If you're happy, do you show it?

Are political conservatives happier than political liberals? Several recent studies have found that greater political conservatism predicts higher levels of self-reported happiness and life satisfaction in the United States (1–4). This “ideological happiness gap” has been attributed to a number of different psychological factors. Some explanations posit that conservatives’ greater happiness is due to a suite of adaptive personal, social, and cultural values (1), such as high levels of personal agency, optimism, and transcendent moral beliefs (2). Others portray conservatism as a protective or even defensive mechanism that serves the palliative function of justifying troubling societal inequalities (3). Although researchers disagree over the mechanism underlying the happiness gap, meta-analytic review has confirmed that it is a small but reliable effect (r = 0.12) (4).

All of the data supporting the relation between political conservatism and subjective well-being rely on self-report measures (e.g., stated agreement with face-valid items such as “In most ways my life is close to ideal”) (4, 5). This is not surprising given that self-reports are the foundation of subjective well-being research (6, 7), but reliance on any single methodology within an area of research can result in systematic methodological artifacts due to common method variance (8). There are many challenges involved in self-report research (9–11), including the influence of self-enhancement motives, which commonly lead to unrealistically favorable self-assessments (12). Self-reports of happiness and life satisfaction, much like self-assessments of other valued characteristics, are susceptible to self-enhancing distortions (13). Both individual differences in, and experimental manipulations of, self-enhancement motivation predict the tendency to report happiness and life satisfaction at unrealistically positive levels (i.e., the “happier-than-average effect”) (13).

Self-enhancing tendencies are not evenly distributed across populations (14–16), and there are reasons to suspect that liberals and conservatives may self-enhance to differing degrees. Conservatism (17) has been characterized as an ideology grounded in ego defensiveness (18), enhanced sensitivity to negativity (19), and personality dimensions related to defensive forms of motivated social cognition (20, 21). Self-enhancement is also associated with a number of factors related to politically conservative ideologies: It is more pronounced among individualistic cultures (14), religious people (15), and competitive, hierarchically oriented groups (16). Given that self-reported happiness is related to political conservatism (4) and self-enhancement (r = 0.10) (13) at comparable magnitudes, it is possible that ideological happiness differences may simply be an example of conservatives’ stronger tendency to evaluate the self favorably.

In study 1, we examined whether conservatives’ reports of greater subjective well-being, relative to liberals, could be attributed to self-enhancing tendencies. Visitors to YourMorals.org, a psychological research Website, reported their political ideology and completed the Satisfaction With Life Scale, the most frequently used measure of subjective well-being (5), as well as the Balanced Inventory of Desirable Responding, a well-validated measure of the tendency to engage in self-deceptive enhancement (22). As expected, increasing political conservatism predicted greater reported life satisfaction [r(1433) = 0.10, P < 0.001; adjusting for demographic characteristics: β = 0.09, P = 0.002]. This happiness gap was similar in magnitude to that found in past research (4). Importantly, we also found that self-deceptive enhancement was higher among conservatives than liberals [r(1433) = 0.16, P < 0.001; adjusting for demographics: β = 0.18, P < 0.001]. A bootstrapped mediation analysis revealed that, as hypothesized, self-deceptive enhancement fully mediated the ideology–life satisfaction association [indirect effect:b = 0.05, P < 0.001, 95% confidence interval (CI) = (0.03, 0.07)] (see Table 1).

General

The first column in this table presents all the predictors that were used in either Model 1 or Model 2. For each model, the statistical analyses examine whether or not these predictors can accurately predict self-reported life satisfaction. The asterisks (*) are indicators of statistically significant predictors.

Model 1

This is a simple model, using only two predictors: political conservatism and self-deceptive enhancement.The predictors are entered into the model one at a time (i.e., in multiple steps), so that their effects can be examined independently from one another.

In Step 1, we see that political conservativism is a significant predictor of life satisfaction, such that higher levels of conservatism predict greater self-reported life satisfaction. In Step 2, we see that self-deceptive enhancement is also a significant predictor; in this case, increased self-deceptive enhancement predicts greater self-reported life satisfaction.

This is initial evidence in support of the authors’ hypothesis that self-deceptive enhancement can explain the relationship between political conservatism and life satisfaction.

Model 2

This is a more complicated model, using quite a few additional predictors.

In Step 1, the only predictor that is not included in the model is self-deceptive enhancement. Here, again, we see that political conservatism does predict life satisfaction. Step 2 “controls for” all of the predictors from Step 1, which means that any new significant predictors (in this case, self-deceptive enhancement) haves a bigger effect on the outcome variable (life satisfaction) than any of the Step 1 predictors.

The fact that political conservatism is no longer a significant predictor in Step 2 means that the effect between conservatism and life satisfaction is “fully mediated” by self-deceptive enhancement. Put another way, this means that if self-deceptive enhancement were held statistically constant, there would be no relationship between conservatism and life satisfaction. This is stronger support for the authors’ hypothesis.

R²

The bottom row of this table reports R² values for each step. What do these values mean? Each value is an indication of how much of the variance in life satisfaction (i.e., the range of participants’ responses) is explained by the predictors in that model/step. If R² increases from Step 1 to Step 2, that is an indication that Step 2 does a better job of explaining the relationship between the variables.

So, in the first step of Model 1, political conservatism on its own only explains 1% of the variance in responses concerning life satisfaction. In Step 2, R² increases to 1.2%. This means that including self-deceptive enhancement in the model is a better explanation of the data. In Model 2, we see that, in general, including variables like participants’ age and gender is valuable, because it boosts the explanatory power of the model up to 5%.

Finally, adding self-deceptive enhancement into Model 2 results in a large improvement of the model. Now, 21% of the difference in responses can be explained by the model. Step 2 of Model 2 is therefore the “best” of the four steps that are displayed in this table.

These results indicate that conservatives’ greater reports of life satisfaction were attributable to their stronger tendency to engage in self-enhancement. But there are two possible interpretations of these findings. First, conservative participants may be experiencing greater happiness and life satisfaction than liberals as a beneficial consequence of their self-enhancing predispositions (23). Alternatively, conservatives’ elevated reports may simply be an example of their stronger tendency to provide unrealistically favorable self-assessments. This interpretative ambiguity highlights a major limitation of relying on self-reports alone: They do not distinguish between genuine and superficial presentations of happiness.

Behavioral indicators of happiness provide important information that can clarify this ambiguity. If we take self-reports of well-being at face value, rather than as examples of self-enhanced assessments, we would expect to observe conservatives’ greater happiness in unobtrusive measures of happiness-related behavior—for example, in the emotional content of liberals’ and conservatives’ speech, or in the frequency and intensity of their smiling behavior. Assessing smiling behavior also allows us to distinguish between genuine and superficial expressions of happiness. Intense, genuine smiling, known as Duchenne smiling, involves the muscles lifting the corner of the mouth as well as those orbiting the eye (24). Non-Duchenne (also known as social, deceptive, or standard) smiling involves only the muscles lifting the corners of the mouth and is less often related to genuine feelings of happiness or enjoyment. Non-Duchenne smiling is also less predictive than Duchenne smiling of beneficial long-term psychological and physical health outcomes (24, 25).

In study 2, we examined happiness-related behavior among the United States’ most salient liberals and conservatives: members of the U.S. Congress. We assessed two behavioral indicators of happiness within this group: the use of positive and negative emotional language from a text analysis of the 2013 U.S. Congressional Record and the smiling behavior exhibited in their publicly available photographs. For our linguistic analysis, we recorded word-use frequencies for members of the 113th U.S. Congress using relevant terms from one of the most frequently used emotion scales in the psychological literature, the Positive and Negative Affect Schedule: Expanded Form (PANAS-X) (26). We assessed political ideology using continuous liberalism-conservatism scores based on analysis of each politician’s voting record (27) and controlled for each member’s legislative chamber, status as a political minority/majority member in their chamber, overall word usage, and demographic characteristics. Greater conservatism was associated with a small but significant decrease in positive affect word use (β = –0.16, P < 0.001). Conservatism was not significantly associated with the use of negative affect words, joviality-related words, or sadness-related words (Table 2). We replicated this analysis with political ideology defined by party identification instead of voting record and found a similar pattern of results.

Linguistic analyses

The main relevant takeaway point from this portion of the table is that political conservatism is a significant predictor of positive affect word use, but not of negative affect or joviality word use.

The negative value of the coefficient under positive affect (-0.88) means that for every increase of 1 on the conservatism scale (e.g., going from a 3 to a 4 on a scale where higher values mean more conservatism), positive affect usage tends to decrease by 0.88 words. More conservatism is correlated with less positive affect word usage.

Conservatism is also marginally (p < .10) related to use of sadness words. In this case, conservatives were slightly less likely to use sadness-related words.

Smile analyses

Here, the results indicated that political conservatism is significantly (but negatively) related to overall smiles and contraction of the orbicularis oculi muscle, as well as marginally related to contraction of the zygomatic major muscle. This means that conservatives were less likely to smile in general, and of those who did smile, conservatives’ smiles were less intensely “genuine.” These are fairly small effects (e.g., an increase of 1 on the conservatism scale relates to 0.03 fewer smiles), but they are statistically significant.

Going beyond conservatism

This table includes quite a few other predictors, in addition to political conservatism. Conservatism is the predictor that relates directly to the authors’ hypotheses, but it is not the only significant predictor in these analyses. Indeed, some of the strongest effects (i.e., smaller p-values) are these other predictors.

What do you think of the other significant findings here? For example, why might party status (majority versus minority party) be significantly related to positive affect word usage? Or what about the correlation between sex (female versus male) and contraction of the zygomatic major muscle?

Note that, for dichotomous predictors like these, when it says something like “1 = Majority,” that refers to how the two possible values of that predictor were coded in the analysis. If majority status = 1, then minority = 0. In turn, this means that as we look from minority to majority status (an increase of 1 on the scale), Congress members are likely to use 7.06 more positive affect words; those in the majority party display more positive affect than those in the minority party.

We also tested the reliability of the relationship over time by assessing emotional language at the party level across 18 years of Congressional Record data, including more than 432 million words. Members of the liberal-leaning Democratic Party used a higher ratio of positive to negative affect words (M = 13.65:1) than members of the conservative-leaning Republican Party (M = 11.50:1), including a higher frequency of positive affect word usage in 17 of 18 years.

This pattern of emotional expression was not limited to linguistic analyses. We next analyzed smiling behavior in photos from the Congressional Pictorial Directory of the 113th U. S. Congress (28) using the Facial Action Coding System (FACS) (29). A FACS-certified coder assessed the intensity of two action units (AUs) associated with genuine smiling behavior: activity in the orbicularis oculi (AU6) and the zygomatic major (AU12). We found that increasing political conservatism predicted less intense overall smiling (β = –0.12, P = 0.032), controlling for the same variables as the linguistic analysis above (Table 2). We observed only marginally significant differences in the intensity of smiling behavior in the muscles lifting the corners of the mouth (AU12: β = –0.10, P = 0.096), but conservatism predicted significantly less intense facial action in the muscles around the eyes that indicate genuine happiness (AU6: β = –0.13, P = 0.031). The odds of displaying non-Duchenne smiles (i.e., action in AU12 but not AU6) were slightly higher for conservatives than for liberals, but this did not reach statistical significance [controlling for demographics: odds ratio (OR) = 1.04, P = 0.206].

Study 2 revealed no evidence of greater emotionally positive behavior among conservative elected officials. Instead, it was liberal politicians who tended to more frequently express positive emotional language, smile more intensely, and smile more genuinely. These effects were modest in size, but in the opposite direction of the previously observed self-report–based happiness gap (4). Of course, elected political leaders are not representative of liberal and conservative individuals more generally, and it is unclear how well speech and facial expressions occurring within the confines of Capitol Hill reflect similar happiness-related behaviors in less overtly political contexts.

Accordingly, study 3 assessed the linguistic content of 47,257 Twitter status updates (“tweets”) from liberal and conservative members of the general public. We analyzed the statuses of individuals who subscribed to (“followed”) the official Twitter pages of either the Democratic or Republican Party, excluding those following both, under the assumption that users who followed one party exclusively were likely to share that party’s political views. We assessed the emotional content of each tweet using word lists from the PANAS-X, the Linguistic Inquiry Word Count software (LIWC) (30), and lists of “happy” and “sad” emoticons. Logistic regressions predicting the presence or absence of emotion words/emoticons at the tweet level were conducted, with political party followed as the independent variable. Relative to Democratic Party subscribers’ updates, Republican Party subscribers’ updates were significantly less likely to contain positive emotion words, joviality words, and happy emoticons, and significantly more likely to contain negative emotion words (all Ps < 0.05) (Table 3). Marginal effects also revealed similar patterns on the PANAS-X negative affect and sadness subscales.

Positive emotions

The table’s caption tells us that “ORs above 1.00 indicate that conservative tweets contained a higher percentage than liberal tweets; values below 1.00 indicate the opposite.” The key values here, therefore, are the top row (OR values) and the third row (p-values). Based on these results, it appears that liberals were significantly (p < .05) more likely to tweet using joviality words, positive emotion words, and happy emoticons. Positive affect word usage was in the same direction (OR < 1.00), but this difference was nonsignificant.

Negative emotions

The only significant relationship in this portion of the table is that conservatives were more likely than liberals to tweet using negative emotion words (OR > 1.00 and p < .05). Tweets from followers of the Republican Party were also marginally more likely (compared with tweets of followers of the Democratic Party) to include negative affect words sadness-related words. The negative emotion category looks at other emotions, in addition to sadness, such as anger and fear.

Correlation and causation

You may have heard this one before … correlation does not imply causation. Because all the experiments in this paper are correlational (using nonrandomized samples), it is not valid to conclude from these results that being conservative (or liberal) causes anything related to happiness/well-being. Just because two variables co-occur—in this case, political liberalism and displays of positive emotion—that does not mean that one variable causes the other. How might this influence the kind of conclusions that you can make from the data? What might an experimental study add to the current findings?

In study 4, we analyzed 457 publicly available photographs of individuals from LinkedIn, a business-oriented social networking Web site. We selected participants who publicly self-identified as employees at organizations strongly associated with ideologically liberal or conservative values (e.g., Planned Parenthood versus the Family Research Council), under the assumption that the majority of employees at these organizations were likely to share the organizations’ ideological views. As we found among Democratic and Republican Congress members in study 2, smiles were marginally more intense among employees at ideologically liberal organizations (M = 1.98, SD = 1.35) than among employees of ideologically conservative organizations [M = 1.75, SD = 1.24 ; t(455) = 1.91, d = 0.18,P = 0.057; controlling for gender and ethnicity: β = –0.07, P = 0.143]. We again did not observe ideological differences in the intensity of smiling behavior in the muscles lifting the corners of the mouth [AU12; liberals: M = 2.18, SD = 1.45; conservatives: M = 1.99, SD = 1.32; t(455) = 1.42, d = 0.13, P = 0.157; controlling for gender and ethnicity: β = –0.03, P = 0.458]. However, individuals at conservative organizations expressed significantly less intense facial action in the muscles around the eyes that indicate genuine feelings of happiness [AU6; liberals: M = 1.78, SD = 1.43; conservatives: M = 1.50, SD = 1.38; t(455) = 2.14, d = 0.20, P = 0.033; controlling for gender and ethnicity: β = –0.09, P = 0.051]. We also found that the proportion of individuals displaying non-Duchenne smiles was significantly higher at conservative organizations (n = 41 of 217, or 18.89%) than at liberal organizations (n = 24 of 240, or 10.00%; χ² = 7.39, P = 0.007; controlling for demographics: OR = 2.35, P = 0.003).

Together, our studies found that political liberals exhibited more frequent and intense happiness-related behavior than political conservatives. Contrary to the pattern found in self-reports of happiness and life satisfaction (1–4), linguistic analyses of massive archives of text from both elected politicians and the general public revealed a modest but consistent tendency for liberals to use more positive emotional language than conservatives. In addition, we found that liberal politicians and employees at organizations promoting liberal values smiled more intensely and genuinely than their conservative counterparts. Although the effects in these studies were small, they consistently revealed greater happiness-related behavior among liberals, rather than conservatives. These behavioral data support our finding from study 1 that differences in liberals’ and conservatives’ reports of happiness can be attributed to conservatives’ stronger tendency to provide flattering self-assessments.

These findings add nuance to past research suggesting that subjective well-being is a multifaceted construct composed of both hedonic or experiential aspects (e.g., reports of current affect and observed behavior) and eudaimonic or evaluative ones (e.g., global assessments of happiness or life satisfaction) (31, 32). What does it mean when self-reports and behavioral indicators provide contradictory data? Happiness lacks an objective “gold standard” validation criterion, and both self-report (6) and behavioral (25, 33) measures of happiness have been linked to beneficial short- and long-term outcomes. For this reason, it would be a mistake to infer from our data that liberals are “objectively” happier than conservatives or that conservatives’ self-enhancing tendencies are necessarily maladaptive. It is not currently known whether claims of happiness bolstered by self-enhancement motivation are any less advantageous than more “genuine” reports of positive well-being. In fact, there is evidence that self-enhancement tendencies can often facilitate psychological adjustment, including the capacity for productive work, satisfying relationships, and resilience to psychological threat (23). This possibility raises important questions for future research, including the potential existence of “defensive” forms of happiness similar to other forms of defensive self-evaluation, such as those uncovered in research on self-esteem (34).

Our studies did not directly test previously proposed mechanisms for the ideological happiness gap, such as differences in ideological values (1), system justification motivation (3), or judgments about personal agency, optimism, and transcendent moral beliefs (2). However, it seems plausible that these explanations share common motivational mechanisms with self-enhancement (21, 23). A key difference is that these previous accounts would predict happiness-related behavior to correspond with self-report evidence of greater conservative happiness. Our self-enhancement-based account explains this discrepancy. Future research should continue to examine whether these motivational mechanisms truly lead to genuinely positive emotional experiences and greater psychological adjustment or if they primarily motivate the superficial report of one’s positive qualities.

The questions raised by this research are important because of a growing interest in using self-report measures of happiness to inform public policy (1, 6). Our research supports those recommending caution about promoting any particular ideology or policy as a road to happiness (32). Research investigating self-report–based happiness differences between nonrandomized groups (e.g., cultures, nations, and religious groups) may inadvertently capture differences in self-reporting styles rather than actual differences in emotional experience. Both behavioral measures and self-reports of subjective well-being are valuable tools, but any comprehensive assessment of subjective well-being should involve multiple methodological approaches (6, 8). Reliance on any single methodology is likely to lead to an oversimplified account of not only who is happier than whom but also what it means to be happy at all.

Supplementary Materials

www.sciencemag.org/content/347/6227/1243/suppl/DC1

Materials and Methods
References

References and Notes

1. C. Brooks, Gross National Happiness: Why Happiness Matters for America—and How We Can Get More of It (Basic Books, New York, 2008).

2. R. Schlenker, J. R. Chambers, B. M. Le, Conservatives are happier than liberals, but why? Political ideology, personality, and life satisfaction. J. Res. Pers. 46, 127–146 (2012).

3. J. L. Napier, J. T. Jost, Why are conservatives happier than liberals? Psychol. Sci. 19, 565–572 (2008).

4. E. Onraet, A. Van Hiel, K. Dhont, The relationship between right-wing ideological attitudes and psychological well-being. Pers. Soc. Psychol. Bull. 39, 509–522 (2013).

5. E. Diener, R. A. Emmons, R. J. Larsen, S. Griffin, The satisfaction with life scale. J. Pers. Assess. 49, 71–75 (1985).

6. The term “subjective well-being” is used interchangeably with “happiness” and consists of life satisfaction, positive and negative affect, and domain satisfaction (7).

7. E. Diener, Subjective well-being: The science of happiness and a proposal for a national index. Am. Psychol. 55, 34–43 (2000).

8. T. Campbell, D. W. Fiske, Convergent and discriminant validation by the multitrait-multimethod matrix. Psychol. Bull. 56, 81–105 (1959).

9. Zou, U. Schimmack, J. Gere, The validity of well-being measures: A multiple-indicator–multiple-rater model. Psychol. Assess. 25, 1247–1254 (2013).

10. R. E. Nisbett, T. D. Wilson , Telling more than we can know: Verbal reports on mental processes. Psychol. Rev. 84, 231–259 (1977).

11. B. Gawronski, B. K. Payne, Eds., Handbook of Implicit Social Cognition: Measurement, Theory, and Applications (Guilford Press, New York, 2010).

12. J. D. Brown, Understanding the better than average effect: Motives (still) matter. Pers. Soc. Psychol. Bull. 38, 209–219 (2012).

13. S. P. Wojcik, P. H. Ditto, Motivated happiness: Self-enhancement inflates self-reported subjective well-being. Soc. Psychol. Personal. Sci. 5, 825–834 (2014).

14. S. J. Heine, S. Kitayama, T. Hamamura, Which studies test whether self-enhancement is pancultural? Reply to Sedikides, Gaertner, and Vervea, 2007. Asian J. Soc. Psychol. 10, 198–200 (2007).

15. K. Eriksson, A. Funcke, Humble self-enhancement: Religiosity and the better-than-average effect. Soc. Psychol. Personal. Sci. 5, 76–83 (2014).

16. S. Loughnan, P. Kuppens, J. Allik, K. Balazs, S. de Lemus, K. Dumont, R. Gargurevich, I. Hidegkuti, B. Leidner, L. Matos, J. Park, A. Realo, J. Shi, V. E. Sojo, Y. Y. Tong, J. Vaes, P. Verduyn, V. Yeung, N. Haslam, Economic inequality is linked to biased self-perception. Psychol. Sci. 22, 1254–1258 (2011).

17. Political ideology is a complex and nebulous construct that is most frequently operationalized in this literature using self-report measures of ideology, party affiliation, or scores on psychological constructs associated with self-reported conservatism such as authoritarianism and social dominance orientation (4, 20). The meaning ascribed to terms like “liberal,” “progressive,” and “conservative” also varies over time and across national contexts. Given this conceptual complexity, we have restricted our analyses to U.S. samples where possible and tested our effects across multiple operationalizations of liberalism-conservatism, including self-identified ideology (study 1) and party affiliation (study 2), but also using behavioral measures such as congressional voting records (study 2), social media subscriptions (study 3), and involvement with liberal and conservative organizations (study 4).

18. G. D. Wilson, Ed., The Psychology of Conservatism (Academic Press, London, 1973).

19. J. R. Hibbing, K. B. Smith, J. R. Alford, Differences in negativity bias underlie variations in political ideology. Behav. Brain Sci. 37, 297–307 (2014).

20. J. T. Jost, J. Glaser, A. W. Kruglanski, F. J. Sulloway, Political conservatism as motivated social cognition. Psychol. Bull. 129, 339–375 (2003).

21. J. T. Jost, I. Liviatan, J. Van der Toorn, A. Ledgerwood, A. Mandisodza, B. A. Nosek, in A. C. Kay et al., Eds., The Psychology of Justice and Legitimacy: The Ontario Symposium (Erlbaum, Hillsdale, NJ, 2010), pp. 173–203.

22. L. Paulhus, Balanced Inventory of Desirable Responding: Reference Manual for BIDR Version 6 (University of British Columbia, Vancouver, Canada, 1994).

23. S. E. Taylor, J. D. Brown, Illusion and well-being: A social psychological perspective on mental health. Psychol. Bull. 103, 193–210 (1988).

24. P. Ekman, R. J. Davidson, W. V. Friesen, The Duchenne smile: Emotional expression and brain physiology. II. J. Pers. Soc. Psychol. 58, 342–353 (1990).

25. L. Harker, D. Keltner, Expressions of positive emotion in women’s college yearbook pictures and their relationship to personality and life outcomes across adulthood. J. Pers. Soc. Psychol. 80, 112–124 (2001).

26. D. Watson, L. A. Clark, The PANAS-X: Manual for the Positive and Negative Affect Schedule-Expanded Form (University of Iowa, Ames, Iowa, IA, 1994).

27. That’s My Congress, Legislative Scorecards, http://thatsmycongress.com/index.php/legislative-scorecards

28. United States Congress Joint Committee on Printing, Ed., Congressional Pictorial Directory: One Hundred Thirteenth Congress (Government Printing Office, Washington, DC, 2013).

29. P. Ekman, W. V. Friesen, Facial Action Coding System: Investigator’s Guide (Consulting Psychologists Press, Washington, DC, 1978).

30. J. W. Pennebaker, M. F. Francis, R. J. Booth, Linguistic Inquiry and Word Count: LIWC-2001 Operations Manual (LIWC.net, Austin, TX, 2001).

31. L. Deci, R. M. Ryan, Hedonia, eudaimonia, and well-being: An introduction. J. Happiness Stud. 9, 1–11 (2008).

32. B. Krueger, A. A. Stone, Progress in measuring subjective well-being. Science 346, 42–43 (2014).

33. S. D. Pressman, S. Cohen, Positive emotion word use and longevity in famous deceased psychologists. Health Psychol. 31, 297–305 (2012).

34. M. H. Kernis, Toward a conceptualization of optimal self-esteem. Psychol. Inq. 14, 1–26 (2003).

35. Acknowledgments: The authors received no financial support for the research, authorship, and/or publication of this article and declared no potential conflicts of interest. We thank J. Haidt and A. Stone for their feedback on an earlier draft of the manuscript and R. Idrogo-Lam and J. Larsen for their assistance in data collection. The data used in this paper are available from Open ICPSR: http://doi.org/10.3886/E26078V1.