Drunk Walking, Halloween Edition

Photo: Tabbyn0ds

Freakonomics has reported at length on the human tendency to worry about rare problems. And perhaps because of this, we’ve  been in the vanguard of the campaign against drunk walking, which on a per-mile basis is, believe it or not, 8 times more likely to result in a death than drunk driving. (Please note: this is not an excuse to drive drunk; rather, it is a caution against thinking that walking drunk is safe.)

And so a recent article from Christopher Shea at the WSJ Ideas Blog caught our eye, because it so perfectly combines these two obsessions. Shea writes that a razor blade in an apple on Halloween is a myth, and has probably never happened. Pedestrian deaths, however, are four times higher on October 31 than an average day, because so many more people are wandering around outside. We at Freakonomics would like to add that some of those Halloween revelers (the adults at least) are also more likely to be inebriated, which no doubt explains some of the accidents. 

Happy November.

Ken Arromdee

Drunk walking may be more likely to result in your death than drunk driving, but the big concern with drunk driving is the deaths of third parties. Drunk walking doesn't lead to many deaths of third parties.


Despite meeting strangers wearing masks, Halloween is also the safest day of the year in terms of child molestation.


"Pedestrian deaths, however, are four times higher on October 31st than an average day"
It would also seem logical that the total number of Pedestrians is way higher than four times on Oct 31. as numerous Car slugs who would not otherwise walk trudge out of their homes for festivities.
When one considers that isn't a four times increased death rate some what low or at the very least expected?


Yes, but a drunk walker is not 8 time more likely to kill someone else.


To the people making the argument: "Yes, but a drunk walker is not 8 time more likely to kill someone else."

I don't think that this observation implies that drunk walking is better than drunk driving. Let's make the large assumption that every drunk driving accident involves the death of one other person (unrealistically high, but we'll assume it anyways). That means that the drunk driving statistic is underreported by half, or effectively double. But doesn't this just mean that drunk walking kills quadruple the number of people rather than eight times? The substance of Dubner's argument still stands.


This blog keeps harping on how drunk walking as if it is a big problem. The reality though is that a person's death caused by his or her own choices is, at the very least, perceived to be less of a problem than if his or her death was caused by someone else. So while drunk walking may be more dangerous (although I doubt this to be true) it shouldn't be stigmatized or punished in the same way drunk driving should.

This blog continues to write continuously about how dangerous drunk walking is by comparing it to drunk driving. From my point of view, this does nothing but de-stigmatize drunk driving. It's annoying and frustrating.


I agree because on halloween people think that they can get drunk at bars and start fighting and god forbid they pull out a gun and start fireing at people and not get in trouble


This comparison of drunk driving and drunk walking fails to control for a wide range of characteristics. If nothing else, people walking home are probably far more drunk than the typical driver who went afoul of the .08 limit.
And if drunk drivers were only a threat to themselves, we wouldn't have such a strong campaign about it.


This gives math a bad name. Please do not drive drunk because of this post!

Here is a partial list of flaws in the argument:

1. Drunk walkers may have different characteristics than non-drunk walkers, e.g. maybe some of them are mentally ill and homeless.
2. Drunk walkers may be more drunk than drunk drivers.
3. The dangers of the walk might not be proportional to the # of miles. Maybe the pedestrians who die aren't walking a mile home alone from the bar (which I have done, and usually I find that the exercise and cold air wakes me up a bit), but are simply standing in a crowd on the sidewalk near the bar. These would enter the numerator, while the denominator is the ridiculously small amount of walking that happens in our car-based society.

More on these lines from

You'll notice that argument was made on blogs and in their book, but not published in a place where other social scientists would have to agree with the reasoning. That's because their point is *interesting*, but far too thinly evidenced to support over-confident claims like "drunk walking is 8 times more dangerous than drunk driving."


Ryan King

I think you need to carefully consider the externalities of your argument, such as devaluing degrees from the University of Chicago with claims ignoring endogeneity.

Ryan King

Sorry, last night was a bit emotional at your post. What I should have said
1) You are aware that the assumptions going into the relative risk calculation are silly, why are you presenting it as fact?
2) Alcohol related car accidents are not a terribly rare event; looking only at deaths ignores the large number of non-fatal accidents and the age-specific rates. Minimizing them in this way is offensive to the many people who have been impacted by drunk driving.
3) Asserting that the larger number of pedestrian fatalities is due to their drunkenness is also silly to the point of offensive. That article points out that 4x is young pedestrians (children) who are presumably not drunk, and the major cause of mortality of kids on Halloween is being struck by a car; the higher intoxication of drivers could be rather important. Anyone who's worked in an ER on Halloween / new years / 4th of July would tell you this.

I know that you don't mean to come off saying that drunk driving is awesome, but you should consider how people will interpret your posts. Given that you are famous, a bit of social responsibility is a must.



This assertion about drunk walking appears at the start of superfreakonomics at the same time as asserting the point of the book is that you should be careful of assertions made by people with a lot of influence but not much expertise or data passing off anecdotes as facts.

I thought the dangers of drunk walking was the authors being ironic to make that very point. However the joke is a little tired now.

The assumption here is that drunken injuries are a function of distance. Are they not a function of time? You spend longer walking, you are more at risk, and drunken people walk slower, more time at risk. If the mortalities are caused by road accidents, then distance has nothing to do with it. If you cross the road, you will do that once per journey, whether that's 100 yards or 10 miles. Is that not the greatest risk?