This is when you chain correlations of correlations, like a categorical syllogism, to draw a statistical conclusion. For example:
Height is correlated with IQ.
Men are taller than women.
Men have higher average IQs than women.
The conclusion may or may not be true, but the argument is invalid. The reason this isn’t valid is that correlations aren’t transitive. So if A correlates with B and B correlates with C, it doesn’t follow that A correlates with C. This causes a lot of nonsense because we tend to think of qualities using the shorthand language of categories, like “men are taller than women”. Taken literally, this statement means “height(men) > height(women)”, and statements like this have the property of transitivity. But probabilities are not transitive, which means that this logical fallacy can be used to draw statistical conclusions which are categorically contradictory. For example:
High educational attainment is correlated with liberalism.
Being a conservative pundit is correlated with high educational attainment.
Therefore, conservative pundits are likely to be liberal.
This enthymeme makes use of the human tendency to internally translate statistical descriptions into categorical statements:
Educated people are A.
B are educated.
Therefore B are A.
This syllogism is valid. The error was committed in the process of accepting a statistical statement as categorically true. I call this categorical gerrymandering because, as in political gerrymandering, the rhetorician uses statistical majorities tactically to “claim” categorical territory. Because you can produce a chain of correlations between any two statistical quantities, in this way it’s always possible to justify belief that a variable correlates with its opposite, which is a categorical contradiction. It is therefore a method of choice for conceited people who prefer deception by misleading technical honesty, because they can defend themselves by pointing to a list of correct statistical statements and say “show where I’ve made an incorrect statement”.
It may be because I have more exposure to thinkers on the right than on the left, but this appears to be a weakness more common among rightists. I believe this is because they have a greater local coherence drive (autism, colloquially) so that they feel a need to “justify” their beliefs. I love watching his videos, but Ed Dutton is an example of someone who commits this logical fallacy frequently. For example, in his video on whether people in power are more likely to be pedophiles, he makes the argument that:
Pedophilia is correlated with psychopathy.
Psychopathy is correlated with political success.
Therefore pedophilia is likely to be correlated with political success.
Whether he’s correct or not, this argument is invalid. We could make the argument that following such chains of correlations is just “suggestive for future research”, but this is merely kicking the can down the road to maintain plausible deniability. It’s bad mental hygiene to use heuristics which as often as not produce type 1 errors (false positives) while producing more confidence in their likelihood than stream-of-consciousness free association, which at least has no reassurances that one has relied purely on statements of scientific fact.
The worst example of this I’ve seen is the redefinition of K-selected humans as having all of the traits of fast life history strategists and none of the traits of slow life history strategists, which is the literal opposite of the truth. This is almost certainly because the traits associated with a fast life history strategy (e.g. “Chad” boldness, sexual desirability, and low neuroticism in chaotic environments) are considered prosocial, virtuous, and moral in the modern West while traits associated with a slow life history strategy (e.g. “Virgin” risk-aversion, divestment from sexual competition, and dependence on high-investment nurturing social environments for mental health) are considered antisocial, insecure, and parasitic, and k-selected people place a high value on being prosocial, virtuous, and moral.
Leftists, in my experience, are far more likely to simply ignore apparent contradictions and therefore less likely to reach for justifications. For example, an SJW coworker was arguing one time that Michigan should try to be more progressive like California, to which another coworker responded that California has flesh-eating viruses. The SJW simply said that he didn’t believe it, Michigan has bigger problems than California, therefore California is better, therefore if you really think about it California doesn’t really have any truly big problems like Michigan and has better weather to boot, and we should thus try to be more progressive like California. This is like the Jewish method of selective amnesia where you repeatedly forget anything that’s inconvenient to your preferred narrative frame. (You have to admit, these techniques have the strength of being simpler.)