New, preliminary data testing the "more abortion, less crime" hypothesis

I've been very interested in the hypothesis from the book "Freakonomics" that mothers who abort their pregnancies are often poorly situated to provide their children with a good upbringing. The hypothesis goes on to state that where abortion is illegal, the children are born and due to inadequate upbringing end up committing crime at a higher rate than the background average. With abortion, you get less crime. The book authors refer to certain data from Europe as well is from the United States, but I am particularly interested in Romania.

The authors state that the Romanian dictator Ceaucescu banned abortion in the 1960s, resulting in increased crime rates when those children reached their teenage years. The authors go onto state that abortion was quickly legalized and widely used starting immediately in the year following Ceaucescu's overthrow, which would be 1990. It occurred to me that we are reaching a point where the children born after the revolution are getting old enough so that we can test the hypothesis and see whether it applies to them.

I trekked over to the Stanford University Library yesterday, with great hopes of finding the Romanian crime rate for 13 year olds and for 16-year-olds from the years 2000 through 2004. All 13-year-olds in 2004 will have been born after the revolution, while all 16 year olds will have been born prior to the revolution.

It turns out that information is not as easy to get as one might hope. Two wonderful Stanford research librarians and I spent over an hour trying to find the data but were unable to do so. I had to leave, but then received an e-mail seven hours later (!) from one of librarians with some pretty good information. It is the number of crimes committed in Bucharest from 2000 to 2004 by adults and also by minors, separated into the categories of those less than 14 years old and those from 14 to 18 years. So it isn't a national figure and it isn't a crime rate, but assuming that the teenage population in Bucharest did not change dramatically in those five years, then it seems to be a reasonable substitute for a crime rate. It's also in Romanian, but seems to be fairly understandable.

These are the total crimes committed from 2000 to 2004 separated by age category [my notes in brackets]:

V.PERSOANE INVINUITE PENTRU COMITEREA DE INFRACTIUNI

a. pina la 14 ani: 124 [in year 2000], 47, 91, 60, 34 [year 2004]

b. 14-18 ani: 1594 [year 2000], 1767, 1545, 1158, 2016 [year 2004]

- tineri [apparently means "youths"] (18-30ani): 9607, 9579, 9003, 7818, 9238

So in this data, we have three categories. Everyone in the first category was born after the revolution, the second category is mixed (although we can guess that it is strongly dominated by people born before the revolution), and everyone in the third category was born before the revolution. The relevant data is that from 2003 to 2004, crime went down in the first category and up in the latter two categories. This tends to reinforce the argument from Freakonomics that abortion reduces the number of children growing up in social situations more likely to produce criminal behavior.

A couple of cautionary notes: I assume this is the age when arrested, not age at some later point like when sentenced. Also, I am not a statistician so I can't say that change from 60 to 34 is significant, although I suspect it is. The strong swings in the data over the years could also make it harder to identify a particular trend, and there is always a possibility of confounding factors. Despite all this, it seems to be some preliminary evidence in support of the thesis.

Equally important, we will be able to continue to test the hypothesis with increasing confidence over the next few years. The first category should have a different trend in its crime rates than the third category, and the second category should increasingly resemble the first category in the trends for its crime rates.

I think this is relevant to a discussion that came up in Tim Lambert's blog over whether it is possible to do an honest scientific analysis of a politically charged issue. It is hard to cherry pick your data sets if what you are doing is making predictions for the future. When I started researching this, I had no idea what the data would show, nor do I have much invested in the outcome. I obviously have no control over will happen in the future in Romania, so again by making these predictions, I think it is possible to make some progress on politically-controversial scientific questions.