State Resident Neuroticism Accounts for Life Satisfaction Differences Between Conservative and Liberal States of the USA

Abstract

Past research indicates associations between higher conservatism and higher life satisfaction, lower neuroticism and higher life satisfaction, and higher conservatism and lower neuroticism. Qualified deduction led to the following hypothesis: Neuroticism can account for the association between higher conservatism and higher life satisfaction. The 50 American states served as the units of analysis. Responses of 619,397 residents to the 44-item Big Five Inventory in an internet survey conducted from 1999 to 2005 provided mean neuroticism scores for each state. Conservative-liberal leaning of over 84,000 respondents to CBS News/New York Times polls from 1999 to 2003 and the percent voting Republican in each state in the 2000 to 2008 presidential elections combined to form a conservatism score for each state. The Gallup-Healthways Well-Being Index provided life satisfaction scores for over 1,000,000 respondents, transforming to a 2008 to 2010 composite score for each state. In a sequential multiple regression equation with life satisfaction as the criterion, state socioeconomic status and white population percent entered first as a block, conservatism entered second, and neuroticism entered third, the demographic controls accounted for 45.7% of the variance, conservatism accounted for another 10.4%, and neuroticism accounted for an additional 10.6%. However, with the entry order of conservatism and neuroticism reversed, neuroticism accounted for another 19.6% but conservatism accounted for only an additional nonsignificant 1.4%. Therefore, the hypothesis was supported. Three alternative explanations suggested by other researchers were not supported in the state-level analysis.

Keywords

Life satisfaction happiness neuroticism conservatism big five personality

Conservatism related to life satisfaction

Various studies with individuals as the units of analysis have shown that conservatives tend to be more satisfied with life compared to liberals (e.g., Burton, Plaks, & Peterson, 2015; Gritching, 1983; Jetten, Haslam, & Barlow, 2013; Napier & Jost, 2008; Onraet, Van Hiel, & Dhont, 2013; Schlenker, Chambers, & Le, 2012; Taylor, Funk, & Craighill, 2006; Vigil, 2010). The association is evident for both conservative-liberal ideological orientation (e.g., Napier & Jost, 2008) and political party preference (e.g., Vigil, 2010). It also exists elsewhere (e.g., Gritching, 1983) as well as in the USA (e.g., Napier & Jost, 2008). Stavrova and Luhmann (2016) found that liberals were happier in only 5 of 92 nations and never in the separate analyses of 28 years between 1974 and 2014 in the USA. Nevertheless, in the only study located that used the American states as units of analysis, Valdmanis (2015) did report a negative correlation between state-level conservatism and well-being, although certain methodological issues might have somewhat limited the results.¹

Why might higher conservatism be associated with greater happiness and life satisfaction? Napier and Jost (2008) have put forth what is perhaps the most intriguing explanation. They conclude that the tendency to rationalize inequality, a key aspect of conservatism, serves as a buffer against the negative emotional impact of societal inequality. In a different vein, Jetten et al. (2013) conclude that conservatives tend to have higher socioeconomic status (SES) which allows access to more group memberships and the group memberships lead to enhanced life satisfaction. In an approach that resonates somewhat more with the present study, Schlenker et al. (2012) conclude that conservatives have greater satisfaction with life because they tend to have a variety of characteristics implicated in superior adjustment and mental health such as higher personal agency, greater transcendent moral beliefs, broader belief in fairness, and a more positive outlook. However, the present research takes a different tack to explain why conservatism is associated with higher life satisfaction which involves neuroticism as an underlying predictor that can account for the relation between conservatism and life satisfaction.

Neuroticism related to life satisfaction

Nomothetic scientific inquiry using individuals as the units of analysis also has demonstrated a clear link between lower neuroticism and life satisfaction (e.g., Burton et al., 2015; DeNeve & Cooper, 1998; Diener, Oishi, & Lucas, 2003; Joshanloo & Afshari, 2011; Kampfe & Mitte, 2010; Keyes, Shmotkin, & Ryff, 2002; Schimmack, Oishi, Furr, & Funder, 2004; Schlenker et al., 2012). The negative correlation occurs in the USA (e.g., Keyes et al., 2002; Schimmack et al., 2004) as well as in other countries (e.g., Joshanloo & Afshari, 2011; Kampfe & Mitte, 2010). In fact, Diener et al. (2003) considered neuroticism to be a primary factor in the relation of personality to subjective well-being, and, in a meta-analysis involving 137 personality traits, DeNeve and Cooper (1998) found that neuroticism indeed was the most potent predictor of happiness and life satisfaction. As well, with states as the units of analysis, Rentfrow, Mellander, and Florida (2009) and McCann (2010) reported that lower state levels of resident neuroticism were associated with higher state levels of well-being and life satisfaction, and McCann (2014c) found that happy Twitter tweets were more likely to emanate from American states with less neurotic residents.

Why should lower neuroticism be associated with higher life satisfaction? As operationally defined in the present study by scores on the Big Five Inventory (BFI; John & Srivastava, 1999), neuroticism refers to one of the five core factors of the influential Big Five personality model (e.g., Costa & McCrae, 1995; Goldberg, 1990; John & Srivastava, 1999). Those higher on neuroticism tend to display higher levels of vulnerability, self-consciousness, depression, angry hostility, impulsiveness, and anxiety. Items on the BFI indicate that a person high on neuroticism tends to see himself or herself as someone who is not “emotionally stable,” does not “handle stress well,” “gets nervous easily,” “worries a lot,” is “easily upset,” is not “relaxed,” does not remain “calm in tense situations,” “can be tense,” “can be moody,” and is “depressed” or “blue” (John & Srivastava, 1999, p. 132). It seems conceivable that such characteristics should tend to foster unhappiness and lower life satisfaction. Therefore, the nature of the neuroticism construct itself tends to suggest why it correlates negatively with life satisfaction, and various studies have corroborated such a relation.

Conservatism related to neuroticism

There also is growing empirical evidence at the individual level of analysis that higher conservatism is associated with lower neuroticism (e.g., Carney, Jost, Gosling, & Potter, 2008; Gerber, Huber, Doherty, Dowling, & Ha, 2010; Iyer, 2011; Kabanoff & Ashton, 1984; Mondak, 2010; Mondak, & Halperin, 2008; Verhulst, Eaves, & Hatemi, 2012). However, conservatism occasionally also has been positively correlated with neuroticism (e.g., Jost, West, & Gosling, 2009; Kabanoff & Ashton, 1984), and others have found no association (e.g., Alford & Hibbing, 2007; Barbaranelli, Caprara, Vecchione, & Fraley, 2007; Carney et al. 2008; Jost et al., 2009; Mondak, 2010; Schlenker et al., 2012; Verhulst et al., 2012). Nevertheless, research with American states as the units of analysis has shown that lower state resident neuroticism is associated with higher state levels of social, economic, and political conservatism (McCann, 2014a) and with several different indicators of Republican preference (McCann, 2014b). Rentfrow, Jost, Gosling, and Potter (2009) reported no relation but see the reanalysis by McCann (2014b).

Why should one expect lower neuroticism to be associated with higher conservatism? There are various reasons (McCann, 2014a, 2014b). For example, Iyer (2011) found that conservatives scored lower on all BFI neuroticism items than liberals, Peterson and Maiden (1993) found that lower neuroticism was related to higher support for the status quo, Breakwell, Fife-Schaw, and Devereux (2001) found that those more worried about events and issues were less conservative, and Vigil (2010) found that republicans showed lower experiential hardship and emotional distress than democrats. Generally, those lower on the neuroticism dimension tend to lead a calmer, more contented, and secure life (e.g., Mondak & Halperin 2008) and have a stronger “capability of controlling irritation, discontent, and anger” (Caprara, Barbaranelli, & Zimbardo, 1999, p. 183). Furthermore, Gerber et al. (2010) noted that the elevated emotionality, anxiety, and worry proneness of more neurotic persons may push them to prefer “liberal economic policies that create ‘safety nets’ and reduce exposure to market risks” (p. 116) and “to identify with those who seek redress through social interventions” (p. 116). As well, Mondak (2010) has stated that “liberalism corresponds with a willingness to see government tackle new and varied problems, while conservatism implies a more cautious approach in which presumption favors the status quo” (p. 127). Consequently, those more neurotic may find greater compatibility with liberalism while those less neurotic may be better suited to conservatism.

The present study

As the preceding review of the literature at the individual level of analysis indicates, and the few studies at the state level of analysis also partially demonstrate, there is empirical evidence grounded in theory for associations between higher conservatism and higher life satisfaction, lower neuroticism and higher life satisfaction, and higher conservatism and lower neuroticism. Neuroticism correlates negatively with both conservatism and life satisfaction, while conservatism and satisfaction correlate positively. Therefore, it is reasonable to speculatively deduce that levels of neuroticism might account for the association between higher conservatism and higher life satisfaction. Such a deduction formed the basis for the general hypothesis of the present study: State resident neuroticism can account for the association between higher state-level conservatism and higher state-level life satisfaction.

To reiterate, three relations are pertinent to the hypothesis of the present study: conservatism and life satisfaction, neuroticism and life satisfaction, and neuroticism and conservatism. It appears that higher conservatism leads to higher life satisfaction at the individual level (e.g., Jetten et al., 2013). As stated earlier, various potential explanations have been articulated for this relation (e.g., Jetten et al., 2013; Napier & Jost, 2008; Schlenker et al., 2012). It appears also that lower neuroticism leads to higher life satisfaction at both the individual level (e.g., Schimmack et al., 2004) and the state level (e.g., Rentfrow et al., 2009). Such a relation is assumed to spring directly from the characteristics of a person on the neuroticism dimension of personality. Characteristics at the high end such as not handling stress well, being emotionally unstable, and being easily upset are thought to promote lower life satisfaction while characteristics at the low end such as not being depressed or blue, being calm in tense situations, and not being moody seem conducive to higher life satisfaction. As well, it appears that lower neuroticism leads to higher conservatism at the individual level (e.g., Verhulst et al., 2012) and at the state level (e.g., McCann, 2014a). Various potential explanations also have been suggested for this relation (e.g., Gerber et al., 2010; McCann, 2014a; Vigil, 2010).

Given the nature of these three relations, it is possible that lower neuroticism not only fosters higher life satisfaction but also might account for the higher satisfaction stemming from higher conservatism. The theoretical and empirical connections between neuroticism and life satisfaction seem to have stronger foundation than those between conservatism and life satisfaction, and the theoretical and empirical links between lower neuroticism and higher conservatism also appear to be valid. If this is so, then statistically controlling for neuroticism might eliminate or substantially reduce the relation between conservatism and life satisfaction. However, conversely, one would not expect statistically controlling for conservatism to eliminate or reduce to the same degree as that of the relation between neuroticism and life satisfaction. In effect, neuroticism might be mediating the relation between conservatism and life satisfaction.

The overarching purpose of the present study is to determine the nature of the potential relations between neuroticism, conservatism, and satisfaction at the state level of analysis and how they might be explained. If a positive association between conservatism and life satisfaction exists at the state level of analysis, the present study seeks to provide data on which to evaluate the applicability of the following four different explanations for such a state-level connection: (a) the level of state resident neuroticism accounts for the relation between state levels of conservatism and life satisfaction, (b) the Napier and Jost (2008) view that the conservative tendency to rationalize inequality buffers the negative emotional impact of societal inequality, (c) the Jetten et al. (2013) view that higher conservative SES allows access to more group memberships and it is these memberships that enhance life satisfaction, and (d) the Schlenker et al. (2012) view that conservatives have higher life satisfaction because they have various characteristics implicated in superior mental health and adjustment.

The emerging perspective of geographical psychology (Rentfrow, 2010, 2014; Rentfrow et al., 2008) provided a theoretical foundation and spurred the present state-level study. Some social science researchers in disciplines such as political science, sociology, and economics base their inquiries on variables aggregated at the state level (e.g., Akai & Sakata, 2002; Erikson, Wright, & McIver, 1993; Vasseur, 2014). Geographical psychology can do the same but in the interest of using psychological theory and research to understand state-level variables and their relations. A macro-level research strategy in psychology is important because it has the potential within a specific context to broaden our knowledge of human behavior and foster greater integration of the results of aggregate-level and individual-level inquiry both within the discipline of psychology and across disciplinary boundaries.

Geographical psychology (e.g., Rentfrow et al., 2008) promotes the possibility of understanding aggregate-level relations as cumulative results of individual-level dispositions and processes. A fundamental assumption of geographical psychology (e.g., Rentfrow et al., 2008) is that an area’s aggregate position on a dispositional dimension reflects the central tendency of the area’s individuals on that dispositional dimension and is associated with the pervasiveness in that area of the psychological and behavioral tendencies associated with that dispositional dimension. However, Rentfrow et al. (2008) were aware of risks involved in cross-level extrapolation. When it is assumed without empirical verification that aggregate-level results generalize to the individual level, the “ecological fallacy” (Robinson, 1950) occurs; when it is assumed without empirical verification that individual-level results generalize to the aggregate level, the “compositional fallacy” (Pettigrew, 1997) occurs. In other words, there is a real possibility that relations at the aggregate level reflect and are dependent upon relations at the individual level but one ultimately should draw such conclusions from empirical verification rather than assumption.

There is no necessary correspondence between research results at the individual level and the aggregate level. However, psychologists often tacitly assume that such correspondence exists. Little effort is often focused on empirical verification for such assumptions. For example, most psychologists probably assume that prejudice and discrimination are likely to be higher in aggregates in which individuals tend to be more authoritarian, or for that matter, in aggregates in which individuals tend to be more prejudiced and discriminatory. However, we can only assume that this will be so in the absence of supportive aggregate-level data. Other aggregate-level factors might minimize, moderate, or eliminate such an association.

In addition to the key variables—life satisfaction, conservatism, and neuroticism—and state-level demographic indicators—SES, white population percent, and urban population percent—Gini coefficients of net income inequality for each state (Noss, 2010) also were included in the database because of the part that income inequality is thought to play in the relation between conservatism and life satisfaction in the explanation by Napier and Jost (2008). Gini coefficients range from 0 (perfect equality) to 1 (perfect inequality). In the present study, they serve as indicators of the gap between the rich and the poor in each state. The state Gini coefficients allowed the opportunity to determine whether state-level inequality related to state-level satisfaction. But more importantly in the present context, they permitted an exploration of the degree to which support for the conservative tendency to rationalize inequality might be operative in the present data. A significant interaction between state levels of conservatism and Gini Index values in which there was a tendency for residents of more conservative states to be similarly satisfied in low and high Gini conditions, for residents of more liberal states to be much less satisfied in high Gini conditions than in low Gini conditions, and for residents of conservative and liberal states to be similarly satisfied under low Gini conditions would lend support to the contention of Napier and Jost.

Of course, it also was desirable in the present study to control other state-level variables that might relate to state-level life satisfaction. Differences in life satisfaction at the individual level of analysis have been found to relate to SES (e.g., Luhmann, Murdoch, & Hawley, 2015), being white or not (e.g., Taylor et al., 2006), and living in an urban environment (e.g., Kron, 2012). State-level research also has shown that these variables sometimes correlate significantly with each other (e.g., McCann, 2011). Therefore, state SES, white population percent, and urban population percent were considered as potential covariates when testing the current hypothesis.

In a broad sense, it is important to know whether there is something inherent in conservatism that gives rise to greater life satisfaction or whether other variables correlated with conservatism and satisfaction can account for such a relation. It also is beneficial to understand how these dynamics operate at the aggregate level because we often refer to differing levels of conservatism and life satisfaction between geographical entities such as states and nations. For example, political commentators, pundits, analysts, and researchers (e.g., Erikson et al., 1993) tend to refer to such aggregates rather than individuals. It also is becoming increasingly common and important to refer to such aggregates when assessing and discussing variables gauging quality of life such as the Gallup-Healthways Well-Being Index (e.g., Gallup-Healthways, 2016).

The present inquiry was expected to be the first to predict and empirically verify that neuroticism could account for a relation between conservatism and life satisfaction at either the state level or the individual level of analysis. However, individual-level research with the same general thrust by Burton et al. (2015) was published while the present article was being completed. Burton et al. (2015) collected data from several hundred respondents in each of two online surveys conducted through Amazon’s Mechanical Turk service. In each study, higher conservatism was associated with greater life satisfaction, lower neuroticism was associated with greater life satisfaction, and lower neuroticism was associated with higher conservatism. Furthermore, in each study, the correlation between conservatism and life satisfaction was not significant when they controlled for neuroticism through partial correlation, thereby offering parallel support to the general hypothesis of the present state-level research in the context of individual-level research.

Method

Measures

Life satisfaction 2008–2010

Mendes (2012) provided state Life Evaluation Index (LEI) scores from the Gallup-Healthways Well-Being Index for 2008. Gallup-Healthways (2012a) provided scores for 2009 and Gallup-Healthways (2012b) provided scores for 2010. Gallup-Healthways surveyed 1000 or more persons daily through live phone interviews conducted each day except for major holidays with representative samples of Americans at least 18 years of age beginning in 2008 (Harter & Gurley, 2008). Responses to two items that asked respondents to rate their current life and their anticipated life five years in the future formed the basis for the LEI scores. Gallup-Healthways (2012c) provides further information about the LEI. In the present study, the mean of the three annual LEI values for each state formed a composite life satisfaction measure. The resulting 2008–2010 life satisfaction variable had a Cronbach’s alpha of .92.

Conservatism

Two sources provided data for a state conservatism composite. The first was the state ideological identification values determined from the answers of 84,119 respondents in all states except Alaska and Hawaii to national telephone polls conducted live from 1999 to 2003 by CBS News and the New York Times (CBS/NYT). The second was the presidential election results of 2000, 2004, and 2008. CBS/NYT pollsters asked the following question (Erikson et al., 1993, p. 14): “How would you describe your views on most political matters? Generally, do you think of yourself as liberal, moderate, or conservative?” Erikson et al. (1993) and Erikson, Wright, and McIver (2007) present further details regarding the data collection methodology, as well as reliability, validity, and temporal stability information. Wright (2012) has tabled the state values for 1976 to 2003 but only data for the years 1999 to 2003 were selected because they pertain to the most recent five-year period closest to the years of the data for the other variables in the present study. Only respondents indicating “liberal” or “conservative” were included in the present ideological variable computation. The conservative ideology variable was the mean percentage of conservatives minus the mean percentage of liberals over the years 1999 to 2003 in each of the 48 states. The resulting conservative ideology variable for the 48 states had a Cronbach’s alpha of .92. The mean of the conservative ideology composite for the 48 states served as a substitute value for Alaska and Hawaii.

The percentages in each state who voted Republican in the presidential elections of 2000 and 2004 were taken from the Statistical Abstract of the United States (U.S. Census Bureau, 2009) and the percentages for 2008 were taken from Leip (2012). The mean of the three percentages formed a mean Republican percent score for each state. This composite had a Cronbach’s alpha of .97. The created conservative ideology variable and the Republican percent election composite were highly correlated (r = .79, p < .001). Therefore, the two were converted to z scores and the mean was computed to form the state conservatism composite used in the analyses of the present study.

Neuroticism

Rentfrow et al. (2008) created state-aggregated z scores for neuroticism from the responses of 619,397 residents to the 44-item Big Five Inventory (John & Srivastava, 1999) in an Internet survey conducted between December of 1999 and January of 2005. Rentfrow et al. (2008) reported that the sample was generally representative based on comparisons to census data. State-level neuroticism scores from three random subsamples showed a high average inter-subsample correlation of .85. A test–retest correlation of .86 based on two temporal subsamples also demonstrated high convergence for the neuroticism dimension. The factor structure at the state level for the Big Five variables closely mirrored that usually reported for the individual level. The factor congruence coefficient for neuroticism was .95.

SES 2005–2010

The following five variables were used as constituents of a simple additive composite state SES variable: population percent 25 years and over with at least high school graduation, population percent 25 years and over with at least a Bachelor’s degree, personal income per capita in constant dollars, percent of individuals below the poverty line, and unemployment rate. Data for 2005 and 2010 were obtained from the U.S. Census Bureau (2007, 2008, 2012a). For 2010, data were not available for high school graduation, Bachelor’s degree, or poverty line, so the 2009 values served as substitutes. For the 50 states, all correlations between the 2005 and 2010 (or 2009) values were significant (p < .001): .94 for high school education, .95 for Bachelor’s degree, .93 for personal income, .92 for poverty level, and .50 for unemployment rate. Each of the five variables for each state in 2005 and in 2010 were converted to z scores and the sign was reversed for the poverty line and unemployment variables. The mean scores from the five respective z score variables for each state for 2005 and for 2010 were then calculated. Cronbach’s alpha was .86 for the 2005 composite and .84 for the 2010 composite. The two composites correlated highly, r(48) = .92, p < .001. The resulting state composite scores then were standardized and the 2005 and 2010 state composite z scores were added together and divided by 2 to produce an SES composite variable for the period 2005 to 2010.

White percent 2005–2010

The 2005 white population percent in each state was taken from white and total population figures (U.S. Census Bureau, 2007) and 2010 white percent was obtained from the U.S. Census Bureau (2012a). The 2005 and 2010 percents correlated highly, r(48) = .95, p < .001, so the mean of the two years served as the white percent of the population in each state in the present analyses.

Urban percent 2000–2010

The percent of each state’s population that was urban in the 2000 and 2010 census years was taken from the U.S. Census Bureau (2007, 2012b). The two variables rounded to a perfect positive correlation. The mean of the two formed the urban percent variable in the present work.

Gini Index 2008–2010

Noss (2010) provided Gini coefficients for each state for 2008, 2009, and 2010. The mean value over the three years served as the Gini variable in the present study. The Cronbach’s alpha of the Gini composite was .98. Higher scores indicate greater inequality of net income.

Analytic strategy

The analytic strategy had four planned steps: (1) The first step was to compute descriptive statistics for the 50 states on the seven state-level variables as well as a Pearson correlation matrix for the seven variables. (2) The second step was to use simultaneous multiple regression to determine which of the three demographic variables could independently account for variance in state levels of life satisfaction. (3) The third step was to determine through sequential multiple regression whether neuroticism could account for the relation between life satisfaction and conservatism with appropriate demographic controls. (4) The fourth step was to use sequential multiple regression to explore the relation of the state Gini Index values to life satisfaction alone and in interaction with conservatism with demographic variables controlled.

Two-tailed tests and an alpha level of .05 were used throughout the analyses. The sample and the population of the contiguous states are isometric. Therefore, descriptive statistics for the sample pertain directly to the population. Errors of inference from sample to population in this analysis are not a concern. Confidence intervals would be misleading because there cannot be a range of values in this context. However, p values have been reported largely as benchmarks for a sample of this size randomly drawn from a larger population.

Results

Table 1 displays the means, standard deviations, and Pearson correlations for the seven variables for the 50 states. Higher life satisfaction was significantly associated with lower neuroticism (r = −.53, p < .001). Higher conservatism also significantly related to lower neuroticism (r = −.28, p < .05). Life satisfaction did not correlate significantly with the Gini Index (r = −.22, p = .12) but did correlate significantly with SES (r = .40, p < .01), white percent² (r = −.48, p < .001), and urban percent (r = .43, p < .01). Neuroticism also correlated significantly with the Gini Index (r = .47, p < .001) but conservatism did not (r = −.18, p = .20). Most unexpected was the lack of correlation between life satisfaction and conservatism (r = −.01, p = .94).³
Table 1.
Means, standard deviations, pearson correlations, and partial correlations for the 50 states.

Variable M SD 1 2 3 4 5 6 7

1. Life satisfaction 45.87 3.45 1.00

2. Neuroticism .01 1.01 −.53* 1.00

(−.59)

3. Conservatism .00 .95 −.01 −.28 1.00

(.47) (−.54)

4. SES .00 .98 .40 −.25 −.51* 1.00

5. White percent .00 .99 −.48* −.04 .07 .15 1.00

6. Urban percent 72.64 14.71 .43 −.19 −.37** .31* −.38 1.00

7. Gini Index .45 .02 −.22 .47* −.18 −.34* −.29* .22 1.00

(−.38) (.50*) (−.38**)

Note. Partial correlations with SES, white percent, and urban percent controlled are in brackets.

*
p < .05. p < .01. p < .001.

A further examination of the directions of correlation for the demographic variables with life satisfaction and conservatism shows a pattern suggesting that the chosen demographic dimensions might serve as suppressor variables (Cohen, Cohen, West, & Aiken, 2003; Lowry, 2012; UNC, 2008) in the correlation between state levels of conservatism and life satisfaction. When a predictor correlates negatively with a criterion and positively with a second predictor, or when a predictor correlates positively with a criterion and negatively with a second predictor, the correlation between the second predictor and the criterion is likely to be suppressed and surface when the first predictor serves as a control in partial correlation (UNC, 2008, p. 5). Generally, this was the pattern of the correlations in the present data, especially for SES and urban percent. Furthermore, “material suppression effects are likely to be found in analyses of aggregate data … because of the small error variance that results in these conditions” (Cohen et al., 2003, p. 78).

Therefore, a partial correlation strategy tested and confirmed the suspected suppression of the correlation between conservatism and life satisfaction (UNC, 2008). With state differences in SES, urban population percent, and white population percent controlled, partial correlation revealed the expected relation between state levels of life satisfaction and conservatism, r_p(45) = .47, p < .001. The correlation between life satisfaction and neuroticism also increased to r_p(45) = −.59, p < .001, and the correlation between conservatism and neuroticism rose to r_p(45) = −.54, p < .001 (see Table 1).

Additional partial correlations were computed to determine which of the three demographic variables, singly or in a pair, were most central as suppressors of the correlation between conservatism and life satisfaction. Controlling for SES was most effective for a single demographic control model but still resulted in a nonsignificant correlation, r_p(47) = .24, p = .09. Controlling separately for white and urban population percent resulted in smaller correlations of r_p(47) = .03, p = .86, and r_p(47) = .18, p = .23, respectively. The best pair of demographic controls was SES and white population percent, which resulted in a correlation of r_p(46) = .44, p < .01. The SES and urban population percent pair also resulted in a significant correlation of r_p(46) = .37, p < .01. However, the pairing of white and urban population percent resulted in a nonsignificant correlation of r_p(46) = .16, p = .29. These partial correlation results also strongly suggested that such suppression effects would be in play with the same variables in a multiple regression framework.

As a first step in the regression analysis, a simultaneous equation was used to determine whether each of the three demographic variables was a significant independent predictor of state levels of life satisfaction (Equation 1, Table 2). The three together accounted for 46.4% of the variance in life satisfaction, F(3, 46) = 13.27, p < .001. The standardized β coefficients were .44 (t = 3.73, p < .001) for SES, − .51 (t = −4.17, p < .001) for white percent, and .10 (t = .79, p = .43) for urban percent. SES and white percent were significant independent predictors of life satisfaction, while urban percent was not.
Table 2
Sequential multiple regression equations with life satisfaction as the dependent variable.

Equation Step Entry Predictors df R² change F Final independent predictors β t

1 1 block 3 demographic variables 3, 46 .464 13.27 SES .44 3.73*

White Percent −.51 −4.17*

2 1 block 2 demographic variables 2, 47 .457 19.74* SES .47 4.04*

2 forced Conservatism 1, 46 .104 10.87 White percent −.58 −6.49*

3 forced Neuroticism 1, 45 .106 14.31* Neuroticism −.39 −3.78*

3 1 block 2 demographic variables 2, 47 .457 19.74* SES .47 4.04*

2 forced Neuroticism 1, 46 .196 25.91* White percent −.58 −6.49*

3 forced Conservatism 1, 45 .014 1.91 Neuroticism −.39 −3.78*

4 1 block 2 demographic variables 2, 47 .457 19.74* SES .40 3.67*

2 forced Gini Index 1, 46 .056 5.31* White percent −.62 −5.72***

Gini Index −.26 −2.31*

5 1 block 2 demographic variables 2, 47 .457 19.74* SES .50 3.53*

2 forced Conservatism 1, 46 .104 10.87 White percent^a −.67 −6.62*

3 forced Gini Index 1, 45 .013 1.35

4 forced Conservatism X Gini 1, 44 .032 3.52

a
Large significant β coefficients for conservatism (4.35) and the conservatism × Gini interaction variable (−4.10) also occurred because of their high collinearity but are not reported here because it is not logical to assess the contribution of conservatism after the interaction term is in the regression equation.

*
p < .05. p < .01. p < .001.

Multiple regression test of the general hypothesis

Two sequential multiple regression equations determined whether neuroticism could account for the demonstrated relation between state levels of conservatism and life satisfaction. In the first equation (Equation 2, Table 2), SES and white percent were entered as a block on the first step and controlled, conservatism was entered on the second step, and neuroticism was entered on the third step. The two demographic controls jointly accounted for 45.7% of the variance in life satisfaction, F(2, 47) = 19.74, p < .001, and conservatism accounted for another 10.4%, F(1, 46) = 10.87, p < .01. Neuroticism accounted for a final increment of 10.6%, F(1, 45) = 14.31, p < .001.

In the second sequential equation (Equation 3, Table 2), the two demographic variables were entered first, neuroticism was entered second, and conservatism was entered third. Now neuroticism accounted for 19.6% of the variance in life satisfaction, F(1, 46) = 25.91, p < .001, but conservatism only accounted for an additional non-significant increment of 1.4%, F(1, 45) = 1.91, p = .17. Therefore, the results supported the general hypothesis. That is, with state resident neuroticism controlled, there was no significant relation between state-level conservatism and life satisfaction.^4–6

Analyses involving the Gini Index in relation to the Napier and Jost (2008) explanation

The potential role of the Gini Index was explored as a sole predictor of life satisfaction and as a predictor interacting with conservatism in relation to life satisfaction. A sequential multiple regression equation was computed with SES and white population percent entered as a block on the first step and the Gini Index entered on the second (Equation 4, Table 3). Above the 45.7% of the variance in life satisfaction accounted for by the two demographic controls, the Gini Index accounted for another 5.6%, F(1, 46) = 5.31, p < .05. The β coefficient for the Gini Index was − .26 (t = −2.31, p < .05), indicating that state resident life satisfaction tended to be higher in states where income inequality was lower.
Table 3.
Predicted life satisfaction with the mean of 0 for SES and white percent, and values of +1 SD and −1 SD for conservatism and the Gini Index.

Low Gini coefficient High Gini coefficient

High conservatism 8.66 .04

Low conservatism −8.24 −.46

To provide critical information regarding the Napier and Jost (2008) explanation, another sequential multiple regression equation determined whether there was an interaction between state conservatism and the Gini Index when SES and white population percent were controlled. SES and white population percent entered as a block on the first step, conservatism entered second, the Gini variable entered third, and the product of the conservatism variable and the Gini Index entered on the final step (Equation 5, Table 3). Beyond the 45.7% of the variance in life satisfaction accounted for by the two demographic controls and the 10.4% accounted for by conservatism, the Gini Index accounted for only an additional non-significant 1.3%, F(1, 45) = 1.35, p = .25. The interaction effect was near significance and accounted for a final 3.2% of the variance in life satisfaction, F(1, 44) = 3.52, p = .07.

This nearly significant interaction prompted an exploration of its nature. The β coefficients were 4.35 (t = 2.02, p < .05) for conservatism, −.21 (t = −1.73, p = .09) for the Gini Index, and − 4.10 (t = −1.88, p = .07) for the product term. Predicted values on life satisfaction were calculated with the mean of 0 for SES and white percent, and high values of +1 SD and low values of −1 SD for conservatism and the Gini Index. Table 3 displays the results. Contrary to expectation based on the work of Napier and Jost (2008), it appears that there was a tendency for residents of more conservative states to be happiest in low Gini conditions and for residents of less conservative states to be least happy in low Gini conditions. However, under high Gini conditions, there was relatively little difference in the life satisfaction of residents of more conservative or less conservative states. Generally, these state-level data provided no support whatever for the position of Napier and Jost (2008).

Discussion

The present state-level results support the hypothesis that neuroticism can account for the association between higher conservatism and higher life satisfaction. Partial correlation demonstrated that state SES, white population percent, and urban population percent suppress the correlation between conservatism and life satisfaction at the state level. However, conservatism accounted for another 10.4% of the variance in life satisfaction when SES and white population percent—the two non-redundant predictors of life satisfaction—were controlled in a sequential multiple regression equation. More importantly, when neuroticism entered the equation directly after state SES and white population, and conservatism entered on the final step, neuroticism accounted for another 19.6% of the variance in life satisfaction but conservatism only accounted for an additional non-significant 1.4% of the variance. Therefore, with states as the cases, state resident neuroticism clearly had the capacity to displace resident conservatism as a predictor of state resident life satisfaction. Conversely, conservatism could not account for the neuroticism-satisfaction link.

The present study did not offer state-level support to three other suggested alternative interpretations of why conservatives tend to have greater life satisfaction. For example, the exploratory Gini-centered analysis did not produce results consistent with the Napier and Jost’s (2008) conclusion based on system justification of inequality. In their view, inequality takes less of a toll on conservatives because of their ability to rationalize it. This results in the greater life satisfaction of conservatives, especially in conditions of high economic inequality. In the present research, there was evidence that life satisfaction was higher in conservative states and in states with lower Gini coefficients. However, the critical interaction between conservatism and income inequality was not significant at the .05 level and the interactive pattern (p = .07) generally was contrary* to that found by Napier and Jost. The results suggested that residents of more conservative states are much less satisfied under high inequality conditions than under low inequality conditions and that residents of more liberal states are much more satisfied in high inequality conditions than in low inequality conditions. As well, under low inequality conditions, residents of more conservative states are much more satisfied compared to residents of less conservative states. Of course, extreme caution in interpretation is necessary because this analysis is purely exploratory and based on a non-significant interaction.

The Jetten et al. (2013) position is that the generally higher SES of conservatives makes more group memberships possible and the greater number of group memberships enhances life satisfaction. However, in the present study, the relation between state conservatism and life satisfaction was negative and persisted when the state SES composite based on three economic and two educational variables was statistically held constant. Therefore, this explanation does not stand up to empirical scrutiny with the present dataset. Of course, the present analysis does not contain a state-level assessment of the number of group memberships but it does control state-level SES which suggests state-level control of the number of group memberships, as suggested by the individual-level work of Jetten et al.

The Schlenker et al. (2012) explanation too received little support. Their view is that conservatives’ greater life satisfaction occurs because they have characteristics involved in superior mental health and adjustment. They suggest that these differences between conservatives and liberals involve dimensions such as optimism, personal control, trust, religiosity, and moral clarity. However, neuroticism also is an element, or contains elements somewhat implicated in inferior mental health and adjustment (e.g., McCann, 2011), but it was the sole variable necessary to account for the relation between state resident conservatism and life satisfaction in the present state-level study.

It should be noted that the Pearson correlation between conservatism and life satisfaction in the present state-level analysis was virtually zero but rose to .47 when SES, white population percent, and urban population percent were controlled in a partial correlation. The three demographic variables, especially SES, acted as suppressors of the relation between conservatism and life satisfaction at the state level of analysis. More specifically, SES in combination with white population percent mostly accounted for the suppression effect, yielding a .44 between conservatism and life satisfaction. Controlling SES alone produced a non-significant partial correlation of only .24.

The results suggest that the correlation between conservatism and life satisfaction is masked in state-level analyses primarily because higher SES is associated with higher life satisfaction and lower conservatism, and because a higher white population percent tends to be associated with lower life satisfaction and somewhat higher conservatism. The combination of these factors results in the suppression of the positive correlation between conservatism and life satisfaction. Therefore, when SES and white population percent are controlled in multiple regression, the masked relation between conservatism and life satisfaction is revealed. These outcomes emphasize the crucial nature of controlling for SES and white population percent in any study involving life satisfaction and related constructs in state-level analyses.

The relation between SES and life satisfaction probably reflects the corresponding relation which robustly occurs at the individual level (e.g., Luhmann, Schimmack, & Eid, 2011), while the relations involving SES, life satisfaction, conservatism, and white population percent at the state level have probably resulted from an historical combination of factors including but not limited to migration patterns, ethnic and racial differences, distribution of natural resources, and economic development opportunities. However, to the extent that such relations exist in a sample of individuals, a somewhat similar pattern of suppression may exist if there is a sufficient range of conservatism and life satisfaction scores and the error variances are not too large. Therefore, it is possible that the demonstrated individual-level relations may be considerably higher than often reported. Perhaps a suppressed relation at either level of analysis can hide a rather different real relation between conservatism and life satisfaction.

The recent individual-level study by Burton et al. (2015) bolsters confidence in the position advocated here that individual-level dynamics concerning neuroticism are responsible for the presently found state-level relations. Individual-level results from Burton et al. (2015) and state-level results from the present study support parallel relations between conservatism, neuroticism, and life satisfaction at each level of analysis. However, Rentfrow et al. (2008) have cautioned that although relations in an aggregate-level and an individual-level analysis may be consistent, it is still possible, but much less likely, that the parallel relations remain logically independent.

Strengths and limitations

The present study has apparent strengths and limitations. For example, large representative state samples and official government state data constituted the basis for the state-aggregated variables. Assessment samples were sometimes different but this should not be perceived as detrimental to the study. If individual-level variable relations are at the root of the state-level relations, sample membership incongruence could have reduced the chances of significant state-level results. Also, personality and life satisfaction data were collected a few years apart and the impact of the economic recession on life satisfaction could have altered and dampened the relation between personality and life satisfaction. However, the fact that the study produced significant state-level relations despite such deficiencies of sample and temporal congruence should bolster confidence in the robustness of these relations.

Some readers may think that the small sample does not seem appropriate for conducting multiple regression analysis. Inference concerns regarding the instability of regression coefficients tend to be associated with such analysis on small samples with several predictors. However, the sample and the population of 50 states are isometric in this study. The descriptive statistics for the “sample” pertain directly to the “population.” Potential errors of inference based on the size and stability of regression coefficients is not really a concern here because whatever the coefficient in the “sample,” the estimate for the “population” is the same. Therefore, confidence intervals for this research would be misleading but the p values have been retained as a benchmark for what would be found with a sample of this size randomly drawn from some much larger population. Based on the rationale for inferential statistics, some even doubt the need for their computation in this circumstance (e.g., see Simonton, 1984). Nevertheless, such multiple regression strategies have been successful in past small-sample studies in which the units of analysis have been individuals (e.g., McCann, 1992, 1997; Simonton, 1986) and states (e.g., McCann, 2008, 2014a; Pesta, Bertsch, McDaniel, Mahoney, & Poznanski, 2012).

Of course, the usual warning that one cannot infer causation from correlation-based research pertains here. Although the present report suggests that higher neuroticism fosters lower life satisfaction, that lower neuroticism promotes higher conservatism, and that higher conservatism consequently leads to greater life satisfaction, one cannot draw sound directional conclusions from the reported correlations and multiple regressions. It remains theoretically possible that unspecified and unknown concomitants of neuroticism, conservatism, and life satisfaction are responsible for relations between these variables. Genetic endowment shared by neuroticism, ideological orientation, and life satisfaction also could lie at the causal root of these relations (e.g., Loehlin, McCrae, Costa, & John, 1998; Verhulst et al., 2012).

Some caution also is in order in the interpretation of the effect sizes of relations between conservatism and life satisfaction. Burton et al. (2015) noted that such correlations at the individual level tend to be relatively small. In contrast, the relations found here in the partial correlations and the multiple regressions tend to be rather large. This difference is at least partially a common finding resulting from the process of aggregation (e.g., Erikson et al., 1993; Ostroff, 1993; Rushton, Brainerd, & Pressley, 1983). Measurement errors tend to cancel out during aggregation and this leads to correlations after aggregation that are larger than correlations based on the original individual cases.⁶

Future research

Empirical inquiry in this area could take several paths. For example, at the aggregate level of analysis, researchers could examine the degree of generalization to other countries and their internal geographical divisions (e.g., the 380 Local Authority Districts of Great Britain, see Rentfrow, Jokela, & Lamb, 2015). Researchers also could conduct a similar analysis with a sample of nations as the geographical units of analysis.

Of course, more studies utilizing these same variables at the individual level of analysis are encouraged as well to demonstrate the speculated link between the aggregate and state levels. At present, the only individual-level study of parallel relations is the one by Burton et al. (2015). Furthermore, to be more conclusive about individual-level dispositional dynamics being directly responsible for the results found in the present study of the states, researchers could eventually conduct a multilevel analysis. This would involve a hierarchically integrated state-and-individual level design utilizing individual-level neuroticism, conservatism, life satisfaction, and control variable data pertaining to the same persons in a large sample with adequate representation from each of the 50 states. It then would be statistically possible to demonstrate empirically the degree to which state-level relations are dependent upon individual-level relations. However, a suitable database does not yet exist. Such a large-scale study would be costly and time consuming but it could be conducted more economically if it were planned well to also contain assessments of other variables that researchers could use to test other unrelated hypotheses.

Conclusion

In recent years, substantial strides have been made in understanding the dispositional roots of ideological preferences (e.g., Jost, Glaser, Kruglanski, & Sulloway, 2003; Mondak, 2010). Contrary to the common attributions of political pundits, electorate dispositional factors do predict ideological orientation and political party preference to a greater degree than previously thought (e.g., Jost et al., 2009; McCann, 2014a, 2014b; Rentfrow et al., 2009). The present study extends this important line of empirical inquiry to show that the lower neuroticism of residents in more conservative states can account for the phenomenon of elevated life satisfaction among the residents of more conservative states, and the results of Burton et al. (2015) show essentially the same dynamics at the individual level of analysis. The results of the present study show that conservatives, compared to liberals, may be happier and proclaim greater satisfaction with life simply because they tend to have higher levels of abiding emotional stability, or to express it in a different manner, lower levels on the neuroticism dimension of the Big Five personality factors.

Variable	M	SD	1	2	3	4	5	6	7
1. Life satisfaction	45.87	3.45	1.00
2. Neuroticism	.01	1.01	−.53***	1.00
			(−.59***)
3. Conservatism	.00	.95	−.01	−.28*	1.00
			(.47***)	(−.54***)
4. SES	.00	.98	.40**	−.25	−.51***	1.00
5. White percent	.00	.99	−.48***	−.04	.07	.15	1.00
6. Urban percent	72.64	14.71	.43**	−.19	−.37**	.31*	−.38**	1.00
7. Gini Index	.45	.02	−.22	.47***	−.18	−.34*	−.29*	.22	1.00
			(−.38**)	(.50***)	(−.38**)

Equation	Step	Entry	Predictors	df	R² change	F	Final independent predictors	β	t
1	1	block	3 demographic variables	3, 46	.464	13.27***	SES	.44	3.73***
							White Percent	−.51	−4.17***
2	1	block	2 demographic variables	2, 47	.457	19.74***	SES	.47	4.04***
	2	forced	Conservatism	1, 46	.104	10.87**	White percent	−.58	−6.49***
	3	forced	Neuroticism	1, 45	.106	14.31***	Neuroticism	−.39	−3.78***
3	1	block	2 demographic variables	2, 47	.457	19.74***	SES	.47	4.04***
	2	forced	Neuroticism	1, 46	.196	25.91***	White percent	−.58	−6.49***
	3	forced	Conservatism	1, 45	.014	1.91	Neuroticism	−.39	−3.78***
4	1	block	2 demographic variables	2, 47	.457	19.74***	SES	.40	3.67***
	2	forced	Gini Index	1, 46	.056	5.31*	White percent	−.62	−5.72***
							Gini Index	−.26	−2.31*
5	1	block	2 demographic variables	2, 47	.457	19.74***	SES	.50	3.53***
	2	forced	Conservatism	1, 46	.104	10.87**	White percent^a	−.67	−6.62***
	3	forced	Gini Index	1, 45	.013	1.35
	4	forced	Conservatism X Gini	1, 44	.032	3.52

	Low Gini coefficient	High Gini coefficient
High conservatism	8.66	.04
Low conservatism	−8.24	−.46

Footnotes

Article Notes

References

Akai

Sakata

(2002) Fiscal decentralization contributes to economic growth: Evidence from state-level cross-section data for the United States. Journal of Urban Economics 52: 93–108.

Alford

J. R.

Hibbing

J. R.

(2007) Personal, interpersonal, and political temperaments. Annals of the American Academy of Political and Social Science 614: 196–212.

Barbaranelli

Caprara

G. V.

Vecchione

Fraley

C. R.

(2007) Voters’ personality traits in presidential elections. Personality and Individual Differences 42: 1199–1208.

Breakwell

G. M.

Fife-Schaw

Devereux

J. D.

(2001) The relationship of self-esteem and attributional style to young peoples’ worries. Journal of Psychology: Interdisciplinary and Applied 122: 207–215.

Burton

C. M.

Plaks

J. E.

Peterson

J. B.

(2015) Why do conservatives report being happier than liberals? The contribution of neuroticism. Journal of Social and Political Psychology 3: 89–102.

Caprara

G. V.

Barbaranelli

Zimbardo

P. G.

(1999) Personality profiles and political parties. Political Psychology 20: 175–197.

Carney

D. R.

Jost

J. T.

Gosling

S. D.

Potter

(2008) The secret lives of liberals and conservatives: Personality profiles, interaction styles, and the things they leave behind. Political Psychology 29: 807–840.

Cohen

West

S. G.

Aiken

L. S.

(2003) Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.), Mahwah, NJ: Lawrence Erlbaum.

Costa

P. T.

McCrae

R. R.

(1995) Domains and facets: Sequential personality assessment using the Revised NEO Personality Inventory. Journal of Personality Assessment 64: 21–50.

10.

DeNeve

K. M.

Cooper

(1998) The happy personality: A meta-analysis of 137 personality traits and subjective well-being. Psychological Bulletin 124: 197–229.

11.

Diener

Oishi

Lucas

R. E.

(2003) Personality, culture, and subjective well-being: Emotional and cognitive evaluations of life. Annual Review of Psychology 54: 403–425.

12.

Erikson

R. S.

Wright

G. C.

McIver

J. P.

(1993) Statehouse democracy: Public opinion and policy in the American states, New York, NY: Cambridge University Press.

13.

Erikson

R. S.

Wright

G. C.

McIver

J. P.

(2007) Measuring the public’s ideological preferences in the 50 states: Survey responses versus roll call data. State Politics and Policy Quarterly 7: 141–151.

14.

Goldberg

L. R.

(1990) An alternative “description of personality:” The Big-Five factor structure. Journal of Personality and Social Psychology 59: 1216–1229.

15.

Gallup-Healthways. (2012a). 2009 WBI aggregate by state. Retrieved from wbi.meyouhealth.com.

16.

Gallup-Healthways. (2012b). 2011 city, state, and congressional district well-being reports. Retrieved from www.well-beingindex.com/stateCongressDistrictRank.asp.

17.

Gallup-Healthways. (2012c). Gallup-Healthways well-being index. Retrieved from www.well-beingindex.com/files/Gallup-Healthways Index Methodology Report FINAL 3-25-08.pdf.

18.

Gallup-Healthways. (2016). Gallup-Healthways well-being index. Retrieved from www.well-beingindex.com.

19.

Gerber

A. S.

Huber

G. A.

Doherty

Dowling

C. M.

S. E.

(2010) Personality and political attitudes: Relationships across issue domains and political contexts. American Political Science Review 104: 111–133.

20.

Gritching

W. L.

(1983) Domain, scope and degree of happiness. British Journal of Social Psychology 22: 247–260.

21.

Harter, J. K., & Gurley, V. F. (2008). Measuring well-being in the United States. Observer, 21, 17–20. Retrieved from https://www.psychologicalscience.org/observer/getArticle?id=2394.

22.

Iyer, R. (2011). Are liberals more neurotic than conservatives? Retrieved from http://www.polipsych.com/2011/02/11/liberals-conservatives-neuroticism-happiness/.

23.

Jetten

Haslam

S. A.

Barlow

F. K.

(2013) Bringing back the system: One reason why conservatives are happier than liberals is that higher socioeconomic status gives them access to more group memberships. Social Psychological and Personality Science 4: 6–13.

24.

John

O. P.

Srivastava

(1999) The Big Five Trait taxonomy: History, measurement, and theoretical perspectives. In: Pervin

L. A.

John

O. P.

(eds) Handbook of personality: Theory and research, 2nd ed. New York, NY: Guilford, pp. 102–139.

25.

Joshanloo

M. A.

Afshari

(2011) Big Five personality traits and self-esteem as predictors of life satisfaction in Iranian Muslim university students. Journal of Happiness Studies 12: 105–113.

26.

Jost

J. L.

Glaser

Kruglanski

A. W.

Sulloway

F. J.

(2003) Political conservatism as motivated social cognition. Psychological Bulletin 129: 339–375.

27.

Jost

J. T.

West

T. V.

Gosling

S. D.

(2009) Personality and ideology as determinants of candidate preferences and “Obama conversion” in the 2008 U.S. presidential election. DuBois Review 6: 103–124.

28.

Kabanoff

Ashton

(1984) Conservatism, locus of control and life satisfaction: A person/environment fit analysis. Australian Psychologist 19: 19–23.

29.

Kampfe

Mitte

(2010) Tell me who you are and I will tell you how you feel? European Journal of Personality 24: 291–308.

30.

Keyes

C. L. M.

Shmotkin

Ryff

C. D.

(2002) Optimizing well-being: The empirical encounter of two traditions. Journal of Personality and Social Psychology 82: 1007–1022.

31.

Kron, J. (November 30, 2012). Red state, blue city: How the urban-rural divide is splitting America. The Atlantic. Retrieved from http://www.theatlantic.com/politics/archive/ 2012/11/red-state-blue-city-how-the-urban-rural-divide-is-splitting-america/265686/.

32.

Leip, D. (2012). Dave Leip’s atlas of U.S. presidential elections. Retrieved from uselectionatlas.org.

33.

Loehlin

J. C.

McCrae

R. R.

Costa

P. T.

John

O. P.

(1998) Heritabilities and common and measure-specific components of the Big Five personality factors. Journal of Research in Personality 32: 431–453.

34.

Lowry, R. (2012). Subchapter 3a. Partial correlation. Retrieved from vassarstats.net/textbook/ch3a.html.

35.

Luhmann

Murdoch

J. C.

Hawkley

L. C.

(2015) Subjective well-being in context: County- and state-level socioeconomic factors and individual moderators. Social Psychological and Personality Science 6: 148–156.

36.

Luhmann

Schimmack

Eid

(2011) Stability and variability in the relationship between subjective well-being and income. Journal of Research in Personality 45: 186–197.

37.

McCann

S. J. H.

(1992) Alternative formulas to predict the greatness of U.S. presidents: Personological, situational, and zeitgeist factors. Journal of Personality and Social Psychology 62: 469–479.

38.

McCann

S. J. H.

(1997) Threatening times, ‘strong’ presidential popular vote winners, and the margin of victory (1824–1964). Journal of Personality and Social Psychology 73: 160–170.

39.

McCann

S. J. H.

(2008) Societal threat, authoritarianism, conservatism, and U.S. state death penalty sentencing (1977-2004). Journal of Personality and Social Psychology 94: 913–923.

40.

McCann

S. J. H.

(2010) Subjective well-being, personality, demographic variables, and American state differences in smoking prevalence. Nicotine & Tobacco Research 12: 895–904.

41.

McCann

S. J. H.

(2011) Emotional health and the Big Five personality factors at the American state level. Journal of Happiness Studies 12: 547–560.

42.

McCann

S. J. H.

(2014a) Big Five personality differences and political, social, and economic conservatism: An American state-level analysis. In: Rentfrow

P. J.

(ed.) Geographical psychology: Exploring the interaction of environment and behaviour, Washington, DC: American Psychological Association, pp. 139–160.

43.

McCann

S. J. H.

(2014b) Neuroticism and state differences in partisanship in the USA: Emotional stability, ideological stability, and Republican preference. Journal of Social and Political Psychology 2: 242–267.

44.

McCann

S. J. H.

(2014c) Happy Twitter tweets are more likely in American states with lower levels of resident neuroticism. Psychological Reports 114: 891–895.

45.

Mendes, E. (2012). Wellbeing: Hawaii tops Utah for nation’s best. Retrieved from http://www.gallup.com/poll/125849/hawaii-tops-utah-nation-best.aspx.

46.

Mondak

J. J.

(2010) Personality and the foundations of political behavior, New York, NY: Cambridge University Press.

47.

Mondak

J. J.

Halperin

K. D.

(2008) A framework for the study of personality and political behavior. British Journal of Political Science 38: 335–362.

48.

Napier

J. L.

Jost

J. T.

(2008) Why are conservatives happier than liberals? Psychological Science 19: 565–572.

49.

Noss, A. (2010). Household income for states: 2008 and 2009. Retrieved from www.census.gov/prod/2010pubs/acsbr09-2.pdf.

50.

Onraet

Van Hiel

Dhont

(2013) The relationship between right-wing ideological attitudes and psychological well-being. Personality and Social Psychology Bulletin 39: 509–522.

51.

Ostroff

(1993) Comparing correlations based on individual-level and aggregate-level data. Journal of Applied Psychology 78: 569–582.

52.

Pesta

B. J.

Bertsch

McDaniel

M. A.

Mahoney

C. B.

Poznanski

P. J.

(2012) Differential epidemiology: IQ, neuroticism, and chronic disease by the 50 U.S. states. Intelligence 40: 107–114.

53.

Peterson

S. A.

Maiden

(1993) Personality and politics among older Americans: A rural case study. International Journal of Aging and Human Development 36: 157–169.

54.

Pettigrew

T. F.

(1997) Personality and social structure: Social psychological contributions. In: Hogan

Johnson

J. A.

Briggs

S. R.

(eds) Handbook of personality psychology, San Diego, CA: Academic Press, pp. 417–438.

55.

Rentfrow

P. J.

(2010) Statewide differences in personality: Toward a psychological geography of the United States. American Psychologist 65: 548–558.

56.

Rentfrow, P. J. (Ed.). (2014). Geographical psychology: Exploring the interaction of environment and behavior. Washington, DC: American Psychological Association.

57.

Rentfrow

P. J.

Gosling

S. D.

Potter

(2008) A theory of the emergence, persistence, and expression of geographic variation in psychological characteristics. Perspectives on Psychological Science 3: 339–386.

58.

Rentfrow

P. J.

Jokela

Lamb

M. E.

(2015) Regional personality differences in Great Britain. PLoS One 10: e0122245.

59.

Rentfrow

P. J.

Jost

J. T.

Gosling

S. D.

Potter

(2009) Statewide differences in personality predict voting patterns in 1996–2004 U.S. presidential elections. In: Jost

J. T.

Kay

A. C.

Thorisdottir

(eds) Social and psychological bases of ideology and system justification, Oxford, England: Oxford University Press, pp. 314–347.

60.

Rentfrow

P. J.

Mellander

Florida

(2009) Happy states of America: A state-level analysis of psychological, economic, and social well-being. Journal of Research in Personality 43: 1073–1082.

61.

Robinson

W. S.

(1950) Ecological correlations and the behavior of individuals. American Sociological Review 15: 351–357.

62.

Rushton

J. P.

Brainerd

C. J.

Pressley

(1983) Behavioral development and construct validity. Psychological Bulletin 94: 18–38.

63.

Schimmack

Oishi

Furr

R. M.

Funder

D. C.

(2004) Personality and life satisfaction: A facet-level analysis. Personality and Social Psychology Bulletin 30: 1062–1075.

64.

Schlenker

B. R.

Chamber

J. R.

B. M.

(2012) Conservatives are happier than liberals, but why? Political ideology, personality, and life satisfaction. Journal of Research in Personality 46: 127–146.

65.

Simonton

D. K.

(1984) Genius, creativity, and leadership: Historiometric inquiries, Cambridge, MA: Harvard University Press.

66.

Simonton

D. K.

(1986) Presidential personality: Biographical use of the Gough Adjective Check List. Journal of Personality and Social Psychology 51: 149–160.

67.

Stavrova

Luhmann

(2016) Are conservatives happier than liberals? Not always and not everywhere. Journal of Research in Personality 63: 29–35.

68.

Taylor, P., Funk, C., & Craighill, P. (February 13, 2006). Are we happy yet? Retrieved from http://www.pewsocialtrends.org/2006/02/13/are-we-happy-yet/.

69.

UNC. (2008). Partial correlation and suppressor variables. Retrieved from http://www.unc.edu/courses/2008spring/psyc/270/001/partials.html.

70.

U.S. Census Bureau (2007) Statistical Abstract of the United States, Washington, DC: U.S. Government Printing Office.

71.

U.S. Census Bureau (2008) Statistical Abstract of the United States, Washington, DC: U.S. Government Printing Office.

72.

U.S. Census Bureau (2009) Statistical Abstract of the United States, Washington, DC: U.S. Government Printing Office.

73.

U.S. Census Bureau (2012a) Statistical Abstract of the United States, Washington, DC: U.S. Government Printing Office.

74.

U.S. Census Bureau. (2012b). 2010 census urban and rural classification and urban area criteria. Retrieved from http://www.census.gov/geo/www/ua/2010urbanruralclass.html.

75.

Valdmanis

V. G.

(2015) Factors affecting well-being at the state level in the United States. Journal of Happiness Studies 16: 985–987.

76.

Vasseur

(2014) Convergence and divergence in renewable energy policy among US states from 1998 to 2011. Social Forces 92: 1637–1657.

77.

Verhulst

Eaves

L. J.

Hatemi

P. K.

(2012) Correlation not causation: The relationship between personality traits and political ideologies. American Journal of Political Science 56: 34–51.

78.

Vigil

J. M.

(2010) Political leanings vary with facial expression processing and psychosocial functioning. Group Processes and Intergroup Relations 13: 547–558.

79.

Wright, G. C. (2012). Zip file of the CBS/New York Times national polls, ideology party identification, 1976-2003 [Stata 7.0 format]. Retrieved from http://php.indiana.edu/∼wright1/.