BLM and Covid: Dissecting a Questionable Study

“Scientists” don’t impress me. I’m the progeny of one of the best of ’em, and I know how the sausage is made. A smart enough “scientist” can finesse any study to reflect a predetermined conclusion. Scientists go where the grant money is. And they go where the government and corporate jobs are. And nowadays, in the Cancellation Age, they go where the lynch mob demands.

I’m not saying that all scientists are bought and paid for; I’m just saying that they’re more likely to be tempted into corruption than a roofer or bricklayer or whore, all of whom have a relationship with customers where poor results are immediate and obvious and unlikely to be explained away by fancy verbiage.

Now, before I get to the meat of this week’s column, a brief diversion, just to put an even finer point on what’s to come. Back in the early 1990s, U.S. cities were experiencing an unprecedented wave of violent youth crime. Reluctant to blame the epidemic on certain demographic populations, “scientists” and assorted experts soon caught on that the best way to get an op-ed on the topic in a major newspaper was to pen a “study” pinning the violence on movies and TV shows.

An analysis of aggressive behavior in ghetto kids from broken homes? No takers. “Hollywood is hypmotizing our children into committing murder”? Hello, New York Times!

In 1994 I was invited by a Hollywood libertarian group to speak on the subject. On the face of it, the “violent media” crusade had problems. If violent movies or cartoons “make” otherwise normal kids violent, Japan would have the highest youth crime rate in the world. But I wanted to go beyond that, and pick apart some of the specific studies that were at the time being widely cited in the press.

Professor Leonard Eron of U Michigan was the elder statesman of the “blame Hollywood” crowd. Eron, a Fulbright scholar and diplomate of the American Board of Professional Psychology and a fellow of the Academy of Clinical Psychology, the American Psychological Foundation, and the American Association for the Advancement of Science, had recently opined in the L.A. Times that TV shows “turn young girls violent.” To prove his point, he interviewed female juvenile offenders and asked if they were fans of the Bionic Woman TV show when they were younger. As many of them were, Eron declared causation.

I called Eron at the university. To his credit, he agreed to speak on the record. I asked him a very simple question:

“Have you ever actually seen an episode of The Bionic Woman?”

He said no, but he “knew it was violent.”

Now, I’d gotten to know Lindsay Wagner back in the ’80s (she was affiliated with a leftist org my AP history teacher introduced me to). And I knew, from Lindsay’s own mouth (and from having watched her crappy show as a kid), that—as a vegan hippy-dippy peacenik—she had specifically instructed the producers to never have her character engage in “aggressive” violence. Two years on the air, and she never socked anyone. Or killed anyone. Her bionics were primarily used to help nuns and orphans; every episode had a moral lesson at the end.

To that extent, it was an oddly nonfeminist show, as the “superheroine” mainly used her powers to make sandwiches and clean floors.

As I explained to Eron, if those young delinquents he studied had all been big fans of the program, in a way it proves that TV messaging doesn’t influence real life. They’d been fed weekly messages of love, kindness, and nonviolence from a show they enjoyed, but it didn’t stop them from taking the opposite path in life.

Eron had no response. Literally, none. He said he’d try to watch the show for himself. We exchanged info…and I never heard from him again!

Eron wasn’t necessarily corrupt or stupid. He was the product of an environment wherein when a scientist says the “right thing,” whatever that might be at the moment, he gets published and never critiqued. Hence fatal flaws in major studies can result. Never be cowed by credentials or tenure. To hell with the fact that they’re “scientists.” Always call them out when needed, especially as some of these lazy fakers and frauds have the power to influence policies that the rest of us have to live under.

Last June, a “scientific study” about Covid and the BLM riots exploded onto the scene with stunning speed and widespread attention.

Black Lives Matter Protests, Social Distancing, and Covid-19” was commissioned by the National Bureau of Economic Research (NBER). The paper had five authors: Dhaval Dave (Bentley University), Andrew Friedson (UC Denver), Kyutaro Matsuzawa (San Diego State), Joseph Sabia (San Diego State), and Samuel Safford (Center for Health Economics & Policy Studies).

The study made a sensational assertion: BLM unrest, which saw hundreds of thousands of people take to the streets in cities all over the country in violation of Covid lockdowns, didn’t just not increase Covid spread, it reduced Covid spread!

“Regardless of its flaws, the NBER paper will continue to be cited as authoritative by ‘friendly’ sources till the end of time.”

The riots and protests made the pandemic…better!

The authors, who did not have their study peer-reviewed, based their astounding claim on cell-phone data, which, they held, proved that the violent riots forced people off the streets out of fear. In other words, the BLM violence made ordinary folks afraid to leave home, thus Covid was not spread. The authors claim that the cell data shows that enough people stayed home to offset any spread caused by the people in the streets.

All throughout the summer, this study was everywhere. Every magazine, every newspaper, every TV news program, every news website. It soon became yet another piece of received wisdom. “Of course BLM didn’t spread Covid. We have a study that proves it!”

The authors did the media circuit like celebrities! And no one asked them a single pointed question…even though the study had a central flaw so massive, so obvious, only a “scientist” could’ve missed it. Let’s very specifically spell out the authors’ methodology: employ cell ping data to establish the “home base” of users—the place where they sleep—and then examine how much “away time” those users spent pre–BLM riots, contrasted with how much they spent during the riots. “Away time” decreased in riot-affected cities, thus “proving” that thousands of people who, based on the pre-riot stats, would be outside were now staying home.

Catch the flaw? The cell data doesn’t differentiate between driving and walking, or being in a crowd or by oneself. The authors assume that the pre-riot “away time” always involved doing in-person activities in rooms or venues filled with strangers. But in the case of L.A.—one of the cities that was studied—that’s simply not true. In May, in the weeks leading up to the riots, our malls and bars were closed, entertainment venues and museums were shuttered, indoor dining and worship banned. Strict distancing rules were in place regulating park and beach visits.

In other words, that missing away-time traffic—the traffic that was there pre-riot but not during it—would very likely have consisted of people in their cars, alone or with the same folks they live with, driving, perhaps to get takeout, or maybe delivering for Uber Eats or Grubhub. There’s no proof that the people scared off the street by the riots would have been “mixing” had the riots not happened, because in L.A., there was practically no place to “mix.” To claim that the riots prevented spread by scaring people away from crowds, and to attempt to back that up by saying “the weekend of May 16 cell pings show that more people were outside the home than on the weekend of May 30,” doesn’t wash, unless you can prove that the May 16 “away activities” that in theory were curtailed by riot fear involved crowds and mixing.

And in L.A., they didn’t. They couldn’t have.

In June, I attempted in a most friendly and respectful manner to contact the authors, to run that perceived flaw past them. As the study was being used to excuse rioting in my city, I felt I had a small right to ask a question.

I emailed all five guys. Only Safford got back to me, explaining that he’s not one of the corresponding authors, so I’d have to “reach out to Drs. Dave, Friedson, and Sabia.”

And I did. Repeatedly. All summer long. No reply from any of them.

In July I reached out to the people at NBER. After all, it was their study. So I found the names of the responsible admins, and I emailed them. No reply. A week later, I tried again. And I found that my email address had been blocked by all of them!

No joke. Here’s a screenshot of my inbox, filtered to show all communications with NBER.

It may not be a joke, but it sure is funny.

The thing is, though, I have a small problem with being told “no.” So I waited, patiently. The whole rest of the year. And this month, I had an old friend—a really good guy who works for a leftist radio station—act as my front. We couldn’t email all five authors again, as that might’ve raised suspicion that it was me. So we chose one: UC Denver’s Dr. Friedson.

And guess what? Because the email was now from a “friendly” media organ, Friedson replied swiftly!

Sadly for him, he didn’t get the expected hand-job; he got some tough questions. As my friend told him that the questions were for an upcoming on-air segment, Friedson actually answered them, so as not to appear uncooperative to a receptive audience.

Friedson pointed out that the study had recently been updated. Interestingly enough, even though the authors never took my 2020 emails, they appear to have taken my advice. They tried to deal with my point about car traffic by amending the study to factor in specific foot traffic at bars and restaurants, to show that the riots specifically curbed foot traffic at locations where people otherwise would have been mixing.

I’ll let the conversation play out as it occurred:

Friedson: The data has been updated to cover a longer time-period, but the findings remain the same. The new version also has some information on foot-traffic at businesses around the times of the protests, and shows that one possibility for our findings is that the protests kept people away from locations where spread of COVID-19 might have been more severe (such as bars and restaurants). The paper is still undergoing peer review.

Me: If I understand correctly, you measured foot traffic at establishments like bars and restaurants by studying cell-phone “pings” at those locations prior to the beginning of the BLM protests, and during the protests. Is that correct?

Friedson: That is correct

Me: Regarding Los Angeles, at the time of the BLM protests here, no bars or indoor-dining restaurants were open due to Covid. Bars were closed entirely, and any “pings” at restaurants before or during the protests would have had to represent orders being picked-up for takeout or sent out for delivery. Was that taken into account?

Friedson: Yes, our methods take this into account.

Me: Assuming, as is reasonable, that people continued to eat during the days of protest (i.e., even if away-from-home behavior patterns were altered, the need to eat was not), is it possible that the foot traffic that might have been reduced at restaurants was redirected to supermarkets/grocery stores? Were those types of establishments monitored for their foot traffic? (purely anecdotally, I can attest that foot traffic and a measure of “panic buying” were up significantly in my West L.A. neighborhood markets during the week of the protests)

Friedson: No, this project did not use any data on grocery stores.

Me: Also, in a city like L.A., where a curfew was imposed during the week of the protests (and the week after), leading to all grocery stores cutting back their hours, is it possible that foot traffic at grocery stores became more dense during that period (i.e., with fewer hours to shop due to the curfew, the stores became more crowded)?

Friedson: That is possible, but I have no empirical evidence either way.

Me: Several areas in L.A./Beverly Hills saw instances of looting from May 30th through June 1st (Beverly Hills, Melrose, The Grove shopping mall, the Fairfax district, Santa Monica, Hollywood). During that time, retail stores saw fast but massively dense instances of large-crowd indoor gatherings (i.e., a crowd of looters would be packed very densely into a location, for as long as it took to clean the location out). While those looters certainly spent less time in the store than if they were shopping legitimately, for the briefer time they were indoors, they were packed together much more tightly (the looters didn’t observe the social distance markers at the checkout line, for example). Were the possible effects of this dense indoor packing factored into your study?

Friedson: If their cellphones were on and pinged in the location, then they would show up in our data. We have no way to discern between a customer and a looter, we merely know the number of pings.

Me: Finally, the November 2020 Journal of Public Health study Black Lives Matter Protests and COVID-19 Cases: Relationship in Two Databases states that there was a “statistically significant” increase in Covid cases following the protests. Do you dispute that finding?

Friedson: I find our analyses to be more convincing.

Things went so well, we picked it up again the next day!

Me: Do you think it weakens your study that grocery stores were not factored in for foot traffic? Looking at L.A., where, in late May/early June, no bars, indoor dining, malls, etc. were open (regardless of the protests), if we examine the ways in which residents of the areas that experienced unrest (Beverly Hills, Century City, Fairfax, Melrose, Santa Monica) would have left their homes for food, we have two possibilities:
A) drive to pick up takeout from a restaurant
B) go to a grocery store

Those were the ONLY two options (I’m not counting delivery, as that would not entail leaving the home).

Now, if we compare the process of picking up food from curbside service (one person hands bag to another person open-air) vs. going to a supermarket (hundreds of people mixing indoors), there’s no question that the latter puts people in greater close-quarter contact with a greater number of people than the former.

So for your study to be as accurate as possible, shouldn’t you have gauged if supermarket traffic was up in the cities you measured?

Friedson: Not having grocery stores is certainly a limitation. That said, we have plenty of data on at home behavior (which was up) and COVID spread which was flat. My read of our results is that anything going on at grocery stores is part of these net estimates (albeit an unobserved part) so if protests were causing huge infection spread in grocery stores it would have shown up in the COVID estimation.

Me: I’ve been in touch with Dr. Greg Neyman of the Journal of Public Health study. He stands by his results as you stand by yours. Dr. Neyman states that there was a “statistically significant” rise in Covid cases in cities that experienced BLM unrest, but he told me that he doesn’t blame the protests, but rather the other things going on in those cities at that time that attracted crowds. Your study, on the other hand, states that nothing else was going on during those days that attracted crowds.

You see my dilemma? I have one study that says, “BLM protests kept people from mixing in close quarters outside of the protests, and therefore drove down infection rates,” and another study that says, “there was a rise in Covid cases during the BLM protest weeks, but only because people were mixing in close quarters outside of the protests.”

Study one: “Covid cases were lessened by the protests because the protests kept people from doing non-protest activities that led to mixing.”

Study two: “Covid cases rose but only because people were doing non-protest activities that led to mixing.”

Both of those realities cannot be true at the same time in the same city.

Friedson: I can’t solve your issue of who’s study to believe for you. I do not find the other study particularly convincing – the authors themselves give reasonable alternative explanations for their findings in the paper: their control group of cities is much smaller population than the protest cities so it is possible that they’re just picking up pre-existing differences in the trend of COVID cases, our statistical methods explicitly check against this potential problem.

But my honest advice is to just ask a credible third-party expert to give their read of the papers, ideally someone who knows how to evaluate these types of studies.

Third party? Naw, I prefer to go to the source. Next week, I’ll tackle the aforementioned Greg Neyman/Journal of Public Health study. Expect some additional hilarity! And maybe even some edification.

For now, though, I’ll leave you with my brief summation of the NBER paper: Without grocery stores factored in, it’s worthless. Whether or not the BLM riots caused a rise in cases, the notion that they kept non-rioters from “mixing” cannot be accurately ascertained without data from the main places that people in a locked-down, curfewed city like L.A. would have gone for food.

Absent that data, the study is valueless and its central thesis unprovable. Curfew behavior would also need to be studied. For example, did the curfews (which wouldn’t have happened without the riots) lead to larger crowds in grocery stores, as nighttime shopping was curbed, preventing a normal daily distribution of customers and forcing more shoppers into fewer shopping hours (and thus creating longer checkout lines and more “mixing”)? Maybe the rioters forced a few people off the streets. Maybe the curfews they caused forced a lot more people into crowded supermarkets.

Sadly, regardless of its flaws, the NBER paper will continue to be cited as authoritative by “friendly” sources till the end of time.

But at least now, if you’re confronted with it, you can cite something in return.



Columnists

Sign Up to Receive Our Latest Updates!

SIGN UP

Daily updates with TM’s latest