Myth #3: The count is GOING UP!… No WAIT! It’s GOING DOWN!
I love Italy. Every moment I’ve spent there has been a joy; the history and people are fantastic… and now I have a new reason. From a data perspective, their numbers are a perfect example of what you’d expect to see from a country that was punched in the stomach by COVID-19 and took aggressive countermeasures to contain the spread. I selected the virus-related deaths to use as an example, but know that the case count graph follows the same pattern.
Let’s look at the red points. The pre-red death count on May 2nd was 269, and then on May 3rd it jumped to 474. I can’t speak to how that info was received in Italy, but I can imagine how we would have handled it in the USA. I can see a headline of “DEATH TOLL SPIKES!!” accompanied by interviews with experts, with reporters requesting an explanation for the horrific 57% increase in deaths. Then the next day there were 174 deaths, so we’d do the whole thing in reverse. Envision a “Time to Reopen America!” headline based on the ‘dramatic’ 63% reduction. Cigars all around… until the next day.
The first critical principle here is that two points do not make a trend. A mentor of mine once asked “what grade do you need to be in to tell whether one dot is higher than another dot?” Not a useful way to look at information. Instead, you have to put the most recent point in context; take a step back and look at the entire two-month progression. If you do that, you see a sharp increase to the peak, followed by a gradual reduction over time. This is what should happen if the steps taken to contain the virus are effective.
In my opinion, overreacting to / trying to explain insignificant day-to-day fluctuations in data is the biggest waste of time in the business world… but the virus data is unique because it takes overreaction to a whole new level. Why? Several reasons:
- This is literally about life and death,
- People are hungry / desperate for explanations, and
- News outlets need to fill 24 hours of airtime every day.
On that last point, imagine an alternate news report on May 3rd after the case count increase: “the count is up, but we need to wait for better context before we know if this is a momentary blip or a pronounced and prolonged change of direction.”
That won’t sell- I almost fell asleep while I was typing it.
To learn how Ralph broke down the numbers, check out Orion’s Using Graphs and Charts to Analyze Data web class.
Second critical principle: no matter how many data points you have, there will be a highest one and a lowest one. Significance is often attached to the ‘finding’ that one point was higher or lower than the others. Having a ‘highest day’ or ‘lowest day’ isn’t a revelation, it’s a mathematical certainty. So if you hear statements like:
“The number of deaths on April 23rd in the United States was the highest in the entire month of April!”
Remember that all that is really being reported is an understanding of the > sign from first grade math class.
Next up, Myth #4: How quickly can we defeat COVID-19?