I'm inclined to think that this explanation is shaky, at least standing alone. Petroleum engineers, clergy in some faiths, military officers, and corporate lawyers, for example, tend to be conservative but are also highly educated, and neither military officers nor clergy are particularly well paid. Highly educated specialist physicians are not particularly liberal. Mormons have above average education levels but tend to be conservative regardless of profession. College educated people leaned GOP until recently but major party coalitions changed.

I love the regurgitation of education being linked to liberalism. I've yet to see a link explained. There is a correlation, but the causation remains completely unexplained. @ohwilleke points out a few counter-examples to education leading to liberalism. If you consider ideas to be like diseases, it explains liberal ideas roosting in academia better than education itself causing liberalism.

Great answer! So then the question becomes: why do those highly educated people are more often leaning left?

@DavidS My guess would be that the correlation is the other way around, namely that liberalism usually promotes universal higher education while conservatives are usually fine with elitism where they are educated and their followers ... just follow. So broadening education and sharing wisdom rather than safeguarding it is probably the intrinsically liberal idea.

@ohwilleke I notice the first three of these examples are groups that financially benefit from conservative ideology, and the last one is literally a cult.

@ohwilleke Citing negative confounders like you're doing isn't an argument against causation, the fact that the correlation still exists despite them only strengthens the argument for causation. You should prove a lack of causation by citing positive confounders.

The 54-24% liberal lean of people with postgrad degrees is significant, but doesn't in itself explain the overwhelming 97-3% bias of professors claimed by OP. It also doesn't explain journalists.

@dan04 I do not assume y = 1 * x relationship between the variables here, or even any linear relationship at all. In the most extreme case, this can be a feed-forward system (rather than feed-back), where a slight imbalance in ideology leads to self-selection by hiring people with similar ideology. Thus, a system that is 51% liberal gradually becomes nearly 100% liberal.

The portion of people with liberal views across all educational categories went up between 1995 to 2015 according to that graph. I wonder whether particular questions asked were more likely to get favorable responses in 2015 than 1995, or if the favorable responses went up evenly across the questions. The portion of hard right and hard left also seemed to increase across all categories, which may indicate a societal change that isn't solely due to education.

@ohwilleke It is the simplest explanation possible. Plus given the non-linear self-reinforcing effects of ideological self-selection in hiring, a small difference can grow into near-complete dominance of a single ideology. But I would be interested in other explanations, provided they are supported by data.

@DavidS Re: "I've yet to see a link explained. There is a correlation, but the causation remains completely unexplained." I am not claiming I have an explanation for that association. The association is the proximal cause of the facts in the OP. Beyond that, I do not know. But association is a fact supported by data. I invite you to question the association - with data. And no number of anecdotal counterexamples is convincing. Statistics is welcome here, not a list of exceptions.

@IllusiveBrian You can see the particular questions they asked here, and that page also has a report with detailed results for many years besides those shown in the chart. Some specific questions saw consistent shifts, some stayed mostly the same, and some saw significant movement in both directions over the course of the various survey runs.

@TimurShtatland "It is the simplest explanation possible." It is simple but it isn't a sufficient explanation, particularly when 20 years ago the proportion of college educated people who identified as Republicans was much higher. Also as another example of non-generality, in Turkey's 2023 national election less educated people were more liberal. An economic interest in liberal ideology does make sense, for example, for academics just as it does for petroleum engineers. The recent populist anti-intellectual shift on the right could be another factor.

@FrederikVds "Citing negative confounders like you're doing isn't an argument against causation, the fact that the correlation still exists despite them only strengthens the argument for causation." Confounds are absolutely an argument against causation. Correlation does not imply causation although it implies some cause in most cases, either in the opposite direction or from unstated other reasons. Confounds suggest unstated third reasons that affect some but not all cases giving ris to the correlation.

@ohwilleke Positive confounders are an argument against causation. Negative confounders are an argument for causation. A negative confounder means that without that confounder, the correlation would be even stronger, making it even more likely that there might be a causal link. They're not absolute proof of causation, but they strengthen the case for causation instead of weakening it. Every exception to the rule that you can explain away based on a third factor, makes it more likely that the rule exists, just like every adherence to the rule that you can explain away makes it less likely.

@ohwilleke - Clergy in the United Methodist Church in fact have a tendency to be much less conservative than the rest of the denomination. That's in fact been a major driver behind the current troubles. The Conservatives actually have more votes in the UMC conference, but are choosing to leave anyway because they can't get the church hierarchy behind enforcing the (IMHO harsh) conservative stuff they pass in conference.

@DavidS : there was a similar question on SE with an answer which cited sources of conservative professors being fired for very slightly "offensive" acts (by that I mean not openly advocating a liberal cause), while liberal professors could glorify openly racist views without any problem, hinting at universities getting rid of conservatives, but despite plenty of sources the answer... jut... vanished... and the user blocked. Therefore in order to not get that treatment I'll refuse to answer. Or maybe if I answered with "liberals smart, conservatives stoooopid" I could get some cheap upvotes?

@vsz: Do you remember which SE site it was?