pageok
pageok
pageok
[James Q. Wilson, guest-blogging, June 9, 2008 at 5:20pm] Trackbacks
Deterring crime:

Several readers have asked whether prison deters crime, some saying that correlation is not the same as causation. Social scientists have known that dictum for the better part of a century and have worked hard at finding out whether prison deters crime among would-be-offenders (it obviously prevents it among people already locked up). Steven Levitt, principal author of Freakonomics, discusses the issue in that book and summarizes the evidence, now very strong, in the chapter he wrote from a book I and Joan Petersilia edited entitled Crime. The econometric technique is simple: construct an equation that asks what can explain the crime rate using, as explanations, everything that we think causes crime (for example, urbanization, unemployment, and the like) and adds the chances of going to prison. When you do this, you find that the higher the chances of going to prison in a state, the lower the crime rate, other things being equal. (You also discover that the unemployment rate has very little effect on crime.)

hattio1:
How do you determine "the chances of going to prison?" It seems that if you just look at the percent of the population in prison, you are going to wind up with the state with the highest percentage of crime. If you look at anything else, you are letting other factors creep in. For example, if you look at the number of cases filed vs convictions achieved, you would be including some who are convicted, but don't go to jail. Similarly a state that tended to charge less serious crimes that were easily proven vs. speculative charges they then plea bargain down would also effect the percentages. In short, it would seem impossible to tease this out of data from different states.
6.9.2008 6:27pm
T. Gracchus (mail):
What is the causal mechanism? How is it that the deterrence effect actually runs in these cases?
6.9.2008 6:38pm
Tyrant King Porn Dragon (mail):

(it obviously prevents it among people already locked up).

Are crimes committed by prisoners against each other, and by guards against prisoners (or vice versa) taken into account in your rubric?
6.9.2008 6:43pm
R Gould-Saltman (mail):
I'm also curious about what, for lack of a better word, would be a "rebound" effect on those not imprisoned for life: i.e., is there any reliable data from which it is possible to tease out the effect of prison time (versus, say a well-run probation/retraining system, if such can be found anywhere) on rates of recidivism?

Is there any validity to anecdotal claims that prison teaches a significant number of offenders, particularly relatively youthful ones, to be "better criminals", with better skills at evading detection and conviction in future, new alliances to gangs,and more economically rational selection of offenses/victims?


Richard Gould-Saltman
6.9.2008 7:17pm
jgshapiro (mail):

When you do this, you find that the higher the chances of going to prison in a state, the lower the crime rate, other things being equal.

When you say chances of going to prison, that could mean two things. It could mean the chance that a convicted person gets a prison sentence, and it could mean the chance that they are caught and convicted in the first place. At least to me, the chance of going to prison means, if I knock over a liquor store today, what are the chances that I will (1) get caught, (2) get convicted, and (3) get a prison sentence that is not suspended. (The third factor could also include the chance that the sentence would later be reduced for good behavior or replaced with parole.)

I think the first two factors are being conflated into the third factor in this discussion. You can specify any length of sentence, but if the prospective criminal thinks it unlikely he will get caught, or if caught thinks it is unlikely he will be convicted, I wonder how much more he would be deterred by the potential of 10 years of hard time than he would be by a potential sentence of 1 year.

Aren't we better off focusing on catching them and convicting them (so we deter more crime in the first place) than we are on lengthening sentences and building more prisons to house more inmates serving longer sentences?
6.9.2008 8:00pm
Dr. T (mail) (www):
Does imprisonment deter crimes, or does it just translocate crimes to within the prisons? Do released prisoners commit just as many crimes per year as case-matched criminals who had never been caught?

My suspicions are that violent criminals become violent prisoners, drug dealing criminals become drug dealing prisoners, thieves become master-thieves-in-training, etc.

I believe that most U.S. prisons are counterproductive. If we want them to be effective, then they must be redesigned to minimize within-prison crimes. That means better security, heavy use of cameras, no large areas for congregation, separation of violent and nonviolent prisoners, tight screening for intoxicants (which means close scrutiny of guards and other prison employees), etc. If prisons become tough but safe and offer education and useful job training, then we might get lower rates of recidivism. The best way to afford this: release the nonviolent drug offenders and use the space and extra funds to redesign the prisons.
6.9.2008 8:44pm
Perseus (mail):
does it just translocate crimes to within the prisons?

Even if that's true to some extent, that is still a significant improvement in the quality of life for law-abiding citizens (i.e. "Escape from NY" sounds pretty good to me).
6.9.2008 9:11pm
rick rick (mail):
Despite all the efforts to be "tough on crime", a criminal has to be arrested, and plead guilty for something in front of a judge about 10 times before going to prison. Generally speaking, people in this category commit many other crimes, for which they are not caught.

There is a whole subculture that is permanently involved with the court system, with multiple cases dragging out over years, and with periodic spells behind bars only delaying the other cases. The crimes are usually for theft, robbery, burglary, DUI, drug possession, drug dealing, and drunken brawling. Much less often do these types commit murder, rape, or other severe violent crimes.

Unfortunately, I know a number of people like this very well, and I am acquainted with a large number of their friends, who are all the same way. This is not just an inner city problem. I have seen this exact same pattern in several suburban areas, and in smaller citys as well. Nobody in this category wants to go to jail, however, they are not willing to stop drinking and using drugs, and they are not willing to discipline themselves to hold a job and keep their finances in order.

When people in this category are finally sent away for a long time, they are prevented from committing crimes as long as they are locked up. I believe these types are the majority of the prison population. Keeping people like off the streets is the primary way that crime rates are driven down.

I would like to think there is a system of rehabilitation and job programs that will help people like this. However, I don't think these programs work unless a person wants to go straight. Most of the people in this category are not willing to accept the straight life.
6.9.2008 9:37pm
Stacy (mail):
I guess not a lot of lawyers take econometrics in college. To expand (a bit) on the post above, the method involves adding together all imaginable variables and assigning a coefficient to each. I haven't done it in awhile but if memory serves it goes much like this:

1 = A*X + B*Y + C*Z and so on.

The number 1 represents 100% of whatever phenomenon we're interested in. A, B, C are the coefficients and X, Y, Z represent the factors we think contribute to the phenomenon. If A is a huge number but B and C are tiny, then X is the most important contributing factor. It's one of those areas where statistics can make something plainly obvious that's just impossible to see accurately by simple intuition.

Admittedly, I don't know the statistics on recidivism and habitual criminals, but I have the sense that the majority of serious violent and property crimes are committed by a relatively small number of repeat offenders. If true, then it makes perfect sense that locking these offenders up and throwing away the key would make a dramatic dent in crime.
6.9.2008 11:20pm
Demosophist (mail):
I haven't read the book in question, but I should think you'd be able to make a fair stab at "imprisonment odds" by computing a ratio of crimes committed to prison population. Provided you use the same metric for the first variable the correlation between the actual odds and the computed odds should be pretty high. (Correlation means that you'd come up with the same rank ordering.) The main problem would be deciding whether to use actual crime stats, or a "victim survey," but you could cover that base by using both and comparing the results. Each has advantages and disadvantages.
6.10.2008 11:05am
LarryA (mail) (www):
Aren't we better off focusing on catching them and convicting them (so we deter more crime in the first place) than we are on lengthening sentences and building more prisons to house more inmates serving longer sentences?
How does catching and convicting a criminal deter crime, if there is no penalty? (Yes, there is the social consequence of being convicted, but particularly for repeat offenders it's minimal. In some parts of society a criminal record is actually seen as a social benefit. Check the British policy of issuing ASBOs to teens. [Anti-Social Behaviour Order])

And I read the question as concerned going to prison, not necessarily staying there a long time.
I would like to think there is a system of rehabilitation and job programs that will help people like this. However, I don't think these programs work unless a person wants to go straight.
Bingo. As I remember a significant percentage of young people commit one crime, get caught, and figure out that crime is not a good lifestyle choice. A more generous approach to restoring rights after a suitable time with a clean record would go a long way toward encouraging people to go straight.

We could also end the war on drugs, which provides so much of the funding for violence in the U.S. and many other countries.
6.10.2008 4:31pm
Andy Freeman (mail):
It would be nice to know what the "that's not enough to show causality" folks would accept as adequate proof or have accepted as adequate proof in other cases.
6.10.2008 4:39pm