Lies, Damned Lies and Statistics
Worried about your
investments? Concerned that your IRA and
401-K plans have tanked, leaving you with the prospect of taking in boarders to
fund your retirement? Well, good news! I am going to share with you an absolutely
100% guaranteed plan to make you 100 times more likely to win the Lottery. That’s a 10,000% increase in your
chances! But first, a few words on the
exciting world of statistics.
Mention
“statistics” and most people immediately associate it with such pleasant
memories as the crusty old professor who waxed eloquent about the intrinsic
beauty of correlation coefficients and multivariate equations, or perhaps
thoughts of their last root canal. But
in truth, statistical analysis is part of our daily life. We just don’t call it that.
When we say
that something is “likely” to happen, we usually mean that the chances are
greater than 50% that it will. Or if we
say it is “probably” going to happen, we mean 7 chances out of 8, or perhaps 9
out of 10. You get the idea.
But it’s the
misuse of statistics that makes life fun.
We see it every day and rarely recognize it. One wag pronounced statistics as the science
of producing unreliable facts from reliable figures. Politicians love it. Say that you ask a random group of voters to
rate Mr. X, a rather bland, middle-of-the-road elected official. You give them 3 choices: poor, adequate, and good. Assuming that he hadn’t offended too many
people, you might find that 15% consider his record “poor,” the same percentage
“good,” and the vast majority, 70%, as “adequate,” whatever that means. When election time rolls around you’ll have
his opponent screaming that only one out of seven people (or about 15%) think X
is doing a good job. X’s handlers, on
the other hand, will say that 85% of his constituents endorse his record. Same numbers, different spin.
With a bit of
ingenious thinking, one can distort all sorts of numbers. Take averages for example. Ever hear the story about the man who drowned
crossing a river that averaged only one foot in depth? Perhaps you’d like to do some good by
raising the average income of a poor African country? All you’ve got to do is convince Bill Gates
and Warren Buffet to move there. Even in
the good old USA, as many as half of all workers earn a below-average
income. Averages ignore extremes, and
it’s the extremes of life that give us our most intense pleasure and pain.
One joyful
source of statistical misinformation is the concept of relative versus absolute
numbers. I saw an ad on television the
other day for one of the “buttery spreads” (read vegetable oil) that advertises
“Zero Trans-Fats.” Now remember, the
average American consumer has no idea what a trans-fat is, much less how much
is bad for you. Add this on to the fact
that we think in pounds and ounces, not grams.
The “buttery spread” warned us that under government rules, products
with “zero” trans-fats can actually have up to 0.49 grams per serving. Do you have any idea how much that is? There are 454 grams in a pound, so 0.49 grams
less than 1/900 of a pound. The proud
product had an even smaller fraction.
The advertiser was no doubt banking on consumer ignorance, hoping that
by tossing around some vague numbers, health conscious housewives would be
intimated into buying his product. As my
grandmother would say, “Weren’t none there in the first place.”
And there’s the
association-causation conundrum. Just
because things seem to be linked doesn’t necessarily mean that one causes the
other. Or does it? It won’t surprise you to know June is the
most popular month for weddings. But
it’s also the most popular month for suicides.
And when you think of where to get married, there’s Las Vegas with its
dozens of wedding chapels. The city with
far and away the highest suicide rate in America? Ditto.
That’s how rumors get started.
Folks with
products to promote love percentages, most of which sound great but have no
meaning. Remember Ivory Soap’s slogan,
“99 and 44 one-hundredths percent pure”?
Pure what? Ever see a bottle of
this or that cleaning solution that “contains 25% more”? As compared to what, a smaller size at less
cost?
The percentages
game even laps over into medical nostrums.
Have you ever heard of the “placebo effect”? Numerous studies have shown that pills
containing no active medicine will give partial or complete relief half to
two-thirds of the time for conditions ranging from hypertension to hysteria. Knowing this, the purveyors of vitamins and
magical “tropical fruit and berry juices” make millions despite the almost
complete lack of objective scientific evidence supporting their claims. Noni juice anyone?
Some scientists
and pollsters are tripped up by their own incompetence. A classic example is the Literary Digest poll done for the 1936 presidential election
contest between Landon and Roosevelt.
The Digest had picked the
election winner before, and 1936 was to be no exception. They sent out some 10 million ballots to
voters, most of whose addresses were chosen from telephone and automobile
registration records. Landon, the
Republican, was predicted to be the winner by a landslide. The final results, however, only gave him 37%
of the vote. What happened? In the post-mortems that followed, the
pollsters realized that those who could afford a phone or a car in
Depression-era America were likely to be Republicans anyway, effectively ignoring
the huge number of voters who put Roosevelt back in office. Lessons were learned, but polling is still a
very inexact science and one that is easily manipulated to yield the desired
result.
Sometimes the floods
of spurious number-backed claims are almost enough to drive you mad. Which brings to mind a supposedly reliable
statistic from the National Institute of Mental Health. They state, unequivocally, that more than one
in four American adults “suffer from a diagnosable mental disorder in a given
year.” Think of your three best
friends. Any of them a bit crazy? If not, have a look in the mirror—you could
well be that one in four.
It gets
worse. If you live in south Georgia, you
have about a one in three chance of being obese. By the time you’re age 50, there’s a one in
two chance that you’ve got hemorrhoids, which are the same odds as your having
a doctor-given diagnosis of arthritis by age 65. Given these numbers, you have a pretty good
chance of running into a fat, mentally ill senior citizen with arthritis and
hemorrhoids on your next trip to the mall.
Better hope they’re in a good mood not carrying a gun. (By the way, 4
out of every 10 Georgians admit to being gun owners….)
The truth is
there’s no truth in statistics. They’re
often used like a drunk uses a lamp post, more for support than
illumination. Mark Twain said it best
when he observed that, “Facts are stubborn.
Statistics are more pliable.”
So, back to the
lottery. Want to win the MEGA Millions
game currently being offered by the Georgia Lottery? You can increase your chance of winning by
10,000% simply by buying a hundred tickets instead of one. Your chances go from one in 175,711,536 to
one in 1,757,115—more or less. Before
you get your hopes up though, even with those odds you’re about four and a half
times more likely to be struck by lightning during any given year. Don’t give up quite yet. There must be a way to spin that; with the
creative use of statistics we can all be winners.
About the author:
William Rawlings is a physician and author from Sandersville,
Georgia. During a moment of temporary
insanity when he was young and foolish, he earned a Masters Degree in
Epidemiology and Biostatistics, starting him on a cynical course in life that
persists to this very day.