Abortion/Crime: Where Do Ideas Come From?


It’s always interesting to see where smart people get their ideas. Often, especially in the creative arts, it’s impossible to trace an idea down to its roots. But it’s easier in the social sciences.

I, for one, believe that Steve Levitt has had an awful lot of good research ideas, and it’s good to hear how a particular idea germinated. In our latest Freakonomics video, Levitt talks about the research that led to his most controversial paper, “The Impact of Legalized Abortion on Crime,” co-authored by John Donohue. We wrote about this subject at some length in Freakonomics, and there was more than a little bit of blowback. This video is the first of a two- or three-part series.


The video states that the U.S. went from "almost no abortions" to "1 million a year". But all the pro-abortion crowd would say that there were tons of abortions performed in back alleys.

First, are there enough back alleys to support this theory?

Second, how did you measure the amount of abortions before the Supreme Court found abortion in the constitution?


So the church doesn't want abortion because....why?

Is it because murder is immoral?

Or not aborting produces a populations who are more inclined to criminal activity, which gives them more people to save, while spreading fear amongst the general population and giving a large slice of the rest of the people cause to believe in something greater when bad stuff happens?

No money, no mission.

Born & raised a Catholic, go figure.....


"So the church doesn't want abortion because....why?"

Because they feel it is murder. I don't think there is any church conspiracy to oppose abortion in order to maintain a large number of "sinners" or whatever you are suggesting.

I would argue that the religious still have a right to oppose abortion, even if it lowers the crime rate. They feel that it is murder, plain and simple, so in their eyes the amount of wrong-doing has not gone down.

In an extreme example, requiring males from low-income households to be imprisoned from age 17 to 26 might greatly reduce crime, but that would not mean it was right to do so.


I remember the very moment I read about the abortion/crime correlation in Freakonomics. It was both jaw dropping and eye opening. It continued to linger around the back of my mind for weeks.

William Tanksley

I don't want to believe that abortion causes low crime rates. Yes, that's a declaration of bias. At the same time, I found the initial argument compelling. There are severe holes, though. The biggest is that a similar crime drop happened in Canada, in which abortion wasn't legalized until much later. A smaller one is that during that time period, population didn't shrink -- suggesting that other effects (perhaps derived from immigration) might be drowning out the alleged effects of abortion.


I don't think this is necessarily so. I am very much pro-life and have absolutely no problem with what Mr. Levitt has presented over the past several years on this issue.

Char Lyn

@ William Tanksley

I don't know what time frame you are referencing with your stats of Canada, but could it be that the similar crime drop in Canada corresponded to the legalization of abortion in the states? The border with Canada has always been easier to cross and a large majority of the Canadian populations lives within driving distance of US cities.

Just a thought.


I did a calculation once for a campus political group I was part of--I assumed more liberal people are both slightly more likely to both have an abortion, and to raise liberal (i.e., Democratic) children--by, say, a 55-45 margin. This means that Democrats have lost a permanent electoral advantage of 2.25 million potential voters, which is growing every year. (Note that ALL young voters went for Kerry in 2004 by 55-45, and this excludes their parents' political orientations' effect--55-45 is probably low).

To make a conservative estimate, say 2 million potential voters, at 50% turnout. That's a 1% swing--enough to have given Kerry Ohio and New Mexico, put Ohio in play, and given the Democrats 2-4 Senate seats.

Brett S

I'm curious. Did your statistical model have predictive accuracy after it's 2001 publication?

Has the recent rise in crime in certain US cities such as Boston and Phoenix been accurately predicted by changes in US abortion rates in the early 90's? If not, what explains the recent rise in felonious crime in these cities?

I understand that abortion rates are only one highly correlated variable. If the model doesn't fully explain these rising crime rates, what other variables should we consider? Perhaps, the rise of population through net immigration and hence increase in male juvenile populations?

This would be an interesting data mining project in relation to the immigration debate and the population changes it brings.

If these significant changes in crime rates were not predicted by the abortion model, could it simply be hotter than usual summers, changes in incarceration patterns do to overcrowding, or something else?

I'm curious and searching for a model with predictive accuracy - a high R-squared on data through 2007.

- Posted by Brett S


Rita: Lovely Meter Maid

What Stephen Levitt's thoughts and research about abortion and the link to reduced crime say to me is that those who are most likely to become criminals come from extremely compromised beginnings, especially ones that are mired in poverty. It's a mistake if people get too caught up in abortion or race when considering this. Abortion is not the bad guy here, (or race, of course) even for those who are against abortion. If negative, impoverished conditions were greatly improved for those who are born into them, I believe that crime rate would drop through this, as well.

William Tanksley

"could it be that the similar crime drop in Canada corresponded to the legalization of abortion in the states?"

That's a great speculation. I'd think it deserves analysis. A similar speculation would address the observation that the drop in criminality began with those born before Roe v Wade: perhaps the older siblings of aborted children found themselves (statistically speaking) in a better home environment.

There's a lot more tough analysis to be done, and I don't think any solid conclusions at the end.

As Levitt implies, even if there were strong conclusions to be drawn, there can't possibly be enough to defuse the argument -- after all, capital punishment is controversial; predictive capital punishment is a fiction ("Minority Report"); how much worse is statistical predictive capital punishment! ("We're going to kill you now because your kind tends to murder more often when you grow up.")


Brendan Dornan

Many studies show that crime cycles right along with economic cycles. when things are good, there is less crime, when the economy falters, crime increases. A properly constructed study would need to control for these things in some way. will leave it up to you do determine if freakenomics does this well.

Dash D.

Levitt's my hero...

Steve Sailer

Here's the abstract of a new paper in press by economist Ted Joyce. It's one of several by social scientists who have _not_ been able to confirm Levitt's celebrated abortion-cuts-crime theory.

A Simple Test of Abortion and Crime
Ted Joyce
Baruch College and Graduate Center
City University of New York
National Bureau of Economic Research

Forthcoming in "Review of Economics and Statistics"


I replicate Donohue and Levitt's results for violent and property crime arrest rates and then apply their data and specification to an analysis of age-specific homicide rates and murder arrest rates. The coefficients on the abortion rate have the wrong sign for two of the four measures of crime and none is statistically significant at conventional levels. In the second half of the paper, I present alternative tests of abortion and crime that attempt to mitigate problems of endogeneity and measurement error. I use the legalization of abortion following the 1973 decision in Roe v. Wade in order to exploit two sources of variation: between-state changes in abortion rates pre and post Roe, and cross-cohort differences in exposure to legalized abortion. I find no meaningful association between abortion and age-specific crime rates among cohorts born in the years just before and after abortion became legal.



I would also be interested in knowing how the rate of child abuse has changed since RvW...once you legally establish that chidren are property, as RvW did, doesn't it follow that it is a parent's right to do anything they "choose" to do with them, at least up until a baby is considered to be self-aware? Isn't that the what the arguments in favor of abortion hinge upon...sentience? A newborn is not any more sentient five minutes after birth than they are five minutes before birth, yet they are killed during late term regularly for no reason at all by Dr. George Tiller.
We are officially now slipping down that slippery slope, people. Recognize it. Stop it now.

Andrew Leigh

Unless I'm missing something, I thought the point of the post was to draw attention to how Levitt thinks up new ideas - not to have the 1001st re-hash of the abortion paper.

I thought the video was terrific. Incidentally, I think the phrase SL was looking for in the opening of Part II might have been "pastiche of evidence".

David Kessler

Has anyone does a study of the correlation between crime and abortion in the United Kingdom?

Prior to 1967 abortion in the UK was only available in extreme cases of danger to the life or health of the mother. From 1967 onwards abortion for a variety of reasons - including "social reasons" - was legalized. It was even provided free on the National Health Service. (I mention this because Freakonomics tacitly treats "legalized abortion" in the US as synonymous with "affordable abortion.")

If the correlation between crime and abortion is correct then we would expect a big drop in crime in Britain starting round about 1985 or so. Can anyone tell me if this happened?

Parethentically, does anyone know what happened to the murder rate in the UK when they abolished the death penalty in 1965?


I'd have to say that poverty/economic conditions play a bigger part in the crime rates than anything else. Incidentally, it also has a huge impact on the abortion rates.

Solution: Put as much energy into fighting poverty and homelessness as people do in fighting crime and abortion and you will see the two latter conditions decrease on their own.

Attack the problem, don't just treat symptoms.


Something else he might want to investigate is that the number of number of swimming pool drownings goes up in direct proportion to how many lawns are being mowed. Let's do all we can to prevent those drownings (the solution seems obvious).


I believe abortion is wrong for the following reasons. Every unborn child has the following criteria that makes it different from the born.

But if you take the same look and compare the following criteria to the people are living, you will see how ludicrous abortion really is.

S - Size - just because somebody is smaller than you, does that make it right for you to kill it?

L - Level of Development - If somebody is retarded, or another mental disorder, should that person be killed off.

E - Environment - Should you kill off everybody in Alaska because of their location.

D - Dependency - If you are dependant on drugs, or dependant on another to provide you with care, does it make it right for you to be killed off because of that?