top of page

Disparity-Mining: Advocacy Journalism's Statistical Hype

December 2022

Advocacy journalism is the errant descendant of 1970s Gonzo Journalism. The difference between the two is that Gonzos never pretended to be objective.

Today's advocacy journalists are devotees of disparity-mining. This is the practice of digging up statistical nuggets that are used to exaggerate the socioeconomic inequality between groups. More often than not, the journalists' analyses are woke exposés disguised as empirical science. They exaggerate and mislead.

Here are three typical examples of their statistical deception...

Diversity scandal in the Census Bureau's executive suite.

In a recent article, an NPR reporter agonizes about the glaring whiteness of the 47 senior executives at the Census Bureau. The author's smoking gun is this statistic:

People of color make up 40% of the U.S. population, but they represent only 23% of the Bureau's senior executives.

This huge 17 point disparity (40-23) is objective evidence of serious social injustice, right?

Not really. The fraud lies in using "percent of U.S. population" as the benchmark, as though nothing less than 40% representation can achieve equity. This benchmark is a journalistic favorite because it usually maximizes the disparity.

But it's invalid because eligibility for executive jobs in the government requires citizenship and a college degree. So, applicants for those jobs are not drawn from the entire U.S. population, but from a subset of Americans who are college educated citizens. And within that subset, only 25% are people of color. The appropriate benchmark for racial equity is 25%, not 40%.

In other words, in the absence of racial discrimination, people of color should expect to capture 25% of the executive seats. Since 23% of the Bureau's honchos are non-white, the real disparity is only 2%, nothing more that statistical noise.

Advocacy journalism is quite adept at detecting grand inequities where they don't exist.

Racial Bias in the FBI

For years the FBI has been scolded for its failure to gather a special agent corps that "looks like the people they serve." Currently, only 4.4% of special agents are Black. When compared to Blacks' 13.4% share of the US population, the result is an extravagant disparity of 9 percentage points, which of course, insinuates racial bias.

Again, "percent of the U.S. population" is a fraudulent benchmark. The candidate pipeline into the special agent bailiwick does not draw from the entire population, but from a subset of citizens that satisfy the FBI's eligibility requirements: young adult, college educated, felony-free, and paid-up child support.

Within this pool of non-felonious, paid-up, college educated citizens, only 7.4% are Black, not 13.4%. The drop to 7.4% is mostly due to Blacks' lower rate of college completion. However, the fact that Black men are much more likely to have felony convictions and unpaid child support also matters a bit.

The appropriate benchmark of 7.4% shrinks the racial disparity to only 3% (7.4-4.4)

The most likely explanation for the 3% gap is NOT racial bias in hiring and promotion, but the perpetual fact that Blacks score significantly lower on every postgraduate standardized test - GRE, LSAT, ITE, MCAT, and GMAT. Since part of the FBI's acceptance exam is similar to those tests, it's safe to assume that Blacks are less likely to meet the cut-off score.

So, contrary to the wishes of advocacy journalists, Blacks' 4.4% representation among special agents is not a scandalous case of racial injustice.

Rather, this case clarifies why the FBI's Diversity Officer has been unable to significantly boost Black representation among special agents. The reason is, he's powerless to change the structural impediments to recruiting qualified Black applicants. That is, he is powerless to increase the college completion rate of Black citizens and powerless to reduce the prevalence of felony convictions and unpaid child support among Black men. So, the best he can do is to institute advertising and outreach programs in order to drum-up more applicants in the hope that enough of them are qualified. That's a little better than grasping at straws.

80% Racial Gap in the Mortgage Market

Under stress to appear woke, even the most respectable of news programs can be taken in by statistical exaggeration. Such is the case with The Marketplace Morning Report. It dedicated a segment to reviewing a "complex statistical analysis" of mortgage applicants who met or exceeded standard financial criteria for loan approval. The headline finding: "Lenders were 80 percent more likely to reject Black applicants than comparable White applicants." Wow, 80% signifies flagrant racism, no?

No! Rejection data was chosen as the focus precisely because it magnifies disparities. Since the rejection rate numbers shown in the table are very small, the resulting percentage difference between them is huge.

And that is why the segment never examines the approval rates, the flip-side of rejections. Approval rates produce a drastically smaller disparity, as shown in the table. It reveals that almost everyone who is financially qualified gets approved: 95% of Whites and 91% of Blacks. So in terms of approvals, the racial disparity is just 4%.

mortgage approval and rejection rates

But the most serious indictment of this report that the 4% disparity isn't real; it's an artifact of missing data on a critical variable. The authors admit that they could not secure applicants' FICO scores which happen to be the most powerful predictor of loan-defaults. Had that information been included in the estimation of applicants' qualifications for a mortgage loan, the observed disparity would surely shrink.

So, the Morning Report's exposé of racism in mortgage lending turns out to be "sound and fury signifying nothing." But it does score woke-points.



bottom of page