Lewandowsky, S., Woike, J. K., & Oberauer, K. (2020). Genesis or Evolution of Gender Differences? Worldview-Based Dilemmas in The Processing of Scientific Information. Journal of Cognition, 3(1), 9, Apr 30 2020. DOI: http://doi.org/10.5334/joc.99
Abstract: Some issues that have been settled by the scientific community, such as evolution, the effectiveness of vaccinations, and the role of CO2 emissions in climate change, continue to be rejected by segments of the public. This rejection is typically driven by people’s worldviews, and to date most research has found that conservatives are uniformly more likely to reject scientific findings than liberals across a number of domains. We report a large (N > 1,000) preregistered study that addresses two questions: First, can we find science denial on the left? Endorsement of pseudoscientific complementary and alternative medicines (CAM) has been anecdotally cited as being more consonant with liberals than conservatives. Against this claim, we found more support for CAM among conservatives than liberals. Second, we asked how liberals and conservatives resolve dilemmas in which an issue triggers two opposing facets of their worldviews. We probed attitudes on gender equality and the evolution of sex differences—two constructs that may create conflicts for liberals (who endorse evolution but also equality) and conservatives (who endorse gender differences but are sceptical of evolution). We find that many conservatives reject both gender equality and evolution of sex differences, and instead embrace “naturally occurring” gender differences. Many liberals, by contrast, reject evolved gender differences, as well as naturally occurring gender differences, while nonetheless strongly endorsing evolution.
Discussion
Relationship to previous results
Our results coordinate well with multiple precedents in the literature, which we take up for each of the constructs examined. Considering first religiosity, we replicated the substantial association between stronger religious beliefs and conservatism in the American population (Malka et al., 2012; Schlenker, Chambers, & Le, 2012). In our study this association generalized across a broadly-defined socio-political conservatism construct as well as a specific construct targeting endorsement of laissez-faire free-market economics. We also replicated the long-standing strong negative association between religiosity and acceptance of evolution (e.g., Ecklund, Scheitle, Peifer, & Bolger, 2017; Tom, 2018) and the modest negative association between religiosity and analytic thinking (i.e., CRT performance) reported previously (Jack, Friedman, Boyatzis, & Taylor, 2016; Shenhav, Rand, & Greene, 2012; Stagnaro, Ross, Pennycook, & Rand, 2019). Likewise, the correlations between religiosity and the gender constructs (e.g., Table 5) are consistent with previous reports that religiosity predicts sexism (Van Assche et al., 2019). Our results go beyond previous findings because our scales did not probe discriminatory sexism but the origin of presumed gender differences. We find that religiosity makes it less likely that people believe that gender differences have evolved.
The negative association between religiosity and CAM rejection is also unsurprising in light of previous research that has shown acceptance of CAM to be driven by intuitive thinking, paranormal beliefs, and ontological confusions (Lindeman, 2011). At least one of those variables (intuitive thinking) is also known to be a predictor of religiosity (e.g., Shenhav et al., 2012). The positive correlation between CAM rejection and acceptance of vaccinations replicates much previous research (e.g., Attwell, Ward, Meyer, Rokkas, & Leask, 2018; Browne, Thomson, Rockloff, & Pennycook, 2015; Bryden, Browne, Rockloff, & Unsworth, 2018; Ernst, 2002).
However, our findings concerning religiosity also deviate from aspects of other recent research (Rutjens, Sutton, & van der Lee, 2018). Unlike Rutjens et al., we found no evidence of a link between religiosity and rejection of vaccinations. Given that Rutjens et at. observed this link only in some of their studies and only for some measures of religiosity (mainly measures of religious orthodoxy), we are not concerned about this apparent departure from previous results. Indeed, in another recent as-yet unpublished study involving identical constructs, we did observe a negative association between vaccination and religiosity, suggesting that this relationship may well be real but is only observable in certain circumstances.
Turning to the associations involving CRT performance, the observed modest but significant negative correlation with religiosity replicates previous results (Gervais & Norenzayan, 2012; Shenhav et al., 2012; Stagnaro et al., 2019). Jost (2017) reported a meta analysis of 13 studies that related CRT performance to political views. The vast majority of those studies showed that liberals exhibited more cognitive reflection than conservatives. In the present data, this is echoed by the modest negative correlation with free market, although it was not reflected in the socio-political conservatism measure. The positive associations of the CRT with endorsement of all three scientific constructs, vaccination, CAM rejection, and evolution replicate similar previous findings (Shtulman & McCallum, 2014; Wagner-Egger et al., 2018). The association also coordinates well with recent findings that analytical thinking is associated with better differentiation between “fake news” and valid information (Pennycook & Rand, 2018).
Rejection of science on the political left?
Our findings provide little or no evidence that people on the political left reject vaccinations. On the contrary, to the extent that worldviews determined vaccination attitudes, it was free-market endorsement that predicted rejection. This result parallels a similar association observed by Hornsey, Harris, and Fielding (2018), albeit using a different instrument to measure Libertarian attidudes (hierarchical-individualism as opposed to free-market endorsement). The result is also consonant with the notion that libertarians object to the government intrusion arising from mandatory vaccination programs (Kahan et al., 2010). It also meshes well with the pattern observed by Lewandowsky, Gignac, and Oberauer (2013), who showed that when socio-political conservatism was removed from a model, free-market endorsement on its own predicted rejection of vaccinations (whereas the converse was not true). Overall, our results thus converge with other recent findings that have found an association between right-wing politics and rejection of vaccinations (Baumgaertner, Carlisle, & Justwan, 2018; Kahan et al., 2010; Rabinowitz et al., 2016). In a recent cross-sectional analysis of voting behavior and vaccination rates across European countries, Kennedy (2019) found a strong relationship between the vote share for populist parties and vaccine hesitancy.
Similarly, contrary to reports that CAM use and left-wing ideas have a natural affinity for each other (see, e.g., Keshet, 2009), we found that CAM rejection was negatively, but modestly, associated with the All conservatism factor that subsumed all three of our worldview constructs; namely, religiosity, free market endorsement, and socio-political conservatism. Moreover, in our data, none of the gender constructs were associated with CAM attitudes. This runs counter to the idea that CAM use is “feminist” (Scott, 1998). To our knowledge, our results constitute the first empirical examination of the links between political views and CAM attitudes. Our results that conservatives are more likely to embrace CAM is consonant with historical analyses that have found strong links between right-wing organizations, such as the John Birch Society in the U.S., and endorsement of “alternative” cancer treatments (Markle, Petersen, & Wagenfeld, 1978). The present result adds to the list of failed attempts to discover science denial on the political left (e.g., Hamilton, 2011, 2015; Hamilton, Hartter, & Saito, 2015; Hamilton, Hartter, Lemcke-Stampone, et al., 2015; Kahan et al., 2010; Lewandowsky, Gignac, & Oberauer, 2013; Tom, 2018).
Attitudes towards gender differences
We observed an intriguing interplay of the attitudes towards general Darwinian evolution, gender differences, and how those gender differences might have arisen. At a coarse level of analysis, we observed three unsurprising associations: The idea that men and women differ naturally was highly correlated with the idea that they evolved differently, but was negatively correlated with the construct that proclaimed gender equality. The equality construct was also negatively correlated with the idea that men and women evolved differently, although that correlation was smaller than for natural differences.
At a more detailed level of analysis, several intriguing associations emerged. First, acceptance of general Darwinian evolution was positively associated with two seemingly conflicting constructs; namely, that men and women evolved differently and that they are the same. Moreover, evolution was negatively correlated with the idea that men and women are naturally different, even though evolution is one way in which such “natural” differences might have emerged. A similarly nuanced pattern obtained when the worldview constructs were used to predict gender attitudes. Although the over-arching All conservatism factor functioned as expected, with negative weights for gender equality and positive weights for the two constructs insisting on gender differences, there was an additional selective effect of religiosity on the rejection of evolved gender differences.
Further analysis revealed that the involvement of evolution, either on its own or in explaining gender differences, served as a “wedge issue” that disrupted otherwise straightforward associations between right-wing politics and opposition to gender equality (and, vice versa, rejection of gender differences and left-wing politics) and—as foreshadowed in Figure 1—created dilemmas for participants of all political persuasions. As noted in connection with Figure 6, conservatives who strongly rejected Darwinian evolution resolved their dilemma by endorsing “natural” gender differences while rejecting evolved gender differences. Those participants were thus willing to forego endorsement of gender differences to maintain consistency with their opposition to evolution. Conversely, liberals who are strongly committed to gender equality tended to reject the idea of evolved gender differences, even though those participants were demonstrably committed to accepting evolution. Those participants were thus willing to forego endorsement of a specific manifestation of evolution to maintain consistency with their commitment to equality. Thus, partisans of either stripe can agree in their rejection of the idea that men and women evolved differently, but they do so for entirely different reasons. Conservatives do so when they are committed to reject evolution, and liberals do so when they are committed to gender equality. Both groups therefore resolve the dilemmas posed by our gender constructs by “sacrificing” endorsement of evolved gender differences.
Conclusion
Our results contribute to two seemingly conflicting streams of outcomes in the literature on how worldviews moderate people’s responses to scientific issues. On the one hand, there is much evidence for pervasive attitudinal asymmetry, at least in the United States, with conservatives being more likely to reject well-established scientific propositions than liberals. To date, little or no evidence for left-wing science denial has been reported. We add to this stream by showing that, contrary to previous largely anecdotal reports, liberals are more likely to reject complementary and alternative medicines, in line with the scientific evidence, than conservatives.
On the other hand, there is considerable evidence that liberals and conservatives process scientific data in a symmetrical fashion. That is, liberals and conservatives alike resort to the same cognitive shortcuts when data conform to their biases, giving rise to a symmetric set of errors (Kahan, Peters, Dawson, & Slovic, 2017; Washburn & Skitka, 2018). We also add to this stream of research by showing that, when confronted by worldview-triggered dilemmas, both liberals and conservatives resolve those dilemmas in an equally “rational” fashion, by selectively “sacrificing” endorsement of a specific construct about gender differences. Liberals, who generally endorse evolution, believe that for some reason it did not affect differences between the sexes; this could be rationalized perhaps by assuming that evolution causes differences only between but not within species. Conservatives, who frequently reject evolution, believe that men and women differ naturally without having evolved differently; this could be rationalized by assuming, for instance, that those natural differences were the result of divine intervention.
A final contribution of our study is that it points to the advantages of a more nuanced analysis of political worldviews, beyond a convenient but simplistic classification of people into left and right, or liberals and conservatives. While this classification is sufficient to explain some scientific attitudes—for example, it matters little how one measures political worldviews to explain rejection of climate science (e.g., Hornsey, Harris, Bain, & Fielding, 2016; Kahan, 2015)—there are other circumstances in which a more nuanced differentiation between different aspects of worldviews provides considerably greater explanatory power.
Posts
