Numbers make people stupid

As soon as you introduce numerical data into an informational vector, you are going to confound people and turn them from rational into irrational.

Some people get numbers. Some people understand what math means in a practical sense. For others, statistics turn them into idiots.

Ex. 1 – IMPACT

Coronavirus is 98% survivable (ignoring distribution). No big deal, right? Why is everyone freaking out?

Would you play Russian roulette? The odds of winning are pretty good – 80%!

Most people would say no, because they aren’t idiots. How many cylinders would it take for you to gamble? 20? 50?

98% survival is the same as playing Russian roulette with a 50 cylinder revolver. It’s a stupid risk.

Why are people making this error? 98% coverts into “sure thing” in people’s minds due to the stupefying magic of numbers. You’d be an idiot to not plop down money on a 98% bet if all you would lose is your money. If you lost your life instead? It’s gonna have to be a BIG payoff.

Ex 2 – The Magic of Big Numbers

Coronavirus again – 98% survival means 1/50 people who gets it will DIE. If as many people got COVID19 as got the flue this year, you are looking at 500,000 – 600,000 deaths (assuming equal distribution among groups).

That’s like having the city of Fresno, CA wiped off the map.

Pretty significant. That’s 100 9/11 attacks.

The error in thinking occurs because we aren’t wired to think of tribes of millions, but of dozens. Combine that with “98% survival” meaning a sure thing and you reckon that means nobody will die.

Ex 3 – Ignoring or misunderstanding distribution

People get used to statistics being “normalized” – basically being reduced to a percentage with no other groupings showing how data is actually distributed among categories or areas.

This can take two forms – a percentage or a normalization of some other factor, such as time (example: “Every 15 minutes somebody dies in a drunk-driving collision). People end up believing that something is random or regular, when it fact it is concentrated or at least somewhat predictable. Drunk driving isn’t a regular occurrence, but one where the incidents are concentrated around specific times and places.

Heart attacks are not distributed uniformly, but are concentrated in specific types of people.

98% survival in the Coronavirus isn’t 98%. It’s just about 100% survival for some, and drops down to around 85% for others. This means if you are young it really isn’t very deadly, but much more so for your parents, and considering they could get it from you, the virus is far more dangerous than the “98% sure thing” survival rate would lead you to believe.

Ex 4 – Believing Statistics Have Predictive Value

This is the one that I have the hardest time teaching others about.

Most statistics do not convert into future probability.

They are a collection of past events, and these events will be predictive of the future only if they are regular and subject to some degree of randomness, such as figuring the annual rainfall for a given area.

Things like divorce rates or rates of criminality DO NOT HAVE PREDICTIVE VALUE OVER AN INDIVIDUAL.

The fact that 1/3 of African-American Males end up being incarcerated at some point in their lives does not mean a given African American male has a 1/3 chance of being incarcerated. Why? Because incarceration is not a random event. In other words, it’s subject to the choices and actions of the individual. The idea that such a statistic holds predictive value falls apart if you yourself are black, or can at least imagine being black. You can decide not to go to jail by not engaging in criminal behavior.

This also applies to things like divorce rates. The total number of divorces has no bearing on whether or not you and your spouse get divorced, because divorce is not a random event, whatever you might have heard from bitter men and women online.

Ex. 5 – Assuming binary results.

Returning once again to Coronavirus, there is a built-in assumption created by the “98% survival” stat that there are two outcomes: death or survival.

In reality, there could be another one – long-term health impact of significant severity. It’s too soon to know for sure, but the virus might cause lung damage, which means you don’t want to get it even if you are in one of those low-risk groups when tracking death.

People got this instinctively with things like Polio – it wasn’t a disease that necessarily killed you, but it could cripple you to the point where you would need to be in an iron lung to breathe.

Getting in a car accident doesn’t necessarily mean death either – but it could mean some significant pain.

When rationality returns

People often return to rationality in the absence of data. Common sense is quite underrated.

You can avoid getting into a collision with a drunk driver by staying off the roads late at night on weekends. You can avoid going to jail by obeying the law. You can avoid getting coronavirus by avoiding crowds and washing your hands, just like the flu, and you ought to, because it can be dangerous to have for yourself and deadly for your parents.

Paranoia is a survival trait. Avoiding risks with high impact is natural and appropriate.

Buy my newest books. BUY THEM, I SAY!

Audiobook coming soon:

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.