Levitt and Dubner respond

The authors of Freakonomics have responded to  my last post.

www.freakonomics.com/blog/2006/03/09

The gist of their argument is as follows:

One thing to consider, however: the theory put forth in Freakonomics examined why crime had fallen all over the country, not just in New York, and one of the many arguments against “broken windows” as a major cause was the fact that such innovative policing wasn’t being practiced elsewhere—and yet crime was falling in those places as well.

Fair enough. But that still doesn't solve the puzzle that is New York City—which is the problem I set out to try and solve in the Tipping Point.

In the period from 1994 to 2004, violent crime in the United States declined 32.8 percent. In New York City, it declined 54.3 percent. In the same period, property crime in the United States declined 23.2 percent. In New York City, it declined 46.4 percent. All of the same things happened in New York City as happened in the rest of the country to reduce crime—there were more cops on the streets, a better economy, high rates of imprisonment, a declining crack trade and yes, maybe even the beneficial effects of freely available abortions—but over and above those factors something else clearly happened in New York to cause crime to drop much further and much faster than in any other city in the United States. What is that something else? Freakonomics doesn't have an answer. I do: I say it's Broken Windows.

It's important to note, once again, that I'm not disagreeing with the core of the Freakonomics argument. I'm just pointing out that there nothing inconsistent between what that book says and what my book says. They are looking at national crime trends. I was interested in explaining the anomaly that is New York.

Comments

[NIck Davis]

I present a unified theory (or at least the dust jacket):

http://www.nickdavis.com/index.php?q=node/152

[Peter Klaus]

Glad to see you're blogging. Did you enjoy your time at Blackboard's Bb World '06 Conference? As an fyi, they've got a fantastic new blog of their own called Educate Innovate -- www.educateinnovate.com -- that explores all issues related to e-learning...it's authored by four Blackboard employees with varying backgrounds in education and technology. And you, MG, are featured in one of their first posts.

Congrats on the blog...I'll look forward to reading regularly.

[Jay Matthews]

Right on, Malcolm. Now if I can convince to stop paying sidelong respect to the abortion theory. The abortion theory comes from a "root causes" worldview that perceives certain individual preconditions as being the primary determinant of whether a particular person becomes a criminal.

The broken window data show that society's response to crime (even in its tiniest emergences) is the predominant factor in the crime rate.

Criminals are a "lost volume" product; there is an ample supply of potential criminals which will emerge if the community demonstrates it will "tolerate" the behavior.

[thm]

I don't really get why the social sciences--or their popularizations, at least--always seem so obsessed with finding one cause for what are very complex phenomena. In my field, physics, we find multiple causes all the time to what are much simpler phenomena. (Physics might seem more complex than, say, criminology, but it's not--the things that physics studies are so simple that they can be fruitfully understood through very complex mathematics.)

As an example, I've recently been looking at the theory for the thermal conductivity of electrically insulating materials. For a single sample of a single substance, this quantity can vary by more than a factor of 1000 depending on the temperature. But there are a half-dozen or so different phenomena that go into a full description, whose relative contributions vary tremendously with temperature. These all add up to a temperature dependence that is not only non-linear but also non-monotonic. The only reasonable answer to the question "what sets the thermal conductivity of a piece of silicon" is "it depends on the temperature" and even if you pick a temperature, there might not be a single most dominant cause.

So for questions that are much more complex, such as the crime rate--or,say, of why Al Gore lost--and where the data are nowhere near as well characterized as in a physics experiment, why is there the insistence on a single answer?

[Siddiq]

I didnt see this article referenced in either the post or the comments. THought it might be an interesting read: http://www.boston.com/news/globe/ideas/articles/2006/02/19/the_cracks_in_broken_windows/?page=full
The cracks in 'broken windows'
A crime-fighting theory that says stopping major crimes begins with stopping small ones has influenced policing strategies in Boston and elsewhere since the 1980s. But scholars are starting to question whether fixing broken windows really fixes much at all.

Here is also a rebuttal:
http://www.nationalreview.com/comment/bratton_kelling200602281015.asp

Both finds via Arts and Letters Daily at: http://www.aldaily.com/

[Craig]

I find this whole "debate" extremely interesting, and tend to agree with THM and, it seems, Malcolm on the fact that there are a multitude of factors at play when considering social trends. Just out of curiousity...anyone aware if there's any research on juvenile detention rates (rather that juvenile crime rates and overall detention rates, as Levitt examines)versus the violent crime decline?

[Lorena]

No tolerance to crime works. Just look at Singapore where as you arrive in the airport you get a card that says "Death by Hanging to Drug Traffickers" in red letters... Drugs are not a problem in that society. That does not mean that ignoring the underlying issues (poverty, lack of education, hunger, etc.) and practicing brutal repression will result in a peaceful society.

[Aaron]

Mr. Gladwell,

You say that NYC's statistics were markedly better than the national average. But were they better than the other largest cities in the country that didn't (or that did) implement a Broken Windows-type program? A national average could reflect any number of things: e.g., weaker gains in smaller cities.

Thanks,
Aaron

[Jim Jones]

Such debates get muddy fairly quickly when trying to assess issues with such a large number of complex input parameters. Maybe NYC cops just tried harder that year to stop crime? Maybe there was some unmentioned officer that led the way to upping the standard of the department?

Could it be possible that one of these immeasurable factors was large, contributing factor in the decline. Absolutely.
-- Jim http://www.runfatboy.net

[WOTS]

M,

The flaw in your argument is that New York's crime rate has been in steep decline since 1990, a few years before "Broken Windows" and Compstat policing began in New York, and four years before the statistics you cite. You leave readers with the impression that the drop in the crime rate began when Broken Windows policing did. That is not true.

What did happen? 1989 was the tip year of a long peaking trend for crime in NYC -- it was bound to go down. 1990 began a decade long expansion of wealth coupled with the gentrification of New York, particularly Manhattan and some of its worst neighborhoods (Viacom moved into Times Square in 1990), and particularly some of its most dangerous areas. I'd posit that Broken Windows policing rode this wave, rather than caused it.

Best,
Worst of the Slate

[c&d]

A debate on Broken Windons with Bernard E. Harcourt and David E. Thacher:
http://www.legalaffairs.org/webexclusive/debateclub_brokenwindows1005.msp

...I remember the crack trade being very bad in NYC as well, or was that an Ice-T movie I saw?

thm writes,
"why is there the insistence on a single answer?"
thm, I suggest you re-read the relevant posts by Malcom and Levitt. Neither person insists on a single answer, both speak of three or four main causes. It may seem like Levitt is saying abortion was the key, but he is not; he is arguing it is a significant factor.

[Karan]

I'd be curious to see how much of an impact NY's statistics have on the national statistics - if say you took out NY's bigger leaps, or changed it into 0 sum year for NY, would the national statistics have looked as good?

I think NY probably won't account for a large chunk, but then again, result might be surprising.

[Steve No_]

An obvious test (besides targeted gallery showings of broken windows in a series of crime bellweathers) would account for covariance in NYC and say, NJS and other environs in NY and PA, assuming I could ever get the types of crime or Public Servant Stop classifications to line up.
The less obvious but sensible continuation would be to hire a designer to make still less-broken seeming windows or still more contrived barriers between public and private spaces, and see whether crime went down as a result and whether it went down in the city before anyone tried not doing it anywhere else, try to contrive 'interviews' (haha, short grease pens on heavy barb chains or abuse of traffic cameras) etc.
I think the fiberglass homonculus thing has been dangled in that direction, and then of course the blue-sworl-lines-in-the-intersection experiments; so how's your NYC rating for those? Should Zaha Haddadi be doing the interchanges to upstate, or what?

[AAwoken]

This is the beauty of blogging. You are having a conversation that would never occur under normal circumstances. This is the essence of modern communication. The way the world converses has changed forever.

[Mike Moore]

I was reading about Freakonomics before and it seems that there overall idea is that the crime rate has gone down due to the Abortion statistic and how there where less babies born into bad homes and there was less of a chance to grow up with a sub par environment.

[TEB]

Malcolm,

I agree it's "Broken Window," but I've often wondered if the fall of communism helped as well.

After 1990 the immigration patterns changed in this country. There were far more eastern europeans and africans in NYC. Put simply, they are fairly well behaved. I've often wondered if they changed behavior patterns as well.

[Aidan]

"something else clearly happened in New York to cause crime to drop much further and much faster than in any other city in the United States"

What about immigration? I've read a lot of articles that highlight the role of new immigrants in revitalizing formerly troubled neighbourhoods. NYC attracts more than its fair share of immigrants - perhaps this should be factored in.

[Darrel]

"I'm just pointing out that there nothing inconsistent between what that book says and what my book says."

This is America and this is a blog. Two combinations that make explaining to folks that the world doesn't exist as a pure 'this or that' argument is going to be tough. ;o)

Great blog, BTW. I'm hooked and will be searching out your books this weekend.

[Victoria]

Interesting commentary on "broken windows" from the good folks at Civic Strategies:

http://www.civic-strategies.com/resources/news/007.htm

They cite some research that suggests that the drop in crime in NY (& etc) came from changes in the crack market.

And happily, the research comes from one of my graduate advisors!

[TBTAM, MD]

Interesting discussion here. I'd like to weigh in with the point of view of someone who has provided both gynecologic and obstetrical care to the women who, presumably are the mothers, or potential mothers, of the unwanted children that become the criminals of the future.

These women, for many reasons, are not effective or consistent users of the pill. They may come to the clinic for a prescription, but return within months pregnant again. Perhaps it is because they have no money for the pills. Or because their lives are too chaotic for daily pill taking and monthly refills and mandated doctor visits to get precriptions. Some are drug or alcohol addicted, with habits that preclude the stability of schedule and residence needed to practice effective contraception. Many are afraid of the pill. Some come from communities where myths about the pill abound and frighten them. Some have side effects of the pill, and their partners are uusually unwilling to use condoms. Many don't trust the medical establishment. Many are teenagers, in the phase of their lives where contingency planning (which is what contraception really is) is beyond their developmental stage.

In my opinion, the availability of abortion would be more likely to impact birth rates in this group of women than the availability of oral contraceptives.

I side with the authors of Freakenomics on this one.

[TBTAM, MD]

Just to clarify my first comment, I realized it may have come off wrong.

The women of whom I speak are those who live in the poor communiities where crime rates were highest, and who I presume were the ones addressed in Freakonomics. I actually don't know enough about crime to know where the criminals are born, just wanted to address the contraception part of the issue.

I don't mean to say that all criminals come from poor communities - I know of at least one criminal who grew up in my neighborhood (a middle class community) and even have a distant relative who was raised in a very nice upper class family who became drug addicted and actually murdered someone. The common factor in these cases was drugs, not unplanned pregnancy.

I also have taken care of many responsible, hard working women from poor, crime infested communities raising womderful children.

I just wanted to weigh in on the pill vs abortion impact on unplanned pregnancies, and affirm that in poor women, the pill is just not as well used.

[Dave]

What weight do you give to the argument that the rise in popularity of 1st person shooting games like DOOM, Castle Wolfenstein, and Halo is the true reason for the falling crime rate? Male teenagers started spending a significant of their time inside playing games instead of going outside and causing trouble.

[Tracy]

the FBI’s compilation of crime stats shows that last years murder rate hit a 40 year low. All incidents of violent crime including murder, rape and armed robbery are lower now than they were in the 1970’s. The trend is expected to continue as the U.S. population ages.

[Overworm]

I'm very excited to see that you have a blog. Your insightful writing is refreshing in this world of narrowmindedness.

[Abobtrader]

An excellent debate. One factor that hasn't been considered (or maybe it has been considered and rejected) is the possibility that recorded rates for certain crimes may change simply due to changes in measurement. In the UK, for example, government policy has attempted to incentivise improvements in the public sector but this has led to many instances of 'massaging' of data to give better results.

I saw Malcom Gladwell on UK tv a few days ago, talking about Blink. Excellent stuff indeed.