Lying with Statistics – 101

“Our case fatality rate has continued to decline and is lower than the European Union and almost everywhere else in the world… If you watch American television, you’d think the United States was the only country involved with and suffering from the China virus. Well, the world is suffering very badly. ... We’ve done much better than most. And with the fatality rate lower than most, it’s something we can talk about.”

Trump Press Briefing, 7/22 (above)

When you want numbers to help you make a decision, or analyze where you made a mistake, you probably should frame the statistical question properly. For example, if you want to know how the earning power of middle of the bell curve of Americans is faring (most of us), looking at the stock market or analyzing gross domestic product numbers (where the richest in the land pull up the numbers for everybody) is a misdirection. The numbers you need to look at are those that apply to the target you are trying to understand. Not looking at “everybody” or links that do not correlate to the performance level you are targeting. 

But for leaders from both sides of the aisle, stock market performance and GDP numbers are the measure… if they look good. That 70% of Americans, even before the pandemic, had not had a meaningful increase in buying power in four decades would tell you the real story… and the dramatic and very quantifiable erosion in of the middle class and upward mobility coupled with the rising costs of housing, food, education, etc. 

Likewise, if you want to know what is happening in a rapidly changing situation – like right now – examining the numbers that skew heavily from “a while ago,” aggregating them into the numbers you are looking right now does not help you understand the immediate situation. Apples to apples. Recent numbers, immediate numbers need to be your focus. You want to know about now, look at now. 

Finally, you can skew the results by using words that people think have one meaning but actually have another. Take the word “average,” for example. If I have nine people who make $10,000 a year, and one person who makes $10 million at year – a total of $10,090,000 – the average is the that total divided by the number of people; hence, the average individual earnings in that group would be $1,009,000. So, if you use an average, they’re all doing pretty well. The median number would be the number dead in the middle, eliminating all the numbers below and above. That would be $10,000, which is not such a good number. 

Donald Trump is still trying to convince the world that he is doing a great job at containing the pandemic, even though everyone knows he has done the worst job of any leader in the developed world. He aggregates numbers from the earliest outbreak of the virus and presents cumulative results. For those nations that stared badly but currently have the very acceptable infection and mortality rates, who are not spiking as is the United States, it suggests that they are failures. But if you just look at current results, we are absolutely among the worst performers on earth. Those that have contained the virus get no credit for their success, and Trump just makes us appear good. We are just the opposite. 

Now, let’s focus on both cumulative numbers and vocabulary. “Let’s start with the ‘mortality rate.’ This is a per capita measure of how many residents of a particular country or region have died of the disease. Think of it as ‘COVID-19 deaths per hundred thousand’ (or million, or any other convenient number). 

“For weeks, Trump was saying America’s mortality rate is the lowest in the world. It’s not. According to John Hopkins University, it’s actually the 10th highest, after San Marino, Belgium, the U.K., Andorra, Spain, Italy, Sweden, Chile and France. To date, the U.S. has suffered 43.07 deaths per 100,000 residents. Brazil has suffered 38.25. Mexico has suffered 31.29. Iran has suffered 17.61. Germany has suffered 10.97. 

“And this actually understates America’s current mortality problem. The mortality rate is typically presented as a cumulative number — that is, how many per capita COVID-19 deaths have occurred so far during the entire pandemic. But you can also measure it daily, which gives you a better sense of how your country is performing right now. 

“Here, the U.S. is faring even worse. In early April, far more people were dying of COVID-19 per capita across Europe than across the U.S. Today, America’s daily mortality rate is 10 times as high as the European Union’s. 

“To assist Trump with his briefing, someone in the White House seems to have told him to stop touting America’s (relatively unflattering) COVID-19 mortality rate and start touting its (ostensibly more flattering) case fatality rate instead, which is really what Trump was trying to do all along. Hence the big ‘Case Fatality Rate’ chart… It didn’t really help him make his point. 

“The case fatality rate (CFR) of COVID-19 is just the ratio between confirmed COVID-19 deaths and confirmed COVID-19 cases. This sounds like a solid metric, but there’s a hitch. COVID-19 deaths are fairly straightforward to count. But the raw number of COVID-19 ‘cases’ a country finds is totally dependent on how many infections it’s able to detect through testing.” Andrew Romano, writing for the July 22nd Yahoo News. And we know there are long lines across the nation for testing, some hours long, and extended periods before results come back (there is a huge backlog), so those numbers are at best questionable. 

The other fly in the ointment is the time lag between rising infections and rising deaths from the infection. We watched the infection rate – which correlates to CV-19 hospital admissions and ICU usage for the doubters – continue to skyrocket. It took a couple of weeks for the mortality rates to begin to rise as a result. But they just did. The numbers will begin to worsen, so it will be interesting to see how Trump will spin that negativity. 

The President also announced a nearly $2 billion vaccine research development payment to Pfizer and a smaller German research company with research showing some positive early results. Their vaccine has yet to be vetted through the required litany of clinical tests, but the contract with them calls for a vaccine to be ready and deployable in 2020. 

Even though there are huge additional steps to certify the effectiveness and safety of that vaccine, I suspect we will soon hear statements from the President that the vaccine absolutely will be ready and available this year. Immediately before the November election, Trump could easily promise the American public that the vaccine is 100% going to be ready… and then if it is not, simply say that there were unanticipated glitches or simply blame overzealous government bureaucrats for doing their job. 

Had we taken the steps that Germany, Taiwan, Korea, etc. had taken when Trump first learned of the virus, our death toll just might have been one third the 140,000 plus we have to date in this country. Trump has failed… dramatically. 

I’m Peter Dekom, and too many people are listening to or watching the Trump-directed headlines on the virus without remotely understanding how profoundly misleading those charts and words really are.