This is an open access article distributed under the terms of the Creative Commons
Attribution License (A 2006 public-opinion survey found that Republicans in the United States have been more likely than Democrats to report being “very happy” every year since 1972 (Taylor, Funk, & Craighill, 2006). Other researchers have reported similar findings (Jetten, Haslam, & Barlow, 2013; Napier & Jost, 2008; Schlenker, Chambers, & Le, 2012), although the associations between measures of political orientation (PO) and different measures of well-being or happiness are typically small, vary according to the specific measure of PO, and appear largely restricted to United States samples (Onraet, van Hiel, & Dhont, 2013a). Nonetheless, the suggestion of a reliable link between conservatism and happiness has drawn efforts to explain this association.
In one influential paper, Napier and Jost (2008) posited that U.S. conservatives report being happier than liberals because conservatives hold beliefs justifying current societal affairs and inequalities. Indeed, Napier and Jost (2008) found that endorsement of system-justifying and meritocratic beliefs accounted for the relationship between measures of happiness and conservatism, even after controlling for income, marital status and church attendance.
Other researchers have focused less on the specific beliefs of conservatives and liberals, and more on endemic qualities of each group that are removed from politics per se. For example, Jetten and colleagues (2013) suggested that the conservative-liberal happiness gap could be better explained by conservative students’ higher socioeconomic class and greater number of group memberships. In other recent work, Schlenker and colleagues (2012) suggested that the happiness gap between conservatives and liberals is due more to conservatives’ higher scores on predictors of mental health and positive adjustment (e.g., higher personal agency, transcendent moral beliefs, positive outlook) than to differences in the motivation to justify inequalities or maintain the status quo.
In this article, we build on such findings to propose that differences between conservatives and liberals in basic personality traits may provide an additional, potent explanation for the link between happiness and political orientation in North American populations. That is, while conservatives and liberals may reliably differ in specific beliefs related to happiness (e.g. system justifying beliefs), they may also differ in broad emotional, cognitive, and behavioral tendencies, independently of specific beliefs or socioeconomic status. These stable differences in personality traits may, in turn, play a powerful role in shaping both individuals’ happiness and individuals’ choice of which side of the political spectrum to endorse.
Previous research has found, for example, that conservatives score higher than liberals in conscientiousness (Caprara, Barbaranelli, & Zimbardo, 1999; Carney, Jost, Gosling, & Potter, 2008; Jost, 2006) and lower in openness (Carney et al., 2008; Jost, Glaser, Kruglanski, & Sulloway, 2003; McCrae, 1996; van Hiel, Kossowska, & Mervielde, 2000). The current studies specifically consider the role of neuroticism in differentiating both liberals from conservatives and people who are more or less satisfied with their lives.
Neuroticism and Conservatism
Neuroticism has been theorized to be negatively related to conservatism (Carney et al., 2008). Although some studies find null associations between the two (e.g. Alford & Hibbing, 2007; Caprara, Barbaranelli, & Zimbardo, 1999; Carney et al., 2008; Schlenker et al., 2012), others report evidence that conservatism, or components of conservative ideology, are associated with lower neuroticism (or higher emotional stability; Mondak & Halperin, 2008; Verhulst, Eaves, & Hatemi, 2012).
Recently, it has been demonstrated that more conservative individuals are more likely to perceive external threat (threats, danger, or chaos perceived in society) but not necessarily internal threat (threats perceived in one’s personal life, based on measures including a neuroticism measure; Onraet, van Hiel, Dhont, & Pattyn, 2013b). In one study (Study 3), when the researchers controlled for external threat, conservatism and internal threat were negatively correlated with one another. The authors concluded that the apparent association between general perception of threat and conservatism is due to the association between conservative PO and external threat rather than internal threat. Thus, these studies constitute indirect support of the idea that conservatism may be negatively related to neuroticism.
Why might one expect there to be a negative association between neuroticism and conservatism? Some have argued that an individual’s level of neuroticism may shape the political attitudes he/she adopts. For example, Mondak and Halperin (2008) suggested that emotional stability at a general level leads people to fret less over societal injustices or inequalities. This, in turn, increases the likelihood of identifying with conservative ideology (Jost et al., 2003). More recently, researchers have reported evidence of a common genetic influence on both personality traits and political attitudes (Verhulst et al., 2012). Whatever the reason for a potential association between conservative ideology and personality traits, to the extent that neuroticism and political ideology are negatively associated, neuroticism constitutes a viable explanatory factor for the association between political ideology and psychological well-being.
Personality Traits and Subjective Well-Being
Subjective well-being is a multi-component construct, comprised of positive affect, negative affect, and life satisfaction (Diener, 2000). Positive and negative affect constitute an emotional component of well-being, and life satisfaction a cognitive-evaluative component; one’s sense of the overall quality of one’s life (Diener, Emmons, Larsen, & Griffin, 1985; Diener, Suh, Lucas, & Smith, 1999). Personality traits are strongly and consistently related to the components of subjective well-being, over and above the influence of specific life events (Magnus & Diener, 1991). For example, trait conscientiousness and extraversion are positively related to life-satisfaction (DeNeve & Cooper, 1998; Hayes & Joseph, 2003). On the other hand, neuroticism is negatively associated with life-satisfaction (Diener et al., 1985; Vittersø & Nilsen, 2002) and positively associated with negative affect (Costa & McCrae, 1980). Thus, happier people are typically higher in extraversion and lower in neuroticism (Costa & McCrae, 1980; DeNeve & Cooper, 1998).
The Present Research
Here, we focus on neuroticism. We hypothesized that the relationship between conservatism and satisfaction with life would be at least partially accounted for by trait neuroticism. Although Schlenker and colleagues (2012) did include the 10-item Big Five Index – Brief Version (Rammstedt & John, 2007) in their first of four studies, we felt it important to test the relationships among the Big Five personality traits, satisfaction with life, and political orientation using more comprehensive and robust measures. We administered a well-validated 100-item measure of personality (Big Five Aspect Scale (BFAS); DeYoung, Quilty, & Peterson, 2007) and assessed how related (if at all) political orientation was to happiness, once the Big Five traits were thoroughly controlled for.
In two studies, we operationalized “happiness” through life satisfaction (Diener et al., 1985). Although satisfaction with life does not encompass all angles and complexities of “happiness” or subjective well-being, the Satisfaction with Life Scale (SWLS) is a widely used and well-validated measure. It arguably constitutes a psychometric improvement from the one-item measures of happiness (general or domain-specific) employed in some previous studies on this subject (e.g. Napier & Jost, 2008; Schlenker et al., 2012, Studies 2 and 3). In the present studies, we also measured religiosity (Study 1), gross household income (Studies 1 and 2) and system justification beliefs (Study 2) to examine these variables’ associations with political orientation and conservatism relative to each other and the Big Five traits. In both studies, the logarithm of income was used in analyses in line with recent recommendations for examining the association between well-being and income (see, e.g., Kahneman & Deaton, 2010).
Study 1
Method
Eight-hundred and sixteen United States community members were recruited to take part in an online survey through Amazon’s Mechanical Turk platform for compensation of $0.53. In preparing data for analysis, we considered the amount of time participants took to complete the survey, and IP addresses from which participants completed the survey. Twenty-four participants were removed for having duplicate IP addresses and matching demographic information (in these cases, the second study completion was removed according to the dates and times of completion), and six participants were removed for participating from Indian IP addresses. Next, we removed 142 participants who did not complete either, or both of, the measure of SWL and the measure of conservatism. These exclusions yielded 644 participants (299 male, 5 unreported; Mage = 32 years, SD = 12). Next, we calculated the 5th percentile for minutes taken, which was 12 minutes (M = 28 minutes, SD = 16; range = 7 to 221). In all analyses, 25 participants who took fewer than 12 minutes were excluded; all other participants were included unless specified otherwise.i
After excluding these participants, 619 remained (279 male, 3 unreported; Mage = 32 years, SD = 12). Participants started by completing demographic information including ethnicity, religious affiliation, gender, age, and gross household income (in $5 000 increments ranging from $10 000 or under to $150 001 or over; Mincome = $45 001 - $50 000). Participants also completed Cohen & Rankin’s (2004) 4-item religiosity scale (sample items include, How religious are you?; How much do you believe the teachings of a religion?; How much do you practice the requirements of a religion?, Cronbach’s α = .93), a 3-item scale measuring general conservatism (Cronbach’s α = .93), and the 5-item Satisfaction with Life Scale (Diener et al., 1985; sample items include, In most ways my life is close to ideal; The conditions of my life are excellent; I am satisfied with my life, Cronbach’s α = .91). Next, participants completed the 100-item Big Five Aspect Scale (Cronbach’s αs for the Big Five traits between .87 and .93); see Table 2 for descriptive statistics and Appendix for sample items from the BFAS, and the entire 3-item general conservatism scale. Participants also completed a measure of disgust sensitivity and subsequently indicated their stance on multiple political issues by responding to items written by the authors, the discussion of which are beyond the scope of this paper.
Results and Discussion
The bivariate correlation between conservatism and SWL was small but significant: r(617) = .08, p < .05 (for all bivariate correlations, see Table 1; descriptive statistics, see Table 2). Next, we tested whether this correlation would hold once we controlled for gender (see Norrander & Wilcox, 2008) and age (see Truett, 1993), followed by religiosity and income, and, finally, the Big Five traits. This sequence of additions was used to test, first, whether the conservatism-SWL correlation would hold when basic demographic variables (i.e. gender and age) were controlled for, and subsequently when variables implicated in explaining the conservatism-SWL association in previous studies were controlled for (i.e. religiosity: Schlenker et al., 2012; and income: Jetten et al., 2013). Finally, because the Big Five traits, and neuroticism in particular, are the unique focus of this study, they were added last to test whether they predicted SWL over and above conservatism and the previous control variables.
Table 1
Correlation Matrix for Study 1 (N = 619)
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Conservatism | – | .08* | .17** | .06 | -.11** | -.15** | .04 | -.11** | .18** | .43** | .04 |
| 2. SWL | – | .28** | .11** | -.45** | .03 | .34** | .06 | 0 | .12** | .27** | |
| 3. Conscientiousness | – | .22** | -.41** | .21** | .34** | .07 | .08 | .19** | 0 | ||
| 4. Agreeableness | – | -.14** | .34** | .21** | .32** | .15** | .25** | -.03 | |||
| 5. Neuroticism | – | -.11** | -.41** | .19** | -.16** | -.13** | -.10* | ||||
| 6. Openness | – | .38** | .11** | -.04 | .10* | -.06 | |||||
| 7. Extraversion | – | .04 | -.03 | .17** | .09* | ||||||
| 8. Gender | – | .05 | .16** | -.10* | |||||||
| 9. Age | – | .22** | .07 | ||||||||
| 10. Religiosity | – | -.05 | |||||||||
| 11. Log Income | – |
Note. Gender was coded such that male = -1 and female = 1.
*p < .05. **p < .01.
Thus, we used hierarchical regression to predict SWL with conservatism in Block 1, age and gender in Block 2, religiosity and the logarithm of income in Block 3, and the Big Five traits in Block 4. One participant had a standardized residual more than three standard deviations above the mean and was excluded from this analysis. In the model (R2 = .03, F(3, 396) = 3.44, p = .02) controlling for age (β = -.05, p = .31) and gender (β = .13, p = .01; male = -1, female = 1), conservatism was a significant predictor of SWL (β = .11, p = .03). In the next model (R2 = .13, F(5, 394) = 12.06, p < .001), which included the logarithm of income and religiosity, conservatism was no longer a significant predictor of SWL (β = .04, p = .50). SWL was related to gender (β = .12, p = .08), age with marginal significance (β = -.09, p = .06), religiosity (β = .14, p = .01), and the logarithm of income (β = .31, p < .001). In the final model (R2 = .34, F(10, 389) = 20.14, p < .001), which included the Big Five traits, conservatism was again not a significant predictor of SWL (β = .01, p = .84), nor was religiosity (β = .08, p = .11), but age (β = -.11, p = .01), gender (β = .21, p < .001), and income (β = .26, p < .001) were. Among the Big Five traits, significant predictors of SWL were neuroticism (β = -.38, p < .001), extraversion (β = .15, p = .004), and openness (β = -.10, p = .04); neither agreeableness (β = -.01, p = .81) nor conscientiousness (β = .05, p = .27) added significantly to the model.
In order to test which variables could individually account for the shared variance between conservatism and SWL, we conducted partial correlations between conservatism and SWL, controlling for individual variables. The control variables selected were those which correlated in the same direction with both conservatism and SWL: neuroticism, conscientiousness, and religiosity. The SWL-PO association was rendered non-significant when controlling for neuroticism (r(568) = .04, p = .38), conscientiousness (r(571) = .04, p = .39), or religiosity (r(616) = .03, p = .43).
These results support two main conclusions: first, there was a small but significant negative correlation between conservatism and neuroticism, indicating that more conservative people reported being more emotionally stable than more liberal people. Second, as in previous studies (Jetten et al., 2013; Napier & Jost, 2008; Schlenker et al., 2012), there was a small but significant bivariate correlation between conservatism and well-being, such that conservative people reported higher life satisfaction. This association, however, was rendered non-significant by accounting for conservatives’ lower neuroticism, higher conscientiousness, or higher religiosity.
Table 2
Descriptive Statistics for Variables in Studies 1 and 2
| Study 1 (N = 619)
|
Study 2 (N = 700)
|
|||
|---|---|---|---|---|
| Mean | SD | Mean | SD | |
| 1. Conservatism (1 to 7) | 3.26 | 1.66 | 3.35 | 1.77 |
| 2. Satisfaction with Life (1 to 7) | 4.24 | 1.52 | 4.26 | 1.54 |
| 3. Extraversion (1 to 5) | 3.32 | .63 | 3.29 | .66 |
| 4. Agreeableness (1 to 5) | 3.84 | .52 | 3.92 | .58 |
| 5. Conscientiousness (1 to 5) | 3.49 | .57 | 3.47 | .62 |
| 6. Neuroticism (1 to 5) | 2.77 | .76 | 2.67 | .80 |
| 7. Openness (1 to 5) | 3.89 | .52 | 3.89 | .57 |
| 8. Religiosity (1 to 9) | 4.32 | 2.70 | - | - |
| 9. System Justification (1 to 9) | - | - | 4.29 | 1.54 |
Study 2
Study 1 indicated that the differences between liberals and conservatives in neuroticism, conscientiousness, and religiosity can each account for the differences in reported life satisfaction. Previous studies have provided contradictory findings regarding the role of system justification beliefs in explaining the association between measures of well-being and political identification (e.g. Jetten et al., 2013; Napier & Jost, 2008). Thus, in Study 2 we included Kay and Jost’s (2003) measure of system justification (SJ) to clarify the role of system justification beliefs in explaining the conservative-liberal happiness gap, relative to personality traits (specifically neuroticism).
Method
One-thousand four hundred and two participants residing in the United States participated on Amazon’s Mechanical Turk. After we removed participants whose data were incomplete on measures of SWL or political orientation, two participants who participated from Indian IP addresses, and 17 who completed the study in 4 minutes or less (5 minutes was in the 5th percentile of time taken; M = 10.5 minutes, SD = 5.3 minutes, range = 2 to 46 minutes), 700 remained (334 male; Mage = 33 years, SD = 12).ii Participants completed information on household income in $5,000 increments ranging from $10,000 or under to $200,000 or over (M = $45,001-$50,000) and completed the SWLS (Cronbach’s α = .91), the BFAS (Cronbach’s αs between .88 and .94), the 3-item measure of conservatism employed in Study 1 (Cronbach’s α = .94), and the complete eight-item System Justification Scale (Kay & Jost, 2003; sample items include, In general, you find society to be fair; Everyone has a fair shot at wealth and happiness; Most policies serve the greater good, Cronbach’s α = .83); see Table 2 for descriptive statistics.
Results and Discussion
Conservatism had a positive correlation with SWL similar in magnitude to that found in Study 1 (r(698) = .08, p = .04); for all bivariate correlations, see Table 3. Therefore, we proceeded with analyses to test for replication of the effects from Study 1, as well as any effects of system justification.
Table 3
Correlation Matrix for Study 2 (N = 700)
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Conservatism | – | .08* | .17** | -.03 | -.11** | -.27** | .01 | -.08* | .18** | .16** | .29** |
| 2. SWL | – | .29** | .11** | -.38** | .02 | .34** | .09* | 0 | .31** | .28** | |
| 3. Conscientiousness | – | .27** | -.40** | .13 | .32** | .09* | .18** | .15* | .17** | ||
| 4. Agreeableness | – | -.15** | .33** | .22** | .32** | .16** | .01 | -.02 | |||
| 5. Neuroticism | – | -.14** | -.42** | .14** | -.22** | -.11** | -.26** | ||||
| 6. Openness | – | .38** | .11* | .01 | -.08 | -.21** | |||||
| 7. Extraversion | – | .05 | .06 | .11** | .10** | ||||||
| 8. Gender | – | .06 | 0 | -.06 | |||||||
| 9. Age | – | .10* | .14** | ||||||||
| 10. Log Income | – | .17** | |||||||||
| 11. System Justification | – |
Note. Gender was coded such that male = -1 and female = 1.
*p < .05. **p < .01.
Similar to Study 1, we conducted a hierarchical regression predicting SWL with conservatism in Block 1, basic demographic characteristics (i.e. age and gender) in Block 2, variables previously implicated in explaining the conservatism-SWL association (i.e. system justification beliefs and the logarithm of income) in Block 3, and the Big Five traits in Block 4. In this analysis, three participants had standardized residuals more than three standard deviations away from the mean and were excluded from the regression analysis.
In the model (R2 = .02, F(3, 563) = 3.16, p = .02) controlling for age (β = .00, p = .99) and gender (β = .10, p = .02), conservatism still predicted SWL (β = .09, p = .03). In the next model (R2 = .18, F(5, 561) = 23.23, p < .001), which included system justification beliefs and the logarithm of income, conservatism was no longer a significant predictor of SWL (β = -.02, p = .63). SWL was related to gender (β = .10, p = .01), the logarithm of income (β = .29, p < .001), and system justification beliefs (β = .25, p < .001), but not age (β = -.05, p = .24). In the final model (R2 = .34, F(10, 556) = 27.88, p < .001), which added the Big Five traits, conservatism was again not a significant predictor of SWL (β = -.04, p = .32), but SWL was associated with age (β = -.13, p = .001), gender (β = .14, p < .001), system justification beliefs (β = .15, p < .001), and the logarithm of income (β = .25, p < .001). Among the Big Five traits, significant predictors of SWL were neuroticism (β = -.31, p < .001), extraversion (β = .20, p < .001), and openness (β = -.12, p = .004); SWL was not associated with agreeableness (β = .02, p = .57) nor conscientiousness (β = .05, p = .25).
In order to test which variables could individually account for the shared variance between conservatism and SWL, we ran partial correlations between conservatism and SWL, controlling for single variables. The control variables selected were those correlated with both conservatism and SWL in the same direction: neuroticism, conscientiousness, the logarithm of income, and system justification beliefs. The SWL-PO association was rendered non-significant when controlling for any of the following: neuroticism (r(697) = .04, p = .32), conscientiousness (r(697) = .03, p = .41), the logarithm of income (r(582) = .03, p = .45), or system justification beliefs (r(697) = .01, p = .88).
This study replicated the small but significant bivariate correlation between conservatism and happiness reported in Study 1, as well as the small but significant negative correlation between conservatism and neuroticism. This study also suggests that neuroticism on its own can account for the association between conservatism and happiness, as can SJ beliefs, income, and conscientiousness. When any one of these variables was accounted for, the correlation between conservatism and SWL was reduced to non-significance.
General Discussion
How Do our Results Relate to Previous Findings?
These studies present several noteworthy findings. First, with data drawn from over 1300 U.S. residents across both studies, we replicated the positive bivariate correlation between conservatism and “happiness” (here, life satisfaction) that has been reported in previous studies. The magnitude of the correlations between PO and SWL, in both of our studies, was low. Taken together with previous studies finding relatively small correlations between PO and measures of happiness, these data send a cautionary message against over-interpretation of the bivariate correlations often reported between political orientation and happiness. Instead, it appears that at least five variables can virtually eliminate the association between political orientation and subjective well-being.
These results are also consistent with a recent meta-analysis which found a small but significant positive association between conservatism and life satisfaction across 24 samples with a total of 7 935 participants (Onraet et al., 2013a). In the context of the Onraet et al. (2013a) findings, our studies fall under the category of those measuring “conservatism” as opposed to authoritarianism or social dominance orientation, and “life satisfaction” as opposed to positive affectivity or self-esteem. In light of their finding that a conservatism-SWL association is not apparent in European and Oceanic countries, we note that our findings are best applied to a North American population; further work is required to determine the nature of the relationship between neuroticism and conservatism, if any, in other parts of the world.
Second, the present data suggest a reliable association between trait neuroticism and political orientation, such that conservatives are lower than liberals in neuroticism. Thus, neuroticism appears to make a noteworthy – and previously under-examined – contribution to the dispositional profile of political orientation.
Third and most important, the present results demonstrated that one can account for the conservatism-happiness association by controlling for any of the following: neuroticism, income (in Study 2 but not Study 1), conscientiousness, religiosity, and system-justification beliefs.
Conservatives are More Emotionally Stable Than Liberals
In some respects, our results are consistent with a positive adjustment explanation for the happiness differential between liberals and conservatives (see Schlenker et al., 2012). Indeed, the higher neuroticism (or lower emotional stability) evinced by more liberal individuals in both studies is related to many negative outcomes and characteristics, including the predisposition to experience negative affect (Diener, Oishi, & Lucas, 2003; Miller, Vachon, & Lynam, 2009), higher sensitivity to negative mood inductions (Larsen & Ketelaar, 1989), lower relationship quality (Karney & Bradbury, 1995), and lower performance motivation (Judge & Ilies, 2002). More succinctly, neuroticism is strongly positively related to negative affect and negatively to other measures of happiness (Costa & McCrae, 1980). Thus, it is understandable that the happiness differential between liberals and conservatives can be largely accounted for by differences in their mean levels of emotional stability.
As noted earlier, there are at least two reasons why conservatives might be higher than liberals in emotional stability. First, emotionally stable people may feel less need to adopt an ideology that advocates changing the status quo. In particular, lower neuroticism may lead people to feel less aggrieved by apparent inequities in the distribution of resources (Mondak & Halperin, 2008). Second, there are likely common genetic factors underlying one’s personality traits such as neuroticism and one’s eventual political orientation (e.g. Verhulst et al., 2012). Clearly, however, more work is required to fully elucidate the nature of the relationships among conservative ideology and anxiety, sensitivity to threat, emotional stability, and satisfaction with life.
Implications for Understanding Social Problems and Promotion of Social Justice
A growing body of research in political psychology has linked political orientation with psychological variables that are, in and of themselves, irrelevant to political positions (e.g., disgust sensitivity: Inbar, Pizarro, & Bloom, 2009; anticipated negative affect: Joel, Burton, & Plaks, 2014). Such results begin to locate the source of political attitudes in more fundamental differences between people in personality and temperament. The present work builds on this literature by not only highlighting the contribution of trait neuroticism to distinguishing liberals from conservatives, but also by demonstrating that conservatives’ lower neuroticism helps to explain the happiness gap.
In recent years, non-academic and academic observers have noted an increase in political polarization across a range of issues (e.g., Brooks, 2007; Dunlap, Xiao, & McCright, 2001). Liberals and conservatives appear less willing and able to communicate meaningfully with one another, often resulting in the failure to advance legislation that requires even small amounts of bipartisan compromise. However, greater knowledge of the underlying psychological foundations of political orientation may begin to suggest ways in which conservatives and liberals may learn to “speak the other side’s language” more effectively. For example, Feinberg and Willer (2013) found that liberal-conservative differences in endorsement of pro-environmental policies were completely eliminated when conservative participants were presented with pro-environmental arguments that emphasized the disgusting aspects of pollution.
The present finding – that liberals are more neurotic than conservatives – may similarly be used to craft messages that more effectively resonate with their intended audience. For example, liberals, rather than decrying the perceived callousness of conservatives, may be better served by reframing their understanding of conservativism in terms of generalized emotional stability. Similarly, conservatives, rather than decrying the perceived “whining” of liberals, might reframe their construal of liberals in terms of generalized proneness to dissatisfaction with the state of affairs, or even “perfectionism”. Both of these reappraisals of the other side imply new language and metaphors that, in theory, may produce appeals that meet with more intuitive understanding and, in turn, willingness to entertain compromise. We recommend that future researchers take advantage of findings such as ours to develop psychologically-informed pathways to productive dialogue.
Limitations
One limitation of these studies is that they both employ online samples involving Mechanical Turk participants. Although we took measures including checking for repeat participation, examining the length of time taken to complete the study and excluding participants with extremely fast completion times, and checking for and removing participants who participated from the second biggest pool of participants, India (Shapiro, Chandler, & Mueller, 2013), some weaknesses of Mechanical Turk sampling remain. Namely, we cannot speak to the nature of participants’ motivation to complete the studies, or what their state of mind or level of distraction may have been while participating. However, the fact that, despite these concerns, we still replicated the effect of neuroticism on the PO-SWL link across two large, independent samples speaks to the robustness of neuroticism’s effect.
Additionally, we cannot speak to the experience of participants with similar studies (see Chandler, Mueller, & Paolacci, 2014), which could impact the fluency with which participants completed our studies. However, given the correlational design and lack of deception or strict requirement of naïveté on the part of participants for the collection of valid data in these particular studies, the issue of non-naïveté in these participants may present less of a threat to validity than in other types designs or with other types of measures.
Finally, it has been pointed out that Mechanical Turk samples tend to result in under-sampling of individuals who identify as conservative, particularly conservative males (Kahan, 2013). It is hard to say to what precise extent our conclusions might be biased by this issue. Our samples did skew conservative. We did, however, arguably obtain adequate representation from conservatives in general, and conservative males specifically, and achieved high statistical power.iii
Conclusions
Overall, this work suggests that the bivariate association between SWL and political orientation should not be over-interpreted. It is small and easily accounted for with several third variables. On the other hand, a link between neuroticism (or emotional stability) and political orientation should perhaps be afforded more attention in future work. The present studies demonstrate a replicable, albeit small, negative association between neuroticism and conservatism. Thus, we suggest that the most fruitful future directions would be those dedicated to better understanding the statistical, construct-level, and causal relationships between conservative ideology, neuroticism/emotional stability, and system justification beliefs.