Climate science and conspiracy theories

Reanalysis of Lewandowsky et al 2012

Joe Duarte (of Political Diversity Will Improve Social Psychological Science fame) has been complaining about some climate science denial/skepticism being linked to conspiracy theories study for years. He posted a new analysis on X along with the data. I don’t know where he got the data from exactly, it doesn’t appear to be on the journal website, but anyway, I had Claude run the analysis quickly. Here’s their questions:

And the distributions:

Duarte writes:

In the sample of 1,145 participants, only ten (10) participants thought that the moon landings were a hoax – effectively zero – and most of those ten did not believe that climate science was a hoax (though we can’t do anything with 10 of 1,145 participants anyway – it wouldn’t matter if all ten thought climate science was a hoax).

We see this above in that the green bars for moon landing are very small indeed, less than 1% of the sample for both types of “agree”. The distributions and small scale of these variables means that ordinary factor analysis is not suitable, since it assumes continuous, normally distributed-ish data. But we can calculate the latent correlations and use these as the basis of our analysis. Latent correlations are based on the assumption that each of these ordinal scales are a discretized (binned) version of a normal distribution. If we can grant this assumption and the math, then we can figure out what the Pearson correlations would have been if the data had been measured better. These are:

There is the usual nearly positive manifold, except for climate change, which has notably weaker correlations with the other variables, and even one negative near-zero correlation. Thus, any factor analysis will give it a weak factor loading:

Note the general factor is ~50% of variance, but climate change only has a loading of 0.29, the weakest by far in their study. If we instead use a 2 factor solution, we get results that depend on method: 1) one method makes factor 2 = aliens (area51 + Roswell), which can also be seen from the correlation matrix (r = 0.90), 2) another method makes factor 2 = climate change (the only variable with salient loading). Factors with just a single variable are pointless. This is just telling us again that this variable correlates relatively poorly with the others. Anyway, regarding the title of the original paper “NASA Faked the Moon Landing—Therefore, (Climate) Science Is a Hoax: An Anatomy of the Motivated Rejection of Science” is fairly derpy. Looking at the correlation matrix, there is no particular relationship between climate science conspiracy theory belief and moon landing one, it has a slightly stronger relationship to AIDS conspiracy belief. But given the relatively few people who endorsed some of these beliefs, these correlations will not differ more than expected by chance.

There are many other studies that have looked at many conspiracy, paranormal, and other unconventional beliefs. Many years ago (2015) I posted one of these, that produced these correlations:

That study is from 1994, so that’s why it has quite different beliefs. Japanese_econ is a belief that the Japanese are deliberately trying to destroy the US economy, apparently a topic of interest back then (the Japanese economic growth crashed in 1990 and has never recovered). There was no question about climate science since there was little public debate about that in 1994.

Another more recent study (2024), had a very comprehensive coverage -- 85 beliefs -- which could be clustering/factored into a few dimensions:

Here we see that the correlations are relatively strong between climate science related and the other domains. All the correlations are positive, as with the other studies. People tend to believe in every conspiracy in equal measure, at least, statistically speaking. This suggests that belief in them share psychological causes. Presumably these include low intelligence, schizophrenia-like personality, know-it-all-ness/epistemic supremacy, and so on. The political correlations of belief in these I think will depend on context, which I wrote about extensively previously. I think the general reason that conservatives are given bad names for such beliefs is that scientists and journalists are eager for findings that portray their political enemies/out-group in a poor light, so a lot of attention is given to Obama birth certificate beliefs, while little attention is given to Black people’s theories about how the police are colluding to enforce White Supremacy, or any number of other false beliefs more common on the left or among its ‘protected classes’. Right now, conservatives are also the slightly less intelligent political wing in the USA, so insofar as low intelligence causes such beliefs, they will have a bit higher for now (and the Democrats should be higher back when they were the dumber party).

R notebook for my/Claude’s analysis above. Data is from the Github repo that Duarte posted.