Income inequality and personality: Are less equal U.S. states less agreeable?

Abstract

Richard Wilkinson’s ‘inequality hypothesis’ describes the relationship between societal income inequality and population health in terms of the corrosive psychosocial effects of social hierarchy. An explicit component of this hypothesis is that inequality should lead individuals to become more competitive and self-focused, less friendly and altruistic. Together these traits are a close conceptual match to the opposing poles of the Big Five personality factor of Agreeableness; a widely used concept in the field of personality psychology. Based on this fact, we predicted that individuals living in more economically unequal U.S. states should be lower in Agreeableness than those living in more equal states. This hypothesis was tested in both ecological and multilevel analyses in the 50 states plus Washington DC, using a large Internet sample (N = 674,885). Consistent with predictions, ecological and multilevel models both showed a negative relationship between state level inequality and Agreeableness. These relationships were not explained by differences in average income, overall state socio-demographic composition or individual socio-demographic characteristics.

Introduction

In recent decades the income gap between rich and poor in the United States has widened considerably. From the end of the Second World War until the mid 1980’s the percentage of income received by the richest 10% of the U.S. population remained relatively stable at around 30–33% (Atkinson & Piketty, 2007). Since the mid 80’s that percentage has steadily increased reaching 45% of total income in 2007 (Atkinson & Piketty, 2007).

This increase in inequality can also be seen in other measures of the income distribution, such as the Gini coefficient. Rather than simply looking at the earnings of the richest fraction of the population, the Gini coefficient compares the population share of each fraction of the population with the share of total income they receive. Perfect equality gives a Gini index of 0, with increasing inequality bringing the index closer to the maximum of 1 (perfect inequality – all of the income accruing to a single individual; Cowell, 2000). Judged on this measure, overall inequality in the U.S. increased by 15% between 1980 and 2007, from 0.40 to 0.46 (U.S.Census Bureau, 2008).

A number of authors, most prominently the epidemiologist Richard Wilkinson (Wilkinson, 2005), have suggested that greater inequality has a significant negative effect on society. More unequal countries and U.S. states have been shown to have poorer general health, higher infant mortality rates, lower average life expectancy, increased levels of obesity, greater illegal drug use, higher levels of homicide and violent crime, a greater prevalence of depression, and lower levels of self-reported wellbeing (Brockmann et al., 2009, Wilkinson and Pickett, 2006).

Wilkinson and others claim that a good part of this relationship between inequality and negative social outcomes is in fact causal; that something about the existence of inequality at the social level changes the way individual people feel and behave so as to contribute to this plethora of negative outcomes (Kawachi and Kennedy, 2002, Wilkinson, 2005). The cornerstone of this theory is the direct relationship between inequalities of income and inequalities of status. One’s income relative to others in society is an important part of social status (Marmot, 2004) and Wilkinson contends that a more unequal income distribution reflects a more unequal distribution of status (Wilkinson, 2005). In other words, the more unequal a society is, the more pronounced its status hierarchy.

The extent of hierarchy in society has profound consequences for the nature of people’s social experience. According to Wilkinson’s theory, in a more unequal society a person (at any level of the hierarchy) is more likely to experience a persistent awareness of the position of others above (or below) themselves in the social order; for example through material indicators of success, such as a newer car or a larger home (Ellaway, McKay, Macintyre, Kearns, & Hiscock, 2004). Everyday social interactions may also be altered by greater inequality, becoming increasingly coloured by considerations of relative status (Wilkinson, 2005).

This experience is said to act as a prompt towards a certain type of social strategy. In a more egalitarian environment, Wilkinson suggests that individual status is less important and a social strategy of mutuality and cooperation might be most appropriate to achieve useful goals. However in a more strongly hierarchical social structure, individual status achievement matters more, and others might be more advantageously viewed as competitors for that status, rather than as potential collaborators (Wilkinson, 2005).

By this view, a more competitive, less trusting, more self-focused mindset is an adaptive response to experiential cues indicating a hierarchical society (Wilkinson, Kawachi, & Kennedy, 1998). Wilkinson labels this attitude ‘the psychology of dominance and subordination’ (Wilkinson & Pickett, 2006, p.192). A society made up of many individuals with this outlook would be one with lower levels of trust, friendliness, cooperation and reciprocity; the elements that together make up a society’s stock of ‘social capital’ (Kawachi & Kennedy, 1999).

A number of authors point to a reduction in social capital as a potential mechanism by which inequality may affect the wide variety of outcomes mentioned above (Kawachi and Kennedy, 1999, Kawachi et al., 1997). There is considerable evidence to support this contention, as low levels of social capital are associated with many of these outcomes including overall health (Kawachi et al., 1997), health related behaviour (Lindstrom, 2005), and homicide and violent crime (Kawachi, Kennedy, & Wilkinson, 1999). However previous investigations have not focused on how inequality may deplete social capital, to lead to other outcomes in turn. The pathway suggested above, through experience of hierarchy to social attitudes and behaviours offers a potential mechanism. This possibility has not previously been investigated directly.

Concepts from personality psychology offer an interesting means of testing this hypothesis. A core endeavour of the field of personality psychology is to identify underlying personality ‘traits’; groups of characteristics that go together within individuals with sufficient regularity to be considered a fundamental dimension along which humans differ. One of the dominant models in the field is the Five-Factor Model (FFM; John & Srivastava, 1999) which breaks personality down into five basic dimensions; Agreeableness, Conscientiousness, Extraversion, Neuroticism, and Openness to experience. The Agreeableness dimension is a remarkably close conceptual match to Wilkinson’s ‘psychology of dominance and subordination’. Both are concerned with attitudes and behaviours towards others including empathy, trust, altruism, and inclinations towards friendship and cooperation (Digman, 1990, Wilkinson, 2005). It could be argued that Wilkinson’s ‘psychology of dominance and subordination’ simply represents the negative pole of the Agreeableness trait dimension. Regardless of whether one makes this strong claim or not, measures of Agreeableness represent a very good proxy for the kinds of social attitude Wilkinson is claiming are altered by inequality. This leads to the hypothesis that individuals in more unequal societies, responding to this cue to social hierarchy, should, on average display lower Agreeableness.

This inequality-low Agreeableness link has not yet been directly tested in either the epidemiological or social science literature. The closest study examined the correlates of culture-level average personality traits (McCrae & Terracciano, 2005). This study found no significant relationships between the Gini coefficient of inequality and any of the five personality factors as measured by the Revised NEO Personality Inventory (NEO-PI-R; Costa & McCrae, 1992). However there are two main aspects of this study that prevent it from being a definitive test of the inequality-low Agreeableness link we have outlined.

First, in line with the authors’ intentions the study examines a sample of 51 cultures, including both high and low income countries. However the ‘Wilkinson hypothesis’ is explicitly intended to explain differences in health and social outcomes within and between rich countries only (Wilkinson, 2005). As we discuss above, the hypothesised effect of inequality on Agreeableness is through its effects on the psychosocial environment. In poorer countries the psychosocial environment is a far less important component of inequality than the material circumstances of poorer individuals (Wilkinson, 2005). For this reason, most studies examining the effects of inequality restrict themselves to rich, economically developed countries, or to smaller sub-regions within rich countries, most commonly U.S. states (Lynch et al., 2004, Wilkinson and Pickett, 2006).

The second reason the McCrae and Terracciano study is not a strong test of the inequality – low Agreeableness link (McCrae & Terracciano, 2005), is its small and admittedly unrepresentative sample. This consists of roughly 300 respondents in each culture, all of whom are students, either of university age (18–21), or over 40, with no respondents between these ages. This restricted sample is not sufficient to provide a strong test of any specific hypothesised association.

Here we test the prediction that individuals in more unequal societies should be lower in trait Agreeableness using a larger, more representative sample from the U.S. Specifically, we follow Rentfrow, Gosling, and Potter (2008), who established the viability and validity of using a large internet-based sample to measure personality at the U.S. state level. U.S. states offer an ideal test-bed for the inequality-low Agreeableness link because they differ markedly in their levels of inequality; the most equal state in 2000 was Wisconsin with a Gini index of 0.37, the most unequal area was Washington DC with a Gini of 0.56 (U.S. Census Bureau, 2000a, U.S. Census Bureau, 2000b). This difference is larger than that between Denmark and Greece (Smeeding & Grodner, 2000). However as part of a single nation, language problems relating to systematic differences in culture and interpretation of survey items are likely to be minimised, especially when compared with cross-national studies (Huang et al., 1997, Ramirez-Esparza et al., 2008).

There are several factors that may confound any potential relationship between state level inequality and Agreeableness. First, more unequal states in the U.S. tend to be the most urbanised (U.S. Census Bureau, 2000a, U.S. Census Bureau, 2000b). Urban residence is also negatively associated with several factors tapped by the Agreeableness trait dimension, including trust and helping behaviour (House & Wolf, 1978), and social participation (Greiner, Li, Kawachi, Hunt, & Ahluwalia, 2004). It is therefore possible for more unequal states to have lower average levels of agreeableness due to their higher proportion of urban residents; a compositional effect of urban living.

Second, compositional effects of age and education may also serve to confound this relationship because older age (McCrae et al., 1999) and higher levels of education are associated with higher average levels of Agreeableness. Note that in the latter case it is likely that greater Agreeableness promotes higher educational achievement, rather than the other way around (Digman, 1990, Hampson et al., 2007). There is also some evidence that age distribution can affect income inequality through the dispersion of earnings by age (Aigner & Heins, 1967), and considerable evidence that inequality is related to lower average levels of education (Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996).

One further demographic factor, rather than confounding the relationship between inequality and Agreeableness, may in fact obscure it. The literature on the five-factor model has consistently shown that females are significantly higher in trait Agreeableness than males (Terracciano & McCrae, 2001). It is also very plausible that the proportion of women in a state could be positively related to income inequality due to the pay disadvantage of women.

Finally, both state average income and the proportion of the population from a minority ethnicity have been shown to be strongly related to income inequality in U.S. states (negatively and positively, respectively; Daly et al., 2001, Deaton and Lubotsky, 2003). The compositional or contextual links between these factors and personality are not clear, but it nonetheless seems prudent to account for them when examining the relationship between inequality and Agreeableness.