Does the death penalty save lives? (part 2)

To continue the discussion of whether the death penalty is really a deterrent (as some new research has suggested), but to quote someone who probably wasn't thinking about the death penalty at the time:

Researchers have discovered that when people find it easy to imagine an event, they overestimate the likelihood that it will actually occur.

That's from Daniel Gilbert's excellent book Stumbling on Happiness -- another one of those books that I "set aside when I was midway through, but plan to finish eventually."

We're more inclined to do this -- to overestimate the likelihood of easily imaginable events -- with good things than with bad things. A study of college students showed that their beliefs about how likely it was for certain things to happen to them showed that their beliefs were skewed in the direction of the positive. So, for instance, they tended to think it was likely that they'd have a gifted child, but not so likely that they'd have a heart attack.*

But just because we're more inclined to do this with the good doesn't mean we don't also do it with the bad. And indeed, we do.

You're going to a party, and you imagine you'll make some spectacularly buffoonish faux pas when you get there. To use Gilbert's example, everyone else will show up with a nice gift for the host, but you didn't realize you were supposed to bring anything. You'll knock over a whole table of elaborately arranged food and drinks. You'll put your foot in your mouth when you're just trying to make small talk. It's not that you think any of these things are likely to happen. But you're still thinking about it, because you can't get these vivid images out of your head.

To use another of Gilbert's examples, when you go to the doctor, even if you rationally believe you're in good health and don't have any troubling symptoms of anything, you're not just thinking: "Well, it's highly probable that things are fine, so I have nothing to worry about." No, you can't stop thinking about the tiny chance that the doctor's going to tell you you have cancer, or you have six months to live, or you have some rare disease that they've never heard of before. Again, this doesn't make you dumb or irrational. You might know it's not very likely. But you're still going to think about it, because the scenarios are readily available to your brain.

"These dire images make us feel dreadful -- quite literally." But, as Gilbert notes, there are reasons why our minds do this. You might be softening the blow in case the bad thing ends up happening. Then at least it won't take you by surprise.

More relevant to the death-penalty issue, though, is that "forecasts are 'fearcasts.'" If you make yourself aware of the worst possible outcome in some situation, this can be a way of motivating yourself to try to avoid that outcome.

You're worrying about the chance of knocking something over at the party, even way out of proportion with the chances that that will happen ... so you make sure to look where you're going. You're worrying about the chances your doctor will tell you you have some terrible disease at your next appointment, even way out of proportion with the chances that will happen ... so you change little things about your daily routine like exercising more and eating healthier foods.

In short, overestimating how likely it is for something terrible to happen to you isn't just "irrational" -- it could be a good (albeit unintended) strategy to make yourself be more careful by visualizing the worst possible consequences of bad choices.

And that seems to be exactly what's going on with the person who decides not to kill because they can't stop thinking about the vivid image of getting a lethal injection, or just sitting in court and hearing the judge pronounce a death sentence. They might be overinflating the probability that this will occur. But if it stops them from killing someone, then who cares that they got the probability wrong?

Now, do Donohue & Wolfers, or Steven Levitt, cite any of the psychological research backing up these human tendencies? Do they even acknowledge the possibility that our minds might work this way? No. But as I blogged the other day, that doesn't stop them from claiming, in academic articles surrounded by impressive-looking data, that the death penalty is not a deterrent because the death penalty is so rarely applied that it can't rationally deter people. (Wolfers was also quoted making essentially the same point in the New York Times.)

Levitt in particular is usually willing to explore the possibility that people's minds systematically work in irrational or unrealistic ways. I already pointed out that he's willing to say that people overestimate small probabilities like terrorist attacks -- but he assumes this does not happen with people contemplating the probability that they'd receive a death sentence for killing someone. Levitt was also perfectly happy to describe deeply irrational behavior in his book Freakonomics: parents who know they'll be fined for showing up late to pick their kids up from day care are more likely to show up late than those who don't get penalized -- utterly backwards from what you'd expect if people always just neatly calculated the likely costs and benefits of each possible course of action and then did whatever would give them the best outcome.

Yet when it comes to the death penalty, Levitt assumes that people can't be motivated by the fear of death (of all things to dismiss the potential disincentive value of!), as long as a careful reading of the statistics shows that death sentences are rare among all murder cases.

Does he think murderers are more rational than most people? I doubt it. More likely, he has different standards of intellectual rigor for an academic study than an international phenomenon and New York Times best-seller. It's just odd that the intellectual standards are higher for the smash hit than for the academic study.

I still haven't delivered on my promise to explain my other problem with the Donohue & Wolfers study. I'll get to that soon, so stay tuned... [UPDATE: Here it is.]

* Actually, the way Gilbert summarizes this study leaves me wondering whether they really were inaccurate in estimating the likelihood of these things. He says people thought it was more likely for the good things to happen than the bad things, but that doesn't tell you how accurate those estimates are. Also, there's the perennial problem of announcing a sweeping conclusion about human nature based on studying college students. Aren't college students actually better off than average (most people don't have college degrees), so might it not be reasonable for them to think they'll do better than average in all sorts of areas of life? But these quibbles aside, the basic point seems pretty convincing. [back]