The University of Washington’s IHME COVID-19 model is the one most widely cited by policymakers and journalists. Superficially, that model has been relatively stable, with its total projected fatalities nationwide from the Wuhan virus declining from the mid-80,000 range to its current faux precise level of 67,641, which is probably a pretty good guess–one that you or I might have made–based on experience to date.
But if we look at the IHME model’s projections for individual states, we see capricious changes that cast serious doubt on whether the model is based on any coherent scientific principles at all. I have focused on the five states of the Upper Midwest–Minnesota, Wisconsin, Iowa, North Dakota and South Dakota–in posts dated April 11 and April 15. As a result I have a record of IHME’s projections for those five states on those dates, which is significant because, as best I can tell, each version of the IHME model’s output is sent to the memory hole when it is updated.
So here are the wildly divergent fatality numbers the IHME model has predicted for the Upper Midwestern states, on each of three dates, with the most recent update coming on April 22.
Minnesota
April 11: 442
April 15: 656
April 22: 360
Currently, Minnesota has 200 fatalities, over 75% in nursing homes and assisted living facilities. IHME’s estimate is moving in the direction of reality. But what caused the model’s gyrations over a mere 11 days? And wouldn’t a politician have been foolish to rely on the model’s projection of just nine days ago? I should perhaps add that our governor’s extreme shutdown order was predicated on a home-grown model that said there would be 74,000 fatalities without a shutdown, and 50,000 even with a shutdown–a number that was later revised downward, but nowhere near far enough.
Wisconsin
April 11: 357
April 15: 338
April 22: 356
Wisconsin’s numbers are by far the most consistent of this group.
Iowa:
April 11: 743
April 15: 618
April 22: 365
For some reason, the IHME model has it in for Iowa. It was ridiculous to predict Iowa to have more fatalities than Minnesota or Wisconsin, a fact implicitly admitted by the adjustment over the last week. Again, any Iowa policymaker would have been foolish to rely on the IHME model. And according to the IHME’s own tabulation, Iowa had only 83 deaths as of April 21.
North Dakota:
April 11: 369
April 15: 32
April 22: 356
There is no possible scientific justification for these gyrations in the North Dakota projection. Certainly nothing in North Dakota’s experience explains them. This is IHME’s own North Dakota chart, which shows deaths to date as 13. It is notable that with regard to several of these Upper Midwestern states, IHME has continued to project an epidemic, but has repeatedly moved the date back in time.
South Dakota:
April 11: 356
April 15: 181
April 22: 93
As of April 21, according to IHME’s chart, South Dakota had recorded a whopping 8 deaths. As with North Dakota, IHME predicts the epidemic will get underway any day now, only at a much lower level–around 1/4 of what was predicted less than two weeks ago. Once again, a South Dakota policymaker would have been foolish to make decisions in reliance on the IHME model, or any other model I have seen.
Liberals have viciously attacked South Dakota’s governor Kristi Noem because she refused to implement a shutdown order as most governors have done. She said that she has confidence that South Dakotans are competent to protect themselves–which, evidently, they are. The Washington Post attempted an absurd smear against Governor Noem based on the fact that workers at a pork processing plant in Sioux Falls had tested positive for COVID-19, like workers in pretty much every pork plant across the U.S. The same thing happened in a pork processing plant in Minnesota, where our Democratic governor has implemented a particularly harsh shutdown order, but the Washington Post took no notice, nor did other national media outlets. As usual, it is all about the narrative, and drive-by press coverage is the norm.
The reality, which the Washington Post has no desire to report, is that South Dakota is perhaps the nation’s best-governed state. This has actually caused problems for the state in dealing with the federal government’s bailout. Governor Noem explains in this video clip: South Dakota has a fully funded pension plan, no income tax, no corporate tax, a AAA credit rating, and a balanced budget. The problem the state has as a result of COVID-19 hunkering down is a shortfall in sales tax revenue. But that is one thing the federal dollars–more than a billion of them!–can’t be used to make up. Governor Noem explains:
Kristi Noem comes from my home county in South Dakota. I doubt that she has any interest in running for president, being sane and all, but if she decides to run she’s got my vote.
