An expressive voting model of anger, hatred, harm and shame

Abstract

To consider some political implications of angry voters, we alter the standard expressive model in a fundamental way. One result is that an angry voter with a strong sense of shame at the thought of voting to harm others, may still do so, even when the harm is brutal. Indeed, his willingness to vote for harming others may increase if the proposed harm becomes more severe, even though the angry voter is more “decent” (less willing to harm others) than most of us sometimes are. Several examples are given that are consistent with the most troubling implications of the model. An empirical appendix follows the concluding section which tests the implications of the model indirectly.

This is a preview of subscription content, log in to check access.

Fig. 1

Notes

  1. 1.

    See Brennan and Lomasky (1993) for the most definitive analysis and applications of expressive voting theory. For the first explicit application of expressive voting, see Tullock (1971), although he didn’t use the term expressive voting. The term may have been used first by Brennan and Lomasky (1993), although, in response to a question by one author of this paper, Brennan wrote that he is not sure that is true. More recent treatments of expressive voting include Caplan (2007), Brennan (2008), Lee and Clark (2014) and Lee (2015).

  2. 2.

    Most of the time we drop the use of hatred when it is clear that anger is based on hatred.

  3. 3.

    Voting to harm others has been discussed occasionally in the expressive-voting literature; see Brennan and Lomasky (1993, pp. 48–49, 174, 217 and 221). But we are unaware of analysis of expressive voting to harm others that goes beyond an intuitive discussion, which concludes that in large turnout elections, the personal satisfaction from voting to harm others is not offset by any sense of responsibility for the harm.

  4. 4.

    Any expected financial cost the angry voter faces in terms of higher taxes to inflict the harm is ignored since it will become effectively zero as P approaches zero, as it is in the standard expressive–voting model.

  5. 5.

    This simplifies the discussion when considering the sign on the effect of reducing P, since the derivative with respect to P directly indicates the sign on the effect of increasing P.

  6. 6.

    For convenience of exposition, we shall assume that the voter is voting directly on the policy. In this case, if his utility in (1) is positive he will vote for the policy and if (1) is negative he will vote against it. Although fully aware that women vote, we use the male pronoun throughout.

  7. 7.

    The plausibility of those assumptions is discussed in Sect. 3. In Fig. 1, the horizontal axis along which α is measured is broken from 0.80 to 0.95, with the scale being enlarged over that last 0.05 units. It is plausible to assume that anger and shame have independent effects on the voter’s utility, with the zero cross-partials U1,2 = U2,1 = 0. But nothing fundamental changes in Fig. 1 if a small interaction exists between the effects of A and S on the voter’s utility as α increases.

  8. 8.

    As P approaches zero, the sense of responsibility effectively declines to zero, but neither anger nor shame decline to zero, as is discussed in Sect. 3.

  9. 9.

    The authors express concern that we can become more polarized when connecting primarily with others of similar ideologies, while recognizing that such connections can motivate more people to vote.

  10. 10.

    Interestingly, Pinker ties in these comments on violence to the search for “perfect justice” as a source of revenge and strife.

  11. 11.

    This discussion raises a question that seldom gets asked—what is the most effective way to motivate political support; appeals to love or to hate? One obviously can debate this issue. As a political motivator, however, hate has the advantage of scaling far better than love. We can quickly develop an intense hate for a large group of people, indeed an entire nation of them. Our love for others may be more intense than our hate, but the intensity of our love diminishes rapidly when extended beyond a very few. Of course, expressive voting can allow politicians to generate support for costly policies from large numbers whose “love” for such groups as the poor, American workers, the sick and minorities is, for most, rather shallow.

  12. 12.

    Philippine Senator Leila de Lima, former Secretary of the Philippine Department of Justice, was accused by President Duterte of remaining in contact with the head of a drug cartel while serving time inside the New Bilibid Prison, and receiving drug money from narcotics traders. She was indicted on a related charge on December 21, 2016, by the Philippine Department of Justice. See Ray (2016).

  13. 13.

    See Iyengar (2016), Holmes (2016), and Mendez (2017).

  14. 14.

    See Gladwell (2013, p. 236). Most of the information in this subsection is from Chapter 8 of Gladwell’s book.

  15. 15.

    Gladwell (2013, Chapter 8) adds to and expands on these qualifications.

  16. 16.

    Obviously, there have been periods of white anger and hatred toward Blacks in the South, as discussed by Glaeser (2005, pp. 66–71). This anger and hate was less when “little difference in the policies of the [two major political] parties policies toward [Blacks]” existed, which happened when neither party was concerned about blatant violations of the civil rights of Blacks. When both parties “endorsed Jim Crow policies, there was no party that supported African-Americans, so the incentive to supply anti-Black hatred disappeared” (Ibid., p. 70).

  17. 17.

    See Hayek (1960, Chapter 12) for a good discussion of the importance of constitutional constraints on the power of government. Also see Brennan and Buchanan (2000, pp. 8–9).

  18. 18.

    Hitler did call for an election soon after becoming Chancellor and won 43.9% of the vote (his largest vote share ever), although the election was hardly a fair one. For a very concise discussion of Hitler’s rise to power, see Goldhagen (1996, pp. 85–87).

  19. 19.

    However, it should be noted that laws enforcing segregation required it beyond levels that businesses would have imposed voluntarily. See Roback (1986).

  20. 20.

    In his 2012 State of the Union Address. See Eichler (2012).

  21. 21.

    This is partly explained by rent seeking and the political advantage politicians in both parties hope to realize by exaggerating some problems, blaming them on the other party, and promising to fix them.

  22. 22.

    See Ginges et al. (2007) for a broader view of the moral psychology of compromise.

  23. 23.

    As Brennan and Lomasky (1993, p. 175) point out, “even if voting does engage the moral impulses of the citizenry, there can be no guarantee that these moral impulses will in fact generate social goods”.

  24. 24.

    Glazer (2008) develops an expressive voting model in which the motivation to vote for a candidate, say candidate A, is anger at those who vote for candidate B and the desire to please those who also vote for candidate A.

  25. 25.

    It is worth noting that a synonym for disgust is hatred.

  26. 26.

    The data compiled by Rentfrow et al. (2013) use five different samples compiled at different times with different methodologies between the years 1999 and 2010.

  27. 27.

    The sample size is 49 because the “Big Five” data only cover the continental United States, plus the District of Columbia. Accordingly, data were collected only for the “lower 48” states and DC.

References

  1. Brennan, G. (2008). Psychological dimensions in voter choice. Public Choice, 137(3/4), 475–489.

    Article  Google Scholar 

  2. Brennan, G., & Buchanan, J. M. (2000). The power to tax: Analytical foundations of a fiscal constitution. Indianapolis: Liberty Fund.

    Google Scholar 

  3. Brennan, G., & Lomasky, L. (1993). Democracy and decision: The pure theory of electoral preference. Cambridge: Cambridge University Press.

    Google Scholar 

  4. Brown, S., & Taylor, K. (2014). Household finances and the ‘big five’ personality traits. Journal of Economic Psychology, 45(December), 197–212.

    Article  Google Scholar 

  5. Caplan, B. (2007). The myth of the rational voter: Why democracies choose bad policies. Princeton: Princeton University Press.

    Google Scholar 

  6. Carney, D. R., Jost, J. T., Gosling, S. D., & Potter, J. (2008). The secret lives of liberals and conservatives: Personality profiles, interaction styles, and the things they leave behind. Political Psychology, 29(6), 807–840.

    Article  Google Scholar 

  7. Christakis, N. A., & Fowler, J. H. (2011). Connected: How your friends’, friends’ affect everything you feel, think, and do. New York: Little, Brown and Company.

    Google Scholar 

  8. Eichler, A. (2012). State of the Union Address 2012: Obama calls income inequality “the defining issue of our time.” The Huffington Post, January 24.

  9. Eysenck, H. (1977). Crime and personality. London: Routledge.

    Google Scholar 

  10. Fletcher, J. M. (2013). The effects of personality traits on adult labor market outcomes: evidence from siblings. Journal of Economic Behavior & Organization, 89, 122–135.

    Article  Google Scholar 

  11. Gerber, A. S., Huber, G. A., Doherty, D., Dowling, C. M., & Ha, S. E. (2010). Personality and political attitudes: Relationships across issue domains and political contexts. American Political Science Review, 104(1), 111–133.

    Article  Google Scholar 

  12. Ginges, J., Atron, S., Medin, D., & Shikaki, K. (2007). Sacred bounds on rational resolution of violent conflicts. Proceedings of the National Academy of Sciences of the United States of America, 104(18), 7357–7360.

    Article  Google Scholar 

  13. Gladwell, M. (2013). David and Goliath: Underdogs, misfits, and the art of battling giants. New York: Little, Brown and Company.

    Google Scholar 

  14. Glaeser, E. L. (2005). The political economy of hatred. The Quarterly Journal of Economics, 120(1), 45–86.

    Google Scholar 

  15. Glazer, A. (2008). Voting to anger and to please others. Public Choice, 134(3/4), 247–254.

    Article  Google Scholar 

  16. Goldhagen, D. J. (1996). Hitler’s willing executioners: Ordinary Germans and the holocaust. New York: Vintage Books.

    Google Scholar 

  17. Grynaviski, J. D., & Munger, M. (2017). Reconstructing racism: Transforming racial hierarchy from necessary evil into positive good. Social Philosophy and Policy, 34(1), 144–163.

    Article  Google Scholar 

  18. Haidt, J., & Iyer, R. (2016). A truce for our tribal politics. Wall Street Journal, November 5–6, p. C1–2.

  19. Hayek, F. A. (1960). The constitution of liberty. Chicago: University of Chicago Press.

    Google Scholar 

  20. Higgs, R. (1987). Crisis and leviathan: Critical episodes in the growth of American government. Oxford: Oxford University Press.

    Google Scholar 

  21. Holmes, O. (2016). Philippines president Rodrigo Duterte says he personally killed criminals. The Guardian, December 14.

  22. Iyengar, R. (2016). The death toll from Philippine president Rodrigo Duterte’s war on drugs has exceeded 2,400. Time, September 4.

  23. Kajonius, P. J., & Carlander, A. (2017). Who gets ahead in life? Personality traits and childhood background in economic success. Journal of Economic Psychology, 59(April), 164–170.

    Article  Google Scholar 

  24. Lee, D. R. (2015). The beast is not easily starved. Public Choice, 164(3), 275–285.

    Article  Google Scholar 

  25. Lee, D. R., & Clark, J. R. (2014). Buchanan and Tullock ignore their own contributions to expressive voting. Public Choice, 161(1), 113–118.

    Article  Google Scholar 

  26. Manning, J., Hedden, T., Wickens, N., Whitfield-Gabrieli, S., Prelec, D., & Gabrieli, J. D. E. (2014). Personality influences temporal discounting preferences: Behavioral and brain evidence. Neuroimage, 98(September), 42–49.

    Article  Google Scholar 

  27. McGill, A. (2016). Embarrassed to support Trump: A new report says shame might be the reason why the Republican frontrunner has consistently fared worse in live-interview polls. The Atlantic, December 21.

  28. Mendez, C. (2017). Join fight vs. drugs, corruption—Duterte. The Philippine Star, January 1.

  29. Murray, C. (2012). Coming apart: The state of white America, 1960–2010. New York: Crown.

    Google Scholar 

  30. Pinker, S. (2011). The better angels of out nature: Why violence has declined. New York: Viking.

    Google Scholar 

  31. Ray, A. (2016). Philippines justice department files criminal case against Duterte critic Senator Leila de Lima. International Business Times, December 21.

  32. Rentfrow, P. J., Gosling, S. D., Jokela, M., Stillwell, D. J., Kosinski, M., & Potter, J. (2013). Divided we stand: Three psychological regions of the United States and their political, economic, social, and health correlates. Journal of Personality and Social Psychology, 105(6), 996–1012.

    Article  Google Scholar 

  33. Roback, J. (1986). The political economy of segregation: The case of segregated street cars. The Journal of Economic History, 46(4), 893–917.

    Article  Google Scholar 

  34. Simons, H. C. (1951). Economic policy for a free society. Chicago: University of Chicago Press.

    Google Scholar 

  35. Tullock, G. (1971). The charity of the uncharitable. Western Economic Journal, 9(4), 379–392.

    Google Scholar 

  36. Widgier, T. A. (2017). The Oxford handbook of the five factor model. Oxford: Oxford University Press.

    Google Scholar 

Download references

Acknowledgements

We would like to thank the Center for Study of Economic Liberty at Arizona State University, where Lee is a Liberty Fellow, for providing financial support for work on this paper.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Dwight R. Lee.

Appendix

Appendix

The hypothesized relationship in this model is a tricky one to test, even if focusing narrowly on the data-rich issues in contemporary Western politics. Cross-sectional survey data asking voters whether they voted for (or against) Donald Trump out of anger, for instance, does not exist. One alternative means of testing the model is to investigate feelings of anger at the state level and Trump’s vote share by state. One operationalization of “anger” is to make use of the “Big Five” personality traits, specifically what is known as neuroticism. The “Big Five” model of personality, which measures the dimensions of extraversion, agreeableness, conscientiousness, neuroticism and openness, is the primary empirical means by which psychologists evaluate personality (see Widgier 2017), and has been used by economists to establish links between personality and variables such as labor market outcomes (Fletcher 2013), patterns of decisions in household finance (Brown and Taylor 2014), and various general socioeconomic outcomes (Kajonius and Carlander 2017). Data by state on the “Big Five” can be found in Rentfrow et al. (2013).Footnote 26 While differences in personality characteristics intuitively may seem likely to cancel out at the aggregate level, the possibility that they do not, with clusters of neuroticism appearing across the country, is consistent with the narrative presented by Murray (2012).

The general findings between the “Big Five” personality traits and voting behavior are relationships between openness and progressive politics, and conscientiousness and conservative politics (Carney et al. 2008; Gerber et al. 2010). Previous studies are inconclusive regarding the relationship between neuroticism and political beliefs (Carney et al. 2008). Formally, while the dimensions of neuroticism, as understood technically by psychologists, are subtle, a simple definition is emotional instability. Neurotic individuals tend to react to negative emotions poorly, and may lash out vindictively; neuroticism has been linked to antisocial behavior since Eysenck (1977). Neuroticism in the “Big Five” model is well-studied as a psychological concept and captures the inability to deal with negative emotions and at least correlates with a willingness to harm others.

Within the confines of the theoretical model presented in the main text, this would imply that the marginal utility conferred by alleviating anger, given in (2), is greater for neurotics than non-neurotics, owing to the former’s more intense feelings of negative emotions such as anger. Feelings of shame are mitigated, at least in the short run; those exhibiting high degrees of neuroticism also exhibit higher discount rates (Manning et al. 2014). Therefore, an implication of the model is that populations with a greater tendency towards neuroticism are more likely to act on anger and vote to harm others.

The empirical strategy follows. The vote share in each US state won by Republican presidential candidate Mitt Romney in 2012 correlates with Trump’s vote shares in 2016, even though little of the narrative attached to Trump applied to Romney. The error term of a regression explaining which states voted for Trump (after controlling for which voted for Romney) therefore has little to do with party identification or other variables that would drive voters to choose a conventional, establishment conservative. This is the key control variable used to help distinguish what was especially peculiar about the 2016 election and otherwise unrelated to standard American party politics.

All of the “Big Five” personality traits are entered in predicting the Trump vote share, although neuroticism is our variable of interest. In addition, three conventional economic variables are entered, which at times are seen as underlying voters’ interest in choosing an unconventional candidate: educational attainment of people 25 years old or older (2010 data, from the Census), logged Gross State Product per capita (2015 data, from the Bureau of Economic Analysis) and the state unemployment rate (2016 data, from Bureau of Labor Statistics). Descriptive statistics of these variables can be found in Table 1.Footnote 27 The four regressions to be estimated are to first consider the “Big Five” personality traits by themselves (Regression 1), then to consider Romney’s vote share alone (Regression 2), then to consider the “Big Five” and Romney’s vote share together (Regression 3), and finally to add in the three economic controls (Regression 4).

Table 1 Descriptive Statistics

In Table 2 and Regression 1, neuroticism borders on significant but does not achieve the conventional 95% level of statistical confidence, with a coefficient in the expected direction. Conscientiousness and Openness, approximating conservatism and progressivism, are each statistically significant in the expected direction. In Regression 2, the Romney vote share alone can explain 83.5% of the variation in Trump’s vote share, meaning that little variation remains in the data for the other variables to explain. Despite this, the performance of neuroticism actually improves in Regression 3 relative to Regression 1, with the magnitude of the coefficient increasing, while the variable now achieves significance at the 1% level. In Regression 4, the coefficient falls in magnitude, but maintains its significance at the 1% level. Of interest is that the conventional economic variables appear minimally important for the model upon controlling for the personality traits of voters. Using the coefficient calculated in Regression 4, the point estimate of 0.185, a one-standard deviation increase in neuroticism corresponds to a 0.157 standard deviation increase in the state’s Trump vote share. Anger and desiring to harm others, as proxied by the measure of neuroticism in each state, does predict the vote share for at least one candidate who offered a political platform in 2016 devoted in part to expressing such anger.

Table 2 Regression Results—the effect of neuroticism and other “Big Five” personality traits on Trump vote share

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lee, D.R., Murphy, R.H. An expressive voting model of anger, hatred, harm and shame. Public Choice 173, 307–323 (2017). https://doi.org/10.1007/s11127-017-0480-6

Download citation

Keywords

  • Expressive voting
  • Anger
  • Hatred
  • Shame
  • Harm
  • Voting cascades
  • Trump
  • Duterte
  • Criminal sentences
  • Slavery
  • Hitler
  • Constitutions

JEL Classification

  • D72