Tuesday, September 11, 2012

The Chances Are....

Warning: Math ahead. Not a lot, but what there is, matters. I'm hoping to change the way you think about some important things, and maybe help you save your butt in the process.

Most people don't understand probability. That's especially true when it comes to extremely likely or unlikely occurrences, or extended sequences of events. If they did understand, Las Vegas would be a dusty gas station at a desert crossroads. So let's talk about small chances.

What's the chance of your being robbed, assaulted, burglarized, raped? Of being at a convenience store, druggist, bank, supermarket when some disadvantaged urban youths decide to hit it? Of being in a theater, classroom, office, restaurant, mall, church or in your car when some crazoid/terrorist/disgruntled employee/dissatisfied customer/road raging commuter decides to achieve fame/martyrdom/revenge by killing a bunch of strangers?
Small, right? In fact, vanishingly small on any given day or any given visit to any of the places above. So pick an arbitrary number to represent the tiny chance that any of the above will happen to you. How about one in a thousand--- .001? One life-threatening hostile event in 1000 days, commutes, gas fill-ups or what have you. 

This is important: each event, each day, is independent. What happens on one day has no effect on, no implication for, any other. * But the days are repeated. Each year has 365 of them. On each day the chance that nothing bad will happen is .999, 1 minus the .001 chance of a real threat. But...the probability that nothing bad will happen on any of those 365 days is the product of those independent probabilities, or .999 to the 365th power. That's gotta be small, right?  

In fact, it's .6907. Call it 70%. Not terrible. But after two years it's .4817; after 3, .3344; after 4, .2311; after 5, .1611; and after 10 years it's .0259. Those are the chances of no life-threatening event occurring. To find the chances of at least one such event within a given time, subtract each number from one. Looks different, doesn't it? What the numbers say is that within one year 3 of every 10 people will have experienced some sort of crime. In a ten-year period 98 out of 100 will, at least once.

These are just numbers, and you might argue that they don't reflect reality. Fine, look up the statistics and pick your own. My friend Tom Givens, of Rangemaster in Memphis 
(http://www.rangemaster.com/) estimates the yearly odds there as 1 in 50. In Chicago or DC, who knows? Maybe worse.  If you live in Mayberry...Oh, wait. Mayberry's imaginary. 

The principle goes beyond crime, of course. Do you talk on your cell or text while driving? Like to have a drink or two while you're out? What are the odds that something bad, maybe very bad, will happen over some reasonable time? The message here is that they're higher, way higher, than you think.**

It'll be easy for some people to ignore this post. As Barbie famously said, math is hard, and worse, it's boring. Besides, they won't want to believe it, and won't even go to the small trouble of checking my figures or computing their own. When the world happens to them they'll whine about fairness and demand that the government give them some of my property, or take away some of my rights, to make them feel all better. 

Sorry. It may be heartless and insensitive, but for those people I have this message:
You've been told once, and once is all you get. Act accordingly, or don't. The universe goes on and the numbers roll over, uncaring.

Place your bets.

* That may be debatable, but assuming otherwise makes the argument to follow stronger.
** Sticklers will know that the calculations in these cases are different, because we're talking about differences in odds. The principle still applies.

1 comment:

  1. Welcome to blogging, provocateur! Nice article! More about prob & stat, please. -- Carolyn of CO (you know me!)


I welcome your comments.