The Sturgis Statistical Misfire

Blaming a South Dakota rally but giving protesters a Covid pass

WSJ editorial.

"Holy Hell’s Angels. Last month’s motorcycle rally among the deplorables in Sturgis, South Dakota, contributed to 266,796 Covid-19 cases and $12.2 billion in public-health costs—or so claims a new study widely flogged by the media. Here is another example of how flawed statistical models can generate grossly exaggerated projections.

The annual 10-day rally with live performances, races and bike shows drew 460,000 Americans from around the country to the small town of Sturgis (population: 7,000). South Dakota is among the few states that never imposed a shelter-in-place order or face-mask mandate. It has also let restaurants and bars operate without restrictions.

Many in the media and various health sages warned that the motorcycle rally represented a “worst case scenario” for a “super-spreading” event. It would be surprising if the rally didn’t lead to at least some new Covid cases. But it’s impossible to know how many because there’s no data on how many attendees or their close contacts later tested positive.

Never fear, the modelers as always are here. San Diego State University’s Center for Health Economics & Policy Studies sought to quantify the rally’s Covid impact in South Dakota and nationwide by analyzing the (anonymous) cell-phone data of attendees. They then compared case trends in counties with high, moderate and low numbers of attendees.  

They found that cases increased by 3.6 to 3.9 per 1,000 in South Dakota in the three weeks after the rally, translating to between 3,185 and 3,441 more cases statewide. To put this number in perspective, New York state has recorded as many cases in the last week even with its stringent public-health rules. 

South Dakota still has among the lowest per capita death rates in the country (19 per 100,000) and fewer deaths and cases per capita than its neighbors Nebraska and North Dakota. Covid patients currently occupy 3% of state hospital beds and 6% of intensive-care units. So it seems that attendees at least didn’t expose the society’s vulnerable to the virus even if they were putting themselves at risk.

Where the study jumps off the rails is linking all of the relative increase in virus cases in counties with attendees compared to those without rally participants. The modelers multiplied the percent increase in cases for counties with attendees by their pre-rally cumulative cases to get a total of 263,708 additional cases—266,796 including South Dakota’s increase.

But many “high inflow” counties like Los Angeles, Maricopa (Arizona), Clark (Nevada) and El Paso were experiencing flare-ups before the rally. These counties may have shared other characteristics like higher population density that contributed to their increases. There could be other “endogenous” variables—for instance, counties with more people who attended the motorcycle rally may also have had populations less observant of social distancing.

The study’s authors nonetheless assign each of these 266,796 Covid cases a public-health cost of $46,000—ergo $12.2 billion—though the vast majority of all virus cases are mild or moderate. Talk about a case study in statistical overreach—and double standards.

Another study by the same authors concluded that President Trump’s rally in Tulsa in June didn’t lead to an increase in cases. It was ignored by the media, which has dismissed the mere suggestion that this summer’s nationwide Black Lives Matter protests may have contributed to outbreaks. Who knew the deplorables of Sturgis were more vulnerable to the virus than progressives?"