On the Effectiveness of Prison as Punishment




Stuart Henry, Ph.D.





Professor and Chair of Interdisciplinary Studies,

College of Urban, Labor and Metropolitan Affairs

Wayne State University
































Paper presented at the Conference: Incarceration Nation: The Warehousing of America’s Poor, October 24, 2003, Ivy Tech State College, South Bend, Indiana.



The issue of punishment is emotionally charged, misunderstood and nothing if not uncertain in its outcome.  The central question is “Does Punishment Work?”  The answer clearly depends on what is meant by “punishment” and what is meant by “work.”  Also important, is for whom is punishment effective, and by what mechanism?  For example, are potential offenders likely to be affected by, even deterred by the existence of punishment? Are the kinds of behavior that are changeable though punishment the same as those which we classify as crime? In other words, are all behaviors equally able to be reduced or eliminated by the use of punishment?  In addition, is the effective mechanism in the use of punishment its severity or the effectiveness/certainty of punishment as a consequence for the behavior? Finally, is prison an effective method of punishment for crime, or are other methods more effective in bringing about reductions in crime and changes in harmful behavior? 

The empirical question of the effectiveness of punishment is considerably clouded by the commonsense view of the efficacy of punishment. Most people believe punishment works because they use it in their everyday life, with their children, their co-workers and their pets, and, are themselves subject to it. Indeed,

“punishment is what most of us do. . .first. It is our teaching legacy passed down from generation to generation. We are virtually surrounded by punishing strategies used to influence our behavior: From overdue library books to dogs without licenses; fines, penalties and reprimands whirl around us like leaves in a storm. For many of us, to give up punishment as our primary tool with which to influence negative behavior is to leave us empty handed.” (Friedman and Brinker, 2001: 1).


 Not surprisingly, the populist pro-punishment stance is reflected in public opinion surveys on the value of punishment, particularly extreme punishments like prison and the death penalty.  According to punishment advocates like Morgan Reynolds of the Texas-based National Center for Policy Analysis:

The answer is obvious to most Americans -- yes, of course punishment reduces crime. Punishment converts criminal activity from a paying proposition to a nonpaying proposition, at least sometimes, and people respond accordingly. We all are aware of how similar incentives work in our lives, for example, choosing whether or not to drive faster than the law allows . . . Incentives matter, including the risks we are willing to run. This is only a commonsense observation about how people choose to behave . . . . Public opinion strongly supports the increased use of prisons to give criminals their just desserts. The endorsement of punishment is relatively uniform across all groups. More than three-quarters of the public see punishment as the primary justification for sentencing. More than 70 percent believe that incapacitation is the only sure way to prevent future crimes, and more than three-quarters believe that the courts are too easy on criminals.  Three-quarters favor the death penalty for murder. (Reynolds, 2000)


Notice, however, that these beliefs are based on two conservative philosophies: (1) “incapacitation” i.e. that prison stops offenders incarcerated from offending while they are in prison, and (2) “just deserts” that those who commit crime deserve to be punished.  The question of whether punishment is effective, is much more problematic, even though many, particularly economists who study crime, such as Isaac Ehrlich and Gary Becker, and criminological advocates from the political right, such as Charles Murray (1997) claim to have demonstrated evidence that it reduces crime by deterrence.  So what do we know about the effectiveness of punishment, and in particular, what do we know about prison as an effective punishment?  To answer this we need to first define punishment.


Defining punishment and how it works

What is punishment?  Most psychologists define punishment as a process of presenting a consequence, delivered after a behavior, which serves to reduce the frequency or intensity with which the behavior occurs (Lefton, 1991).  The consequence, i.e. the punishment, can either be providing an undesirable stimulus or removing a desirable stimulus.  In either case the idea is that punishment decreases the probability of the behavior occurring again. 

There are two mechanisms that explain how punishment works to suppress unwanted behavior. The economic model is that of rational calculation: punishment is painful and therefore a cost which will be avoided; if the punishment is associated with the behavior, avoidance of pain is accomplished by avoidance of the behavior. The psychological model is that of learning though conditioned responses: repeated associations made between a behavior and aversive stimulus or removal of desired stimulus or positive reinforcer can, over time, lead to an “automatic” learned response to avoid the behavior.

It is also important to appreciate that punishment can vary in intensity along a continuum from mild to severe: “Punishment is not one single strategy but a collection of strategies that exist on a continuum from very mild to highly aversive approaches. Given our definition of punishment as a behavior-reducing technique, it is important to understand the nature of this continuum.” (Friedman and Brinker, 2001)


Punishment effectiveness

The effectiveness of punishment relates to how far it is successful in suppressing the undesired behavior. Effectiveness depends upon practices that work in general, and those that work with specific populations; the effects are not necessarily the same.

Psychological research on punishment in has shown that mild punishment can be effective in changing behavior, but the evidence is less clear about the effectiveness of severe punishment. Effectiveness of punishment is increased by:

1.      Frequency of application

2.      Immediacy of application

3.      Punishment used in conjunction with positive reinforcement of pro-social behavior


However, punishment, especially in its severe form has several negative effects:

1.      Avoidance or escape

2.      Alienation of those punished, to the point of inaction

3.      Aggressiveness, both targeted and generalized, by those punished

4.      Conditioning of the punishers through rewarding them for behavioral change

5.      Reproducing punishment behavior in those punished


Crucial seems to be the relationship between effectiveness and severity.  In criminal justice this translates into whether certainty of apprehension is more effective than the severity of punishment. Clearly if there was a severe punishment but no one got caught, the likely effect on behavior change because of the punishment would be low. Alternatively, having a high certainty of getting caught with no consequences is not likely to prove effective (although, because of the apprehension effect, it still has some ability to depress the occurrence of unwanted behavior). An optimization of a low to moderate punishment combined with a moderate to high certainty of apprehension seems to be the most effective combination. In criminal justice terms this translates to increased levels of policing. Evidence shows that increased levels of policing produce a substantial reduction in crime over time (Kovandzic and Sloan, 2002; Levitt, 1997; Marvel and Moody, 1996), and that this effect is enhanced when police efforts are targeted at certain problem areas.  Indeed, even advocates of punishment, like Reynolds concede this point:

Scholars also ask which provides the greater deterrent, certainty or severity of punishment? One provocative study involving prisoners and college students came down firmly on the side of certainty. When tested, both groups responded in virtually identical terms. Prisoners could identify their financial self-interest in an experimental setting as well as students could. However, in their decision-making, prisoners were much more sensitive to changes in certainty than in severity of punishment. In terms of real-world application, the authors of the study speculate that long prison terms are likely to be more impressive to lawmakers than lawbreakers. . . . Supporting evidence for this viewpoint comes from a National Academy of Sciences panel which claimed that a 50 percent increase in the probability of incarceration prevents about twice as much violent crime as a 50 percent increase in the average term of incarceration. (Reynolds, 2000)


            While research by Charles Murray (1997) and the NCPA has claimed that increasing the probability of severe sentences for crimes can reduce the incidence of crimes committed, there is little consensus among criminologists about this and general agreement that just increasing the severity of punishment for specific crimes does not reduce their occurrence.  Murray, in his 1997 paper “Does Prison Work?” argues that the crime rate declines in relation to the sentence per crime and rises when the sentence per crime falls. He says that in the US the 10-year decline in the crime rate from the period 1990-2000, at 5% per year, resulted in one third less assaults, 50% less burglaries and 66% less robberies, and 75% less auto theft. This correlated with a substantially higher rate of imprisonment per population and a rising rate of imprisonment per recorded crime. He concludes that the crime reduction in the US is a direct result of the policy of imprisonment, which has both a deterrence effect and an incapacitation effect. Moreover, he believe that offenders should be incarcerated even when the crime rate is low, as prison sentences can stop rising crime and reverse the trend.  Others have applied Murray’s findings (Saunders and Billante, 2002-3) and drawn similar conclusions. 

Critics charge that demonstrating a correlation is not establishing a cause; that rates are aggregate for the US and ignore important state differences; that crime rates have risen again after 2000 while the incapacitation and sentence/crime imprisonment/population both remained high; and that the reliance on official crimes known to the police and use of index crime rather than all crime, make the analysis highly suspect.  Others point to at least 14 explanations for the decline in the US crime rate during the period covered including: (1) a strong economy and low unemployment; (2) a decline in the nature of drug markets, especially a change in the crack-cocaine market; (3) a reduction in the number of young males in the population as the 1960s birth cohort matured; (4) increased law enforcement budgets under the Clinton presidency; (5) increased adoption of community policing strategies.  Clearly whether crime was reduced as a result of higher prison sentences or other factors is an open question, although the combination of factors probably had a significant impact. However, we should be extremely cautious since the improved economy argument is potentially very significant, especially for the rational choice argument. Most research has consistently shown that unemployment is correlated with increased levels of crime. In particular, if the rational choice argument is believed, that during full employment potential offenders commit less recorded crime, then why is it not also likely that potential offenders among the formerly unemployed would, when employed, choose to commit their crime in the relatively unpoliced privacy of their work place.  Here they can engage in property offenses, with virtually no risk of prosecution and incarceration, rather than the relatively high risk/cost arenas of the street?  Thus the rational choice model here would predict that there would be a reduction of officially recorded crime during high employment, not because of severe punishments, but because of a shift of crime reward opportunities from the risk-averse public arena to the risk-free workplace.  In addition, the whole question of whether a prison is an effective punishment, begs the question of whether prison is a punishment is perceived by potential offenders as a cost, and if so, to whom and for what kinds of behavior?


Prison as an effective punishment?

There is no question that prison is seen as a severe punishment for most people. The critical question is whether it is an effective punishment for potential offenders.  This depends on what motivates potential offenders. The deterrence argument is based on the arguments of economic rational choice theory and the classical assumption that offenders are self-interested, reasoning, rational cost-benefit calculators. However, much of the criminological literature has demonstrated that there are a variety of motivations that shape criminal activity ranging from biological predispositions, psychological personality traits, social learning, cognitive thinking, geographical location and the ecology of place, relative deprivation and the strain of capitalist society, political conflict and social and sub-cultural meaning.  The result is that most criminologists reject the arguments of pure rationality contained in Ehrlich and Becker’s utility and wealth maximization theories. Even those like Clarke and Cornish, who favor the rational choice argument, advocate the idea of “limited rationality.”  Indeed, as supporters of Murray’s argument are forced to concede: “The economic theory of crime that has developed out of Becker. . .recognizes that different individuals break the law for different reasons, that not all law breakers are rational utility maximizers, and that different offenders will weight the risks of benefits in different ways.” (Saunders and Billante, 2003: 4). So, who are the offenders who are supposedly influenced to reduce their commission of crime by deterrence through the severity of prison as a punishment?  To answer this question we need to examine who are prisoners, and what are their crimes.


Prisoners and their Crimes Punished in US State Prisons

There are four key facts on prisoners and their crimes: (1) demographics, such as the gender, race and age of prisoners; (2) their level of education; (3) the nature of their offenses, and (4) their  criminal history. Data for 1997 shows the following:


Source: BJS, Sourcebook, 2001


Source: BJS, Sourcebook, 2001


Source: BJS, Sourcebook, 2001

Thus the data show us that 67% of the state prison population are male, Black, Hispanic or other non-white and that 86.8% are aged between 18 and 44; in other words, young minority males.  Moreover, the data also shows that this population is relatively illiterate compared with the US population as a whole, with 40% functional illiteracy rate compared with 21% among the population as a whole.

Source: Open Society Institute, Research Brief, Sept 1997


One might conclude that those incarcerated are less likely to be rational, cost-benefit calculators. Indeed, a look at incarcerated offenders criminal history supports exactly this point.  Data shows that the national re-arrest rate is 63%, although can be as high as 84% for juveniles (Open Society, 1997) but that 76% of the state prison population has a previous criminal history of prior convictions. The data for 1997 shows that almost half of those with prior convictions are for violent offenses. Importantly, 59% of recidivists have more than two previous convictions and 43% have more than 3 convictions.



Criminal History of State Prisoners, 1997

Prior Convictions

Percent of State Prisoners













Source: BJS Sourcebook of Criminal Justice Statistics, 2001

Clearly, the threat of prison as punishment did not work for the majority of these offenders. This picture of the deterrence effect of prison as punishment is further undermined when examining the kinds of crimes that those in state prison have committed.


Percent prison population







Public Order


Source: BJS Sourcebook of Criminal Justice Statistics, 2001

Of these offenses, only 22% are property crimes, the most likely to be deterrable; the rest are violent or drugs crimes, the least likely to be deterrable since they are typically motivated by irrational, expressive acts, or are the result of addiction or behavioral and personality problems.  Indeed, only 10.7% of crimes are “burglaries” which are those most frequently cited as “deterrable” by prison as punishment advocates.  In short, neither the majority of offenders nor the types of offenses committed by state convicted prisoners are of a kind that are the outcome of cost-benefit calculations that take account of the potential prison sentence, prior to the act.


If prison is not effective as a disincentive to offending, what is?

Research over the past 10 years has consistently demonstrated that the most effective way to reduce offending, and particularly reoffending is through education, particularly literacy training and GED (Steurer, Smith, and Tracy, 2001). An Arizona Department of Adult Probation Study showed that probationers who received literacy training had 35% rearrest rate compared with a control group that had 46% rearrest, and those who received a GED had a rearrest rate of 24% (Siegal, 1997). Less dramatic but equally encouraging results were received from a Florida study of 18,414 inmates released from prison in FY1996-97 followed up after 2 years, which found that “inmates who earn a GED are 8.7% less likely to recidivate than those who do not complete a program. . . Inmates who receive a GED and improve their TABE score to 9th grade level or higher are 25.0% less likely to recidivate than those who receive a GED and have a TABE level of 8th grade or less.” (Florida Department of Corrections, 2003). The Florida study also found thatAcademic program impacts are found even among offender groups that normally have higher recidivism, for example, males, younger males, black offenders and prior recidivists.”  Importantly, a New York State study found that “young inmates who earned a GED while incarcerated returned to custody at a rate of 40% compared with 54% of inmates under 21 released with no degree” (Staley, 2001). Most dramatic, however, is the data on those in prison: Inmates with at least two years college education have a 10% re-arrest rate, compared to the national rearrest rate of 62%. A Texas study is most revealing showing that the overall recidivism rate for degree holders in the Texas Department of Corrections between 1990-1991 was 15% compared to 60% for the national rate and a two year follow-up study showed that those with associates degrees had a recidivism rate of 13.7%, those with bachelor’s degrees, 5.6%, and those with master’s degrees zero (Tracy and Johnson, 1994):

Source: Open Society Institute, Research Brief, Sept 1997, based on Tracy and Johnson, 1994


So, if the evidence is clear that prison as punishment is ineffective in deterring offenders, but education makes a substantial difference to recidivism, why do we continue to use prison as punishment? Moreover, why did we stop using education, particularly college-level education, for prisoners?  The analogy of criminal justice and social policy as a “toolbox” comes to mind (Einstadter and Henry, 1995). We have many “tools” each refined for serving different functions. Just as a screwdriver, hammer, saw, wrench serves different functions to solve technical problems, so various policy options are available to deal with crime problems, whether this is biologically based treatment, psychologically based therapy, sociologically based education and training, and economically based punishment. However, it seems that policy makers peering into the justice toolbox only see one tool, the hammer of punishment, and they try to use it to fix everything.  Imagine what would happen if your plumber showed up to fix a leak and all he had was a hammer. Imagine if you took your car to be serviced and all they had was a hammer! Why, given the bio-social, psycho-political complexity of human beings do we restrict our policy to this one-dimensional approach. It makes no sense.  Indeed, we seem to have a better appreciation for how to bring behavioral change in parrots than in people. As noted parrot experts say:

There will always be many unknowns about behavior; there will always be important variables that are out of our control. Behavior is just too complex for simplistic cookbook approaches to mentoring our birds . . . Each situation is unique and requires careful analysis and informed consideration. Facilitating well-adjusted, independent, confident companion parrots through the use of positive teaching techniques is more than just a commitment to learning new strategies; it is also a commitment to changing our legacy. The time for such change is now. (Friedman and Brinker, 2001)

So what are the policy implications of the prison-as-punishment does not deter crime conclusion? First, we need to consider ceasing to use prison as punishment. Incapacitating the most seriously harmful offenders is a different argument. Second, we should draw on the research of what we know works to prevent recidivism, especially literacy programs, skills training and GED, as well as educating prisoners to associate’s degree level in higher education and restore financial support for these successful practices. Third, we should train corrections officers to be corrections officers rather than guards, and if that means training them to be effective and qualified teachers, then this will be money well spent.  Fourth, we should invest the money spent on incarceration on ensuring that the illiteracy rate among the nation’s population is reduced dramatically. Doing so will ensure that our general population is equipped to make the very kind of rational choice decisions that will enable them to make better choices in the first place. Finally, we should abandon the discourse of punishment as our response to unwanted behavior. It doesn’t work for parrots and it doesn’t work for people.



BJS (2001). Sourcebook on Criminal Justice Statistics. Washington DC: U.S. Government Printing Office.


Einstadter, W.J and Henry S. (1995). Criminological Theories: An Analysis of its Underlying Assumption, Forth Worth, TX: Harcourt,Brace and Co.


Friedman, S.G., and B. Brinker, (2001). “The Facts About Punishment”

Original Flying Machine, Issue 4: Jan/Feb http://www.parrottalk.com/punishment.html


Florida Department of Corrections. (2003) Academic, Vocational, and Substance Abuse Program Impacts. http://www.dc.state.fl.us/pub/recidivismprog/execsum.html


Kovandzic, T.V. and J. J. Sloan (2002).  “Police Levels and Crime Rates Revisited: A County-level Analysis from Florida (1980-1998).” Journal of Criminal Justice 30 (2002) 65-76


Lefton, L. A. (1991). Psychology. Boston: Allyn and Bacon.


Levitt, S. (1997).  “Using Electoral Cycles in Police Hiring to Estimate the Effect of Police on Crime.” American Economic Review, 87: 70-91.


Marvell, T. and C. Moody (1996). “Specification Problems, Police Levels, and Crime Rates.” Criminology, 34: 609-46.


Murray, C. (1997). “Does Prison Work?” Institute of Economic Affairs. London: Coronet Books.


Open Society Institute (1997). “Education as Crime Prevention: Providing Education to Prisoners” Center on Crime, Communities & Culture Research Brief No 2. http://www.soros.org/crime/research_brief__2.html


Reynolds, M. (2000). “Does Punishment Work to Reduce Crime?” National Center for Policy Analysis, Congressional Testimony before the United States House of Representatives Committee on the Judiciary, Subcomittee on Crime. http://www.ncpa.org/press/transcript/mor100200.html


Saunders, P. and N. Billante. (2002-3). “Does Prison Work?” Policy, 18(4): 3-9.


Siegal, G.R. (1997). A Research Study to Determine the Effect of Literacy and General Educational Development Programs on Adult Offenders on Probation. Tucson, AZ: Adult Probation Department of the Superior Court in Pima County.


Staley, E. M. (2001). “Follow-Up Study of a Sample of Offenders Who Earned High School Equivalency Diplomas (GEDs) While Incarcerated in DOCS” http://www.prisonpolicy.org/scans/ny_ged.shtml


Steurer, S., Smith, L., and Tracy, A. (2001). Three State Recidivism Study. Lanham, MD: Correctional Education Association.

Tracy, C. and C. Johnson, (1994). Review of various outcome studies relating prison education to reduced recidivism. Windham School System: Huntsville, TX.