(Photo: John Foxx)
It’s an embarrassing episode. The opening sentence of James B. Stewart’s Tangled Webs: How False Statements Are Undermining America is:
“We know how many murders are committed each year — 1,318,398 in 2009.”
But this is false. As Jeffrey Rosen notes in a savage New York Times review, there were 15,241 murders in 2009. The cited number isn’t just wrong, it’s wrong by two orders of magnitude. Where did the 1,318,398 come from? It’s the number of violent crimes, which includes robbery, rape and assault. And only a small proportion of all violent crimes — a little more than 1 in 100 — are murders.
And so this provides a useful teaching moment for thinking about numeracy. How can you avoid such errors?
- Don’t use numbers when words will do. The rhetorical point that Stewart was trying to make is simply that there are a lot of murders. Too many. You don’t need numbers to say this.
- Don’t use numbers that are hard to comprehend. We have everyday experience in thinking in dozens. But we’re hopeless when it comes to millions, billions, or at the other extreme, tiny fractions. For instance, no one ever made the mistake of saying there are 12,000 eggs in a typical carton. But plenty of journalists confused the $700 billion TARP bailout, describing it as a $700 million plan.
- Scale matters: Big (and small) numbers only make sense relative to something else. Is 15,241 murders a lot for a country the size of the U.S.? Find a scaling that gives this some meaning (and avoids the artificial precision of 5 significant figures). Perhaps: Last year around 1 in 20,000 people were murdered. But how can you get your reader to picture 20,000 people? Easy, it’s roughly the number of people who attend a typical MLB game.To over-simplify: Look around an average NBA basketball stadium. If the crowd is representative of the streets, someone in this crowd will be killed this year.
- The laugh test: When you really think about your number, does it seem plausible, or is it laughably wrong? My basketball stadium analogy conveys the true extent of the U.S. homicide problem. If Stewart had followed this advice, he would have seen that his wrong number implied that 1 in 230 people is killed each year. Straight away he would realize that he isn’t grieving the murder of one of his Facebook friends every year or two.
and 
. You can also watch it on 
I support an act of congress to make the millions, billions, trillions, NOT rhyme.
The only way of understanding a really big pile of dollars is “Opportunity Cost”; that is, what you could buy with the money.
What could one do with a billion dollars? A billion dollars is more than all the Alaskan gold mined in 2010— a cubic METER of gold.
Just five percent of that billion dollars ($50 million) would get you: A lifetime lease on a Gulfstream G450 at your beck and call. A home, condo or apartment in ten major exotic destinations, with a (leased) luxury fleet of cars at each, a coterie of servants to follow you around, memberships in every exclusive club, all you could possibly eat, wear, play with and see for a lifetime. And Hell, throw in a yacht. Then you could sail around sipping expensive wines imagining ways to spend the other $950-million dollars.
Well-loved. Like or Dislike:
6 
1
Good topic, but bad example. “We know how many murders are committed each year — 1,318,398 in 2009.” Who would really believe such an absurd statement. My original thought was this was a number for the entire world. The point got lost. I would also have to dissagree with point one, there is nothing wrong with numbers. Just provide the numbers in an intuative way. Numbers are provide a higher level of proof, compared to words.
Really? How to avoid being wrong when using numbers… don’t use numbers. This hardly seems like the kind of advice I expect from the Freakonomics blog. Point 3 and 4 are good advice. Point 1 and 2 just seem like they are saying, readers (and writers) don’t understand any math (or really how to count). So, I guess that settles it, people aren’t good at math so we shouldn’t report numbers, just say “a lot” or “not many” and that way we’ll never have to actually look-up or confirm anything.
Hot debate. What do you think?
9 
11
I use those same concepts in my 8th grade classroom all the time! Glad I’m ahead of the curve on something.
“…1 in 230 people is killed each year.”
Actually, he implied that this many people are murdered, not just killed. I think that’s an important distinction. Thanks for this article — very important things to remember.
Well-loved. Like or Dislike:
9 
1
Given the title of his book, do you think that maybe, just possibly, Mr. Stewart was making an ironical point?
Just wondering.
Hot debate. What do you think?
7 
3
I thought that at first, but if you check out the opening page on Google Books, it seems less likely. If he was going for irony, he was being very subtle.
“Statistics are like ladies of the night. Once laid out…you can do anything with them.” [Mark Twain, i believe]
It all comes down to the ideology a person subscribes to because there are plenty of numerical and semantical arguments on every side. of course, ideology is still subject to logic and practicality.
This blog for example produces numeorus (so called solid)numbers and arguments every week and yet every reader leaaves with his/her own conclusions (from keynesians to austrians).
Raw data is incomprehensive; therefore, we always need someone (who is always subject to a bias) to polish them.
Peace out, yo!
I’ll disagree with point 1, for argument’s sake.
15,241 murders were committed last year – that’s a value-neutral statement.
But if I say “a lot of murders were committed last year”, I’ve introduced my values. I’m saying that I think that is a lot of people. But what if it was the lowest number in 30 years, or a decline of 50% from the previous year? It would be easy to criticize me for trying to hide a decline in murders by talking words instead of numbers.
How you choose the words is important. If murders are up 3% over last year, do you say they “rose slightly” or “skyrocketed”?
My point is that if you’re going to use words instead of numbers, it gets much easier to introduce your values into the discussion, and for your core point (whatever that it) to get lost.
Well-loved. Like or Dislike:
34 
0
I find it a little funny that in pointing out how not to make a mistake it seem that link mistakenly links to the MLB attendance not the NBA. I know a little blog link mistake is nothing like a published book where you drop wrong numbers. Everything you said sounded like great advise I just found it funny is all.