Thread by @Flavio_Azevedo_

🔥off the press 🔥

Colleagues and I tackle admittedly contentious debates in ideology scholarship: ideological innocence, nexus between ideology’s most commonly theorized sub-structures (social & economic), and ideological asymmetries of PolPsy (via NeoLib).

Highlights below:

https://twitter.com/TobiasRothmund/status/1098943951007072256

Using a set of high quality nationally representative surveys from professional survey companies (SSI/Research Now/YouGov), in the US and UK, & in which a greater number and range of ideological instruments and psychological constructs were administered, we demonstrated that >>>

1.) contrary to the ideological innocence (K&K) hypothesis, dating back to Converse (1964), *at least as of 2016*, we observed a surprisingly high degree of ideological coherence, when using both operational and symbolic ideological instruments.

We tested the robustness of these findings across N=3620, 3 indicators of symbolic ideology, 5 previously published well cited ideological instruments (ensuing from PolSci (ANES) and SocPsy research), when partitioning these into their social and economic sub-structures >>

and when broken-down by political sophistication (low vs. high & continuous). This is not to say sophistication is irrelevant, no, same data shows it moderates cor magnitudes (ideology more constrained among sophisticates), in agreement with previous studies/literature.

2.) contrary to the common separation of ideology in its sociocultural and economic components, which argue economic conservatism is qualitatively unlike—and in some respects even opposed to—social conservatism, we found these dimensions were highly correlated >>>

which suggests the may not be as independent as previously thought.
Similar to point (1), findings are robust whether measured via symbolic, operational, or both, across datasets 4, and among levels of political sophistication (Table 2, as above).

3.) On a theoretical, practical, and empirical level, we argue that a great many public policy issues, such as those pertaining to welfare spending, redistribution of wealth, public education, health care, affirmative action, military budgets, and law and order involve >>

both economic priorities and social concerns pertaining to race, ethnicity, nationality, immigration status, gender, and sexual orientation. Given that neoliberal economic policies adversely affect some people more than others—including racial and ethnic minorities, poor people >

and many others who are marginalized in society—it seems much more likely to us that social and economic attitudes are intertwined at least in part because both are linked to individual differences in support versus opposition to social and economic forms of inequality.

* personal take/sideline: political actors/pundits/scholars unwilling to recognize their synergy & contingency may risk perpetuating the neoliberal trope -often concealed under the principled commitment to small government.

4.) Based on the dissociation btw social and economic ideologies, previous research found these are linked to very different types of psychological characteristics. Stenner and Haidt (2018) write: "Psychologically speaking, they have nothing in common with authoritarians".

With respect to the US and UK, we tested this and other psychological differences among Social and Economic conservatives reported in previous studies. We found little supporting evidence - and again - irrespective of political sophistication.

More generally, however, and while not reported in this paper, but on another still at a draft stage, @TobiasRothmund and I conduct a thorough test of "The Psychology of Social and Economic Conservatism", in the US, and find a very similar pattern.

We considered including ANES, GSS, VSG data even if these sources don't provide too many options wrt PolPsy constructs. However, more rencely, thanks to @dyudkin @HiddenTribesUS data, we will be able to assess the extent to which these findings can be independently replication :)

5.) Relating to Neoliberalism We also hypothesized that the endorsement of neoliberal attitudes in particular—including support for laissez-faire capitalism and opposition to social welfare programs—would be positively related (rather than unrelated or negatively related) to >>>

to authoritarianism (RWA), social dominance orientation (SDO-7), General, Economic and Gender-specific forms of system justification. As shown in the pictures, we found support for this hypothesis, across datasets and political sophistication, in the US.

💀Caveats 💀
In addition to those discussed in the article (picture):
a. External Validity. These findings pertain to US/UK where, for some, these patterns may be less contested.
b. Operationalizations. An often neglected aspect is that of analytical choices. All findings >>>

presented-here & elsewhere, are contingent (at least in part) to these choices (cf. Dongen et al. 2018; Silberzahn et al 2018 and OSSC - Crowdsourced Replication Initiative @opensocsci). Wherever possible we made efforts to test all forking paths, and reported it.

c. Measures. Findings are contingent on the measures utilized. Had we surveyed different items/instruments/versions or one instead of several ideological scales, it's conceivable results could be different. Operationalizations of ideological instruments is a major issue and >>>

and samples 1 & 2, as well as replications in 2018 midterms, are designed to study just that. I hope to show why exactly in my next work. But here are 3 quick points:

c1. Jingle-Jangle. A non-exhaustive literature review of 389 articles identified more than 87 ideological measures. But while producing multiple instruments for the same construct may be convenient, it is unlikely that they are all equally valid indicators of ideological content.

c2. Item topic/content & N varies a lot across instruments- both SocPsy & PolSci, & by same author(s) in different papers. This may be OK, under certain conditions, but AFAIK there are no studies assessing the conditions under which this doesn't threaten the validity of findings.

c3. Equivalency or Interchability. it is a standard practice to tacitly assume ideological inventories can be used interchangeably. If this tacit and untested assumption does not hold, it may pose a threat to the comparability and generalizability of findings.

Example. Imagine the below picture depicted the mean scores of 5 ideological instruments. They yield different overall distr, but so far so good. Now imagine one is studying polarization, and uses only one of these instruments, which is often the case in ideological research >>>

in this case, findings would depend on utilized instrument. Another example could be about Americans' ideology, which is said to be symbolically conservative, but operationally liberal. If Ellis and Stimson used instrument N.1 (or ANES) their findings would have been different.

🙌.🙏 Acknowledgements 🙌.🙏

This work joins together beautifully a few projects @JohnJost1 and I pursued at @nyuniversity ensuing from a @FulbrightPrgrm scholarship. Thank you John for being such a gracious and welcoming host!

@KirstiJylhae you will have a blast!

This work & almost 80% of my research output wouldn't be possible without the trust from - and support by - @TobiasRothmund who not only financed my first surveys (samples 1 & 2), but also taught me the ropes of academia. Thank you!

Lastly, and foundationally important, thank you @CompPolCologne for having me, I couldn't have found a better place to conduct my PhD and a better supervisor, André Kaiser, who is not on Twitter but Twitter should know about him!