Labor Day and Sturgis

I haven’t written about COVID in a while because up until this weekend the pattern had been pretty steady.  However, over the weekend the 7 day average has dropped considerably, about 5000 cases per day over 5 days, as can be seen in this graph.

There are reasons to be suspicious about this drop, but it’s still a bit early to draw firm conclusions.  Obviously, the fact that this was a holiday weekend is crucial.  The steep drop in reported cases was most likely driven by delays in testing and recording of cases rather than by changes in transmission.  This is underscored by the fact that there is a similar though smaller change in reported deaths at the same time.  What remains to be seen is whether reported cases rise back up to roughly where they were before or whether this is another stair step drop like the two others we have seen since the July peak.  Only time will settle that question.

While I’m writing, I also want to address the report that the Sturgis rally may have been linked to 250,000 cases.  This is a stark claim in and of itself, but has also been compared to a seemingly contrary claim that BLM protests have not been responsible for a surge in cases.

It’s important to recognize that the 250,000 case number isn’t based on contact tracing.  It’s based on a study that used location data to see which states and counties Sturgis attendees went to after the event.  The researchers then looked at case growth rates in those counties and correlated that to the number of Sturgis attendees who had returned there.  It turns out that counties that had higher numbers of Sturgis attendees tended to also fare poorly on COVID growth during the period. From that correlation, it’s possible to estimate how many fewer cases there would have been if these counties had not fared so poorly.  By itself, this is not a terrible study methodology.  However, it makes a lot of assumptions that aren’t fully warranted.  At best, it presents a sort of worst case scenario of possible transmission.  At worst, it’s a gross confusion of correlation and causation – it may have been that Sturgis attendees tended to be from areas that were culturally rebellious about precautions and hence predisposed to faring poorly.  Here’s a link to a more complete analysis of some of the problems with the analysis and its conclusion.  Be sure to click through to the author’s bio (click on her name at top) to understand her credentials for addressing the issues.

Nonetheless, there are good reasons to think that the Sturgis rally was more dangerous than even the largest BLM protests.  First is simply the size of the event, more than 450,000 people attended Sturgis.  While many more people have attended a BLM protest, no single protest has been anywhere near that size.  Second is the duration, Sturgis lasted for 10 days, BLM protests, individually, don’t last anywhere near that size.  If Sturgis had been broken into many small events of a few thousand people, each lasting a few to several hours, and spread out across the country, it would have been far less dangerous.  Third, BLM protests have a much higher incidence of mask wearing than Sturgis did.  Finally, congregation in BLM protests is almost entirely outdoors.  While Sturgis had outdoor events, the 10 days included lots of time in bars, restaurants, and hotels with little social distancing or other precautions.  BLM protests do pose risks as does any large congregation of people.  However the risks of a large outdoor gathering that lasts a few hours is far, far less than one week-long event that includes both indoor and outdoor gathering.

Finally, here’s an updated projection for the next 10 days.  I’ve decreased the growth rate below its current value anticipating further decline in daily cases.  We’ll see if I was overly or underlie optimistic.


It looks like by the end of September we’ll be over 7 million cases and well over 200,000 deaths unless something changes.

Leave a Reply

Your email address will not be published. Required fields are marked *