Author: Darrell Huff
This was a short read. It was a little more simplistic than I expected. It didn’t require any prior statistics knowledge going in. But a good reminder of all the ways stats can be manipulated and how to question and look out for those ways.
First issue he brought up were sample sizes and looking out for bias in those presenting statistics.
He then talked about averages and how stats can be called an average but actually use the median, mode or mean, whichever one best fits the narrative.
He talked a bit about small differences when looking at stats. His point was that in many cases, we should express things within a range as normal, instead of with a specific number. He used an example of IQ’s, which at the time, 100 was said to be average. Two kids were ranked 98 and 104, so the mother obviously assumed the 104 was smarter. But when you dive deeper into the +/-, the gap narrows and even pointed to the 98 child being more likely to be more intelligent because their +/- was more favourable.
Another example he brought up that I think happens often is what he calls semi-attached figures. He says: “if you can’t prove what you want to prove, prove something else and pretend they are the same thing.” He brought up the example of cold medicine. They couldn’t get it to prove that it reduced colds in adults. So instead, they ran a test by putting it in a test tube with “germs” and it killed 80+ percent of all germs. Well that sounds nice, but it doesn’t prove that the medicine helps colds, just that it kills some random germs that are likely not even related.
This led into the classic reminder that correlation does not equal causation.
He had some other common ways stats can be manipulated as well, with examples.
He closed the book with some questions to ask when faced with statistics:
A lot of this book was stuff I had been exposed to in statistics courses I had taken in the past. But like I said, this was a short read and a nice reminder. It was also a bit humorous which was nice. Decent little book.