Probably to support racism. Like the black people crime statistics.
By using unrelated data to prove a point.
By training an algorithm that will have an impact on said statistics. Not only the algorithm can cheat (see Goodhart’s law), but it can repeat biases that led to these statistics (like those law enforcement algorithms that became racists)
My favorite was the one they were training to detect cancer in imaging scans but they forgot to edit out the info stamp in the corner so it just started flagging all the scans from the cancer center!
correlation and causation. even useless stats comparing apples and oranges, the numbers generated are only as good as the study design and methods.
96.3 % of all statistics are misleading.
Averages. They’re almost always a bullshit flag if it’s tied to anything remotely political. If you’re not going to also give the standard deviation and skew then at least use median.
Blindly. People love to list them as evidence as if the numbers stand on their own. Reality is a person had some hand in assembling the numbers and there is no such thing as a bulletproof statistic. Good statistics ought to be scrutinized.
As a math guy, I hate when people say statistics is math. Like yeah, there are equations, and math plays a role, but the results so often speak more to the selection and interpretation choices made by the statistician than to any kind of mathematical rigor.
Truncated graphs. I hate them but they are so often (ab)used, even in professional situations.
…as a drunken man uses lamp posts — for support rather than illumination.
The question makes me remember Daryl Bem, a celebrated social psychologist. He published a much cited article called “Writing the Empirical Journal Article”. About 15 years ago, he used this advice to prove that humans can see into the future. His advice is probably still used to teach. That’s probably the worst thing you can do.
To cherry pick it and use it to promote fascist views
deleted by creator
Not making sure the result even makes sense. There was a real example, where a ~2010 news article said that the number of crimes in their city has been doubling every year since ~1980.
That is not possible. Assume that there was one crime in 1980. In 2010, there must be at least 2^20 crimes.
When you mix statistics with marketing.
Eugenics
More men are arrested for crime than women, proving that cops are sexist.
