Jeremy Hobson: It's Freakonomics time. Every two weeks we explore the hidden side of everything. Today, why the first decision you make in 2012 is riskier than you think. Here's Stephen Dubner.

Stephen Dubner: Happy New Year, everybody! Now, how are you getting home from that party? If you're in New York City, where I live, good luck getting a taxi. And if you've had some champagne and you're even thinking about driving home... well, don't.

Public service announcement: Drinking and driving is not only against the law, but it can be deadly.

Public service announcement: Over the limit, under arrest.

Public service announcement: Friends don't let friends drive drunk.

All right, so maybe you'll walk home. Smart move, right?

Steven Levitt: That's a terrible idea, walking drunk is one of the most dangerous activities you can engage in.

That's Steve Levitt. He's my Freakonomics friend and co-author. He's also an economist at the University of Chicago.

Levitt: Truly, if you're faced exactly with two choices, walking drunk or driving drunk, you absolutely should drive drunk.

Now wait a minute -- Levitt is not advocating that people drive drunk. We know how incredibly dangerous that is. But what about drunk walking? Is that dangerous? Consider a few numbers. In 2009, the most recent year for which we have data, about 34,000 people died in traffic accidents. Roughly half of them were drivers -- 41 percent of whom were drunk. Now, there were about 4,000 pedestrians killed -- and 35 percent of them were drunk. Here's Levitt again:

Levitt: For every mile walked drunk, turns out to be eight times more dangerous than the mile driven drunk. So just to put it simply, if you need to walk a mile from a party to your home, you're eight times more likely to die doing that than if you jump behind the wheel and drive your car that same mile.

Now there are some caveats here. A calculation like this requires some assumptions, because there's no government database on drunk walking. Also, people drive drunk much farther distances than they'd walk drunk. And most important: a drunk walker can't hurt or kill someone else the way a drunk driver can. That said, the death toll from drunk walking is undeniable.

Thomas Esposito: The danger of impaired walking is not insignificant. And certainly when it comes down to you, it's definitely significant.

Thomas Esposito is a trauma surgeon at Loyola University Health System in the Chicago area. He's used to seeing a New Year's Day spike in pedestrians who've been hit by cars. As a matter of fact, January 1st is the deadliest day of the year for pedestrians -- and 58 percent of the people who died were drunk.

Esposito: I'd rather work New Year's Eve than New Year Day. Because a lot of the time on New Year's Day, that's when people start to realize someone's missing, where are they? And then they find them at the bottom of the stairs or on the side of the road, injured.

Esposito also has personal experience with drunk walking. A few years ago, his cousin was hit by a car and killed while walking home from a New Year's party. He'd been drinking, thought it was better to leave his car, and go home on foot. Esposito believes we've done a pretty good job getting out the "don't drink and drive" message -- but we could a lot better with "don't drink and walk." Here's Steve Levitt again.

Levitt: For 20 years, we've been told you should never, ever drive drunk. We should have been told you should never, ever walk drunk and you should never, ever drive drunk. And because nobody thought about it when we were coming up with what was moral and immoral, somehow now, drunk walking just can't find its way into the immoral box.

So listen, have a great New Year's celebration, but if a friend has been drinking and starts reaching for the car keys -- or decides to set off on foot -- don't let him. Because remember: friends don't let friends walk drunk.

I'm Stephen Dubner for Marketplace.

Hobson: Stephen Dubner, our Freakonomics correspondent. He puts out a podcast, too -- you can get that on iTunes and hear more at Freakonomics.com. He will be back in two weeks.