Data Science & Psychology Data Science applied to Values, Morals, Politics, & things that matter.

Draft by Ravi Iyer, Sena Koleva, Jesse Graham, and Peter Ditto

The 2008 Democratic Presidential Primary has featured 2 candidates, Barrack Obama and Hillary Clinton, with very similar political positions who nevertheless have garnered enthusiastic support from different groups of people. The popular media has reported numerous statistics about the supporters of each candidate based on gender, race, age, and education. It is hardly surprising that people tend to vote for candidates who are members of the same “group”, with people voting for the candidate who matches best based on age, gender, and race. Still, within these groups, there are still individuals with vastly different opinions of each candidate and this study will hopefully shed light on some of the individual differences that may have led people to support one candidate or the other, as can be divined from scales used in psychological research.

From October of 2007 to February of 2008, as the primary season was heating up, we surveyed 8,026 people at yourmorals.org as to their opinions of the leading contenders in the democratic primary. A large number of these people also took 1 or more moral psychology scales, most of which have been well validated by previous psychological research. In this study, we explored the data generated in this period and attempted to form a picture of the personality variables which best predict relative favorability ratings toward Barrack Obama and Hillary Clinton.

Our Sample

These findings explore scores on personality tests and their relation to favorability ratings of “the final 2” 2008 Democratic Presidential candidates, Hillary Clinton and Barrack Obama. Over 8,000 people visited yourmorals.org and answered the question “What is your opinion of the politicians listed below?” on a 7 point scale from very unfavorable to very favorable with the midpoint of the scale, 4, being neither favorable nor unfavorable. The dependent variable that we used in our sample was based on the difference in favorability ratings between the 2 candidates with higher scores indicating that the subject had a more favorable opinion of Obama than Clinton.

Most of the respondents are readers of political or science blog and news articles which linked to our website, so conclusions may be specific to this sample. However, many of the correlations mirror previous polls/research (ie. Women have lower psychopathy scores and higher empathy scores. Women favor Clinton. Younger and better educated subjects favor Obama.), so it seems unlikely that our sample is completely unrepresentative.

All results presented are the result of a linear regression analysis, which allows us to control for differences in gender, age, education, and political orientation. We did not control for race statistically, as our sample was relatively homogenous (88% white, 1-2% black and latino) such that we did not have a reasonable enough sample to analyze results by race. However, having such a homogenous sample means that it is unlikely that race drives our results as for the vast majority of our sample, that variable is naturally held constant.

Moral Foundations Questionnaire

The most popular psychology scale on YourMorals.org is the Moral Foundations Questionnaire and 7,458 people took this scale in addition to completing our survey on favorability towards presidential candidates. The scale is a measure of reliance on and endorsement of five psychological foundations of morality that seem to be found across cultures. Each of the two parts of the scale contained four questions related to each foundation: 1) harm/care, 2) fairness/reciprocity (including issues of rights), 3) ingroup/loyalty, 4) authority/respect, and 5) purity/sanctity. The idea behind the scale is that human morality is the result of biological and cultural evolutionary processes that made human beings very sensitive to many different (and often competing) issues. Some of these issues are about treating other individuals well (the first two foundations - harm and fairness). Other issues are about how to be a good member of a group or supporter of social order and tradition (the last three foundations – authority, loyalty, and purity).

Favoring Obama was related to endorsing the moral values of harm (B=.1, p=.003) and fairness (B = .19, p<.001). Favoring Clinton was related to endorsing the moral values of loyalty (B=.08, p=.012) and authority (B=.07, p<.05). While the effect sizes are statistically significant, but not large, this is interesting because Clinton supporters appear to be more morally conservative though they identify themselves as being just as strongly liberal (see Appendix A for simple correlations of controlling variables). Favoring Clinton appears to be related to being a good group member and/or supporting a social order while favoring Obama appears to be related to endorsing moral foundations related to treating individuals well.

Big 5 Personality Scale

2,379 people also took the the "Big 5 Personality Inventory," created by Oliver John at the University of California at Berkeley. This particular scale is a "short form" of a much longer scale originally developed by Paul Costa and Robert McCrae in the 1980s. The scale is a measure of 5 traits that are said to be most indicative of personality. The traits are openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism.

Favorability towards Clinton was related to higher conscientiousness (B=.30, p<.05) with no other relationship approaching significance. High scorers in conscientiousness are often described as people who are well organized with high standards and who always strive to achieve their goals. They sometimes are seen as uptight.

Empathy

1,224 people took the "Interpersonal Reactivity Index," developed by Mark H. Davis. The scale is a measure of empathy, covering four distinct aspects of empathic responding to others: 1) perspective-taking, 2) fantasy, 3) empathic concern, and 4) personal distress. The idea behind the scale is that empathy is a multi-faceted human trait, and that there are different domains in which people may be particularly sensitive to the feelings of others.

Favoring Obama was significantly related to higher scores on perspective taking (B=.20, p=.01) and personal distress (B=.22, p<.01). The relation to empathic concern was of marginal signficance (B=.15, p=.07). According to Davis (1983), a high score on perspective taking indicates “the tendency to spontaneously adopt the psychological point of view of others”. A high score on personal distress indicates “'self-oriented' feelings of personal anxiety and unease in tense interpersonal settings”. A high score on empathic concern indicates 'other-oriented' feelings of sympathy and concern for unfortunate others. Taken together the results paint a picture of favorability towards Obama being related to increased empathy, though this relation is stronger in areas other than concern for unfortunate others, which may be popularly thought of as the central component of empathy.

Psychopathy

1,063 people took the Levenson Primary and Secondary Psychopathy Scale, created by Michael Levenson, currently at Oregon State University. The scale is a measure of psychopathic personality traits in noninstitutionalized populations. Psychopathy is a disorder characterized by a lack of guilt, empathy, and conscience; psychopaths are often manipulative, insincere, and violate social and moral norms. Contrary to popular belief, psychopathy is not limited to the most extreme serial killers, but can be found everywhere, in all types of industries and social classes. The idea behind the scale is that psychopathy is thought to be continuously distributed in the population so that individuals have varying degrees of psychopathic traits which can be measured.

In addition, the concept of psychopathy is often separated into two factors, each with emphasis on a different aspect of psychopathy. The first type, primary psychopathy, more closely resembles the prototype description of a psychopath and emphasizes a lack of emotion as well as other personality features. Primary psychopaths are often described as glib, superficially charming, grandiose, lying, conning, manipulative, lacking empathy or guilt, emotionally shallow, and callous. The second type, secondary psychopathy, describes individuals characterized mostly by deviant behavior. Features include a need for stimulation, proneness to boredom, impulsivity, irresponsibility, risk-taking, and poor behavioral control. Psychopathy has been found to correlate positively with other personality traits such as extraversion, adventure-seeking, and other traits which are often valued in leadership and competitive environments, so higher scores may not necessarily reflect deviant or immoral behavior, but instead reflect more outgoing personality styles.

Favorability toward either candidate was negatively correlated with psychopathy, meaning that it would be incorrect to state that either candidate’s supporters exhibit greater psychopathy. However, relative favorability towards Obama was highly related to lower primary psychopathy scores (B=.54, p<.001) indicating that those who support Obama may have higher degrees of social emotions such as empathy and guilt that prevent them from advancing their cause at the expense of others.

Forgiveness

1,061 people took the the Heartland Forgiveness Scale (HFS), a self-report measure by Laura Yamhure Thompson, C. R. Snyder, Lesa Hoffman, Scott T. Michael, Heather N. Rasmussen, Laura S. Billings, Laura Heinze, Jason E. Neufeld, Hal S. Shorey, Jessica C. Roberts, and Danae E. Roberts at the University of Kansas. The scale is a measure of dispositional forgiveness (with subscales to assess forgiveness of self, others, and situations).

Favorability towards Clinton was significantly related to higher scores on forgiving oneself (B=.15, p<.01) and forgiving situations (B=.17, p<.01), but not to forgiving others (B=.03, p=.6). This might be indicative of an ability to “let go” and not ruminate about negative events.

Preference for the Merit Principle

1,996 people took the the Preference for the Merit Principle (PMP) Scale, by Liane M. Davey, D. Ramona Bobocel, Leanne S. Son Hing and Mark P. Zanna (2004). The scale is a measure of people's preference for allocating outcomes on the basis of the distributive justice principle of merit. The idea behind the scale is that fairness has many meanings to many people. Some people believe that rewards and promotions should be distributed strictly on the basis of merit, even if that means that some people end up with much more than others.

Favorability towards Clinton relative to Obama was not significantly related to scores on this scale (B=.008, p=.915). Neither candidate’s supporters appear to be significantly more concerned with awarding benefits on the basis of merit.

Identification with All of Humanity Scale

2,555 people completed the Identification with All of Humanity Scale developed by Sam McFarland at the University of Western Kentucky. The scale is a measure of how much one identifies with people locally, nationally, and globally. The idea behind the scale is that some people identify with and feel a sense of community primarily with those nearby. Other people seem less interested in what's going on locally, and think about people more generally, even those far away.

Relative favorability towards Obama was related to identifying with humanity more globally (B=.12, p<.05). There were no significant differences in identifying with people in one’s neighborhood or with people in one’s country.

Ethics Position Questionnaire

1,491 people completed the "Ethics Positions Questionnaire," developed by Donelson R. Forsyth. The scale is a measure of how people determine what is moral or immoral along 2 dimensions, idealism and relativism. Idealism is a measure of how willing a person is to consider committing an immoral act in the service of a greater good. Higher scores indicate less willingness to make such sacrifices. Relativism is a measure of how much one subscribes to the idea that morality can vary depending upon situation and culture. Higher scores indicate more willingness to believe that moral principles are not universal. The idea behind the scale is that, independent of subject matter, some people believe that the nature of morality differs.

Favorability towards Clinton was related to higher scores on moral relativism (B=.17, p<.05). For people who think that morals are absolute, moral relativists may be labelled immoral while moral relativists may see absolutists as rigid.

Discussion/Summary

Correlational evidence can be a function of many things. It could be that some 3^rd factor is causing all of these relationships. However, taken as a whole, the pattern of results from these scales does seem to converge on a set of personality characteristics that match well with what has been reported in the press about these candidates during this election cycle.

Clinton is normally portrayed as the more experienced candidate who is more part of the Democratic establishment. Sheila Leslie, a Democratic assemblywoman from Nevada, stated, “Both Ohio and Rhode Island are Democratic machine states. She (Clinton) tends to do well in machine states.” CBS news reported in one article that Obama had to contend with the fact that Clinton has “earned the intense loyalty of core
Democratic partisans.” As such, it is unsurprising that Clinton supporters score higher in endorsing moral foundations that are a strong part of being a good group member, in this case, being a good group member of the Democratic Party. It makes sense that her supporters would value loyalty more and be loyal to the Clinton family, which has led the Democratic Party throughout the 90s. It makes sense that her supporters would also value authority and might feel that her longevity entitles her to higher standing within the party. Obama supporters value individual based morality (higher Moral Foundation results for harm and fairness) and are therefore perhaps do not see any issue in being “disloyal” to Clinton and refusing to recognize the standing that her years in the public eye might entitle her to. Interestingly, Clinton supporters do not appear to be more concerned with the idea of merit than Obama supporters. Politico.com wrote “"Obama has attracted tens of thousands of young supporters who are loyal to him, not to the Democratic Party. Clinton, on the other hand, has strong support among party regulars." Evidence that Obama supporters might be more concerned with people in general rather than groups can also be seen in the fact that Obama supporters scored higher on concern for people in “the World” but not higher on concern for people in their neighborhood or country level groups.

Obama supporters appear to be a bit more emotional. This would fit his reported campaign style where he attempts to play on the emotions of his audience through inspiring rhetoric. More empathic voters might be more easily swayed by such a campaign. In contrast, Clinton supporters appear to be more coldly rational, scoring lower on empathy measures. Relatively higher moral relativism, forgiveness and psychopathy scores indicate an emotionally cool willingness to make tough decisions that might have negative consequences, in the service of some superseding goal. This could be framed as a positive quality (“the ability to make tough decisions”) or as a negative quality (“the willingness to do anything to get ahead”) and both characterizations have been made of Clinton in the election. Social emotions can be seen as a double edged sword and it makes sense that people favor the candidate who is portrayed as being most similar to them in terms of disposition toward these emotions.

Clinton supporters scored higher on conscientiousness, which can be an indication of being a high achiever or an indication of being uptight. Again, both characterizations have been made of Clinton during this election, indicating that again, voters appear to prefer candidates that are most like themselves. Indeed, perhaps the most academically interesting conclusion of this study is the consistency with which our findings replicate popular media stereotypes about each candidate. Just as with race and gender, voters are apparently also trying to find a candidate that matches them on personality.

Appendix A - Correlations