According to one of the classical psychological theories of ideology, conservatism is associated with a simple, intuitive, unsophisticated, rigid, and authoritarian cognitive and psychological style. This rather unflattering portrait of conservatives has been the target of criticism lately. Critics have argued that it is a product of a “liberal bias” and hostility toward conservatism among social and political psychologists. Studies have been designed to show that the associations between the aforementioned characteristics and political ideology are symmetrical—or in other words, that extremists of any ideological persuasion are simple-minded, rigid, authoritarian, and susceptible to cognitive biases.
Some of the criticism of the classical “rigidity-of-the-right” theory is undoubtedly warranted. But the problem with this theory is not that it is all wrong. The problem is that it has proved to be too simplistic, and some of the criticism of it is also too simplistic. There are probably ideological asymmetries, symmetries, and extremism effects. It all depends on what specific aspect of cognitive or psychological style you focus on. There are also considerable differences between different kinds of left- and right-wing ideologies, which have often been lumped together, somewhat misleadingly, under the broad labels of “liberalism” and “conservatism” (see Nilsson, Erlandsson, & Västfjäll, 2019; Nilsson et al., 2020).
In this blog post, I will discuss the case of authoritarianism, which has recently generated heated debates. This is a case for which I believe there is currently pretty good evidence for the existence of ideological asymmetries. Authoritarianism tends to be higher among right-wing conservatives all other things equal (see Nilsson & Jost, 2020).
Authoritarianism on the right and on the left
In 1981, Bob Altemeyer introduced a new instrument designed to measure Ring-Wing Authoritarianism (RWA), which has become tremendously popular. This instrument has been rightly criticized because it confounds authoritarianism and right-wing or conservative ideology (some of the items refer to conservative issues, e.g., “You have to admire those who challenged the law and the majority’s view by protesting for women’s abortion rights, for animal rights, or to abolish school prayer”). This content overlap could produce a spurious or inflated correlation between authoritarianism and conservatism.
To address this problem, Conway et al. (2018) introduced a parallel scale to measure Left-Wing Authoritarianism (LWA) by rewriting RWA items so that they would refer to “liberal” authorities and norms. For instance, in one of the items, “the proper authorities in government and religion” was replaced with “the proper authorities in science with respect to issues like global warming and evolution”. The authors found that LWA was strongly associated with liberal forms of prejudice, dogmatism, and attitude strength in US convenience samples. On this basis, they concluded that authoritarianism exists among left-wingers and right-wingers in essentially equal degrees.
This conclusion may seem appealing at first glance, and the paper has been uncritically cited as evidence of blatant liberal bias in political psychology by some influential figures. But this research is methodologically flawed.
Content validity. The first critical issue in validating a scale concerns the theoretical correspondence between the content of the items that comprise the scale and the definition of the construct that the scale is purported to measure. Consider the following items:
- “Progressive ways and liberal values show the best way of life.”
- “There is absolutely nothing wrong with Christian Fundamentalist camps designed to create a new generation of Fundamentalists.” (reverse-scored)
- “It’s always better to trust the judgment of the proper authorities in science with respect to issues like global warming and evolution than to listen to the noisy rabble rousers in our society who are trying to create doubts in people’s minds.”
- “With respect to environmental issues, everyone should have their own personality, even if it makes them different from everyone else.” (reverse-scored)
The content of these items is a combination of liberal political views and a rejection of epistemological and moral relativism. For a liberal individual to score low on LWA s/he would have to think that all values are equal, scientists are no more trustworthy than non-scientists with respect to scientific issues, and fundamentalist indoctrination is totally unproblematic. If anything, such a person sounds to me to be more (rather than less) susceptible to authoritarian tendencies.
Even if we would accept that the Conway-LWA scale conceptually parallels Altemeyer’s RWA scale (which is debatable), relying on either these scales to assess the empirical association between authoritarianism and political ideology is pointless, because of the problem with content overlap. Fortunately, criticism of the RWA scale has stimulated the development of more psychometrically rigorous measures of authoritarianism that disentangle different aspects of authoritarianism and remove ideological content and language (here is a sample item: “We should believe what our leaders tell us”). So far, these developments have suggested that an association between authoritarianism and right-wing conservatism remains when content overlap is removed (see Nilsson & Jost, 2020).
Structural validity. A second critical issue in validating a scale concerns the correspondence between the theorized structure of the constructs that are purportedly measured and the empirically derived structure in the data. Conway et al. have not reported any tests of structural validity in any of the papers or online supplements that I am aware of. They only report Cronbach’s alpha reliabilities, which say nothing about dimensionality, and their data are not openly accessible, which makes it is difficult to evaluate the scale. This is particularly troubling given the aforementioned problems with the content validity. Although some of the other items may capture elements of left-wing authoritarianism, it is currently impossible to tell which dimensions the instrument does or does not measure.
Nomological network. A third issue in validating a scale (at least according to some methodologists) is whether the scale correlates in a predictable or theoretically meaningful manner with conceptually adjacent and non-adjacent constructs. The main results Conway et al. (2018) invoke to support the validity of their LWA scale are a set of correlations between LWA and measures of dogmatism, intolerance, and attitude strength designed for left-wing liberals. However, there is so much overlap between these scales that finding anything other than a strong correlation between them would be almost impossible (i.e., the hypothesis that is purportedly tested is not falsifiable). For instance, these are items used to measure left-wing dogmatism (LWD) and left-wing authoritarianism (LWA):
- LWD: “I don’t trust the modern-day scientific experts on global warming very much” (reverse-scored)
- LWA: “It’s always better to trust the judgment of the proper authorities in science with respect to issues like global warming and evolution than to listen to the noisy rabble rousers in our society who are trying to create doubts in people’s minds.”
- LWD: “With respect to environmental issues, there is no “ONE right way” to live life; everybody has to create their own way.” (reverse-scored)
- LWA: “With respect to environmental issues, everyone should have their own personality, even if it makes them different from everyone else.” (reverse-scored)
I can’t help thinking that you could produce whatever correlation you want by adjusting formulations and content overlap.
General points. The argumentation by Conway and colleagues is a bit insidious in implying that either you embrace the conclusions of their paper or you think that LWA is a myth or an “invalid” construct.
- Is LWA a “valid” construct?
Validity is not a property of constructs. A construct can be more or less useful, and it can have referents or lack referents.
- Is LWA a “viable” or useful construct?
There is no evidence that the scale introduced by Conway and colleagues is useful. But that does not mean that the theoretical concept of authoritarianism on the left is not useful. A more psychometrically rigorous and unbiased LWA scale could benefit research in psychology.
- Does authoritarianism on the left exist?
Yes, it does. It is common in contexts where the socially sanctioned authorities and norm systems are left leaning.
Research on left-wing and right-wing authoritarianism should be judged according to the same standards of evidence. This should lead us to the conclusion that neither Altemeyer’s RWA scale nor Conway’s LWA scale can be used for assessing the association between authoritarianism and ideology. Altemeyer’s scale serves other purposes (e.g., measuring ideology or predicting prejudice). An LWA scale could too. But the very idea of basing such a scale on an RWA scale that was designed almost four decades ago is misguided. Although we still have a long way to go with respect to improving measurement practices in psychology, there have undoubtedly been major advances in psychometrics since Altemeyer’s days, and better measures of authoritarianism are available today (see Nilsson & Jost, 2020).
None of this is to suggest that the “rigidity-of-the-right” theory got everything right or that liberal bias is a myth—the picture is, as I wrote earlier, probably a lot more complicated, and liberal bias has probably shaped research in some areas. But there is a risk that researchers could go too far in the other direction—from “it’s all asymmetrical”, to “it’s all symmetrical”, or from liberal bias to anti-liberal bias—as the case of authoritarianism appears to suggest. The history of ideas is full of examples of dramatic over-reactions. For instance, the failures of positivist philosophy led many intellectuals to adopt radical forms of relativism, and the scientific failures of “soft” psychodynamic and humanistic personality psychologies led many researchers to adopt a remarkably naïve trait-theoretical reductionism. The truth is rarely found in either of the extremes.