One year ago, on Thursday, March 12 – I left work for what seemed might be an extended two-week ‘snow day’. A year later, much has changed – Massachusetts has yet to reopen, many big cities still do not have open schools, and the lockdowns appear to be worse than the risk of the initial corona virus.
Reader’s will look back at the period from March 2020 – March 2021 as a period of chaos and confusion. Social media has enabled a consensus – to the point that dissent is punished and will result in a platform banning. That consensus isn’t in line with past thought.
One thing I recall time and again during undergraduate studies was how hard it was to control nature, especially micro-organisms. That desire to have control is hard to admit, when it is clear that we don’t have it. Two other laws have defined this time period:
- Farr’s Law: As stated, this law focuses on the death rate, because every other metric is hard to agree on.
- The Law of Small Numbers: States that humans try to explain statistics with small pools of data.
Combined, these two laws show that we’re trying to explain data that isn’t even the data we should be focused on. This adds to the confusion in messaging, in public response, and in how the media reports on this confusion.
Pandemic Status March 2021
- There is a respiratory virus, it has been identified as a Corona virus.
- It originated out of China, although specifics are debated and this topic is verboten.
- There is no mass media consensus on therapies that reduce the severity of a covid-19 infection.
- Global public health (“PH”) response to the virus has been to issue ‘lockdowns’ – proactive quarantines where the outbreak is ‘bad’.
- The effectiveness of lockdowns is debated; the cost on business, personal freedom, and quality of life is high.
- There are all kinds of ways to interpret, mis-interpret, and de-bunk data about the efficacy of lockdowns.
- The same issues around lockdowns exist around masks, which are also ‘mandated’ in many parts of the world. These are cloth masks, as the global nature of the pandemic caused constraints in the mask supply chain.
- Flu cases are down dramatically worldwide.
- Obesity has recently been identified as a co-morbidity in 90% of covid related deaths.
Farr’s Law – Focus on the Death Rate
William Farr was a British epidemiologist and regarded as an early practitioner of medical statistics. He is known for Farr’s Law, which defined two key items – deaths in a pandemic are the only reliable statistic, and the growth of deaths follows a bell curve.
The death rate is a fact; anything beyond this is an inference.William Farr (1807 – 1883) [Link to the Centre for Evidence Based Medicine]
Farr’s Law is immediately relevant because statistics are so hard to agree on in a pandemic. Especially early on when ‘cases’ were growing, and then in the summer of 2020 when the zeitgeist moved away from deaths to ‘cases’ in most media.
Surprisingly, Farr’s Law became verboten. Farr’s Law does wonders in focusing on a statistic everyone can agree on – the death rate. Everything else can be argued about. ‘Cases’ can be gamed, depending on how cases are measured – and as indicated below the PCR tests and cycle counts became something that the public is not allowed to discuss.
Focusing on Farr’s Law leads to an evaluation of excess deaths. Mortality in the US, and most everywhere around the world appears as if it will be on par with past years. It does not appear that age-adjusted excess deaths have increased from 2019 for the 2020 year of the pandemic.
A great deal of the debate around Covid stems from the lack of agreement on how it should be measured. (Link to article talking about Theory of Constraints and measuring Covid.)
Law of Small Numbers
Daniel Kahneman and Amos Tversky improved our understanding of statistics. In Kahneman’s book, Thinking Fast and Slow (Amazon), he introduces the Law of Small Numbers. The law is a version of statistical apophenia, we look to explain the data with stories. In the book, the story explains how a forced ranking of improvements in education across the 3,143 US counties leads to a desire to explain why one county is higher than another. In a forced ranking – isn’t the forced ranking a driver?
“We are far too willing to reject the belief that much of what we see in life is random.”Daniel Kahneman, Thinking Fast and Slow
This human desire to create a coherent narrative has led us to consistently try to explain why some areas have fewer ‘cases’ than others. Ignoring that cases is not a consistent or well defined term, then there are so many variables about why one area is better than another. How do we compare an island nation in a tropical setting to a US State in a cold climate with some of the busiest airports in the world?
In the current state of confusion there are many attempts to compare one territory or area to others in pursuit of a silver bullet solution:
- “They wore masks better.”
- Region A protected the elderly better.
- Region B wore better masks.
- Region M had a better vaccine strategy. Region L had a better vaccine.
- Region C closed down restaurants, but kept schools open.
- Region Z had a ‘super-spreader’ mass gathering.
As Kahneman’s law states – we don’t know what’s really driving the difference. We don’t even really know if the numbers we’re comparing are truly the same if the value is cases. We don’t know where we are in the race – many comparisons between Southern and Northern hemispheres were made. Praise was placed on island nations for pursuing ‘zero covid’ when it isn’t clear at all if that is a valid long term strategy.
Humanity is telling itself stories about why this pandemic is happening to us, like a child self-soothing during a thunderstorm. The child didn’t cause the thunderstorm; and the odds are low that any interpretation of ‘the data’ will somehow bring the pandemic to a close.
This Greenbook Article on the Law of Small Numbers highlights several issues that are similar to what has been encountered with Covid-19;
- Sample sizes are initially small
- Comparisons are difficult to make
- People are pressure to ‘interpret’ and ‘explain’ the data – when there is nothing to explain
- Correlation is not causation
- Scraps of data lead to interesting stories
- Analysis that leads to forced ranks is very prone to ‘story-telling’ to explain why one group is higher than another