In Gambling as in Government, Anecdotal Evidence is Worthless

Here's how it works. Say the president wants Congress to pass a podiatric care bill. He dredges up a horror story about someone in West Overshoe, Idaho, who couldn't get treated for a serious bunion and mentions it on TV to show how badly reform is needed.

Assume that the anecdote is true. And that the nation's foor care system could bear improvement to broaden coverage while containing costs. But no path of inference leads from the one to the president's plan for the other.

Gambling is also rife with anecdotal evidence. Players, casino operators, and solid citizens who oppose this particular leisure activity all use it often. Unfortunately, it's worthless.

A craps shooter throws, dice fly off the table, seven shows on the next toss. Someone snaps, "After they go off the table, 99 percent of the time the next roll is a seven." Actually, after dice go off the table, 16.67 percent of the time over an extended period the next roll will be a seven. That's the probability of a seven on any roll. Previous events are irrelevant. But those who want to believe they have special insight into predicting results take instances when seven occurs as proof and ignore the other cases.

A blackjack dealer hits 20 or 21 hand after hand. Someone bets an extra spot "to change the flow of the cards" and the dealer busts. The player says, "My extra spot did the trick." Those who presume to control the game this way take the specific incident as evidence; they ignore cases when it didn't work, or when nobody did anything and the dealer broke anyway.

Casino bosses have bizarre beliefs, too. One joint recently started offering "five times odds" at two craps tables. They experienced a few shifts when these games lost money, and one day when a high roller really hammered the house. Quintuple odds became history. The casino bosses cited these instances to prove they were being hurt by the drop in "house advantage" from the old to the new odds (0.375 and 0.251 percent, respectively, on $15 line or come bets). Statistically, neither percentage is high enough to significantly affect table "win" on a single day.

Another casino touted six-deck blackjack at all games. Then a card counting "team" made a big score. Shoes abruptly grew to eight decks at three-fourths of the establishment's tables. The bigwigs took this isolated instance as proof they'd left themselves wide open. Card counters raise their bets when the shoe is rich in 10s. Some win when they do; others lose. Management doesn't notice the losses because the game goes on. But they made a major change based on a few hands that could easily have gone the other way.

Civilians opposed to gambling are liberal when it comes to interpretation of anecdotal evidence. It's true that some players go overboard gambling. Everyone's heard about the Toronto bank clerk who "borrowed" money from the vault on Fridays to finance his weekend casino visits, and never won enough to replace it on Mondays. About a former Philadelphia Eagles football franchise owner, too. And about others.

Sad as these examples are, they can't legitimately be generalized to prove that gambling is dangerously addictive. Yet, some opponents of casino gambling get a lot of mileage using the anecdotal evidence of these singular instances to validate their beliefs.

Uses of anecdotal evidence in gambling and in government differ. In the casinos, the deception is mostly inadvertent. And folks are mostly deceiving themselves. In politics... well, Sumner A Ingmark whose poetic evidence is more apt to be apocryphal than anecdotal put it this way: