Review- Naked Statistics, Charles Wheelan
Thought-Provoking Things 2021
Looking back at the books, podcasts, and media I encountered this year that I’ve caught myself thinking about again and again.
If you were tasked to create an updated ten commandments for making sense of life in 2022, there is a good chance that the first commandment should be “Correlation does not equal causation.” Large minarets should be constructed in every city and economists and statisticians should shout that phrase five times a day to the passersby below.
Every day we are bombarded with statistics and fuzzy causal claims and, if we aren’t careful, can be taken for a ride. The fact that basic statistics do not make up a central part of primary school curriculum is a problem, and if like me, you somehow emerged from post-secondary education without taking a statistics class, then I recommend the primer Naked Statistics, by Charles Wheelan. This book is great as a first dose or a booster to inoculate you against spurious theories and internet hoaxes.
Wheelan does a great job making the subject approachable, stripping away the arcane and technical, instead focusing on what data is telling you and what it isn’t. He breaks down the basic types of statistical analysis, how to understand what results are significant, and how confident you should be in the relationships among your data. After reading it, you should be much better equipped to spot poor assumptions and assess whether a statistic passes the smell test.
There are several reasons why we as humans are pretty bad at statistical analysis. The first is that we are biologically wired to make hasty approximations for rapid decision making. “Good enough, now” was usually much better from an evolutionary perspective than “precise but too late.” The second reason is that as a society we are in our adolescence when it comes to data-driven decision-making.
One of my favorite parts of the book is Wheelan’s showing how until very recently, standardized data and processing capacity were so sparse that rigorous analysis wasn’t feasible. For example, in June of 1930, Herbert Hoover boldly declared “the Depression is over” based on some anecdotal evidence of an upturn. In his State of the Union he said that two and a half million Americans were unemployed, but that number was eight months old. In reality, at the time of his speech five million were out of work and hundreds of thousands were losing their jobs by the week. “Washington was making policy in the dark.” Hoover’s mismanagement would cost him the presidency. He’s not the most sympathetic figure, but at least in this instance it is hard not to sympathize with how sparse the toolkit was to navigate a complicated situation.
It wasn’t until 1934 that economists even devised a standardized measurement of national income, what we now know as GDP. We have less than a century of practice doing this stuff, let alone figuring out how to disseminate learnings to the public. This pandemic has showed the perils of having both too much and too little data, and how tricky it is to effectively communicate that to policymakers and the public.
We now live in a world awash in data, and we should all be a little bit more comfortable manipulating it. This book was like a light shining into the dark.
