Torturing Statistics Until They Confess: An RTD Primer

Image credit: Richmond Times-Dispatch

by James A. Bacon

Sabrina Moreno with the Richmond Times-Dispatch has written a three-piece series arguing that disinvestment in the Virginia Department of Health led Latinos to being “the most likely to get infected, hospitalized and die” during the COVID-19 pandemic.

The fourth paragraph of the story makes the following extraordinary assertion:

Three months after Virginia’s first case, Latinos in Richmond were 38 times more likely to be infected than white residents and 17 times more likely to be hospitalized, according to a Richmond Times-Dispatch analysis of COVID cases and hospitalizations.

That would be an extraordinary indictment of Virginia’s public health system, if true. But it’s not. Even if those particular factoids happen to be accurate for a particular place in time, which I question, it is monstrously misleading. The article did not publish the data, taken from the VDH COVID dashboard, that I now present you…

As of today, “Latinos” in Virginia account for 161,708 confirmed cases of COVID-19, or 9.8% of all confirmed cases. According to U.S. Census data, Hispanics in Virginia comprise exactly 9.8% of the population. If the overall proportion of Hispanics being infected by COVID-19 is identical to their proportion of the population, one might seriously question the accuracy of a claim that they were ever 38 times more likely to be infected than White Virginians at any place or point in time.

VDH data also tell us that Latinos accounted for 6,095 hospitalizations, or 12.8% of all hospitalizations. Thus, it is true that they were, over the full two years of the epidemic, about one-third times more likely to be hospitalized than Virginians as a whole. But that’s a tiny fraction of “17 times more likely” — off by a factor of 50.

Here are some more figures that Moreno neglected to mention in her article. According to VDH data, Latinos account for only 986 COVID-related deaths — about 5.2% of all COVID deaths in Virginia. In other words, Hispanics were almost half as likely to die from the virus as other Virginians. When you’re making the argument that Hispanics are the most likely to die, that might be a data point worth mentioning.

Arguably, the most concrete measure one can devise for judging VDH’s performance in the pandemic is how well it administered vaccines to the Hispanic community. And it turns out that 81.2% of all Latinos in Virginia have had a least one dose of the vaccine — compared to 66.6% for Whites and 63.8% for Blacks. Only Asian-Americans have a higher vaccination rate at 90.8%. When you’re criticizing an agency for neglecting an entire ethnic group of people, that, too, might be a point worth mentioning.

The RTD investigation says its five-month investigation found that state leaders allowed the health department to “wither and shrink” for more than 20 years, creating a “severely undermanned and underfunded” public health system. But the RTD’s own data says that per-capita, inflation-adjusted spending on VDH shrank all the way from $86.62 in 2000 to $85.03 in 2020 — a crushing decline of 1.8%. (VDH’s emergency preparedness budget between 2007 and 2020 shrank by 20% inflation-adjusted, the newspaper claims. But it doesn’t tell us what other priorities the agency did increase funding for.)

I’m not saying that VDH is beyond reproach. We’ve argued at Bacon’s Rebellion that much of its COVID-fighting efforts were misguided or misdirected. And the questions the RTD asks are worth asking. But the RTD’s grotesque misuse of statistics has so undermined its credibility that I don’t believe anything it reports in this series. Not. One. Damn. Thing. I can easily check its use and misuse of statistics, but I have no way of verifying the rest of its report, which shows every sign of jamming facts into a pre-determined narrative.

Anyone who uncritically accepts the findings of this series is a fool. Which probably means it is destined for the RTD’s second Pulitzer Prize.