I just finished reading The Numbers Game: The Commonsense Guide to Understanding Numbers in the News, in Politics, and in Life by Michael Blastland. I really enjoyed reading this book.
This book helps ground the reader in sound thinking after what we read and see in the news. As I read this book, I thought about news coverage of the COVID-19 pandemic and the coverage of the 2020 election/polling. I hope to be able to employ the ideas from this book and techniques for number analysis to better understand current and historical events.
The book reminds us that:
"Numbers, pure and precise in abstract, lose precision in the real world."
Numbers appear to be quite concrete but in reality are not. How squishy is the number defined?
If x, y, z are counted, how sure are we that a, b, c are or are not (or should, or should not) be counted?
If it has been counted, it has been defined, and that will often have meant using force to squeeze reality into boxes that don't fit.
Always think of the limitations.
Change big numbers into human relatable numbers. Here are a couple tools to keep in mind to put numbers into perspective:
"We need to think, just a little, and to make sure the number has been properly converted to a human scale that recognizes human experience.
Humdrum Principle: Harm depends on how much of a given danger people are exposed to, and low levels [implies] no harm at all
We all think of this multiple times a day - consider head injuries or toxins in foods, etc.
When we read headlines about toxicity, think, "In what proportions?" The dose is critical! [Ask how the risk is shared out among the affected population?]
Think of a man walking a dog at night on a long leash and fluorescent collar - how do you know how fast the man is walking and when he changes directions?
Numbers go up and down all the time.
Remember regression to the mean: When the dog reaches the end of his leash, he runs back. Things tend to move back to the average.
When asking why something happened, chance is the potential explanation [most often ignored].
Be on guard for chance - be aware of what chance is capable of. Randomness exists everywhere.
Remember that averages play tricks:
"Whenever you see an average, think: "white rainbow," and imagine the vibrancy it conceals."
Types of Averages:
Recognize that [Averages] are an Abstraction
We will be misled if we look at averages without knowing what they are abstracted from.
Measurement itself is not a passive exercise
"...it [the act of measuring] often changes the very thing we are measuring"'
"Numbers are pure and true; counting never is."
Uncertainty is a fact of life.
Size matters to risk: Usually bigger is worse risk than smaller
[Percentage changes] depend entirely on where you start: Look for the number at the beginning or at the end, not just the difference
Try to rephrase things to a natural frequency to help remove some abstraction
Natural frequency = Number of people affected in every 100
Sampling - imperfect but the best we can do
All kinds of characteristics can, and do, cause bias
"Bias is a risk, not a certainty. Being wise to that risk is not cynicism."
"As long as we know something that is relevant to the question, we should be able to have a stab at an answer." - Like how many gas stations are in Florida?
"When a number appears that's out of line with others, it tells us one of three things:
Apply these to any shocking news item - What is most likely?
Watch for comparisons that [look similar] but [are not] actually true.
Correlation does not equal causation.
Plausibility does not mean truth.