In our data-saturated age, when fans of Dr. Fauci collect his bobbleheads and self-proclaimed “nerds” declare on social media that they “fucking love Science!” it’s hard not to encounter arguments rebutted by the popular expression, “correlation does not imply causation.” This maxim—which presumptively debunks seemingly obvious connections between phenomena by suggesting that mere coincidence is often superficial—has almost become a cliché in conversation about any topic, from climate change to COVID, and from inflation to immigration.

Of course, it is true that correlation does not always imply causation, as with the famous example of ice cream sales correlating positively with shark attacks. Indeed, every summer, both phenomena sharply increase, only to fall during the winter. Do increased ice cream sales drive sharks to bite people, or does the news of shark bites create hunger for ice cream? In this case, ice cream sales and shark attacks are independent factors that each increase during the summer, when it is hot and more people go swimming at the beach.

But the rule, which indeed is useful for identifying fallacious logic in one’s own thinking or in the arguments of others, gets misused by people in bad faith who wish to undercut legitimate policy assumptions. And though I have no study to back this up, it does seem as though it is liberals who more frequently use it to dispute or debunk conservative conclusions. 

For example, conservative states like Missouri and Texas have sued the federal government for terminating the “Remain in Mexico” policy, which requires third-country asylum seekers to wait over the border for the adjudication of their claims. Suspension of the program, the states say, encouraged more potential asylees to file claims, on the apparently correct assumption that once they enter the U.S. it is unlikely they will be forced to leave. 

Homeland Security Secretary Alejandro Mayorkas, in a recent statement defending the termination of “Remain in Mexico,” agreed that “the significant decrease in border encounters following the determination to implement MPP [Migrant Protection Protocols] across the southern border” strongly indicated that the program worked. But, he added, “of course, correlation does not equal causation and, even here, the evidence is not conclusive.” Indeed, restricting the unhindered flow of migrants across the border might be a factor in limiting their free entry to the country, but correlation does not equal causation, so we don’t really know if it works or not. 

Another favorite causal effect that is frequently debunked by the Left is the idea of “gateway drugs.” Using marijuana, say prohibitionists, can lead young people to try harder drugs. As evidence, they point out that few people start with “hard drugs” without first trying marijuana. Pot enthusiasts laugh at this simple-minded logical error, as in this article in The Atlantic, which suggests that it makes as much sense as arguing that an appetizer is a “gateway” to a main course.

So right-thinking, data-driven, and empirically-minded newspaper readers have all been trained to identify when other people mistake correlation for causation and to ridicule the fools who think that umbrellas cause rain.

There’s one area in public life, however, which is exempt from this logical rigor, and that’s the question of racial disparate impact. Outcomes among racial groups differ widely along numerous axes: rates of crime, standardized test scores, household income, home ownership, etc. But there’s no doubt among liberal economists, newspaper editors, or professors as to the cause of these disparities. Racism, either overt or “institutional,” is necessarily the reason. In fact, it’s not even necessary to unpack the racism inherent to any disparity: the existence of the different outcome is prima facie evidence of racism. 

And this isn’t just a matter of casual argument in comment threads or among cable talk show disputants. It’s written into the law and has practically constitutional standing. “Disparate impact theory” in civil rights law about hiring, housing, and education puts the onus on the employer, landlord, or administrator to demonstrate fairness. That is, “facially neutral” practices can be shown to be discriminatory in the absence of intention or design. As the Department of Justice explains, “even benignly-motivated policies that appear neutral on their face may be traceable to the nation’s long history of invidious race discrimination in employment, education, housing, and many other areas.” 

Correlation, in other words, does sometimes imply causation. And even if it doesn’t imply causation, correlation in itself may be bad enough to require urgent action. 

So next time someone snickers about your reasoning, ask them if there are any cases when causation can be legitimately inferred from correlation. Could make for an interesting discussion.