« Ernon appeal in the news | Main | Lots of capital news and notes »

September 9, 2007

Another strong editorial on correction costs

Ball_chain This morning's Detroit Free Press has another effective editorial, entitled "Stop the state prison drain: Sentencing, parole reforms are key to getting state Corrections costs in line."  This latest editorial hits similar themes as this earlier one on the costs to Michigan of ever-increasing incarceration rates.  Here is how it starts:

Michigan's super-sized prison system has put the state on financial lockdown. The Department of Corrections is largely responsible for the state's ongoing budget crisis and the nearly $2-billion shortfall it faces for the next fiscal year.  Michigan taxpayers spend $1.9 billion a year -- $5 million a day -- to lock up more than 50,000 prisoners.  That's more than it spends on higher education.  Today, one of three state civil service employees works for Corrections; in 1980, one in 20 did.

Most troubling, Michigan incarcerates at an average rate of 40% higher than the seven other Great Lakes states, which also report lower crime rates. Michigan's higher incarceration rates take an extra $500 million a year from the state's depleted general fund. It's money that could be better used for education, health care, roads or even a tax cut.

Some recent related posts on sentencing costs, politics and increased incarceration:

September 9, 2007 at 12:29 PM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Another strong editorial on correction costs:


"Most troubling, Michigan incarcerates at an average rate of 40% higher than the seven other Great Lakes states, which also report lower crime rates."

Although the First Amendment doesn't allow it (and shouldn't), sometimes I wish we could require journalists to take a qualifying exam like the bar exam. Anyone who doesn't understand the fundamental principle that correlation does not prove causation would flunk.

Posted by: Kent Scheidegger | Sep 9, 2007 1:02:07 PM

While I generally agree that 1) journalists are generally idiots; 2) it is unconstitutional to require them to be licensed; and 3) you can’t prove causation with correlation, I need to point something else out: according to most philosophers, you can’t prove causation with anything, since causation can’t be observed. Instead, causation is a mental construct we have to explain events.

Anyway, any adjudicating or policy-making body can, and in fact does, use correlation to assume “causation” if it wants.

Posted by: S.cotus | Sep 9, 2007 4:15:57 PM

If correlation and causation are never to be linked, what would you have policy makers do? I find it amusing that someone who is adamantly pro-government in criminal matters finds correlation to be irrelevant here, since causation is frequently "assumed" based on correlation in that context. I.e., drug offenses create blight in our inner cities and increase violent crime, etc. Yet when the correlation cuts against pro-criminalization and government advocates, as in deterrence of the death penalty based on a review of murder rates in capital versus non-capital jurisdictions, or this, we have no shortage of justifications or simple criticism of reform proposals.

Also, Michigan's crime rates in relation to other Great Lakes states is fairly compelling evidence that its incarceration policies, and the costs associated with them, do little to actually deter crime. Michigan does not border South Dakota; the Great Lakes states encompass Chicagoland and major metropolitan areas in Ohio and Indiana, to say nothing of New York and Pennsylvania.

Posted by: Alec | Sep 9, 2007 4:43:33 PM

Moreover, people are routinely convicted based on "correlation" evidence. For instance, the disappearance of something from a store, combined with its appearance in someone's bag is usually enough to infer that he carried it out of the store.

Now, some forms of correlation evidence might be unconstitutional. For instance, to determine that people have a certain mental state because of their race -- and its correlation with certain observer behaviors -- is probably unconstitutional, but the government seems to have floated it from time to time in the "terrorism" context.

Posted by: S.cotus | Sep 9, 2007 4:50:26 PM

Oh, what the heck... Kent is just using a typical undergraduate catchphrase. I don't know why it is worth spending that much time responding to it. I guess he lured me in because I agreed with the first part of his post, and most of his posts are comparatively well-researched as compared to the other blindly pro-government people on here that seek to put a greater percentage of people in jail but refuse to name a number.

Posted by: S.cotus | Sep 9, 2007 4:52:13 PM

Kent said, "Anyone who doesn't understand the fundamental principle that correlation does not prove causation would flunk."

Were you talking to yourself in allusion to your recent Three Strikes post?

Posted by: George | Sep 9, 2007 9:08:28 PM

My, my, my. What a load of muck. Time to get out the shovel.

Alec asks, "If correlation and causation are never to be linked, what would you have policy makers do?" Who said never, Alec? I certainly didn't. There are several reasons why A may be correlated with B. A causes B is one of them. B causes A is another. A and B are both caused by C is another.

Correlation is not "irrelevant," and I didn't say it was. It is a useful starting point for analysis. The next step is to examine additional evidence to see which of the possibilities is the most likely explanation. My problem is with people who claim to have proven causation from the correlation alone, particularly when one of the other alternatives is obviously quite possible.

Comparing Michigan's crime rate with Illinois', according to the Sourcebook the violent crime rates per 100k for 2005 are 552.1 and 551.5, respectively. It's literally true to say Michigan's rate is higher, but that's a pretty thin difference to support a claim of "compelling evidence."

The person who calls himself the Supreme Court of the United States says "journalists are generally idiots." Say that if you like, but don't purport to be agreeing with me, because I said nothing of the sort. I criticized one particular editorial and lamented that the fallacy employed there was much too common among journalists, but that is vastly different from what you said.

The permissive inference of theft from possession of recently stolen goods, see, e.g., Wright v. West, 505 U.S. 277 (1992), is an entirely different matter from the use of correlation statistics in studies of the general population. For one thing, when the question is what happened in a particular case, the defendant can refute the inference by simply introducing evidence as to how he legitimately came by the goods.

"Correlation does not prove causation" is not a "catchphrase" but an important principle. Yes, it is "undergraduate" in the sense that it is usually taught in introductory courses. It needs to be repeated, though, as it is so often disregarded.

Posted by: Kent Scheidegger | Sep 10, 2007 1:07:17 PM

Post a comment

In the body of your email, please indicate if you are a professor, student, prosecutor, defense attorney, etc. so I can gain a sense of who is reading my blog. Thank you, DAB