There are lies, there are damned lies, and then there are statistics.
That famous axiom takes on particular meaning when faced with a worldwide pandemic and economic disruption. Statistics are an incredibly powerful tool. But, borrowing Spiderman’s equally-famous maxim: with great power comes great responsibility.
Headlines — and politicians — shout numerous stats. The mortality rate for COVID-19 is 13 percent. Or maybe seven-tenths of one percent. We aren’t sure. Meanwhile, some have attacked public officials for “overstating” COVID-19 deaths. Which then increases the “death rate.”
Publicly announcing these statistics should be done carefully, whoever the messenger. Even when they are technically accurate, critical context is necessary to responsibly convey the information within them.
Overstating fatality rates creates unnecessary, irrational fear. Prescriptions for anti-anxiety medications have greatly increased since the start of the pandemic. Depending on the length of this disruption, it is likely that the stress from the coronavirus situation will be more damaging to many than the disease itself.
Yet underreporting COVID-19 deaths presents its own problems. Imagine a person diagnosed with terminal throat cancer from a lifetime of smoking. It is awful. They are discharged from the hospital to receive hospice care at home.
On the drive back, they get into a car accident and pass on. How should that death be categorized? Vehicular? Even though their life was likely at its end, it was the accident — and not the cancer — that was responsible. The same analysis holds true with COVID-19.
We overstate fatality rates when we simply parrot a fatality rate as a percentage of confirmed cases. It is becoming more and more clear that there are large segments of the population who are asymptomatic carriers of coronavirus. Until we have much better information, the media and officials are likely creating anxiety based on incomplete information.
Yet we cannot simply discount coronavirus-related deaths simply because the individual may have had an underlying condition. Or, in technical terms, a comorbidity. A lung cancer patient with COVID-19 symptoms who passes away may very well have died of coronavirus. We cannot simply wish away the reality that, for certain vulnerable populations, this can be a deadly disease.
The same caution should apply to economic news. The various headlines announcing massive unemployment percentages are probably accurate — but still misleading. Our statistical approach to “unemployment” simply isn’t built for a situation where the economy is intentionally plunged into a recession by government orders issued to protect public health.
“Unemployment” is measured monthly. However, if we measured it daily, we would probably find massive spikes routinely occurring. Those who work Monday through Friday jobs are “unemployed” on Saturday and Sunday.
The headline “unemployment rate” is also just one statistic among many. It counts people who have looked for work in the previous four weeks and are available to work. Of course, none of this takes into account when jobs — in lodging, food service, retail, and other industries — are prohibited from opening up due to a worldwide pandemic.
These nuances are lost when top-line statistics are reported. Comparisons to the “Great Recession” following the subprime mortgage crisis are inapposite. This is not the collapse of an industry based on market forces. It is something very different.
Models — built upon statistics — carry the same risk of misinterpretation. Statisticians know that “all models are wrong.” However, “some are useful.” The caveat is that they are useful only if used responsibly. It becomes easy to discount models when their predictions do not come to pass.
One of the most well-known COVID-19 models is provided by the Institute for Health Metrics and Evaluation. At the beginning of the month, it forecasted nearly 373 Maine deaths through August. As of April 22, that count was down to 53. The model is certainly wrong, but it can still be useful.
And that is the point. These statistical tools — whether measuring or modeling pandemic progressions or economic estimations — are incredibly powerful. But like other powerful tools, from arc welders to wood chippers, they must be used responsibly.
We can kill people from anxiety. We can kill people from underestimating the threat. We can kill people by destroying their ability to make a living or by scaring off economic recovery by inciting fear.
If cooler heads, skeptical eyes, and responsible actions prevail, we will get through this. And just maybe we will all be a little bit smarter when people start spouting statistics at us. I’d give it a 76.4% chance.