Our results also replicated Iyer et al. (2012) previous finding that libertarians think more analytically than both liberals and conservatives

Are neo-liberals more intuitive? Undetected libertarians confound the relation between analytic cognitive style and economic conservatism. Onurcan Yilmaz, S. Adil Saribay, Ravi Iyer. Current Psychology, February 14 2019. https://link.springer.com/article/10.1007/s12144-019-0130-x

Abstract: Previous studies consistently showed that analytic cognitive style (ACS) is negatively correlated with social conservatism, but there are mixed findings concerning its relation with economic conservatism. Most tests have relied on a unidimensional (liberal-conservative) operationalization of political orientation. Libertarians tend not only to identify themselves as conservative on this scale but also to score higher on ACS than liberals and conservatives. The presence of libertarians might be the reason for the above-mentioned mixed findings. We investigated the relation between social and economic conservatism and ACS (operationalized using the Cognitive Reflection Test; CRT) in a large, web-based sample. There was a negative correlation between CRT and social conservatism both when libertarians were included and excluded. However, the correlation between CRT and economic conservatism was significantly reduced in magnitude and became non-significant when libertarians were excluded. The results support the argument that the undetected presence of libertarians may confound the ACS-economic conservatism relation.

Keywords: Analytic cognitive style Libertarians Liberals Conservatives Cognitive reflection test

---
Discussion
We argued that if a study includes a non-negligible proportion of libertarians and measures political orientation on the single-item self-placement scale, then it might produce misleading results regarding the relationship between cognitive style and economic conservatism. Specifically, such a procedure may produce misleading results because the libertarians tend to score relatively higher on both variables (i.e., ACS and economic conservatism) and they also tend to self-place toward the conservative end of the political spectrum (see Iyer et al. 2012; Talhelm et al. 2015). In other words, the existence of different proportions of libertarians might produce mixed findings across those samples. The current effort is a demonstration of the importance of considering political affiliation (e.g., libertarian) beyond self-placement on the liberal-conservative continuum and, specifically, of the utility of repeating analyses including and excluding libertarians.

The results suggested that the negative (but small) correlation between ACS and social conservatism holds whether libertarians (and other non-mainstream groups) are included or excluded. This is consistent with the previous literature showing that there is a negative correlation between social conservatism and ACS (Deppe et al. 2015; Saribay and Yilmaz 2017; Pennycook et al. 2012; Yilmaz and Saribay 2016, 2017b, 2018). This finding also supports the social orientation hypothesis suggesting that social political orientation is a better predictor of cognitive style differences than economic political orientation (Talhelm et al. 2015). However, our finding also suggests that this relation is weak.

On the other hand, the relation between ACS and economic conservatism differs depending on the inclusion versus exclusion of libertarians (and other non-mainstream groups). The positive correlation between CRT and economic conservatism in the whole dataset lost its significance and was reduced to virtually zero when the non-liberal and non-conservative participants were excluded. The results remained constant when we only excluded libertarian participants (instead of all nonliberals and non-conservatives including BDon’t know/not political,^ and BOther^). Thus, the current findings suggest that the presence of libertarians (and other non-mainstream groups) confounds the relation between ACS and economic conservatism. They clarify why previous findings regarding this relation were mixed. As noted earlier, some previous findings suggested a non-significant relation (Deppe et al. 2015; Pennycook et al. 2012; Yilmaz and Saribay 2016, 2017b), some others showed a negative relation (Sterling et al. 2016), and some others even showed a positive relation (e.g., one of four studies of Deppe et al. 2015). Jost et al. (2017) meta-analyzed these findings and showed a negative—albeit weak—correlation between ACS and economic conservatism (unweighted average r = −.08). Our contention that different proportions of libertarians might determine the direction and size of the correlation regarding ACS-economic conservatism relation is compatible with these findings since the Study 2 of Deppe et al. (2015), where the correlation between ACS and economic conservatism is positive, was arguably the most-representative sample of American population (see Baron 2015). Our results also replicated Iyer et al. (2012) previous finding that libertarians think more analytically than both liberals and conservatives, however, there were no such differences between self-reported liberals and conservatives.