additional moral dimensions that might be at play. For example, liberals’ heightened sensitivity to issues of harm might undergird their traditionally negative attitudes toward capital punishment, as well as their more contemporary distaste for the Bush administration’s use of “enhanced interrogation techniques” with terrorist suspects. Similarly, conservatives’ stronger valuation of ingroup/loyalty seems likely to relate to their attitudes toward illegal immigration, and to affronts to patriotic symbols such as burning the American flag. Perhaps most clearly missing from past conceptualizations of the liberal-conservative divide, however, is the role of concerns about spiritual purity/sanctity. Although it is possible to see conservative disapproval of nontraditional sexuality (e.g. casual sex, same-sex relationships, use of pornography) as a function of reverence for traditional practices or a discomfort with change, it seems plausible that an additional and powerful dynamic underlying these attitudes is the propensity to experience disgust (Inbar, Pizarro, & Bloom, 2009) and apprehension about spiritual contamination.
The current research examines culture war attitudes through the lens of MFT. Specifically, we explore the role of individual differences in moral intuitions as psychological predispositions that underlie political attitudes. In two studies we use endorsement of the five moral foundations to predict moral disapproval for controversial political issues as well as specific attitude stands on such issues. Our goal was not just to improve the prediction of political attitudes, but to use MFT to help understand the psychological underpinnings of such attitudes by illuminating the “moral threads” that may underlie Americans’ culture war attitudes.
4. Study 1: Moral Disapproval
For our first study, we began in the most direct way possible: we measured individuals’ moral disapproval for thirteen controversial behaviors and examined the degree to which these disapproval scores were predicted by demographic factors, interest in politics, political ideology (liberal to conservative), and scores on the Moral Foundations Questionnaire (MFQ).
Participants were 10,2223 adults residing in the U.S. who volunteered at http://www.yourmorals.org. All participants had previously registered at the site, providing demographic information including age (mean age = 38 years), sex (62% male), religious attendance (M = 1.37, SD = 1.73 on a scale from 0 = “never” to 5 = “one or more times each week”), interest in politics (M = 1.70, SD = .50 on a scale from 0 = “not much interested” to 2 = “very much interested”), and political orientation (M = 2.80, SD = 1.62, on scale ranging from 1 = “very liberal” to 7 = “very conservative”).
Participants self-select to take one or multiple surveys from a list of 15-20 available at any one time. The majority of visitors to the site begin by completing the MFQ. Many visitors take additional surveys as well. Here we report results only for those who completed both the MFQ and a second morality survey.
The MFQ is a 30-item self-report measure of the extent to which an individual endorses each of the five types of moral concerns: Harm, Fairness, Ingroup, Authority, and Purity (see Graham, Nosek, Haidt, Iyer, Koleva, & Ditto, 2011 for an extensive analysis of its psychometric properties). The scale consists of two sections. In the first, participants rate how relevant each of 15 concerns are to them when making moral judgments, such as “Whether or not some people were treated differently from others” for Fairness. In the second section, participants rate their agreement with statements that embody or negate each foundation, e.g., “It is more important to be a team player than to express oneself” for Ingroup (the items for the MFQ can be found at www.moralfoundations.org). Six items per foundation (three from each section) were averaged to produce a score for each person on each foundation. Cronbach’s reliability statistics were as follows: Harm α = .67, Fairness α = .66, Ingroup α = .71,
Authority α = .76, and Purity α = .85.
In the morality survey participants answered questions concerning thirteen social issues. The instructions and items for this scale were based on a Gallup poll that was conducted in May of 2007 (http://www.pollingreport.com/values.htm). The instructions read: “Here is a list of controversial issues. Regardless of whether or not you think it should be legal, for each one, please indicate whether you personally believe that in general it is morally acceptable or morally wrong.” Responses were measured on a scale from “1 = Morally acceptable in most or all cases” to “5 = Morally wrong in most or all cases.” Items appeared in an order randomized for each participant. The thirteen issues were abortion, the death penalty, medical testing with animals, euthanasia, same-sex marriage, homosexual relations, burning a U.S. flag, having a baby outside of marriage, stem-cell research, pornography, gambling, casual sex, and animal cloning.
To answer our research question – do the moral foundations help explain moral disapproval on culture war issues beyond ideological self-identification – we used multiple regression. Disapproval ratings for each issue were regressed on all five demographic variables --age, sex (dummy coded where 0 = female and 1 = male), religious attendance, interest in politics, and political orientation4 -- and all five moral foundation scores. The simultaneous inclusion of all foundations in the model created a challenging test for the foundations, given that they are all intercorrelated and all correlated with political orientation, age, gender, religious attendance, and interest in politics (see Table 1). To ensure that the integrity of the regression models was not threatened by the interdependence among the predictors, we obtained collinearity diagnostics. The tolerance values tended to be high and none were below .37, thus collinearity was not a serious concern. Betas for each foundation therefore show what each foundation adds to moral disapproval ratings, above and beyond age, sex, religious attendance, interest in politics, ideology, and the other four foundations. Because of our very large sample and resulting statistical power, our interpretations focus on the size of the standardized coefficients instead of p values. The results are summarized in Table 2.
4.2.1. Demographic Predictors of Disapproval
Ratings on these controversial social issues were uniquely associated with one’s political ideology (mean |β| = .22, range -.02 to .38); only medical testing on animals, cloning, gambling, and using pornography had betas less than .15. This is not surprising and confirms that these issues are appropriate for the investigation of culture war opinions. Religious attendance was a moderate unique predictor (mean |β| = .14; range .01 to .28) for most issues, particularly those related to sexuality, but typically weaker than political orientation. More frequent church attendance uniquely predicted stronger disapproval for all issues except for flag burning (no relationship) and animal testing (weak reverse relationship). In contrast to religious attendance, age (mean |β| = .05; range 0 to .12), sex (mean |β| = .07; range .01 to .19), and interest in politics (mean |β| = .02; range 0 to .07) were weak predictors of moral disapproval.
4.2.2. Moral Foundation Predictors
As seen in Table 2, Purity emerged as the foundation that best predicted disapproval on culture war issues. It was most strongly associated with disapproval for issues dealing with sexuality (casual sex and using pornography), relationships and marriage (same-sex relations, same-sex marriage, and baby outside marriage), and the sanctity of life (euthanasia and cloning). Purity was also the strongest predictor of disapproval of gambling. Lastly, unique effects of Purity were generally much higher than those for the other foundations (βs .25 for ten issues).
Harm was the strongest predictor of disapproval of medical testing on animals and the second strongest (after ideology) for disapproval of the death penalty. Harm was also the second best-predicting foundation (after Purity) for disapproval of cloning animals. Lastly, although Fairness, Ingroup, and Authority were significantly associated with many issues, they were not the top predictors for any of them. Ingroup was the second strongest foundation, after Purity, in predicting disapproval of flag-burning. None of the betas for Fairness were above .06, and for Authority only one beta (predicting death penalty) was above .10.
4.2.3. Moral Foundations and Alternative Models
Because visitors to our research platform can choose to complete multiple questionnaires, a small portion of our Study 1 participants (N = 460, mean age = 36 years, 57% male) also provided scores for Right Wing Authoritarianism (RWA, Zakrisson, 2005) and Social Dominance Orientation (SDO, Sidanius & Pratto, 2001). Because both RWA and SDO have been identified as important predictors of political attitudes (e.g. Duckitt et al., 2002; Duckitt & Sibley, 2009; Jost et al., 2003) we ran supplementary analyses to see how their inclusion in our regression models might affect our results.
We should note that both RWA and SDO were used as external criteria in the development and validation of the MFQ (Graham et al., 2011) as we aimed for comprehensive coverage of the moral domain; therefore we expected that the overlap in these constructs will reduce our coefficients. In particular we thought that RWA might share variance with Authority and Purity and that SDO might share variance with Fairness and Ingroup. Furthermore, we expected SDO, and particularly RWA, to outperform (but not eliminate) the foundations as predictors. This is because the former constructs are closer, conceptually and item-wise, to political attitudes than is the MFQ. For example, Duckitt and colleagues (Duckitt et al. 2002; Duckitt & Sibley, 2009) have argued that RWA and SDO are “more appropriately viewed as measuring social attitude or ideological belief dimensions rather than personality.” In contrast, the moral foundations represent a more basic and generalized set of psychological tendencies (characteristic adaptations) and are closer to what Duckitt and colleagues refer to as “motivational goals” (e.g. values) that may in turn predispose individuals to resonate with certain sociopolitical beliefs.
As expected, adding RWA and SDO to the model reduced the foundation effects somewhat but did not appreciably change the overall pattern of associations (a table summarizing these findings can be found in an online supplement posted at http://www.yourmorals.org/MoralThreads/MoralThreads_Supplement.pdf). For 11 out of 13 issues, the top predicting foundation remained the same and statistically significant, and for 10 issues its effect was comparable to or larger than that of both RWA and SDO. These data indicate that moral concerns about Purity – the dominant foundation predictor in this study – are distinct from right-wing authoritarianism and social dominance orientation. Purity concerns are largely missing from existing psychological models of political attitudes but our findings indicate that this dispositional factor plays a key role and deserves further attention and study.