Thread by @absurdistwords

We are really bad at statistics as a species.

This makes it really easy for people to manipulate us to get us to believe things that aren't actually supported by those statistics.

Particularly white supremacists

Here's how to shred their nonsense arguments.

https://twitter.com/absurdistwords/status/895421056315535361

Some takeaways?

If someone tries to prove a group of people is worse than another by comparing the number of arrests between the two... ask a few key questions...

1) What percentage of the population was arrested? This is because there's usually only a teeny tiny sliver who was

2) How many people are represented in those arrests? This is because 10 arrests of the same person shouldn't be presented as 10 different arrested people, inflating numbers.

3) How big are the populations relative to each other? We ask this because we need to understand the scale and proportion of arrested people to each other. Comparing a group of 10 to a group of 10,000 is sketchy

4) Are both groups proportionally policed? We ask this because legal distributions aren't random. It's not pure math. WHO gets policed more is a subjectively influenced variable.

5) How reliable is the underlying data?

Is it gathered evenly?
Is it gathered voluntarily or regulated?
Is it internally reliable data?
Are the sources reputable?
Are there biases at lower levels that get amplifed in aggregate?

None of these questions PRESUMES that the other person is wrong or acting in bad faith.

But they are critical things to know before coming to any sort of sweeping conclusion.

Those trying to inflame and incite won't want to provide these

Because they don't want to give context

Those who are just parroting statistics from a script likely don't know or understand the answers to these questions.

They have just been shown numbers and told to accept that they mean one specific thing.

Either way, asking the questions is key before swallowing the premise.

Context and scope are really important things.

If I ask how tall you are, and you say 18. I don't know what that means unless you give me a unit of measurement. A point of reference.

Without reference points, statistics are meaningless.

If someone tells you a city is dangerous and cites a murder rate.

Ask how much of the city those murders actually took place in.

It's likely a small fraction of the overall city.

"Crime of passion" violence is random.

organized instrumental violence is usually contained

Does it mean that there's not a problem to be addressed? No.

But it does illustrate that the tendency to pretend that an acute set of incidences is characteristic of the whole is problematic.

"But doesn't this apply to bad police stuff too?"

Yep. Murders by police, overall are a tiny fraction of the actions of police as a whole.

The reason this isn't a rebuttal to BLM is because it's not raw numbers that are at issue, but the systemic response to each incident.

When there is an indictment of cops, the problem isn't that MOST cops are murdering people.

It's that the OVERWHELMING MAJORITY of cops and their command structures will cover for the tiny sliver of those who do murder people.

This brings me to false equivalencies.

Absent context, people can present two very similar sets of statistics and argue that each circumstance is the same.

But same in a vacuum isn't same in real life.

Real life is more than the math, it's what underlies it.

Always ask the questions.

The willingness of the person to answer will be a good marker of where they stand with regard to intellectual honesty.