Statistics

Who Do You Trust More: A Sturgis Biker or An Academic Economist Waving His Wee P-value Around?

The only thing right about that new Sturgis Biker paper everybody is talking about is the opening quotation from somebody calling himself Smash Mouth (asterisks original): “Now we’re all here together tonight. And we’re being human once again. F*ck that Covid sh*t.”

Amen, brother.

The paper is “The Contagion Externality of a Superspreading Event: The Sturgis Motorcycle Rally and COVID-19“, with authors Dhaval Dave, Andrew I. Friedson, Drew McNichols, and Joseph J. Sabia.

This year’s Sturgis rally, the paper claims, drew almost half a million souls. Since this is research on how people going to that rally got or didn’t get the coronadoom, and if they got it whether they became ill or died, the paper is therefore chock full of statistics on these very important numbers. Right?

Laugh, because that’s the best joke in this post.

The other knee slapper I leave to our academic friends. Which is that they make the same egregious mistake everybody else does by calling a positive test a “case”, with the implication that all “cases” are bad things. That blunder is alone enough to disqualify this work from serious consideration.

I don’t want you to think I can only be critical. Our authors do have some impressive findings. They discovered, and then reported to us, “that foot traffic at restaurants and bars, hotels, entertainment venues, and retail establishments in CBGs [census block groups] hosting Sturgis Rally events rose by up to 90 percent during the event, relative to nearby CBGs”.

Nobody except tenured academics would have been able to figure this. So have some respect.

Then comes this:

Second, we find that stay-at-home behavior declined among residents of Meade County We find that the Sturgis Rally event was associated with a 9.4 to 10.9 percent decline in median hours spent at home.

That kind of analysis can only be the result of a college education.

“Yeah, Briggs, whatever. They’re economic geniuses. What about the COVID? Are these maskless bikers putting the rest of us at risk with their non-stay-at-home behaviors?”

Wait. Let me tell you how they got the data first.

Which is “anonymized”—the very word they used—“cellphone data”. Yes: “census-block-group-level data from 45 million anonymized cell phones that allow us to measure the residence of individuals and the jurisdictions to which they travel.”

Feel safer?

I checked but could not discover any call-outs to the NSA, CIA, FBI, or other agencies. Maybe these messages were encrypted or encoded. Anyway, I have it on good authority that government spooks are grateful to Americans for voluntarily participating in—and paying for!—this great e-tracking project. Way to go, us!

Back to the science. They started with this bit o’ math:

It’s complexity is how you know it’s good. Next, to measure “the impact of the Sturgis Motorcycle Rally on the local resident population” they undertook “a synthetic control approach”, which is evidently better than a genuine approach. There’s another, even more complex statistical equation for that, for the synthetic thingy.

Then came table after table of wee p-values. One even verified people went to restaurants. Others were about the weird model parameters. From all this they concluded “the Sturgis Motorcycle Rally generated public health costs of approximately $12.2 billion.”

Well, they’re economists and expected to say things like that. But I am a simple man, and decided to do something that required none of the arcane math, which is much too complicated for me.

I started with this admission of theirs:

A recent assessment found South Dakota to have the least restrictive COVID-19 policy environment when assessing mask wearing mandates (none), travel restrictions (none), large gathering restrictions (none), statewide school restarts (district level decisions), reopening of bars and restaurants (full indoor-dining permitted), work-from home requirements (none) and temperature screenings (not required).

Must be pretty bad in old SD! No restrictions? No lockdowns? No masks?! The deaths must be raging out of control! Here, using our usual sources, are the South Dakota daily coronadoom deaths:

Those negatives are really there, and the result of corrections applied after the fact. They were in our big pictures too (in the updates) but you couldn’t really see them. See the y-axis here? Tops out at 5. Deaths per day.

The red dotted lines are when the rally took place. I figured the deaths should have skyrocketed after, given all the mingling bikers are famous for. You decide for yourself whether that’s true.

Not everybody came from, or went home to, South Dakota. According to the cell phone spying (their Figure 1), there was big “inflow” from Wyoming. Here are the Wyoming daily coronadoom deaths.

That y-axis tops out at 4. Curiously, Wyoming never locked down, either.

Well, Colorado, then. More “inflow” from CO than form Wyoming, they said. Lot of hippies and assorted weirdos around Boulder. They should bump them numbers up! Here are the Colorado daily coronadoom deaths.

Rats. Wait. Nevada had to be it!. That was their next biggest “inflower”. Think of all those boogery chips passing between wrinkled hands. Here are the Nevada daily coronadoom deaths.

The deaths did pop up two days after the rally, but it takes, as we have been relentlessly told, two good weeks after infection to start to see the rack up in body count. Here deaths are on the way down at that time. Meaning, if we follow our good academics’ method, the Sturgis affair was protective.

And, hey, I just remembered that these are all attributed deaths, and not actual deaths, which will be lower.

Hold up. Lots more inflow from Minnesota, Michigan, Nebraska, and North Dakota. What about them?

At this point I become bored.

(If you want more criticism, go see the Ethical Skeptic.)

To support this site and its wholly independent host using credit card or PayPal (in any amount) click here

Categories: Statistics

18 replies »

  1. There was more evidence that “peaceful protests” “caused” the “second wave”

    (at least there was more correlation)

    I might be able to come up with wee P’s if you care
    Or am I thinking of “Sweet Peas”? (Same thing?)

  2. Shouldn’t it take MORE than two weeks for deaths?

    Want to see how silly the whole thing is? Use the SAME scale for death scales on all the maps. MI looks highest, so use that scale. Yep, you’ll need a magnifier for the other states. I learned that technique from mocking climate “science”.

    Wow, Briggs, you went a long ways before boredom set in! I was bored at the headline, but just had to know what stupidity the “experts” were peddling like gypsy wagon sellers for this week.

  3. Spain now has full hazmat suits to deal with ebola–no, wait, Covid. They arrested a surfer and are outfitted in tinfoil looking hazmat suits. The suits probably protect from the alien thought machines that make people not afraid of the Covid non-event. Insanity. Maybe the suits will cause heat-stroke….

    How did geometric functions end up in statistics? I really don’t understand that one.

  4. Second wave is a bald faced fiction. There *will* be a second wave of some sort come flu season and it will (again!) be viciously exploited to kill as many old and sick people as possible and foist more tyranny on the healthy.

    I’ll take the Sturgis crowd any day.

    Both my older brothers were/are bikers. Sturgis attending, with big hogs they built/rebuilt themselves for the last umpteen decades, often in their freaking living rooms with the cooperation of their “old lady.” Not the rice burner sissy boy crowd or the well-to-do suburbanite post retirement luxury Harley crowd.

    They have had nothing but contempt for hippies for what, about 4 or 5 decades now. And they’ve been expecting this pile of grief we’re living through now for just as long.

  5. Sturgis is White Supremacist Nazi Hitlercyclists getting together to gas the Choosies.

    The Overlords cannot allow it.

  6. Long-time reader, first time commenter. I’ve got a question for you, Mr. Briggs. Synthetic control is a relatively new development in the field of economics. It seems to me that the synthetic control, given the background noise in the signal, would tend to be created of groups that are “riding” the error term closer to the sought-for mean, which would then amplify the effect size as the elements of the synthetic control return to the mean. Is that a valid objection to the method? And if so, how might one go about proving it? Thanks!

  7. Briggs is deserving of a Nobel Prize in Medicine.

    This morning, the Associated Press reported that a Norwegian lawmaker has nominated the current President of The United States for the Nobel Peace Prize. Even I can accurately calculate the odds of this nomination being reported by NPR, CNN, MSNBC, Pravda (a/k/a the NY Times), the WaPo, NBC, PBS, CBS or the La-La Times.

    https://apnews.com/4ec1ce1ff6cf9d7321d9cad200650e2c

  8. What probably should shock me, but has ceased to do so, is this. What in the hell is wrong with people to believe B.S. like those reports? (I realize that statistics is sort of a black art, but damn…) More importantly, what the hell is driving people to keep stoking fear, like the publishers of the report in question here? I usually opt for “follow the money” in these cases, but even that seems like a stretch. Geez.

  9. Wilt, there is a universal truth behind the old saying: “If you can’t dazzle ’em with brilliance, baffle ’em with bullshit”. An awful lot of people DO fall for the bullshit, so they keep on spreading it.

  10. With the covid paradigm shift we’ve all got to learn to live with lowered expectations;
    empty store shelves, boarded up store fronts, and ever greater concentrations of wealth.
    The new normal will imperceptibly take on the dimensions of Moscow circa 1950. We
    could even see the reintroduction of five year planning to keep all those statisticians happy.

  11. Just to let you know, Smash Mouth is the name of the band that performed at Sturgis, not a specific person in that band.

  12. Um, Hope-Simpson seasonality. South Dakota is a northern (latitude) state; so the true prevalence should be low now, for coronaviruses anyway. Sure, folks from elsewhere can and will bring it with them if they have it (like the northerners did snowbirding to Southern states along the Gulf of Mexico or Atlantic during the northern hemisphere winter and early spring). Coronavirus season in the US South is ending now. This strain probably will become endemic, just as others have before it.

  13. Wait, wait, wait. More than 400,000 people came to this tiny town and “foot traffic increased 90 percent”??? Hundreds of thousands of extra people, and foot traffic to bars didn’t even double? What kind of numbers are these people using?

Leave a Reply

Your email address will not be published.