BJS: Bureau of Justice Statistics

Home  |  About Us  |  Contact Us  |  Help  |  A-Z Topic List
 
 
Bureau of Justice Statistics (BJS)
Home  |  Publications & Products

Crime and Justice in the United States
and in England and Wales, 1981-96

Justice system's impact on crime

Is there a connection between trends in legal punishment and trends in crime in the two countries?

The two countries differ greatly in how their justice systems responded to crime throughout the 1980's and continuing into the 1990's. For example, during that time an offender's risk of conviction rose in the United States but fell in England (including Wales). Such differences in punishment trends might help explain why crime trends since 1981 differed between the two countries. In theory, raising the risk or severity of punishment might lead to crime decreases, and lowering the risk or severity of punishment might lead to crime increases.

To investigate these possibilities, correlations were computed between punishment trends and crime trends in the two countries. Negative correlations (for example, a falling conviction rate and a rising crime rate) were interpreted as possible support for the theory. Correlations dealt with two separate types of punishment trends: trends in the risk of punishment, and trends in the severity of punishment. Two measures of punishment risk are the conviction rate (defined as the number of convictions per 1,000 alleged offenders) and the incarceration rate (defined as the number of incarcerations per 1,000 alleged offenders). Four measures of punishment severity are the percent of convicted offenders sentenced to incarceration, sentence length, time served, and percent of sentence served. A fifth is "days of incarceration at risk of serving," although this measure actually combines elements of both risk and severity.

U.S. trends were based on data for seven points in time (1981, 1983, 1986, 1988, 1990, 1992, and 1994); English trends were based on six (1981, 1983, 1987, 1991, 1993, and 1995). Detecting a statistically significant relationship between crime and punishment trends is difficult when trends are based on so few points in time. Consequently, statistical significance was not given more weight than other criteria for evaluating results. Other criteria used were strength and direction of correlations between punishment trends and crime trends, and consistency of correlations across offense categories.

Major findings were:

  • Negative correlations in England between trends in punishment risk and crime trends offer the strongest support for the theory that links falling risk of punishment to rising crime (table 2). Specifically, since 1981 the conviction rate fell in England, and English crime rates (both police-recorded crime rates and crime rates from victim surveys) rose (figures 1-10 and figures 25-30). Likewise, the incarceration rate fell, and English crime rates (both police-recorded rates and victim survey rates) rose (figures 1-10 and figures 43-48).

  • In England, correlations between punishment severity and crime trends were mixed (table 2). Roughly half were positive and half were negative. The major exception was motor vehicle theft: correlations were fairly consistently strong and negative between trends in punishment severity (however measured) for motor vehicle theft and trends in the rate of vehicle theft (however measured) (figures 4, 10, 36, 54,60, 66, and 72). Specifically, the percent of convicted motor vehicle thieves sentenced to incarceration, their average sentence length, their average time served, the percent of sentence they served, and the number of days of incarceration they were at risk of serving all fell since 1981. At the same time, the motor vehicle theft rate, as measured in both victim surveys and police statistics, generally rose.

  • In the United States, correlations between punishment risk and crime trends were mixed (table 2). About half were positive and half were negative. Moreover, negative correlations were often low. Furthermore, correlations between trends in punishment risk and trends in crime were predominantly negative when crime trends were measured with victim surveys but predominantly positive when measured with police statistics. In short, trends in punishment risk had an inconsistent relationship with trends in crime in the United States. The major exception is burglary, where there were consistent negative correlations: the risk of punishment (whether measured by the conviction rate or the incarceration rate) rose, and the burglary rate (whether measured in victim surveys or police statistics) fell (figures 3, 9, 29, and 48).

  • In the United States, correlations between punishment severity and crime trends were mixed (table 2). Approximately half were positive and half were negative. Moreover, in instances where there were negative correlations, they were often weak. Furthermore, unlike results from England, correlations between punishment severity and survey crime rates often had a different sign than correlations between severity and police-recorded rates for the same crime. In short, trends in punishment severity had an inconsistent relationship with trends in crime in the United States. The major exception is burglary, where there were consistent negative correlations: for most measures of severity (percent of convicted offenders sentenced to incarceration; sentence length imposed; time served), increases in severity of punishment for burglary were associated with decreases in the burglary rate regardless of whether the burglary rate was measured using victim surveys or police statistics (figures 3, 9, 35, 53, and 59).

To summarize, notable consistencies and notable inconsistencies both characterized results relating punishment trends to crime trends. Notably consistent was the close association in England, across the different crime types, between falling risk of punishment (however measured) and rising crime rates (however measured). Notable inconsistent results were those between England and the United States; between trends in punishment risk versus trends in punishment severity; between police statistics and victim surveys; between different measures of punishment severity; and between different offenses. The major exception was burglary, where trends were fairly consistent irrespective of country, source of crime-rate data, or type of punishment trend.

Possible explanations for the inconsistencies are:

  • Victim surveys may provide a more reliable measure of crime trends than police statistics. If so, that may explain inconsistencies between victim surveys and police statistics.
  • Changes in the risk of punishment are widely thought to have a greater impact on crime rates than changes in punishment severity. If so, that may explain why punishment risk trends and crime trends were more consistently associated with one another than were punishment severity trends and crime trends.
  • Most U.S. crime rates fell in the early 1980's, increased until the early 1990's, and then fell again. Yet linear correlation was used to analyze these nonlinear trends. Perhaps nonlinear correlation would show a closer association between punishment trends and crime trends in the United States.
  • The fact that all trends were based on a small number of points in time (seven in the United States, six in England) suggests a more general explanation for inconsistencies. That number of data points may be adequate for documenting a relationship between punishment trends and crime trends only if major changes occur in punishment trends during the study period in both countries, which was not the case here. English conviction rates, for example, declined sharply during the study period. The increase in U.S. conviction rates was modest by comparison. Consequently, the negative correlations between rising U.S. conviction rates and falling crime rates were relatively modest for most crime rates derived from victim surveys, whereas the negative correlations between falling English conviction rates and rising English crime rates were uniformly strong. The implication is that punishment trends and crime trends should not always be expected to have the same relationship in two countries over any period of time.
  • Some crimes (such as burglary) are more rationally motivated than others (assault, for example). Consequently, in comparison with other crimes, those that are committed by more rationally motivated offenders -- by persons who, for example, plan their crime and weigh their chances of being caught -- are probably more influenced by increases or decreases in the likelihood or severity of punishment. The implication is that punishment trends and crime trends should not always be expected to have the same relationship irrespective of type of crime.
  • A positive correlation between punishment and crime trends was interpreted as possible evidence that increasing punitiveness does not reduce crime. Such an interpretation may not always be justified. For example, if the crime rate rose over some period of time but was kept from soaring by increasingly punitive policies over that period, it would be a mistake to interpret the observed positive correlation between punishment and crime trends as evidence that increasing punitiveness had no crime reduction benefit. Perhaps some of the inconsistent findings described above stem from misinterpreting positive correlations. By the same token, perhaps some of the inconsistencies stem from misinterpreting negative correlations. Interpreting a negative correlation as possible evidence that increasing punitiveness reduces crime may not always be justified. For example, crime rates can fall for reasons having nothing to do with increasing punitiveness. To illustrate, demographic changes in the age and race composition of the U.S. population might explain 41% of the drop in the U.S. murder rate from 1981 to 1996; 47% of the drop in the police-recorded U.S. robbery rate; and 19% of the drop in the police-recorded U.S. burglary rate.



Table 2. Correlations between 1981-1994 trends in U.S. crime rates and U.S. trends in legal punishment; and correlations between 1981-1995 trends in English crime rates and English trends in legal punishment
 
                   


  Correlation between crime rate trends and trends in --  


  Risk of punishment Severity of punishment


  Conviction rate:
convictions per
1,000 alleged
offenders
Incarceration rate:
incarcerations
per 1,000
alleged offenders
Percent of
convicted
offenders
sentenced to
incarceration
Average
incarceration
sentence
imposed
Average
time
served
Percent of
sentence   served
Days of
incarceration
at risk
of serving
   


United States' crime rate        


Police-recorded rate of --        


Murder -.031    -.015    .178    .206    -.162    -.252    -.040   


Rape .844*   .835*   -.573    .487    .776*   .389    .865*  


Robbery .215    .288    .370    -.153    -.045    .207    .169   


Assault .950**  .957**  -.347    .267    .277    -.299    .935** 


Burglary -.835*   -.841*   -.830*   -.518    -.334    .632    -.789*  


Motor vehicle theft .774*   .791*   .389    .544    -.100    -.784    .729   
                   


Victim survey estimated rate of --        


Robbery -.606    -.595    -.111    -.581    -.440    .696    -.612   


Assault -.146    -.132    .267    -.421    -.449    .319    -.178   


Burglary -.826*   -.880**  -.899**  -.680    -.524    .687    -.874*  


Motor vehicle theft .546    .583    .409    .518    -.208    -.790    .517   
                   


England's crime rate -        


Police-recorded rate of -        


Murder -.907*   -.483    .782    .641    .638    .241    .482   


Rape -.953**  -.968**  .434    .969*   .933**  .613    -.699   


Robbery -.868*   -.852*   -.811    .896*   .897*   .437    .030   


Assault -.901*   -.684    .654    .775    .614    -.866*   -.475   


Burglary -.934**  -.956**  -.019    .815*   .711    -.686    -.967** 


Motor vehicle theft -.883*   -.899*   -.331    -.672    -.974**  -.792    -.890*  
                   


Victim survey estimated rate of -        


Robbery -.981**  -.964**  -.775    .681    .712    .533    -.317   


Assault -.995**  -.790    .826*   .544    .379    -.724    -.713   


Burglary -.970**  -.964**  .258    .759    .522    -.788    -.982** 


Motor vehicle theft -.943**  -.951**  -.301    -.533    -.987**  -.888*   -.951** 
                   


*Correlation is significant at the 0.05 level (2-tailed)    


**Correlation is significant at the 0.01 level (2-tailed)    
                   


Previous Contents Next