What It Costs, What It Buys SOC 3333 · Crime Lab · Midwestern State University

Every sentence is an argument about why we punish. Chapter 13 gives you five answers to that question, and they do not agree with each other. This page comes at the argument from an angle the textbook does not take. Every sentence is also a purchase. Somebody pays for it, and something comes back. Once you can see the price tag and the return next to each other, the five theories stop being abstractions and start being budget decisions. Nothing you do on this page is recorded, and no guess is graded.

What does one prison cell cost for one year?

California publishes this number every year in its enacted budget, broken into categories, which makes it the cleanest public accounting of what confinement actually costs. Before you look it up, put down a number. Most people are low by a factor of two or more.

California's budgeted annual average cost per incarcerated person, 2025 to 2026.
$40,000

California Legislative Analyst's Office, annual average cost per incarcerated person, 2025 to 2026 enacted budget. Categories as published. Retrieved August 2026.

Read the categories, not just the total. Security is the largest line, which is what you would expect. Health care is the second largest and it is nearly as big, because a prison is also a hospital with a legal obligation to treat everyone inside it, and because the population is aging. That single fact explains a lot of what follows in this chapter, including why incapacitation gets expensive precisely when it is least useful.

Now the question that matters for Chapter 13. Of that total, how many dollars go to the line labeled rehabilitation programs, meaning academic education, vocational training, and in-prison programming?

Dollars per person per year, out of the total above, budgeted for rehabilitation programs.
$15,000
$4,761 out of $127,788, which is 3.7 cents of every dollar. Rehabilitation is one of the five official justifications for punishment in this chapter, and it is the one the textbook says came back after Martinson. In the budget it is the second smallest line, ahead of miscellaneous. Hold onto that ratio. It is the sharpest single piece of evidence in the chapter that what a system says it is for and what it spends money on can be two different things.
Two honesty notes. California is the most expensive state corrections system in the country and is not typical of the other forty nine. The most recent federal figure is $47,162 per person per year, which is $129.21 per day, and most state figures fall between those two numbers rather than near the California one. Also, this line item is what the budget calls rehabilitation, so it undercounts things like mental health treatment that sit in the health care line and arguably do rehabilitative work.

Federal figure: Bureau of Prisons, Annual Determination of Average Cost of Incarceration Fee, fiscal year 2024, published in the Federal Register. California figure: Legislative Analyst's Office. The California total has risen about 161 percent since 2010 to 2011. Retrieved August 2026.

What a dollar buys, program by program

The Washington State Institute for Public Policy has spent two decades doing one specific thing: taking the research on a criminal justice program, estimating how much crime it prevents, converting that into dollars, and setting the result against what the program costs to run. It is the closest thing this field has to a price list. Below are four programs from its adult criminal justice results, each showing what it costs to run against the benefit the model projects.

The comparison to make. A course of cognitive behavioral therapy costs about $1,709 per person. A year in the California cell you just priced costs $127,788, which is about seventy five times as much. That does not make the cell unnecessary, because some of the people in it are there for reasons no program addresses. It does mean that any argument for the cell has to be made on retribution or incapacitation grounds, because on a pure crime-reduction-per-dollar basis it is not close.
Read these numbers honestly. The benefit figures are modeled projections of lifetime benefits in 2023 dollars, not cash that appears in anyone's budget. They include the value of crimes not committed, which counts benefits to future victims and to participants, so they are much larger than the taxpayer-only figure. For cognitive behavioral therapy the total benefit is $7,127 but the taxpayer share is $2,449. A model is an argument with numbers in it, and you are allowed to argue with the model. What you should not do is treat a benefit-cost ratio as if it were a measured outcome.

Washington State Institute for Public Policy, benefit-cost results for adult criminal justice, methods last updated December 2024. Figures per participant in 2023 dollars. Retrieved August 2026.

Predict the rating

CrimeSolutions is the federal government's evidence clearinghouse, and you have met it in four chapters now. It reads the evaluation literature on a practice and assigns one of three labels: Effective, Promising, or Ineffective. Seven practices are below, one from each corner of this chapter. Predict the label before you look. Pay attention to how the set comes out, because the pattern is the lesson.

You matched 0 of 7.
Notice what almost nothing earns. Across a chapter about five theories of punishment, exactly one practice here is rated Effective, and it is a place-based policing strategy rather than anything done to a person after conviction. Most of what works earns Promising, which is a statement about the strength and consistency of the evidence as much as about the size of the effect. This is the honest state of the field, and it should make you suspicious of any sentencing argument, from any direction, that claims certainty.
The restorative justice split is the one to remember. Restorative programs are rated Promising for reducing reoffending and for victims' sense of fairness, and Ineffective for remorse and recognition of wrongdoing. Braithwaite's theory says the mechanism is remorse and reintegration. The measured outcome shows up and the proposed mechanism does not. That is a live problem for the theory even though the policy looks good, and it is exactly the kind of distinction the final exam rewards.

NIJ CrimeSolutions practice profiles, retrieved August 2026. Hot spots policing, focused deterrence strategies, adult drug courts, cognitive behavioral therapy for adults at moderate and high risk of offending, rehabilitation programs for adults convicted of a crime, restorative justice programs for juveniles, and Scared Straight.

Name the theory behind the sentence

Ten real sentencing policies and practices. For each one, name the theory of punishment it primarily serves. The trick is that most policies serve more than one, so pick the theory that has to be true for the policy to make sense. If you removed that theory, the policy would have no argument left.

You named 0 of 10.
The axis that sorts them fastest. Retribution looks backward at the act and asks what is deserved. Deterrence and incapacitation look forward at risk and ask what prevents the next one. Rehabilitation looks inward at causes and asks what can be changed. Restoration looks outward at relationships and asks what repairs the harm. On an exam question, find the direction the policy is looking and you have narrowed five choices to two.

Where the numbers push back

Three places in this chapter where a policy built on a theory met the evaluation literature and did badly. None of these settles the moral argument, because retribution never claimed to reduce crime in the first place. They matter because each of these policies was sold on a crime-reduction promise, and a promise is a testable claim.

One. Three strikes laws

Three strikes is incapacitation written into statute. Beginning with Washington in 1993, twenty three states and the federal government adopted some version within three years. The largest evaluation looked at 188 American cities with populations over 100,000 across 1980 to 2000, comparing 110 cities in three strikes states against 78 cities in states without such a law, and testing separately for a deterrent effect at passage and an incapacitative effect building over time.

The finding. No credible evidence of crime reduction through either mechanism. The coefficients were small and mostly not close to statistical significance. The authors also reported that homicide rates in adopting cities rose about 10.4 percent on average afterward, which is a striking result and should be handled carefully. It is an association in their models rather than an established causal effect, and the authors treat it as a reason to doubt the deterrent story rather than as proof that the laws killed people.

Kovandzic, T. V., Sloan, J. J., and Vieraitis, L. M. (2004). Striking out as crime reduction policy. Justice Quarterly 21(2), 207 to 239.

Two. What happened when California let some of them out

In 2012 California voters passed Proposition 36, which narrowed the three strikes law and allowed people already serving life for a non-serious, non-violent third strike to be resentenced. More than 2,200 people were released. They had served a median of fifteen years, and 88 percent of them were over 45 at release. If selective incapacitation identifies the dangerous few, these were the people the system had identified. Guess how many were convicted of a new crime within three years.

Percentage convicted of a new crime within three years of release. For reference, the figure for California's general release cohort was 42 percent.
50%
Released under Prop 36
25%
General release cohort
42%

Skog, A. and Lacoe, J. (2025). Three Strikes Resentencing under Proposition 36 (2012). California Policy Lab. Three-year outcomes, 2,217 people resentenced and released. Arrest rates were 39 percent against 67 percent, and returns to prison 12 percent against 17 percent. Nearly two thirds of the new convictions were misdemeanors, against 26 percent in the comparison group.

Now say what this does and does not show. It does not show that these men were harmless when they were sentenced. It shows that by the time they were released they were mostly past the age at which people commit crimes, which is the age-crime curve from Chapter 5 doing exactly what it always does. That is the structural problem with selective incapacitation. Prediction happens at sentencing, the sentence runs for decades, and the risk it was predicting decays on a schedule the sentence does not know about. You end up paying $127,788 a year to confine a risk that expired.

Three. The ultimate punishment and the deterrence question

Capital punishment is where deterrence claims are made most loudly, so it is where the evidence has been examined most carefully. In 2012 the National Research Council convened a committee to review three decades of studies on whether the death penalty deters homicide.

The committee's conclusion, in its own words. Research to date on the effect of capital punishment on homicide is not informative about whether capital punishment decreases, increases, or has no effect on homicide rates. The committee went further and said the existing studies should not be used to inform policy at all. That is not a finding of no effect. It is a finding that the question has not been answered by the research that claims to answer it.

Public support has moved anyway. Gallup recorded 52 percent in favor of the death penalty for murder in October 2025, the lowest reading in its trend since 1972, down from a peak of 80 percent in 1994. Among adults aged 18 to 34 it was 41 percent.

National Research Council (2012). Deterrence and the Death Penalty. National Academies Press. Gallup, October 2025, reported by the Death Penalty Information Center. Retrieved August 2026.

Mandatory minimums, for the record. In federal fiscal year 2024, 24.3 percent of cases carried a mandatory minimum penalty and 15.9 percent of people sentenced remained subject to one. Their average sentence was 157 months. For people whose offense carried no mandatory minimum it was 31 months. Nearly seventy percent of mandatory minimum cases were drug trafficking. Whatever else a mandatory minimum does, it moves sentencing power from the judge to the prosecutor who chooses the charge, which is a Chapter 11 point about who gets to apply a label.

Take this to the forum

This week you sentence someone. Use the offense you have been writing about all term, and write as the judge rather than as an advocate. The point is not to be lenient or severe. The point is to be explicit about which theory you are standing on, and honest about what standing there costs you.

Your forum prompt. Take the offense from your term paper. If you were the judge, which two theories of punishment would drive your sentence, and what would the sentence actually look like? Name the theories, justify the mix, and say what each one sacrifices.

Three things that will make the difference between an adequate post and a strong one.

One. Make the sentence concrete. Not more punishment or a focus on treatment, but a number and a structure. How long, in what setting, with what conditions, and what happens at release. A theory of punishment that cannot be turned into a sentence is not doing any work.

Two. Name the sacrifice. Every pair of theories gives something up. Retribution plus incapacitation gives up on the person and buys certainty about the next few years. Rehabilitation plus restoration gives up proportionality, because the sentence now depends on what the person needs and what the victim wants rather than only on what they did. Say which trade you made and why you can live with it.

Three. Put one number in. You have a page full of them. The cost of a year of confinement, the share of that budget that goes to programs, the benefit-cost figure for a program you would order, a CrimeSolutions rating, the Prop 36 recidivism comparison. One well-chosen number, cited, will do more for your argument than a paragraph of conviction.

Worth knowing before you write. Gallup asked Americans in October 2025 whether more resources should go to addressing social and economic problems such as drug addiction, homelessness and mental health, or to strengthening law enforcement. Sixty seven percent chose the social and economic problems and 29 percent chose law enforcement, and that gap has widened since 2020. If you argue for a treatment-heavy sentence, you are not arguing against public opinion, whatever the political rhetoric suggests. That is a useful thing to know and a slightly uncomfortable one, because it means the system's spending pattern is not simply a reflection of what people want.

Gallup, Crime survey, October 1 to 16, 2025, 1,000 adults nationwide. Retrieved August 2026.

One more thing, about the paper. Part 4 asks you to apply at least two course theories to your offense and evaluate which the evidence supports better. That is the same intellectual move as this forum post, made once about causes and once about consequences. If you write the forum post carefully, a good deal of Part 4 will already be drafted.