Univariate Fallacy and the Monocausal Fallacy

A fruitful and extended discussion of the Univariate Fallacy. The Univariate Fallacy:

The claim that if there is no single, defining trait that can be used to separate two or more categories, then those categories do not exist.

1/ The modern far Left has a political agenda to destroy/deconstruct biological realities under the guise of Social Justice. A common way they go about this is by dishonestly applying univariate statistics to multivariate problems. This is called the Univariate Fallacy.

Thread. pic.twitter.com/xi1n4Xkn3s

— Colin Wright (@SwipeWright) May 3, 2019

Click for the thread.

Lots and lots of examples in the mainstream media and even in science publications who ought to know better.

The Univariate Fallacy is a parallel to the Monocausal Fallacy. In much of everyday conversation, people usually speak and behave as if there is a single cause for some single manifestation of a problem. "What is the root cause?"

Which is fine if you are dealing with monocausal, deterministic systems. Which is rarely the case with human related (or natural) systems. The Monocausal Fallacy leads to all sorts of bad policy.

Home ownership low? Make it easier to borrow.

Local schools reflecting local populations are insufficiently diverse? Introduce busing.

There are disparate impacts from school discipline? Get rid of discipline.

A residential road has too much speeding? Reduce the speed limit.

In all these instances, there is a logical connection between the proposed solution and the manifested issue. But in almost all cases, the proposed solution does not actually fix the manifested issue. There are more causal elements than are being acknowledged.

Complex, dynamic, chaotic, power law affected and loosely coupled systems always have multiple causes which are rarely obvious and produce a range of negative trade-offs in addition to the desired outcome. Tugging on any single cause upsets the apple cart. There are too many moving pieces for any single cause to deterministically drive single outcomes. You have to focus at the system level and focus on incentive structures, experiment, and iterate. And maintain humility in the face of uncertainty.