Sage Journals: Discover world-class research

Abstract

Moral foundations theory (MFT) claims that individuals use their intuitions on five “virtues” as guidelines for moral judgment, and recent research makes the case that these intuitions cause people to adopt important political attitudes, including partisanship and ideology. New work in political science, however, demonstrates not only that the causal effect of moral foundations on these political predispositions is weaker than once thought, but it also opens the door to the possibility that causality runs in the opposite direction—from political predispositions to moral foundations. In this manuscript, I build on this new work and test the extent to which partisan and ideological considerations cause individuals’ moral foundations to shift in predictable ways. The results show that while these group-based cues do exert some influence on moral foundations, the effects of outgroup cues are particularly strong. I conclude that small shifts in political context do cause MFT measures to move, and, to close, I discuss the need for continued theoretical development in MFT as well as an increased attention to measurement.

Keywords

Moral foundations partisanship ideology source cues

Moral foundations theory (MFT) posits that individuals’ intuitions on five psychological modules serve as guidelines for moral judgment (Haidt et al., 2009). Two of these modules, harm/care and fairness/reciprocity, are commonly known as individualizing foundations due to their focus on the well-being of individuals in society, and the other three modules, ingroup/loyalty, authority/respect, and purity/sanctity, are referred to as binding foundations due their focus on maintaining group ties. How do preferences on each of the five foundations affect behavior and choice? Haidt et al. (2009: 112) compare preferences on the five foundations to dials on an audio equalizer—each foundation represents a dial, and the combination of the settings on each of the five dials guides moral judgement.

Briefly, research demonstrates important links between MFT and a wide variety of behaviors and attitudes, including consumer choice (Watkins et al., 2016), charitable donations (Winterich et al., 2012), and attitudes related to suicide (Rottman et al., 2014), heroism (Jayawickreme and DiStefano, 2012), and torture (Smith et al., 2014). Specific to the political world, moral foundations have been shown to correspond to ideological preferences (Graham et al., 2009), turnout (Johnson et al., 2014), and attitudes on highly polarized issues including stem-cell research (Clifford and Jerit, 2013), the environment (Feinberg and Willer, 2013), and various “culture war” issues (Koleva et al., 2012). However, while the aforementioned research makes it clear that moral foundations and behavior are correlated, the extent to which moral foundations cause people to adopt certain political attitudes remains an open question.

Currently, some existing research suggests that preferences on moral foundations do, in fact, cause changes in political attitudes. Feinberg and Willer (2013, study 3), for example, demonstrate that pro-environmental communications framed in terms of purity/sanctity (as opposed to harm/care) resonate with conservatives and close the gap between liberals and conservatives on environmental attitudes. In a similar vein, Day et al. (2014) show that framing policies in terms of moral foundations often increases the extent to which individuals support them, even when the policy in question is initially unpopular. Ultimately, this body of literature suggests that there is a causal pathway that leads from moral preferences to political attitudes.

While the causal path from moral foundations to political attitudes is both theoretically justified and empirically supported, new research suggests that it may be weaker than once thought. Working specifically with MFT as a theory of ideology (which can likely be extended to other important political predispositions), Smith et al. (2017) note that if left/right preferences are in fact rooted in moral foundations, three assumptions central to MFT must hold: moral foundations must (a) be stable and dispositional, (b) predict changes in ideology, and (c) be heritable (pp. 426–427). In short, Smith et al. find “considerable within-individual dynamic variability in MFT measures, little evidence that change in MFT accounts for change in ideology, and minimal and sporadic evidence of heritability in the MFT measures” (p. 434). While their data do suggest that “changes in political orientations lead to changes in moral foundations to a greater degree rather than vice versa” (p. 434), they also note that such evidence is inconsistent, thus prohibiting them from concluding, without hesitation, that political orientations shape moral foundations.

In this paper, I take the research of Smith et al. (2017) one step further, and I explicitly test the extent to which politically relevant psychological predispositions (i.e., partisanship and ideology) cause individuals’ moral foundations to change. Employing an experimental design that systematically varies partisan and ideological cues in the moral foundations questionnaire (modeled after that in Goren et al., 2009), I show that moral foundations, as they are traditionally measured, respond in systematic ways.

Partisanship and ideology as “unmoved movers”

Why might partisanship and ideology influence individuals’ moral foundations? Research in social psychology touts the power of group influence and argues that “groups define the very meaning of objects in the world” (Cohen 2003: 808). In short, group membership serves as an important heuristic that eases the cognitive effort needed to evaluate various stimuli by allowing group members to assume that other members of the social group have sensibilities and tastes similar to their own, and further, that “the judgments of group representatives and leaders can…be viewed as diagnostic, if not definitional, of social meaning” (where “social meaning” is defined as the perceived compatibility of an object of judgment with socially shared values—see p. 808). Using a similar theoretical framework, Kahan (2015) notes that citizens often use “identity-protective cognition” when processing information on contentious political issues so that they can minimize cognitive dissonance and maintain social status in their partisan group.

Operating within this theoretical framework, Cohen (2003) demonstrates that group considerations—more specifically, party cues—exert significant influence on public opinion. Related work finds that members of several self-defining groups engage in identity-protective cognitive processes when thinking about political issues (Kahan, 2015; Kahan et al., 2017). Ultimately, this work aligns nicely with research in political science that suggests partisan identity to be among the most influential aspects of mass-level political behavior (Campbell et al., 1960; Converse, 1964; Huddy et al., 2015; Lenz, 2009; Mason, 2015), even to the extent that it influences what are thought to be the most fundamental factors guiding behavior and choice—human values (Goren, 2005; Goren et al., 2009; McCann, 1997; Nelson et al., 1997).¹ Additional research suggests that ideological identification, like partisanship, is often a core aspect of one’s political identity that exerts a considerable amount of influence on how one processes information when forming preferences (Malka and Lelkes, 2010). Taking this research into consideration, I formulate my first hypothesis: Moral foundations of partisan and ideological identifiers, as they are often measured, are influenced by group-based considerations.

Might certain types of partisan and/or ideological messages exert more of an effect on individuals’ moral foundations than others? Recent research indicates that the “forgotten” side of partisanship—negative partisanship—is also important to public opinion and various forms of political behavior (Caruana et al., 2015; Medeiros and Noël, 2014). In the US context, negative partisanship has a profound effect on the partisan composition of both the House and the Senate (Abramowitz and Webster, 2016). Furthermore, it is suggested that many Americans now dislike the opposing party more than they like their own party (p. 21). With this research in mind, I formulate a second hypothesis: Outgroup cues will exert a greater effect than ingroup cues on measures of moral foundations.

Experimental design, data, methods

To test these hypotheses, I run an experiment that systematically varies partisan and ideological cues embedded in a moral foundations questionnaire (the MFQ20).² Briefly, the MFQ20 is a 22-item battery in which 20 of the items measure respondents’ moral attitudes, and the remaining two items serve as attention checks. Of the 20 MFT-related questions, 10 ask respondents about the extent to which a particular concept is relevant/irrelevant when deciding whether something is right or wrong (measured on a 6-point scale from “not at all relevant,” coded 0, to “extremely relevant,” coded 5), and 10 ask about the degree to which respondents agree/disagree with statements related to moral virtues (measured on a 6-point scale from “strongly disagree,” coded 0, to “strongly agree,” coded 5). Theoretically, four items (two relevant/irrelevant items and two agree/disagree items) are meant to tap attitudes on each of the five foundations. And, for each foundation, the four items are meant to be combined into a summated rating scale. In the interest of maximizing scale reliability, I dropped items that made the scales less reliable. Ultimately, attitudes on harm/care, fairness/reciprocity, and authority/respect are measured with three-item scales (values for Cronbach’s alpha are .75, .77, and .76, respectively), and the full four-item scales are used for ingroup/loyalty (Cronbach’s alpha = .68) and purity/sanctity (Cronbach’s alpha = .88). To simplify the interpretation of statistical results, all scale values are rescaled to range between 0 and 1.

The experiment is designed such that, prior to seeing the survey questions, respondents are randomly assigned to one of four experimental groups: (a) control, (b) party cues, (c) ideological cues, (d) or party + ideological cues. For those in the control group, all MFQ20 items are prefaced with “Some people think/feel… .” For those in the treatment groups, the eight questions designed to measure attitudes on individualizing foundations (i.e., harm/care and fairness/reciprocity) are prefaced with “[Democrats/Liberals/Liberal Democrats] think/feel…,” and the twelve questions designed to measure attitudes on binding foundations (i.e., ingroup/loyalty, authority/respect, and purity/sanctity) are prefaced with “[Republicans/Conservatives/Conservative Republicans] think/feel…”³ (the full questionnaire can be found in the supplemental material). Analyses center on whether the various source cues cause differences in responses among partisans and ideological identifiers.

Participants were recruited via Amazon.com’s Mechanical Turk. As indicated by a number of studies, MTurk samples have desirable characteristics, especially with respect to experimental data (Berinsky et al., 2012; Mullinix et al., 2015). Data collection took place in November 2015, January 2016, and June 2016. Participants that took the survey in November or January were paid US$1.50, and respondents that took the survey in June were paid US$1.10. In all cases, participants completed the MFQ20 battery, followed by a series of questions regarding political attitudes (including partisan and ideological identities) and demographics. Respondents that failed either of the attention checks and respondents that gave the same response for all items in the MFQ20 were dropped from the data (final $n = 1, 070$ ).⁴

I estimate treatment effects first by subsetting the data by partisan and ideological groups, then running a series of tobit models (because some distributions are censored at the top or bottom of the scales) where the independent variables are respondents’ treatment assignments (dummy coded with the control group serving as the baseline for comparison), and dependent variables are the rating scales that correspond to respondents’ attitudes on the five foundations.

Results

Prior to examining the results of the experiment, it is worth noting that the distributions of responses by ideological group are in line with previous research (e.g., Graham et al., 2009). Liberals rated harm/care and fairness/reciprocity significantly higher than the remaining three foundations, and conservatives rated all the foundations in a more equal manner, with harm/care and fairness/reciprocity coming out slightly higher than the other three foundations (see Figure 1).⁵ For both the control group and the treatment groups, confirmatory factor analyses suggest that a 5-factor model fits the data better than a 2- or 3-factor model.⁶

Figure 1.

Distributions of Moral Preferences.

With respect to the main research question, results of the tobit models are shown in Table 1.⁷ Looking first at Democrats and liberals, the ingroup cues have no effect on responses—attitudes on the harm/care and fairness/reciprocity foundations are statistically indistinguishable across treatment conditions. The outgroup cues have a much greater effect, however. For each of the three binding foundations, respondents in the party + ideo. condition recorded significantly lower scores on the foundations than those in the control condition. Additionally, those in the party condition scored purity/sanctity lower than those in the control condition. Among Democrats, effect sizes ranged from 3% of the scale on the ingroup/loyalty foundation in the party + ideo. condition to 12% of the scale on the purity/sanctity foundation in the party + ideo. condition.

Table 1.

Tobit results.

	Harm/Care	Fairness/Reciprocity	Ingroup/Loyalty	Authority/Respect	Purity/Sanctity
Democrats ( $n = 464$ )
Constant	0.76*** (.014)	0.81*** (.015)	0.41*** (.014)	0.42*** (.016)	0.37*** (.021)
Party Cue	−0.02 (.020)	−0.00 (.021)	−0.02 (.020)	−0.03 (.023)	−0.07** (.030)
Party + Ideo. Cue	−0.00 (.020)	0.02 (.021)	−0.03* (.020)	−0.08*** (.023)	−0.12*** (.030)
Liberals ( $n = 467$ )
Constant	0.76*** (.016)	0.81*** (.016)	0.40*** (.015)	0.39*** (.016)	0.32*** (.021)
Ideo. Cue	−0.01 (.022)	−0.00 (.022)	−0.03 (.021)	−0.04 (.023)	−0.06** (.029)
Party + Ideo. Cue	−0.00 (.022)	0.02 (.022)	−0.04** (.021)	−0.08*** (.022)	−0.09*** (.029)
Cue Type	Ingroup	Ingroup	Outgroup	Outgroup	Outgroup
Republicans ( $n = 183$ )
Constant	0.70*** (.030)	0.70*** (.031)	0.52*** (.023)	0.56*** (.025)	0.56*** (.031)
Party Cue	−0.06 (.041)	−0.07* (.043)	0.08** (.032)	0.06* (.035)	−0.01 (.044)
Party + Ideo. Cue	−0.11** (.043)	−0.12*** (.044)	0.03 (.033)	−0.01 (.036)	−0.01 (.045)
Conservatives ( $n = 170$ )
Constant	0.72 (.033)	0.74*** (.031)	0.52*** (.023)	0.58*** (.025)	0.59*** (.032)
Ideo. Cue	−0.14 (.048)	−0.17*** (.044)	0.06* (.034)	−0.00 (.037)	−0.02 (.047)
Party + Ideo. Cue	−0.15 (.046)	−0.17*** (.043)	0.01 (.033)	−0.01 (.036)	−0.03 (.046)
Cue Type	Outgroup	Outgroup	Ingroup	Ingroup	Ingroup

$p < . 1$ , ** $p < . 05$ , *** $p < . 01$ ; Cell entries are tobit coefficients with standard errors in parentheses.

Turning attention to Republicans and conservatives, the treatment conditions do exert significant effects on responses regarding the individualizing foundations (i.e., harm/care and fairness/reciprocity). All eight coefficients are in the expected direction, seven are statistically significant, and effect sizes are all quite large (ranging from 7% of the scale on the fairness/reciprocity condition in the party condition to 17% of the scale on fairness/reciprocity in both the party and party + ideo. conditions). The effects of the ingroup cues (see the coefficients on ingroup/loyalty, authority/respect, and purity/sanctity), though they are significant and positive in some instances, are weaker.

Are the effects of outgroup cues significantly stronger than the effects of ingroup cues? To judge the significance of such differences, I estimate a series of seemingly unrelated regression (SUR) models. Because all parameters are estimated simultaneously, and because the variance-covariance matrix of the estimators stretches across models, SUR models allow the researcher to test for differences in coefficients across models (Zellner, 1962).⁸ To be clear, if outgroup cues exert a stronger effect than ingroup cues, then it should be the case that all of the following coefficient comparisons are significant: Harm/Care versus Ingroup/Loyalty, Harm/Care versus Authority/Respect, Harm/Care versus Purity/Sanctity, Fairness/Reciprocity versus Ingroup/Loyalty, Fairness/Reciprocity versus Authority/Respect, and Fairness/Reciprocity versus Purity/Sanctity (printed in italics in Table 2).

Table 2.

Coefficient comparisons by foundation pairs.

Party/Ideo. Cues	Dems	Libs	Reps	Cons
Harm/Care versus Fairness/Reciprocity	1.03	0.62	0.02	0.13
Harm/Care versus Ingroup/Loyalty	0.00	0.16	0.17	1.41
Harm/Care versus Authority/Respect	0.11	0.30	0.01	6.87***
Harm/Care versus Purity/Sanctity	1.30	0.90	1.05	4.17**
Fairness/Reciprocity versus Ingroup/Loyalty	0.34	0.85	0.10	2.21
Fairness/Reciprocity versus Authority/Respect	0.89	1.13	0.00	8.54***
Fairness/Reciprocity versus Purity/Sanctity	2.91*	1.96	0.98	4.63**
Ingroup/Loyalty versus Authority/Respect	0.38	0.08	0.34	0.92
Ingroup/Loyalty versus Purity/Sanctity	2.85*	0.70	1.33	0.35
Authority/Respect versus Purity/Sanctity	1.76	0.41	0.59	0.29
Party + Ideo. Cues	Dems	Libs	Reps	Cons
Harm/Care versus Fairness/Reciprocity	0.04	0.17	0.00	0.04
Harm/Care versus Ingroup/Loyalty	1.02	1.98	1.34	4.41**
Harm/Care versus Authority/Respect	5.83**	5.46**	4.44**	7.07***
Harm/Care versus Purity/Sanctity	7.94***	3.98**	3.67*	4.46**
Fairness/Reciprocity versus Ingroup/Loyalty	0.74	1.02	1.35	5.54**
Fairness/Reciprocity versus Authority/Respect	6.43**	4.77**	3.99**	8.09***
Fairness/Reciprocity versus Purity/Sanctity	8.59***	2.98*	2.83*	4.48**
Ingroup/Loyalty versus Authority/Respect	5.44**	2.77*	0.23	0.00
Ingroup/Loyalty versus Purity/Sanctity	7.77***	1.11	0.17	0.06
Authority/Respect versus Purity/Sanctity	0.93	0.14	0.01	0.24

$p < . 1$ , ** $p < . 05$ , *** $p < . 01$ ; Cell entries are F-statistics. Null is $| b_{o u t g r o u p c u e} | = | b_{i n g r o u p c u e} |$ . Cons: conservatives; Dems: Democrats; Libs: liberals; Reps: Republicans. Alternative hypothesis is $| b_{o u t g r o u p c u e} | > | b_{i n g r o u p c u e} |$ . Democrats/liberals received outgroup cues on ingroup/loyalty, authority/respect, and purity/sanctity. They received ingroup cues on harm/care and fairness/reciprocity. Republicans/conservatives received outgroup cues on harm/care and fairness/reciprocity. They received ingroup cues on ingroup/loyalty, authority/respect, and purity/sanctity. Top half of table displays F-statistics for comparisons on party vs. party cues or ideology vs. ideology cues (depending on the whether the model was fit to data on partisan or ideological identifiers). Bottom half of the table displays F-statistics for comparisons on partisan + ideological cues.

Results are shown in Table 2. Looking first at the comparisons mentioned in the paragraph above and focusing on the top half of the table (comparisons on which cues provided partisan or ideological information), there are very few significant differences across the first three columns (among Democrats, liberals, and Republicans). Among conservatives, however, 4 out of 6 F-statistics are significant (and coefficients are in the expected directions). Unexpectedly, however, the treatment effect on purity/sanctity is larger than that on ingroup/loyalty among Democrats. Turning attention to the bottom half of Table 2, 18 out of 24 of the italicized coefficient comparisons are significant; but also, it is worth noting that treatment effects on ingroup/loyalty are particularly small (and treatment effects on authority/respect and purity/sanctity are quite large), causing 3 out of 8 comparisons to reach significance. Briefly, the hypothesis receives support insofar as the treatment effects, for liberals and Democrats on the individualizing foundations, are small relative to those on authority/respect and purity/sanctity; but treatment effects on ingroup/loyalty do not fit into this pattern. Along the same lines, treatment effects for conservatives and Republicans are relatively large on the individualizing foundations and small on authority/respect and purity/sanctity; but again, those on ingroup/loyalty do not comport.⁹

Conclusions and discussion

Previous research finds strong links between moral foundations and political attitudes (Clifford and Jerit, 2013; Graham et al., 2009; Koleva et al., 2012), and some experimental work provides evidence suggesting moral foundations exert a significant causal effect on such attitudes (Day et al., 2014; Feinberg and Willer, 2013). Smith et al. (2017), however, demonstrate that the causal effect of moral foundations on political predispositions is weaker than once thought, and their research opens the door to the idea that causality may in fact be reversed—that political predispositions actually cause moral foundations to move. The present research builds on this idea and presents evidence that suggests partisan and ideological considerations exert a significant effect on measures of moral foundations, and in addition to aligning with the work of Smith et al., the results also fit nicely with a large body of literature suggesting that group identity (especially partisanship) shapes a wide variety of political attitudes (Cohen, 2003; Lenz, 2009), including preferences on what are often considered the “building blocks” of behavior and choice—human values (Goren, 2005; Goren et al., 2009; McCann, 1997; Nelson et al., 1997). Furthermore, in line with a growing body of literature on negative partisanship (Abramowitz and Webster, 2016; Caruana et al., 2015; Medeiros and Noël, 2014), the results presented here suggest that outgroup cues may exert more influence on measures of moral foundations than ingroup cues.

What do these results mean for MFT? On the one hand, if one assumes that the MFQ produces the best possible measures of individual-level preferences on moral foundations, one might argue that because moral foundations respond predictably to changes in political context, they cannot provide one with broad and general guidelines for moral judgment. And, as such, moral foundations are no more or less foundational than partisanship, ideology, or any other important group identity. Ultimately, this interpretation of the results casts some doubt on the validity of MFT and suggests that future research on MFT ought to consider group identity, in conjunction with the five virtues, as important considerations when making moral judgments.

On the other hand, some readers may find this conclusion to be too bold, and their argument may be that the group-based cues do not cause individuals’ latent attitudes on moral foundations (i.e., a respondent’s true attitude on the harm/care foundation, for example) to change. Rather, the cues only cause measured attitudes to change. In this way, the problem that this study uncovers might not have to be dealt with at a theoretical level. Instead, it uncovers a problem with current measurement practices. More specifically, the present research shows that the MFQ produces context-specific measures.

Why would the embedded cues cause these measures, but not actual latent attitudes, to move? In short, respondents exposed to treatment conditions may be expressing their dislike for the outgroup when they are rating outgroup-endorsed moral foundations, and the differences between respondents in the treatment and control conditions are due to the expression of negative affect given by those in the treatment groups (and not an actual change in respondents’ latent attitudes regarding the foundations in question). In other words, respondents in the treatment conditions are engaging in expressive partisanship (see Huddy et al., 2015). As previously stated, if this is the case, the results of the present study do not indicate a fundamental flaw in MFT, they simply indicate a fundamental flaw in the measures.¹⁰

Thinking about a particular respondent’s score on the harm/care foundation (the measured attitude, or m in equation 1 below), for example, as a function of the latent attitude (i.e., their true attitude, denoted t in the equation below) and measurement error (e)

m_{i} = t_{i} + e_{i}

(1)

it may be the case that the slight change in context brought about by the experimental prime can be thought of as a non-random (i.e. systematic) part of the measurement error. In this way, measurement error is broken down into systematic error (s in equation 2) and random error (r)

m_{i} = t_{i} + \underset{\begin{matrix} e_{i} \end{matrix}}{\underset{︸}{s_{i} + r_{i}}}

(2)

Though some might argue that this contextual component of the measurement error (s above) is brought on by a small change in context and does not last forever, it is worth noting that, in a polarized political environment, candidates and elected officials often mention members of the outgroup to elicit negative emotion among supporters (Brader, 2005). This the case, when thinking about morality in a political context, MFQ items may often exhibit this sort of measurement bias.

With this in mind, scholars interested in improving the measurement characteristics of the MFQ might consider ways to minimize the impact of contextual changes on respondents’ choices. What types of measurement strategies might accomplish something like this? Ranking techniques, which are popular in values research (for a discussion, see Ciuk and Jacoby, 2015), may be of interest here for a variety of reasons. First, unlike rating techniques (e.g., the MFQ), rankings come with a built in anchor. That is, context may cause one to think the harm/care foundation, for example, is more important today than it was yesterday (which would be reflected in the traditional MFQ battery), but the individual’s set of rank-ordered preferences may be “anchored” by the most important foundations (say fairness/reciprocity in this case). This so, if the individual were to complete a rankings battery, s/he might pause and think “while I think harm/care is incredibly important, it is not more important than the most important foundation, fairness/reciprocity.” Here, because the individual’s preferences are anchored by the “most important” foundation (or, more accurately, the foundation that is one position ahead in the respondent’s rank-ordering, as well the foundation that is one position behind), the measure may be less susceptible to small changes in context.

To close, the present study adds to a growing body of research that suggests the causal relationship between moral foundations and political identity is more complicated than once thought. On the one hand, the results of the survey experiment might suggest that political identities cause people to adopt moral foundations, and that moral foundations are no more foundational than partisan or ideological identity. If this is true, and if moral foundations are to act as a root cause of political behavior, then MFT requires further theoretical development to account for these complex causal relationships. On the other hand, the results presented here might suggest simply that there are serious flaws with how moral foundations are often measured. More specifically, it may not be the case that the source cues actually alter latent attitudes; rather, they alter the measures. If this is the case, then future research ought to consider alternative measurement strategies that are not affected by small changes in context. Ultimately, the conclusions reached here call for greater theoretical development in MFT and an increased attention to measurement.

Supplemental Material

Ciuk_MFTSurveyExp_Revision2_SuppMaterial – Supplemental material for Assessing the contextual stability of moral foundations: Evidence from a survey experiment

Supplemental material, Ciuk_MFTSurveyExp_Revision2_SuppMaterial for Assessing the contextual stability of moral foundations: Evidence from a survey experiment by David J. Ciuk in Research & Politics

Footnotes

Acknowledgements

I would like to thank Dawn Ciuk, John Campbell, Joshua Darr, William Jacoby, Stephen Medvic, Jennifer Meyer, Joseph Patton, Matthew Schousen, and Berwood Yost for their helpful comments on this paper. All remaining errors are my own.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplementary materials

The supplementary files are available at .

The replication files are available at: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi%3A10.7910%2FDVN%2F4QCWML&version=DRAFT

Notes

ORCID iD

David J. Ciuk

Carnegie Corporation of New York Grant

This publication was made possible (in part) by a grant from the Carnegie Corporation of New York. The statements made and views expressed are solely the responsibility of the author.

References

Abramowitz

Webster

(2016) The rise of negative partisanship and the nationalization of U.S. elections in the 21st century. Electoral Studies 41: 12–22.

Abramson

Inglehart

(1995) Value Change in a Global Perspective. Ann Arbor, MI: University of Michigan Press.

Almond

Verba

(1963) The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton, NJ: Princeton University Press.

Berinsky

Huber

Lenz

, et al. (2012) Evaluating online labor markets for experimental research: Amazon.com’s Mechanical Turk. Political Analysis 20: 351–368.

Boudreau

MacKenzie

(2014) Informing the electorate? How party cues and policy information affect public opinion about initiatives. American Journal of Political Science 58: 48–62.

Brader

(2005) Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions. American Journal of Political Science 49: 388–405.

Campbell

Converse

Miller

, et al. (1960) The American Voter. New York: Wiley.

Caruana

McGregor

Stephenson

(2015) The power of the dark side: Negative partisanship and political behaviour in Canada. Canadian Journal of Political Science 2015: 771–789.

Ciuk

Jacoby

(2015) Checking for systematic value preferences using the method of triads. Political Psychology 36: 709–728.

10.

Ciuk

Yost

(2016) The effects of issue salience, elite influence, and policy content on public opinion. Political Communication 33: 328–345.

11.

Clifford

Jerit

(2013) How words do the word of politics: Moral foundations theory and the debate over stem cell research. Journal of Politics 75: 659–671.

12.

Cohen

(2003) Party over policy: The dominating impact of group influence on political beliefs. Journal of Personality and Social Psychology 85: 808–822.

13.

Converse

(1964) The nature of belief systems in mass publics. In: Apter

(ed.) Ideology and Discontent. New York: Free Press of Glencoe, pp. 206–261.

14.

Day

Fiske

Downing

, et al. (2014) Shifting liberal and conservative attitudes using moral foundations theory. Personality and Social Psychology Bulletin 40: 1559–1573.

15.

Feinberg

Willer

(2013) The moral roots of environmental attitudes. Pychological Science 24: 56–62.

16.

Goren

(2005) Party identification and core political values. American Journal of Political Science 49: 881–896.

17.

Goren

Federico

Kittilson

(2009) Source cues, partisan identities, and political value expression. American Journal of Political Science 53: 805–820.

18.

Graham

Haidt

Nosek

(2009) Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology 96: 1029–1046.

19.

Haidt

Graham

Joseph

(2009) Above and below left-right: Ideological narratives and moral foundations. Psychological Inquiry 20: 110–119.

20.

Huddy

Mason

Aarøe

(2015) Expressive partisanship: Campaign involvement, political emotion, and partisan identity. American Political Science Review 109: 1–17.

21.

Jacoby

(2006) Value choices and American public opinion. American Journal of Political Science 50: 706–723.

22.

Jayawickreme

DiStefano

(2012) How can we study heroism? Integrating persons, situations, and communities. Political Psychology 33: 165–178.

23.

Johnson

Iyer

Wojcik

Vaisey

, et al. (2014) Ideology-specific patterns of moral indifference predict intentions not to vote. Analyses of Social Issues and Public Policy 14: 61–77.

24.

Kahan

(2015) Climate-science communication and the measurement problem. Advances in Political Psychology 36: 1–43.

25.

Kahan

Peters

Dawson

, et al. (2017) Motivated numeracy and enlightened self-government. Behavioral Public Policy 1: 54–86.

26.

Koleva

Graham

Iyer

, et al. (2012) Tracing the threads: How five moral concerns (especially purity) help explain culture war attitudes. Journal of Research in Personality 46: 184–194.

27.

Lenz

(2009) Learning and opinion change, not priming: Reconsidering the priming hypothesis. American Journal of Political Science 53: 821–837.

28.

Malka

Lelkes

(2010) More than ideology: Conservative-liberal identity and receptivity to political cues. Social Justice Research 23: 156–188.

29.

Mason

(2015) ‘I disrespectfully agree’: The differential effects of partisan sorting on social and issue polarization. American Journal of Political Science 129: 128–145.

30.

McCann

(1997) Electoral choices and core value change: The 1992 presidential campaign. American Journal of Political Science 41: 564–583.

31.

Medeiros

Noël

(2014) The forgotten side of partisanship: Negative party identification in four Anglo-American democracies. Comparative Political Studies 47: 1022–1046.

32.

Mullinix

Leeper

Druckmana

, et al. (2015) The generalizability of survey experiments. Journal of Experimental Political Science 2: 109–138.

33.

Nelson

Clawson

Oxley

(1997) Media framing of a civil liberties conflict and its effect on tolerance. American Political Science Review 91: 567–583.

34.

Rokeach

(1973) The Nature of Human Values. New York: Free Press.

35.

Rottman

Keleman

Young

(2014) Tainting the soul: Purity concerns predict moral judgments of suicide. Cognition 130: 217–226.

36.

Smith

Aquino

Koleva

, et al. (2014) The moral ties that bind… even to out-groups: The interactive effect of moral identity and the binding moral foundations. Psychological Science 25: 1554–1562.

37.

Smith

Alford

Hibbing

, et al. (2017) Intuitive ethics and political orientations: Testing moral foundations as a theory of political ideology. American Journal of Political Science 61: 424–437.

38.

Watkins

Aitken

Mather

(2016) Conscientious consumers: A relationship between moral foundations, political orientation and sustainable consumption. Journal of Cleaner Production 134A: 137–146.

39.

Winterich

Zhang

Mittal

(2012) How political identity and charity positioning increase donations: Insights from moral foundations theory. International Journal of Research in Marketing 2 9: 346–354.

40.

Zellner

(1962) An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association 57: 348–368.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.27 MB