- Email: [email protected]
The Moral Foundations taxonomy: Structural validity and relation to political ideology in Sweden
Personality and Individual Differences 76 (2015) 28–32
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
The Moral Foundations taxonomy: Structural validity and relation to political ideology in Sweden Artur Nilsson ⇑, Arvid Erlandsson 1 Department of Psychology, Lund University, Lund, Sweden
a r t i c l e
i n f o
Article history: Received 18 August 2014 Received in revised form 21 November 2014 Accepted 22 November 2014
Keywords: Moral Foundations Morality Political ideology Political attitudes Resistance to change Preference for equality System justification
a b s t r a c t Although Moral Foundations Theory claims that the foundations of morality are universal, there are still few studies addressing it through non-English measures. In the current research, 540 persons filled out a Swedish translation of the Moral Foundations Questionnaire, and 332 of them filled out political attitude measures. Confirmatory factor analyses suggested that the fit of the five-factor model was better than alternative models but not optimal, replicating previous findings. Concerns with fairness and prevention of harm predicted political identity leftward, mediated mainly by preference for equality, and concerns with loyalty, authority, and sanctity predicted political identity rightward, mediated mainly by resistance to change and system justification, as hypothesized. Fairness and authority concerns were the best predictors of political ideology. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Moral proclivities have recently been getting increased attention as individual difference constructs, within the domain of ‘‘characteristic adaptations’’ (McAdams & Pals, 2006) or ‘‘worldviews’’ (Nilsson, 2014). Moral Foundations Theory (MFT; Graham, Haidt, & Nosek, 2009; Graham et al., 2013; Haidt & Joseph, 2004) is a particularly influential approach, based on the notion that moral psychology has traditionally been biased by a liberal, individualistic, Western perspective, conceiving of morality in terms of protection of the rights and welfare of individuals while ignoring the fact that people in non-Western or non-liberal contexts – perhaps the majority of humanity – tend to intuitively moralize also concerns with protecting the integrity of groups, social systems, and souls. According to this theory, intuitive moral judgments rest upon at least five distinct foundations: (1) care/harm, which involves caring for others and avoiding to inflict harm or suffering upon them, (2) fairness/cheating, which involves concerns about fairness, equality, justice, and the avoidance of cheating others, (3) loyalty/betrayal, which involves loyalty and other obligations to your in-group, and the avoidance of betrayal, (4) authority/ ⇑ Corresponding author at: Gulsångarvägen 11, 24735 Södra Sandby, Sweden. Tel.: +46 736399011; fax: +46 46 2224209. E-mail addresses: [email protected] (A. Nilsson), [email protected]. se (A. Erlandsson). 1 Fax: +46 46 2224209. http://dx.doi.org/10.1016/j.paid.2014.11.049 0191-8869/Ó 2014 Elsevier Ltd. All rights reserved.
subversion, which involves conformity with the social order, through obedience, respect for authority, and traditional role fulfilment, and protection of the social order from subversion, and (5) sanctity/degradation, which involves concerns about physical and spiritual purity, including chastity, wholesomeness, suppression of desires, and the avoidance of contamination and degradation. Graham et al. (2009) suggested that the former two foundations represent an ‘‘individualizing’’ form of morality, focused on the rights and welfare of individuals, and associated with political liberalism, whereas the latter three represent a ‘‘binding’’ form of morality, focused on strengthening groups and institutions and suppressing selfishness by binding individuals into roles and duties, and associated with political conservatism. The hypothesized association between moral foundations and political ideology has generated ample empirical support (Davies, Sibley, & Liu, 2014; Graham et al., 2011; Kim, Kang, & Yun, 2012; van Leeuwen & Park, 2009). Another key tenet of Moral Foundations Theory is that the postulated foundations have a universal evolutionary basis, forming a ‘‘first draft’’ of a person’s moral ‘‘taste buds’’, which is organized prior to experience but also, to some extent, modifiable by experience; it was originally based partly upon reviews of research on morality across cultures, from an anthropological perspective, and on phylogenetic precursors of human morality in primates, from an evolutionary perspective (Graham et al., 2013; Haidt & Joseph, 2004). The presumed universality of the
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
moral foundations implies that it is crucial to investigate the extent to which the hypothesized factorial structure and relation to political ideology can be recovered in different cultures and languages. Indeed, Graham et al. (2013) ‘‘see MFT’s current and future development being one of method-theory co-evolution, with theoretical constructs inspiring the creation of new ways to measure them, and data from the measurements guiding development of the theory’’. They are currently evaluating the potential inclusion of additional foundations within the moral foundations framework, but the original five-factor model has generated most research so far. The five-factor model is today being cross-culturally evaluated through the Moral Foundations Questionnaire. This instrument was evaluated by Graham et al. (2011), who ran confirmatory factor analyses on data from samples of participants from many different parts of the world, who completed the English version of this scale through the yourmorals.org website. They found that the five-factor model had better fit than the individualizing-binding two-factor model and a three-factor model based on the Shweder, Much, Mahapatra, and Park (1997) distinction between ethics of autonomy (i.e. harm and fairness), community (i.e. loyalty and authority), and divinity (i.e. sanctity). Studies from Korea (Kim et al., 2012), Italy (Bobbio, Nencini, & Sarrica, 2011), Germany (Bowman, 2010), and New Zealand (Davies et al., 2014) have shown similar results. But model fit has typically fallen short of conventional criteria of fit, especially in the case of non-English versions of the scale. Our purpose here is twofold. First, we contribute to the crosscultural evaluation of moral foundations theory, by replicating the confirmatory factor analyses from Graham et al. (2011) and the correlation between moral intuitions and self-identified political ideology from ‘‘left’’ to ‘‘right’’ (cf. Van Leeuwen & Park, 2009), on a Swedish translation of the moral foundations questionnaire. Although Sweden is certainly at the liberal, Western end of the cultural spectrum, it is very different from, for example, the United States, in the sense that it has a long history of social-democrat rule and a political discourse defined mainly by opposition between social democrats and socialists (the left-wing) and social-liberals and libertarians (the right-wing), rather than liberalism versus conservatism. Sweden is perhaps the most secular and liberal country in the world, ranking as the most extreme of the advanced post-industrial democracies on the Inglehart and Welzel (2010) global cultural map. Indeed, one might wonder whether loyalty, authority, and sanctity matter at all to the left–right dimension in Sweden, which is defined mainly by attitudes to equality and free-market support (Cochrane & Nevitte, 2009; Nilsson & Jost, 2012). Second, we test a new perspective on how moral intuitions shape political ideology, drawing on an influential framework introduced by Jost, Glaser, Kruglanski, and Sulloway (2003), which divides political attitudes into resistance to change and preference for equality and postulates the need to justify the current system as a key force structuring the left–right continuum. We hypothesize that the effects of the individualizing foundations on left–right political identity are mediated primarily by preference for equality, because these foundations are concerned with the rights and welfare of individuals, and we hypothesize that the effects of the binding foundations on political identity are mediated primarily by resistance to change and system justification, because they are concerned with the protection of social, cultural, and religious systems. Similar to this, Federico, Weber, Ergun, and Hunt (2013) found that harm and fairness are aligned with social dominance orientation (SDO) and competitive-jungle beliefs and that loyalty, authority, and sanctity are aligned with right-wing authoritarianism (RWA) and dangerous-world beliefs. But SDO conflates preference for equality with group-based dominance and RWA conflates
29
resistance to change with patriotism, religiosity, and deference for authority. By focusing on resistance to change, preference for equality, and system justification per se we hope to further clarify the relationship between morality and political ideology. 2. Method 2.1. Participants Participants were 540 adults (mean age 24.0 years, SD = 4.6 years; 63.2% women) studying the social sciences, humanities, law, or engineering. A subset of 332 persons completed political attitude measures (mean age 23.7 years, SD = 3.5 years, 69.6% women). 2.2. Material 2.2.1. The Moral Foundations Questionnaire (Graham et al., 2009) The Moral Foundations Questionnaire was translated into Swedish by a researcher and back-translated into English by a professional translator (freely available at www.moralfoundations. org). It measures each of the foundations with three items assessing the perceived relevance of moral concerns and three items assessing agreement with moral judgments. Participants respond to the relevance items on a Likert scale ranging from 0 (Not at all relevant) to 5 (Extremely relevant) and to the judgment items on a Likert scale ranging from 0 (Strongly disagree) to 5 (Strongly agree). The latter response bar was substituted for a Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree) for 207 of the participants; these responses were linearly transformed onto the 0–5 scale. Sample items include ‘‘Justice is the most important requirement for a society’’ (fairness) and ‘‘People should not do things that are disgusting, even if no one is harmed’’ (sanctity). The lowest Cronbach alpha reliability estimates were obtained for harm, a = .57, which had two corrected item-total correlations lower than .3, and loyalty, a = .58, for which all three judgment items had corrected item-total correlations lower than .3 with the entire scale and lower than .2 with each other. Fairness, authority, and sanctity all had a = .66 and one corrected item-total correlation lower than .3. 2.2.2. Political ideology We measured political identity (‘‘Where would you place yourself on the following scale of political orientation?’’) on a Likert scale ranging from 1 (Extremely left-wing) through 5 (Neither/ neutral) to 9 (Extremely right-wing) (N = 540). Because few participants chose the extreme response options 1 (N = 6) and 9 (N = 2), we collapsed the 1 and 2 (N = 69) and the 8 and 9 response options (N = 9). We measured system justification with a Swedish translation (Nilsson & Jost, 2012) of the eight-item (e.g. ‘‘Society is set up so that people usually get what they deserve’’) system justification scale (Kay & Jost, 2003), which includes two reversed items, a = .83. We measured resistance to change with eleven items (e.g. ‘‘If you start changing things very much, you often end up making them worse’’), including two reversed items, a = .75, and preference for equality with fifteen items (e.g. ‘‘Prosperous nations have a moral obligation to share some of their wealth with poor nations’’), including six reversed items, a = .86, on the basis of previous attempts to isolate these components of ideology (Nilsson & Jost, 2012). 2.3. Statistical procedure Structural equation modeling was run in AMOS 20.0. All calculations were based upon the covariance matrix and the maximum
30
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
.57 .66 .49 .45 .30 .26
EMO WEA CRU COM ANI KIL
.63 .54 .48 .62 .54 .36
TRE UNF RIT FAI JUS RIC
Loyalty
.55 .67 .72 .27 .35 .18
LUV BET LOY HIS FAM TEA
Authority
.78 .54 .56 .63 .37 .43
RES TRA CHA KID SEX SOL
.74 .64 .30 .55 .51 .40
DEC DIS GOD HDG UNN CHA
Harm
.70 .25
Fairness
-.06 .13
-.08
.18
.76
-.01 .79
.87
Sanctity
.56 .62 .50 .47 .32 .29
EMO WEA CRU COM ANI KIL
.64 .54 .47 .63 .55 .35
TRE UNF RIT FAI JUS RIC
Loyalty
.54 .63 .69 .28 .34 .16
LUV BET LOY HIS FAM TEA
Authority
.73 .51 .53 .59 .32 .39
RES TRA CHA KID SEX SOL
.64 .61 .28 .54 .49 .38
DEC DIS GOD HDG UNN CHA
Harm
.92
Individualizing intuitions .76
Fairness
.09
.84
Binding intuitions
.87
.90
Sanctity
Fig. 1. Moral Foundations Questionnaire confirmatory factor analysis of the five-factor model and the hierarchical two-factor model. Item abbreviations are drawn from Graham et al. (2011).
5
likelihood method. We report v2 tests of goodness of fit, the Comparative Fit Index (CFI), which estimates fit relative to a baseline model, and the Root Mean Square Error of Approximation (RMSEA), which estimates parsimony-adjusted fit in relation to a perfect model. According to widespread rules of thumb a CFI above .95 and an RMSEA below .06 (Hu & Bentler, 1999) indicates adequate model fit. Three models were included in the confirmatory factor analyses: (1) the five-factor model, with one latent factor for each moral foundation, the respective scale items as manifest variables, and estimated relations between all latent factors, (2) the hierarchical two-factor model, substituting the relations between the latent factors for two related superordinate factors (Fig. 1), and (3) a hierarchical three-factor model, treating sanctity as a separate superordinate factor and estimating relations between the three superordinate factors. To model the effect of moral foundations on political ideology, we constructed a path model, using only manifest variables in order to keep model complexity manageable. We divided it into three levels: (1) moral foundations, which predicted. (2) resistance to change, preference for equality, and system justification, which predicted. (3) left–right political identity (Fig. 3). All relations between error terms of variables within each level were estimated.
3. Results 3.1. Structural validity The five-factor model, v2 (396) = 1517.8, CFI = .679, RMSEA = .072 (95% CI [.069, .076]), was significantly better, Dv2 (4) = 23.0, p < .001, than the hierarchical two-factor model, v2 (400) = 1540.8, CFI = .674, RMSEA = .073 (95% CI [.069, .077]). The three-factor model, v2(399) = 1538.3, CFI = .674, RMSEA = .073
4 Harm
3
Fairness Loyalty
2
Authority Sanctity
1
0 1 Strong left
2
3
Moderate left
Slight left
4 Neutral neither
5 Slight right
6 Moderate right
7 Strong right
Self-reported political identity Fig. 2. Moral foundations across political identity from left to right.
(95% CI [.069, .077]), was not significantly better than the twofactor model, Dv2(1) = 2.5, p = .11. These results are consistent with Graham et al. (2011) findings, although our fit estimates are lower in absolute terms. The five-factor and two-factor models are displayed in Fig. 1. In order to investigate the sources of misfit, we looked at potential model modifications post hoc. When the error terms of items within the same factor were allowed to covary, the fit could be increased substantially, Dv2(24) = 374.5, p < .001, to v2(372) = 1143.3, CFI = .779, RMSEA = .062 (95% CI [.058, .066]). These item covariations appeared to stem not just from superficial similarities in item formulation but also from similarities in item content, possibly indicating that the taxonomy is oversimplified: Items relating to disgust sensitivity, religiosity (sanctity), respect for authority, conformity with traditions (authority), and socio-political justice
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
Harm
.33*** .49***
Fairness
31
effect of authority concerns on political identity was partly mediated by low preference for equality. The moral foundations accounted for 33.8% of the variance of system justification, 46.4% of the variance of resistance to change, and 63.6% of the variance of preference for equality.
Preference for equality -.48***
-.23*** -.20***
Loyalty
Authority
.12** .19*** -.24*** .36*** .36***
Sanctity
System justification
.38***
Political identity
.09**
Resistance to change
.19***
Fig. 3. Preference for equality, resistance to change, and system justification as mediators of the relationship between moral foundations and political identity. Covariances between residual terms and non-significant estimates (p < .05) are not displayed. ⁄⁄⁄p < .001. ⁄⁄p < .01. ⁄p < .05.
(fairness) covaried especially strongly, respectively.2 We tried, moreover, to remove items with low factor loadings, which allowed us to further improve CFI (.84 with seven items removed) but not RMSEA. 3.2. Relation to political ideology Concerns with fairness, r = .44, and prevention of harm, r = .25, decreased, and concerns with loyalty, r = .29, authority, r = .50, and sanctity, r = .29 (p < .001), increased continuously as political identity moved from left to right, as shown in Fig. 2. When entered into a regression model, the five foundations accounted for a total of 38.5% of the variance of political identity. Fairness and authority concerns were the strongest predictors; bharm = .09, p = .046; bfairness = .33, p < .001; bloyalty = .10, p = .045; bauthority = .35, p < .001; bsanctity = .03, p = .57. When the five foundations were collapsed into two superordinate categories (aindividualizing = .73; abinding = .82), these categories accounted for 34.6% of the variance of political identity; bindividualizing = .39 and bbinding = .42 (p < .001). Harm and fairness concerns correlated positively with preference for equality (rharm = .47; rfairness = .62), and negatively with resistance to change (rharm = .25; rfairness = .39) and system justification (rharm = .26; rfairness = .37). Loyalty, authority, and sanctity concerns correlated positively with resistance to change (rloyalty = .32; rauthority = .53; rsanctity = .40) and system justification (rloyalty = .22; rauthority = .46; rsanctity = .29), and negatively with preference for equality (rloyalty = .18; rauthority = .50; rsanctity = .30; p < .001). The fit of the path model, which is shown in Fig. 3, was adequate, v2(5) = 9.70, p = .084, CFI = .998, RMSEA = .042 (95% CI [.000, .081]). Post hoc tests revealed that only harm had a significant independent effect on political identity, k = .08, p = .012, which increased fit, Dv2(1) = 6.26, p = .001, to v2(1) = 3.44, p = .49, CFI = 1.00, RMSEA < .001 (95% CI [.000, .061]). As predicted, most of the effect of harm and fairness concerns on political identity was mediated by preference for equality and most of the effect of loyalty, authority, and sanctity concerns on political identity was mediated by resistance to change and system justification. But the effect of fairness concerns on political identity was partly mediated also by low system justification and resistance to change and the 2 ‘‘Whether or not someone acted in a way that God would approve of’’ did not correlate above .2 with any item addressing disgust and ‘‘Whether or not someone conformed to the traditions of society’’ did not correlate above .2 with any item addressing respect for authority.
4. Discussion The current research provides further cross-cultural support for the central claims of Moral Foundations Theory. The five-factor model was indeed significantly better than the alternative twoand three-factor models, replicating previous findings (Graham et al., 2009, 2011). The five-factor model also had incremental validity, when compared with the two-factor model in terms of the prediction of political ideology. The advantage of the five-factor model over the two-factor model in terms of fit was, however, very small, similar to previous findings (Bobbio et al., 2011; Davies et al., 2014; Graham et al., 2011). This suggests that, although the five foundations are distinct, it is useful to think of them in hierarchical terms, with two superordinate factors: individualizing or liberal intuitions, subsuming harm and fairness as facets, and binding or conservative intuitions, subsuming loyalty, authority, and sanctity as facets. As in the case of personality trait models, the superordinate factors bring coherence and parsimony to the theoretical system while the facets may afford greater predictive utility in specific domains of inquiry (Costa & McCrae, 1995). The results also supported the theorized links between all five moral foundations and political ideology in the Swedish context, despite the low levels of conservatism and religiosity in Sweden. The strength of the links seems, however, to be affected by cultural factors. When comparing the current results with results from other countries (Davies et al., 2014; Graham et al., 2011; Kim et al., 2012), concern with fairness was a particularly strong predictor of political identity, which can be explained in terms of the centrality of attitudes to equality within the Swedish division between left and right (Cochrane & Nevitte, 2009), and concern with sanctity was a particularly weak predictor of political identity, which can be explained in terms of the low degree of religiosity across the ideological board in Sweden (Inglehart & Welzel, 2010). Although concern with authority was also a strong predictor of political identity in the current study, this result was similar to those obtained in previous studies. Moreover, the current results provide new insight into the connection between morality and political ideology. As hypothesized, the effects of the individualizing foundations on political identity were mediated primarily by preference for equality and the effects of the binding foundations on political identity were mediated primarily by resistance to change and system justification. There were, however, a few findings qualifying this picture. First, the effect of authority concerns on political identity was substantially mediated by both resistance to change and system justification, which makes sense insofar as it involves both respect for authority, and thus conformity with the system, and conformity with societal traditions and norms. But it was also partly mediated by low preference for equality, which was unexpected, but is intelligible, nevertheless, in terms of the association between respect for authority and justification of current hierarchical arrangements. Second, the effect of fairness concerns on political identity was partly mediated not just by preference for equality, but also by low system justification and resistance to change, suggesting that the system and the traditions were perceived as violating people’s rights. Third, the effect of sanctity concerns on political identity was mediated only by resistance to change, which may be explained in terms of the fact that sanctity incorporates preservation of principles that serve to uphold purity and decency rather
32
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
than conformity with the system per se, which is highly secular in the case of Sweden. It is plausible that the effect of sanctity on political identity is mediated by system justification in non-secular countries. In countries where the left–right dimension is defined by the opposition between liberalism and conservatism, the effect of the binding intuitions on political identity should, furthermore, be more strongly mediated by resistance to change than it was in the current study. Despite providing support for the hypothesized relations between moral intuitions and political ideology, the results do, however, also add to a literature suggesting that the moral foundations questionnaire is not psychometrically optimal. The amount of misfit was rather large in the current study, and the reliabilities were low, similar to other studies using non-English versions of this questionnaire (cf. Bobbio et al., 2011; Bowman, 2010; Kim et al., 2012). Although lower fit can be expected when a measure of ideology is translated, and common fit criteria may be too strict when applied to complex domains of psychological inquiry (Hopwood & Donnelan, 2010), misfit may also indicate that the taxonomy is oversimplified. Some of the items measuring loyalty concerns in particular did not work in our Swedish sample, and we noted several content-based item covariances within the moral foundation factors, which were responsible for some of the misfit. Further research is needed to determine the extent to which these findings can be replicated in other languages or attributed to specific features of the Swedish language or socio-political system. It is, for example, not surprising that some of the loyalty items did not work, given that loyalty with in-groups is a contentious issue in Sweden, associated with nationalism, racism, and fascism, which may make Swedes sensitive to what group is implicitly intended. It is also possible that religious purity and disgust are not associated in the same way in the most secular parts of the world, including Northern Europe and parts of Asia (Inglehart & Welzel, 2010), as they are in the majority of countries in the world. In conclusion, although Moral Foundations Theory offers a powerful framework for understanding variations in moral and political proclivities, more work is needed to test its theoretical claims, evaluate and refine the taxonomy, and develop reliable, structurally valid measurement across the full spectrum of cultures and languages.
References Bobbio, A., Nencini, A., & Sarrica, M. (2011). Il Moral Foundation Questionnaire: Analisi della struttura fattoriale della versione italiana [The Moral Foundations Questionnaire: Analysis of the factor structure of the Italian version]. Giornale di Psicologia, 5, 7–18.
Bowman, N. (2010). German translation of the Moral Foundations Questionnaire – Some preliminary results..Retrieved from 2014 11 21 Cochrane, C., & Nevitte, N. (2009). Policy disagreement in advanced industrial states: The content and structure of left/right opinions. In Y. Esmer, H.-D. Klingemann, & Puranen B. (Eds.). Religion, democratic values, and political conflict: Festschrift in Honor of Thorleif Petterson (pp. 119–145). Uppsala, Sweden: Acta Universitatis Upsaliensis. Costa, P. T., & McCrae, R. R. (1995). Domains and facets: Hierarchical personality assessment using the revised NEO personality inventory. Journal of Personality Assessment, 64, 21–50. Davies, C. L., Sibley, C. G., & Liu, J. H. (2014). Confirmatory factor analysis of the Moral Foundations Questionnaire: Independent scale validation in a New Zealand sample. Social Psychology. http://dx.doi.org/10.1027/1864-9335/ a000201 (Advance online publication). Federico, C., Weber, C., Ergun, D., & Hunt, C. (2013). Mapping the connections between politics and morality: The multiple sociopolitical orientations involved in moral intuition. Political Psychology, 34(4), 589–610. Graham, J., Haidt, J., & Nosek, B. A. (2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96(5), 1029–1046. Graham, J., Nosek, B. A., Haidt, J., Iyer, R., Koleva, S., & Ditto, P. H. (2011). Mapping the moral domain. Journal of Personality and Social Psychology, 101(2), 366–385. Graham, J., Haidt, J., Koleva, S., Motyl, M., Iyer, R., Wojcik, S. P., et al. (2013). Moral Foundations Theory: The pragmatic validity of moral pluralism. Advances in Experimental Social Psychology, 47, 55–130. Haidt, J., & Joseph, C. (2004). Intuitive ethics: How innately prepared intuitions generate culturally variable virtues. Daedalus: Special Issue on Human Nature, 133(4), 55–66. Hopwood, C. J., & Donnelan, M. B. (2010). How should internal structure of personality inventories be evaluated? Personality and Social Psychology Review, 14, 244–254. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Inglehart, R., & Welzel, C. (2010). Changing mass priorities: The link between modernization and democracy. Perspectives on Politics, 8(2), 551–567. http:// dx.doi.org/10.1017/S1537592710001258. Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychological Bulletin, 129(3), 339–375. Kay, A. C., & Jost, J. T. (2003). Complementary justice: Effects of ‘‘poor but happy’’ and ‘‘poor but honest’’ stereotype exemplars on system justification and implicit activation of the justice motive. Journal of Personality and Social Psychology, 85, 823–837. Kim, K. R., Kang, J.-S., & Yun, S. (2012). Moral intuitions and political orientation: Similarities and differences between Korea and the United States. Psychological Reports: Sociocultural Issues in Psychology, 111, 173–185. McAdams, D. P., & Pals, J. L. (2006). A new Big Five: Fundamental principles for an integrative science of personality. American Psychologist, 61(3), 204–217. Nilsson, A. (2014). Personality psychology as the integrative study of traits and worldviews. New Ideas in Psychology, 32, 18–32. Nilsson, A., & Jost, J. T. (2012). Revisiting Tomkins’ Polarity Theory: How Humanism and Normativism shape political ideology. In Paper presented at the 35th annual scientific meeting of the International Society of Political Psychology, Chicago, IL. Shweder, R. A., Much, N. C., Mahapatra, M., & Park, L. (1997). The ‘‘big three’’ of morality (autonomy, community, and divinity), and the ‘‘big three’’ explanations of suffering. In A. Brandt & P. Rozin (Eds.), Morality and health (pp. 119–169). New York: Routledge. Van Leeuwen, F., & Park, J. H. (2009). Perceptions of social dangers, moral foundations, and political orientation. Personality and Individual Differences, 47, 169–173.
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
The Moral Foundations taxonomy: Structural validity and relation to political ideology in Sweden Artur Nilsson ⇑, Arvid Erlandsson 1 Department of Psychology, Lund University, Lund, Sweden
a r t i c l e
i n f o
Article history: Received 18 August 2014 Received in revised form 21 November 2014 Accepted 22 November 2014
Keywords: Moral Foundations Morality Political ideology Political attitudes Resistance to change Preference for equality System justification
a b s t r a c t Although Moral Foundations Theory claims that the foundations of morality are universal, there are still few studies addressing it through non-English measures. In the current research, 540 persons filled out a Swedish translation of the Moral Foundations Questionnaire, and 332 of them filled out political attitude measures. Confirmatory factor analyses suggested that the fit of the five-factor model was better than alternative models but not optimal, replicating previous findings. Concerns with fairness and prevention of harm predicted political identity leftward, mediated mainly by preference for equality, and concerns with loyalty, authority, and sanctity predicted political identity rightward, mediated mainly by resistance to change and system justification, as hypothesized. Fairness and authority concerns were the best predictors of political ideology. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Moral proclivities have recently been getting increased attention as individual difference constructs, within the domain of ‘‘characteristic adaptations’’ (McAdams & Pals, 2006) or ‘‘worldviews’’ (Nilsson, 2014). Moral Foundations Theory (MFT; Graham, Haidt, & Nosek, 2009; Graham et al., 2013; Haidt & Joseph, 2004) is a particularly influential approach, based on the notion that moral psychology has traditionally been biased by a liberal, individualistic, Western perspective, conceiving of morality in terms of protection of the rights and welfare of individuals while ignoring the fact that people in non-Western or non-liberal contexts – perhaps the majority of humanity – tend to intuitively moralize also concerns with protecting the integrity of groups, social systems, and souls. According to this theory, intuitive moral judgments rest upon at least five distinct foundations: (1) care/harm, which involves caring for others and avoiding to inflict harm or suffering upon them, (2) fairness/cheating, which involves concerns about fairness, equality, justice, and the avoidance of cheating others, (3) loyalty/betrayal, which involves loyalty and other obligations to your in-group, and the avoidance of betrayal, (4) authority/ ⇑ Corresponding author at: Gulsångarvägen 11, 24735 Södra Sandby, Sweden. Tel.: +46 736399011; fax: +46 46 2224209. E-mail addresses: [email protected] (A. Nilsson), [email protected]. se (A. Erlandsson). 1 Fax: +46 46 2224209. http://dx.doi.org/10.1016/j.paid.2014.11.049 0191-8869/Ó 2014 Elsevier Ltd. All rights reserved.
subversion, which involves conformity with the social order, through obedience, respect for authority, and traditional role fulfilment, and protection of the social order from subversion, and (5) sanctity/degradation, which involves concerns about physical and spiritual purity, including chastity, wholesomeness, suppression of desires, and the avoidance of contamination and degradation. Graham et al. (2009) suggested that the former two foundations represent an ‘‘individualizing’’ form of morality, focused on the rights and welfare of individuals, and associated with political liberalism, whereas the latter three represent a ‘‘binding’’ form of morality, focused on strengthening groups and institutions and suppressing selfishness by binding individuals into roles and duties, and associated with political conservatism. The hypothesized association between moral foundations and political ideology has generated ample empirical support (Davies, Sibley, & Liu, 2014; Graham et al., 2011; Kim, Kang, & Yun, 2012; van Leeuwen & Park, 2009). Another key tenet of Moral Foundations Theory is that the postulated foundations have a universal evolutionary basis, forming a ‘‘first draft’’ of a person’s moral ‘‘taste buds’’, which is organized prior to experience but also, to some extent, modifiable by experience; it was originally based partly upon reviews of research on morality across cultures, from an anthropological perspective, and on phylogenetic precursors of human morality in primates, from an evolutionary perspective (Graham et al., 2013; Haidt & Joseph, 2004). The presumed universality of the
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
moral foundations implies that it is crucial to investigate the extent to which the hypothesized factorial structure and relation to political ideology can be recovered in different cultures and languages. Indeed, Graham et al. (2013) ‘‘see MFT’s current and future development being one of method-theory co-evolution, with theoretical constructs inspiring the creation of new ways to measure them, and data from the measurements guiding development of the theory’’. They are currently evaluating the potential inclusion of additional foundations within the moral foundations framework, but the original five-factor model has generated most research so far. The five-factor model is today being cross-culturally evaluated through the Moral Foundations Questionnaire. This instrument was evaluated by Graham et al. (2011), who ran confirmatory factor analyses on data from samples of participants from many different parts of the world, who completed the English version of this scale through the yourmorals.org website. They found that the five-factor model had better fit than the individualizing-binding two-factor model and a three-factor model based on the Shweder, Much, Mahapatra, and Park (1997) distinction between ethics of autonomy (i.e. harm and fairness), community (i.e. loyalty and authority), and divinity (i.e. sanctity). Studies from Korea (Kim et al., 2012), Italy (Bobbio, Nencini, & Sarrica, 2011), Germany (Bowman, 2010), and New Zealand (Davies et al., 2014) have shown similar results. But model fit has typically fallen short of conventional criteria of fit, especially in the case of non-English versions of the scale. Our purpose here is twofold. First, we contribute to the crosscultural evaluation of moral foundations theory, by replicating the confirmatory factor analyses from Graham et al. (2011) and the correlation between moral intuitions and self-identified political ideology from ‘‘left’’ to ‘‘right’’ (cf. Van Leeuwen & Park, 2009), on a Swedish translation of the moral foundations questionnaire. Although Sweden is certainly at the liberal, Western end of the cultural spectrum, it is very different from, for example, the United States, in the sense that it has a long history of social-democrat rule and a political discourse defined mainly by opposition between social democrats and socialists (the left-wing) and social-liberals and libertarians (the right-wing), rather than liberalism versus conservatism. Sweden is perhaps the most secular and liberal country in the world, ranking as the most extreme of the advanced post-industrial democracies on the Inglehart and Welzel (2010) global cultural map. Indeed, one might wonder whether loyalty, authority, and sanctity matter at all to the left–right dimension in Sweden, which is defined mainly by attitudes to equality and free-market support (Cochrane & Nevitte, 2009; Nilsson & Jost, 2012). Second, we test a new perspective on how moral intuitions shape political ideology, drawing on an influential framework introduced by Jost, Glaser, Kruglanski, and Sulloway (2003), which divides political attitudes into resistance to change and preference for equality and postulates the need to justify the current system as a key force structuring the left–right continuum. We hypothesize that the effects of the individualizing foundations on left–right political identity are mediated primarily by preference for equality, because these foundations are concerned with the rights and welfare of individuals, and we hypothesize that the effects of the binding foundations on political identity are mediated primarily by resistance to change and system justification, because they are concerned with the protection of social, cultural, and religious systems. Similar to this, Federico, Weber, Ergun, and Hunt (2013) found that harm and fairness are aligned with social dominance orientation (SDO) and competitive-jungle beliefs and that loyalty, authority, and sanctity are aligned with right-wing authoritarianism (RWA) and dangerous-world beliefs. But SDO conflates preference for equality with group-based dominance and RWA conflates
29
resistance to change with patriotism, religiosity, and deference for authority. By focusing on resistance to change, preference for equality, and system justification per se we hope to further clarify the relationship between morality and political ideology. 2. Method 2.1. Participants Participants were 540 adults (mean age 24.0 years, SD = 4.6 years; 63.2% women) studying the social sciences, humanities, law, or engineering. A subset of 332 persons completed political attitude measures (mean age 23.7 years, SD = 3.5 years, 69.6% women). 2.2. Material 2.2.1. The Moral Foundations Questionnaire (Graham et al., 2009) The Moral Foundations Questionnaire was translated into Swedish by a researcher and back-translated into English by a professional translator (freely available at www.moralfoundations. org). It measures each of the foundations with three items assessing the perceived relevance of moral concerns and three items assessing agreement with moral judgments. Participants respond to the relevance items on a Likert scale ranging from 0 (Not at all relevant) to 5 (Extremely relevant) and to the judgment items on a Likert scale ranging from 0 (Strongly disagree) to 5 (Strongly agree). The latter response bar was substituted for a Likert scale ranging from 1 (Strongly disagree) to 7 (Strongly agree) for 207 of the participants; these responses were linearly transformed onto the 0–5 scale. Sample items include ‘‘Justice is the most important requirement for a society’’ (fairness) and ‘‘People should not do things that are disgusting, even if no one is harmed’’ (sanctity). The lowest Cronbach alpha reliability estimates were obtained for harm, a = .57, which had two corrected item-total correlations lower than .3, and loyalty, a = .58, for which all three judgment items had corrected item-total correlations lower than .3 with the entire scale and lower than .2 with each other. Fairness, authority, and sanctity all had a = .66 and one corrected item-total correlation lower than .3. 2.2.2. Political ideology We measured political identity (‘‘Where would you place yourself on the following scale of political orientation?’’) on a Likert scale ranging from 1 (Extremely left-wing) through 5 (Neither/ neutral) to 9 (Extremely right-wing) (N = 540). Because few participants chose the extreme response options 1 (N = 6) and 9 (N = 2), we collapsed the 1 and 2 (N = 69) and the 8 and 9 response options (N = 9). We measured system justification with a Swedish translation (Nilsson & Jost, 2012) of the eight-item (e.g. ‘‘Society is set up so that people usually get what they deserve’’) system justification scale (Kay & Jost, 2003), which includes two reversed items, a = .83. We measured resistance to change with eleven items (e.g. ‘‘If you start changing things very much, you often end up making them worse’’), including two reversed items, a = .75, and preference for equality with fifteen items (e.g. ‘‘Prosperous nations have a moral obligation to share some of their wealth with poor nations’’), including six reversed items, a = .86, on the basis of previous attempts to isolate these components of ideology (Nilsson & Jost, 2012). 2.3. Statistical procedure Structural equation modeling was run in AMOS 20.0. All calculations were based upon the covariance matrix and the maximum
30
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
.57 .66 .49 .45 .30 .26
EMO WEA CRU COM ANI KIL
.63 .54 .48 .62 .54 .36
TRE UNF RIT FAI JUS RIC
Loyalty
.55 .67 .72 .27 .35 .18
LUV BET LOY HIS FAM TEA
Authority
.78 .54 .56 .63 .37 .43
RES TRA CHA KID SEX SOL
.74 .64 .30 .55 .51 .40
DEC DIS GOD HDG UNN CHA
Harm
.70 .25
Fairness
-.06 .13
-.08
.18
.76
-.01 .79
.87
Sanctity
.56 .62 .50 .47 .32 .29
EMO WEA CRU COM ANI KIL
.64 .54 .47 .63 .55 .35
TRE UNF RIT FAI JUS RIC
Loyalty
.54 .63 .69 .28 .34 .16
LUV BET LOY HIS FAM TEA
Authority
.73 .51 .53 .59 .32 .39
RES TRA CHA KID SEX SOL
.64 .61 .28 .54 .49 .38
DEC DIS GOD HDG UNN CHA
Harm
.92
Individualizing intuitions .76
Fairness
.09
.84
Binding intuitions
.87
.90
Sanctity
Fig. 1. Moral Foundations Questionnaire confirmatory factor analysis of the five-factor model and the hierarchical two-factor model. Item abbreviations are drawn from Graham et al. (2011).
5
likelihood method. We report v2 tests of goodness of fit, the Comparative Fit Index (CFI), which estimates fit relative to a baseline model, and the Root Mean Square Error of Approximation (RMSEA), which estimates parsimony-adjusted fit in relation to a perfect model. According to widespread rules of thumb a CFI above .95 and an RMSEA below .06 (Hu & Bentler, 1999) indicates adequate model fit. Three models were included in the confirmatory factor analyses: (1) the five-factor model, with one latent factor for each moral foundation, the respective scale items as manifest variables, and estimated relations between all latent factors, (2) the hierarchical two-factor model, substituting the relations between the latent factors for two related superordinate factors (Fig. 1), and (3) a hierarchical three-factor model, treating sanctity as a separate superordinate factor and estimating relations between the three superordinate factors. To model the effect of moral foundations on political ideology, we constructed a path model, using only manifest variables in order to keep model complexity manageable. We divided it into three levels: (1) moral foundations, which predicted. (2) resistance to change, preference for equality, and system justification, which predicted. (3) left–right political identity (Fig. 3). All relations between error terms of variables within each level were estimated.
3. Results 3.1. Structural validity The five-factor model, v2 (396) = 1517.8, CFI = .679, RMSEA = .072 (95% CI [.069, .076]), was significantly better, Dv2 (4) = 23.0, p < .001, than the hierarchical two-factor model, v2 (400) = 1540.8, CFI = .674, RMSEA = .073 (95% CI [.069, .077]). The three-factor model, v2(399) = 1538.3, CFI = .674, RMSEA = .073
4 Harm
3
Fairness Loyalty
2
Authority Sanctity
1
0 1 Strong left
2
3
Moderate left
Slight left
4 Neutral neither
5 Slight right
6 Moderate right
7 Strong right
Self-reported political identity Fig. 2. Moral foundations across political identity from left to right.
(95% CI [.069, .077]), was not significantly better than the twofactor model, Dv2(1) = 2.5, p = .11. These results are consistent with Graham et al. (2011) findings, although our fit estimates are lower in absolute terms. The five-factor and two-factor models are displayed in Fig. 1. In order to investigate the sources of misfit, we looked at potential model modifications post hoc. When the error terms of items within the same factor were allowed to covary, the fit could be increased substantially, Dv2(24) = 374.5, p < .001, to v2(372) = 1143.3, CFI = .779, RMSEA = .062 (95% CI [.058, .066]). These item covariations appeared to stem not just from superficial similarities in item formulation but also from similarities in item content, possibly indicating that the taxonomy is oversimplified: Items relating to disgust sensitivity, religiosity (sanctity), respect for authority, conformity with traditions (authority), and socio-political justice
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
Harm
.33*** .49***
Fairness
31
effect of authority concerns on political identity was partly mediated by low preference for equality. The moral foundations accounted for 33.8% of the variance of system justification, 46.4% of the variance of resistance to change, and 63.6% of the variance of preference for equality.
Preference for equality -.48***
-.23*** -.20***
Loyalty
Authority
.12** .19*** -.24*** .36*** .36***
Sanctity
System justification
.38***
Political identity
.09**
Resistance to change
.19***
Fig. 3. Preference for equality, resistance to change, and system justification as mediators of the relationship between moral foundations and political identity. Covariances between residual terms and non-significant estimates (p < .05) are not displayed. ⁄⁄⁄p < .001. ⁄⁄p < .01. ⁄p < .05.
(fairness) covaried especially strongly, respectively.2 We tried, moreover, to remove items with low factor loadings, which allowed us to further improve CFI (.84 with seven items removed) but not RMSEA. 3.2. Relation to political ideology Concerns with fairness, r = .44, and prevention of harm, r = .25, decreased, and concerns with loyalty, r = .29, authority, r = .50, and sanctity, r = .29 (p < .001), increased continuously as political identity moved from left to right, as shown in Fig. 2. When entered into a regression model, the five foundations accounted for a total of 38.5% of the variance of political identity. Fairness and authority concerns were the strongest predictors; bharm = .09, p = .046; bfairness = .33, p < .001; bloyalty = .10, p = .045; bauthority = .35, p < .001; bsanctity = .03, p = .57. When the five foundations were collapsed into two superordinate categories (aindividualizing = .73; abinding = .82), these categories accounted for 34.6% of the variance of political identity; bindividualizing = .39 and bbinding = .42 (p < .001). Harm and fairness concerns correlated positively with preference for equality (rharm = .47; rfairness = .62), and negatively with resistance to change (rharm = .25; rfairness = .39) and system justification (rharm = .26; rfairness = .37). Loyalty, authority, and sanctity concerns correlated positively with resistance to change (rloyalty = .32; rauthority = .53; rsanctity = .40) and system justification (rloyalty = .22; rauthority = .46; rsanctity = .29), and negatively with preference for equality (rloyalty = .18; rauthority = .50; rsanctity = .30; p < .001). The fit of the path model, which is shown in Fig. 3, was adequate, v2(5) = 9.70, p = .084, CFI = .998, RMSEA = .042 (95% CI [.000, .081]). Post hoc tests revealed that only harm had a significant independent effect on political identity, k = .08, p = .012, which increased fit, Dv2(1) = 6.26, p = .001, to v2(1) = 3.44, p = .49, CFI = 1.00, RMSEA < .001 (95% CI [.000, .061]). As predicted, most of the effect of harm and fairness concerns on political identity was mediated by preference for equality and most of the effect of loyalty, authority, and sanctity concerns on political identity was mediated by resistance to change and system justification. But the effect of fairness concerns on political identity was partly mediated also by low system justification and resistance to change and the 2 ‘‘Whether or not someone acted in a way that God would approve of’’ did not correlate above .2 with any item addressing disgust and ‘‘Whether or not someone conformed to the traditions of society’’ did not correlate above .2 with any item addressing respect for authority.
4. Discussion The current research provides further cross-cultural support for the central claims of Moral Foundations Theory. The five-factor model was indeed significantly better than the alternative twoand three-factor models, replicating previous findings (Graham et al., 2009, 2011). The five-factor model also had incremental validity, when compared with the two-factor model in terms of the prediction of political ideology. The advantage of the five-factor model over the two-factor model in terms of fit was, however, very small, similar to previous findings (Bobbio et al., 2011; Davies et al., 2014; Graham et al., 2011). This suggests that, although the five foundations are distinct, it is useful to think of them in hierarchical terms, with two superordinate factors: individualizing or liberal intuitions, subsuming harm and fairness as facets, and binding or conservative intuitions, subsuming loyalty, authority, and sanctity as facets. As in the case of personality trait models, the superordinate factors bring coherence and parsimony to the theoretical system while the facets may afford greater predictive utility in specific domains of inquiry (Costa & McCrae, 1995). The results also supported the theorized links between all five moral foundations and political ideology in the Swedish context, despite the low levels of conservatism and religiosity in Sweden. The strength of the links seems, however, to be affected by cultural factors. When comparing the current results with results from other countries (Davies et al., 2014; Graham et al., 2011; Kim et al., 2012), concern with fairness was a particularly strong predictor of political identity, which can be explained in terms of the centrality of attitudes to equality within the Swedish division between left and right (Cochrane & Nevitte, 2009), and concern with sanctity was a particularly weak predictor of political identity, which can be explained in terms of the low degree of religiosity across the ideological board in Sweden (Inglehart & Welzel, 2010). Although concern with authority was also a strong predictor of political identity in the current study, this result was similar to those obtained in previous studies. Moreover, the current results provide new insight into the connection between morality and political ideology. As hypothesized, the effects of the individualizing foundations on political identity were mediated primarily by preference for equality and the effects of the binding foundations on political identity were mediated primarily by resistance to change and system justification. There were, however, a few findings qualifying this picture. First, the effect of authority concerns on political identity was substantially mediated by both resistance to change and system justification, which makes sense insofar as it involves both respect for authority, and thus conformity with the system, and conformity with societal traditions and norms. But it was also partly mediated by low preference for equality, which was unexpected, but is intelligible, nevertheless, in terms of the association between respect for authority and justification of current hierarchical arrangements. Second, the effect of fairness concerns on political identity was partly mediated not just by preference for equality, but also by low system justification and resistance to change, suggesting that the system and the traditions were perceived as violating people’s rights. Third, the effect of sanctity concerns on political identity was mediated only by resistance to change, which may be explained in terms of the fact that sanctity incorporates preservation of principles that serve to uphold purity and decency rather
32
A. Nilsson, A. Erlandsson / Personality and Individual Differences 76 (2015) 28–32
than conformity with the system per se, which is highly secular in the case of Sweden. It is plausible that the effect of sanctity on political identity is mediated by system justification in non-secular countries. In countries where the left–right dimension is defined by the opposition between liberalism and conservatism, the effect of the binding intuitions on political identity should, furthermore, be more strongly mediated by resistance to change than it was in the current study. Despite providing support for the hypothesized relations between moral intuitions and political ideology, the results do, however, also add to a literature suggesting that the moral foundations questionnaire is not psychometrically optimal. The amount of misfit was rather large in the current study, and the reliabilities were low, similar to other studies using non-English versions of this questionnaire (cf. Bobbio et al., 2011; Bowman, 2010; Kim et al., 2012). Although lower fit can be expected when a measure of ideology is translated, and common fit criteria may be too strict when applied to complex domains of psychological inquiry (Hopwood & Donnelan, 2010), misfit may also indicate that the taxonomy is oversimplified. Some of the items measuring loyalty concerns in particular did not work in our Swedish sample, and we noted several content-based item covariances within the moral foundation factors, which were responsible for some of the misfit. Further research is needed to determine the extent to which these findings can be replicated in other languages or attributed to specific features of the Swedish language or socio-political system. It is, for example, not surprising that some of the loyalty items did not work, given that loyalty with in-groups is a contentious issue in Sweden, associated with nationalism, racism, and fascism, which may make Swedes sensitive to what group is implicitly intended. It is also possible that religious purity and disgust are not associated in the same way in the most secular parts of the world, including Northern Europe and parts of Asia (Inglehart & Welzel, 2010), as they are in the majority of countries in the world. In conclusion, although Moral Foundations Theory offers a powerful framework for understanding variations in moral and political proclivities, more work is needed to test its theoretical claims, evaluate and refine the taxonomy, and develop reliable, structurally valid measurement across the full spectrum of cultures and languages.
References Bobbio, A., Nencini, A., & Sarrica, M. (2011). Il Moral Foundation Questionnaire: Analisi della struttura fattoriale della versione italiana [The Moral Foundations Questionnaire: Analysis of the factor structure of the Italian version]. Giornale di Psicologia, 5, 7–18.
Bowman, N. (2010). German translation of the Moral Foundations Questionnaire – Some preliminary results.