Software Used To Predict Who Might Kill

eldavojohn writes "Richard Berk, a University of Pennsylvania criminologist, has worked with authorities to develop a software tool that predicts who will commit homicide. I could not find any papers published on this topic by Berk, nor any site stating what specific Bayesian / decision tree algorithm / neural net is being implemented." From the article: "The tool works by plugging 30 to 40 variables into a computerized checklist, which in turn produces a score associated with future lethality. 'You can imagine the indicators that might incline someone toward violence: youth; having committed a serious crime at an early age; being a man rather than a woman, and so on. Each, by itself, probably isn't going to make a person pull the trigger. But put them all together and you've got a perfect storm of forces for violence,' Berk said. Asked which, if any, indicators stood out as reliable predicators of homicide, Berk pointed to one in particular: youthful exposure to violence." The software is to enter clinical trials next spring in the Philadelphia probation department. Its intent is to serve as a kind of triage: to let probation caseworkers concentrate most of their effort on the former offenders most likely to be most dangerous.

Reference

[Quixote]

This paper was published in the June 2006 issue of "The Journal of Quantitative Criminology".
Here are the pertinent details:
Title: Forecasting Dangerous Inmate Misconduct: An Application of Ensemble Statistical Procedures
Journal: Journal of Quantitative Criminology
Issue: Volume 22, Number 2 / June, 2006
Pages: 131-145

Abstract:
In this paper, we attempt to forecast which prison inmates are likely to engage in very serious misconduct while incarcerated. Such misconduct would usually be a major felony if committed outside of prison: drug trafficking, assault, rape, attempted murder and other crimes. The binary response variable is problematic because it is highly unbalanced. Using data from nearly 10,000 inmates held in facilities operated by the California Department of Corrections, we show that several popular classification procedures do no better than the marginal distribution unless the data are weighted in a fashion that compensates for the lack of balance. Then, random forests performs reasonably well, and better than CART or logistic regression. Although less than 3% of the inmates studied over 24 months were reported for very serious misconduct, we are able to correctly forecast such behavior about half the time.

Unfortunately, you've got to pay $30 to get this paper. Maybe some slashdotter with a school/corp subscription to Springer will put up the text? ;-)

Tag as "precrime"...

[Kadin2048]

Yeah except in Minority Report, they were using psychics to predict who would kill; here it's just an overgrown spam filter.

Re:Reference

[Anonymous Coward]

http://download.yousendit.com/CFD415EB623F072C [yousendit.com]

Re:Reference

[dysk]

Interesting stuff. Here's a link to the full text:

http://130.58.240.179:8080/~erek/minorityreport.pd f [130.58.240.179]

Re:Reference

[dch24]

Thanks! However, reading the paper, it seems that this paper is about the California Department of Corrections, and is not actually about who will commit homicide "on the outside." It's about which prisoners are "likely to engage in serious misconduct while incarcerated" (from the abstract). I don't know if this is the right paper. In fact, I'm going to guess that Berk hasn't published a paper on his new method. This paper may be a similar method, but there's no way to know that.

I also wonder in yousendit.com can handle a slashdotting. I guess we'll know soon!

Edit: Bad Idea.

[SynapseLapse]

I thought Phillip K. Dick already explained why this was a bad idea...

There, I edited that for you buddy.
Let's just leave it at that's what you really intended, because otherwise I'll destroy all of my karma in spewing forth a slur of obscenities about how...
well, let's just leave it at that.

Moderators on drugs?

[Anonymous Coward]

Wow, murder is considered funny?

With the short time I spent in the US I saw just how horrible of a place it is. There are way too many gun owners. Something needs to be done about that. I know I will never return until the US finally puts into place some common sense gun control.

This already happens....

[robot5x]

actually there are many tools like this already in existence... modern probation work has been scientificalised and statisticalised to the extent that you can't do anything with an offender until you know what their various scores are. In the UK the risk of general reconviction is calculated statistically in the OGRS programme based on age, conviction, prison sentences etc. (http://www.homeoffice.gov.uk/rds/probation1.html [homeoffice.gov.uk]) . This also produces a level of risk that that person will commit a violent offence. There are other specialist tools for domestic violence - the Spousal Assault Risk Assessment which is a 20 item checklist. Also, for sex offenders their risk of reconviction is assessed by using the Thornton Risk Matrix 2000. Every offender who comes into the probation system also has an OASys assessment completed on them - which asks the assessor to score factors from 14 different areas such as accommodation, lifestyle, substance misuse etc. (http://www.probation.homeoffice.gov.uk/files/pdf/ Info%20for%20sentencers%203.pdf [homeoffice.gov.uk]).

Re:Utter BS

[Llywelyn]

1) Convicted criminals are the only ones that concern probation officers.

2) Convicted criminals are the only ones they are likely to have the data to fill most of the fields for.

3) Probation officers have a job to do that does not involve tracking random citizens.

Thus, it seems unlikely it could be used for anything *but* the intended purpose without a fairly serious rework.

In other news lately...

[kan0r]

Scotland Yard agrees:

It looks like Scotland Yard [timesonline.co.uk] is also looking for scary new tactics in fighting crime. The latest idea of Laura Richards, head of analysis of the Metropolitan Police's Homicide Prevention Unit, sounds like a strangely familiar concept to those who have seen Minority Report. She aims to create a database of people who could supposedly commit a crime in the future, based on their psychological profile.

Even though preventing crimes is a noble motivation, this idea raises serious privacy issues.

As a sidemark it should be mentioned that Laura Richard also seems to be part of the team that "revealed" Jack the Ripper's face some time ago.

Oh, stop it.

[Kadin2048]

many places in europe (with virtually no legal gun ownership) are in fact much safer than the usa

http://en.wikipedia.org/wiki/Correlation_implies_c ausation [wikipedia.org]

I find it amusing that Europeans love to bemoan Americans for thinking, particularly when they travel, that Europe should be just like America; however, whenever a European or Euro-phile analyzes crime in the U.S., the only difference that ever gets brought up between the two places in question is the difference in gun control. Really ... so that's the most significant difference between "many places" in Europe and the U.S.? You don't think there are, just perhaps, some more significant social, economic, and cultural contributors to the difference in crime?

Europe and the U.S. are not the same place, and you'd have to control for a whole lot more variables than "gun control" in order to start comparing something as high-level as per-capita murder rates.

Re:Reference

[monteneg]

I haven't read the paper, so I may be wrong in interpreting the abstract. However, it looks to me like the cocky people joking about this paper might think before writing. There is a 3% chance of someone committing a serious misconduct while in prison in said 24 month period. The method of the paper can guess about 50% of these. This is much better than flipping a coin, because if you flip a coin then yes you'll get half of the 3%, but the other 48.5% of people that you finger will be people who wouldn't have committed serious misconduct. The abstract doesn't tell the rate of false positives, but presumably it'll be much smaller than 48.5% of the prison population.

Re:And yet again

[inviolet]

Take care to maintain context here. This project is not about individual judgments of other individuals. This project operates on the macro level, directing limited resources where they are most likely to have the greatest benefit. Only after all the likelihoods have been maximized do we re-introduce individual attention and individual treatment. (How else would you apportion too few workers to too many cases?)

And don't worry, just because you're predictable, doesn't mean you don't have free will. You can always assert yourself and behave unpredictably, if you so will to. The point of this project is that most of the time, most of the people will not assert themselves... and this fact can help us allocate resources more efficiently.

It's like those AI rock-paper-scissors programs: you can focus your effort and behave randomly enough to trick the program, but as soon as you divert your attention, you fall back on your own internal rules, and become predictable, and the program starts kicking your ass.

Re:Reference

[btellier]

A violent criminal usually falls either below 85 IQ or above 185 IQ. The frequency of this below IQ 85 is about 85% of the population of such persons with about 13% above IQ 185. Only a tiny fraction falls in the middle.

I defy you to find me a single study which supports this ridiculous claim. According to the IQ bellcurve [geocities.com] only one in 1,000,000 people will have an IQ of 174-200. So what you claim means that there are 300 people in the United States committing 13% of the violent crimes? Nonsense.