One of my favorite jokes involves an election in a mythical town in which there are 1,000 Christian voters and 250 Jewish voters. There is an election for mayor featuring a Christian candidate and a Jewish candidate, who get, respectively, 1,000 votes and 250 votes.
“Boy,” says one Christian voter to another after the results are announced, “those Jews sure stick together, don’t they?”
That’s just a silly, throwaway gag, but there is a real-life example, cited by John Allen Paulos in his invaluable book, “A Mathematician Reads the Newspaper.”
In the New York City mayoral race between Rudolph Giuliani and David Dinkins, claims were made that Blacks voted along racial lines more than whites did. The evidence cited was that 95% of Blacks voted for Dinkins, the Black candidate, while only 75% of whites voted for Giuliani, the white candidate.
“This failed to take into account, however, the preference of most black voters for any Democratic candidate,” Paulos wrote. “Assuming that 80 percent of blacks usually vote for Democrats and only 50 percent of whites usually vote for Republicans, one can argue that only 15 percent of blacks voted for Democrat Dinkins based on race, but 25 percent of the whites voted for Republican Giuliani based on race.”
The point of both the fictional story and the real one is that we interpret numbers the way we do everything else, through the prism of our own experiences and preconceptions. And because innumerate journalists pass along without scrutiny numbers that advocates use to deceive as much as illuminate, we should be especially wary of numbers in the news.
I advise this as an addendum to last week’s column, in which I, one of those innumerate journalists, threw out numbers willy-nilly. To show that politicians and the media have tried to scare us witless over COVID, I listed a number of other ways to die with worse odds than dying of the virus.
And was quickly taken to task by an Indiana University emeritus professor of biology, who urged me to recheck all my alarming statistics, specifically calling attention to my numbers on the flu (1 in 63 chance of dying from it) and plane crashes (1 in 205,552).
“According to the CDC,” she wrote, “the death rate for the flu in the U.S. is 1 in 6,579, not one in 65.” And, “Only about 400 Americans die in plane crashes per year, and in all but a very few years, they are all in private planes, not commercial.”
The air crash odds I must confess to. I found the 1-in-205,552 number in more than one place, but no citations for the source. Other, more realistic odds I found ranged from 1 in 5 million to 1 in 11 million to “too few deaths in 2019 to calculate the odds.”
But the other numbers I got from the National Safety Council, where it was noted that these are “lifetime risks” calculated by dividing the “2003 population by the number of deaths, divided by 77.6, the life expectancy of a person born in 2003.”
So I will leave those numbers, noting only that they served my purpose. I wanted to make a point about misperceiving risk, so I deliberately chose some of the highest risks I could find.
I had an agenda.
So do all those in the news stories we read who use numbers for anything but relaying the specifics of a particular incident. They are trying to tell a bigger story, which means they are trying to sell us something — a position, a philosophy, a world view.
So we should be mindful of what they’re not saying, search for the hidden context, the missing perspective. If City A has more crime than City B, is that in raw numbers or a per capita percentage? If we’re supposed to be alarmed about a new “surge” in COVID, what numbers are being surged from and to, compared to the initial surge? If a family of four can’t live on the minimum wage, do we know how many families of four actually have to try to do that?
If there is a 1 in 2,535 chance of choking on food, are those the lifetime odds or is it the chance we take with every bite?
No, don’t discount numbers. They can be interesting, entertaining and even informative when individual ones are used to show us a potential pattern. But view them with great skepticism when that supposed pattern is used in an effort to prescribe or proscribe our activities.