Criminal Justice System

Previous sections of this entry have described the structure and purposes of criminal justice systems, but it is equally important to examine how systems function in practice—which may be quite different from how they are commonly assumed to function. Perhaps the most striking finding derived from the available data on system operations is the very substantial case attrition between the start of the process (crimes committed) and the final stages of execution of sentence. This is not a new phenomenon, nor one limited to the United States; studies of American criminal justice in the early part of the twentieth century (Cleveland Foundation), and in several other Western countries (Frase, 2001, Table 3; Zeisel, p. 24), have reported similarly high rates of case attrition.

Citizens and their political leaders must keep these findings in mind when making criminal justice policy decisions. For most crimes, nothing even remotely approaching "full enforcement" has ever been achieved, nor is it likely to be achieved; it may even be undesirable. The high level of case attrition in all modern systems suggests that the criminal law has very substantial limitations as an instrument of direct crime control (by means of deterrence, incapacitation, and rehabilitation). Such limitations suggest that, in most cases, the primary value of criminal prosecution is symbolic. Criminal convictions and penalties impose deserved punishment (retribution), and reinforce important societal norms (denunciation)—but only if these penal consequences are imposed according to procedures that are widely perceived as fair and just (Tyler).

Case attrition results from the exercise of discretion by all system actors, both public and private, and from the nature of the criminal process itself. Several factors justify and require the exercise of discretion by public officials: (1) the practical inability of legislatures to specify in advance all of the conditions that properly bear on issues of criminal liability and the scope of criminal statutes, punishment, and criminal procedure (particularly when, as was noted earlier, basic goals and values often conflict with each other); (2) the need for case-specific assessments of the available evidence to convict an offender, as well as the available resources for investigation, conviction, and punishment of this and other offenders; (3) the desirability of taking into account local crime problems, community values, and the desires and needs of crime victims.

Citizens involved in the criminal justice process also exercise substantial amounts of discretion. Crime victims and witnesses may choose not to report a crime, or not to cooperate with prosecution. Suspects and defendants also have choices: to exercise their legal rights, rather than waiving them; to supply information helpful in convicting other offenders; and to comply with conditions of release or prison rules of conduct.

Apart from the exercise of discretion, another major reason why case volumes decline substantially as cases move through the system is that different standards apply at these stages: probable cause may be sufficient to justify an arrest or a search, with the hope of uncovering more evidence. But if such evidence does not become available within a reasonable time, cases must be dismissed or charged down; imposition of the severe social stigma and sanctions of the criminal law requires morally convincing evidence (proof beyond a reasonable doubt).

Statistics on case attrition at various stages. Many of the decisions made at crucial points in the criminal process are not easily observable, so it can be very difficult to determine what decisions are made, let alone the rationale for these decisions. For example, there is very little information regarding police and prosecutorial decisions to investigate. Likewise, there are few available statistics on decisions made in the correctional realm, such as parole and probation revocations, and sentence-reductions based on an inmate's good behavior in prison. However, data is available on decisions at several of the most important stages of the process.

Victim reporting of crime. Victim surveys reveal that substantially less than half of crimes committed are reported or otherwise made known to the police. The reporting rate is especially low for attempts and minor completed crimes, and is also probably very low for so-called victimless crimes (i.e., those with no immediate victim or witnesses other than the offender[s] and the police, such as drug offenses). On the other hand, the proportion of homicides that become known to the police (either by contemporaneous reporting or by discovery of the body) is probably close to 100 percent. For obvious reasons, neither victimless crimes nor homicide are included in victim surveys. These surveys also do not measure crimes against the environment, frauds, or crimes against businesses and government agencies.

In 1999 only 44 percent of violent crimes and 34 percent of property crimes were reported to the police (Rennison, p. 11). Table 1 illustrates the percent of crimes that victims claimed to have reported to the police, by crime type (including attempts).

Table 1 shows that the likelihood of a victim reporting an offense to the police varies significantly by crime type. Victim reporting is highest for auto theft (84 percent), probably because most auto insurance policies require police notification of a stolen vehicle. By contrast, personal thefts outside the home (e.g., a stolen bicycle) are reported to the police only about one-fourth of the time.

Police recording of crime. If the police doubt that any crime was actually committed they will "unfound" the report, take no further investigative action, and decline to include the report in their statistics of crimes known to the police. There are no official statistics on the rates Table 1 SOURCE: Rennison, p. 11. of police unfounding, overall or for particular offenses.

Proportions of crimes solved (clearance rates). The only official statistics on police crime-solution (clearance) rates are for the eight "index crimes" (listed in Table 2) that the Federal Bureau of Investigation (FBI) uses as an overall measure of crime levels. For 1999, the FBI reported that 21 percent of index crimes were cleared—including 50 percent of violent crimes and 17 percent of property crimes (FBI, 2000, p. 203). Regarding specific crime types, Table 2 presents the percentages of index crimes known to police departments in the United States that were cleared by arrest, another charging procedure (e.g., a summons or citation) or, occasionally, by attribution to a dead or otherwise unavailable suspect.

As is the case with victim reporting, there is great variation in clearance rates across types of index crime. More than two-thirds of murder and non-negligent manslaughter cases and almost three-fifths of aggravated assaults were cleared by arrest, whereas less than one in six burglary and motor vehicle theft offenses was cleared. These differences are attributable both to the higher priority that the police give to more serious crimes, and to typical offense patterns: violent crimes often involve offenders who are already known to the victim.

It is likely that clearance rates are lower for other (non-index) crimes, since the latter are generally less serious, and receive less victim, witness, and police attention. On the other hand, the proportion of "victimless" crimes cleared is probably very high, since such crimes are rarely known to the police unless they are committed in Table 2 SOURCE: F.B.I. (2000), p. 203, Table 25. the presence of a police officer or undercover informant.

It should be noted that clearance rate statistics are only roughly comparable to the victim-survey data reported previously. In particular, the surveys exclude offenses against victims under the age of twelve, exclude most property offenses committed against businesses, and include sexual assaults other than forcible rape.

Pretrial detention. Statistics on this critical issue are only available for defendants charged with felonies in a sample of the nation's seventy-five largest counties. In 1996 these counties accounted for 37 percent of the nation's population, 50 percent of serious violent crimes known to the police, and 40 percent of serious property crimes known to the police (Hart and Reaves, p. 1). Sixty-three percent of these defendants were released before case disposition and 37 percent were detained; one-sixth of those detained were denied bail, whereas five out of six failed to post bail. Those charged with a violent offense were less likely to be released (55 percent) than those charged with a property offense (65 percent) or a drug offense (66 percent). Table 3 shows the percent of those released and detained, and the type of release or detention, by the most serious arrest charge.

As Table 3 illustrates, the decisions whether to release suspects before trial, and whether to guarantee their appearance in court by setting bail, varied greatly across felony charges. Only 16 percent of those charged with murder were released before disposition, and 84 percent were detained; 61 percent were denied bail. In comparison, those charged with theft were released two-thirds of the time, and more than half of these releases were nonfinancial (not requiring Table 3 SOURCE: Hart and Reaves, pp. 16, 17, Tables 13 and 14. deposit of bail or other security). For all charges except murder, the majority of those who were released before trial were released within a week of being arrested.

In regard to the relationship between pretrial detention and subsequent case processing, separate analysis of this data by the present authors confirms the results of previous studies: detained defendants were more likely to be found guilty, and were also more likely to receive a custody sentence. At the same time, a significant portion of the detained did not receive a formal conviction and custodial sentence. For example, 34 percent of those charged with burglary who were detained for more than two days before trial either were not convicted or received a non-custodial sentence. For defendants charged with theft or with drug violations, the proportions were 38 and 36 percent, respectively. Thus, even where the system has officially decided that an individual should not receive custodial punishment, a de facto custody "sentence" is often imposed.

Prosecutorial screening and diversion. Despite the critical importance of prosecutorial discretion, there is almost no current national data on charging decisions (for the most recent data, covering eleven states as of 1990, see Perez). Table 4 presents rough estimates of prosecution rates, for selected offenses in 1996, derived by comparing the total number of felony cases filed in a sample of the nation's seventy-five largest counties with the total number of adult arrests for each offense in those counties that year.

Table 4 indicates that offense-specific prosecution rates vary a great deal—one out of five theft cases, three out of four cases of murder and non-negligent manslaughter, and over nine out of ten forcible rape cases. It is important to recognize that these numbers are aggregates, and thus do not represent the disposition of individual cases as they flow through the system. These figures also exclude offenders who were prosecuted for a different crime (usually a less serious one) than their arrest offense. Yet these admittedly inexact numbers are the only available estimates, given the paucity of national data tracking individual cases beginning at the arrest stage (compare Perez, cited above). (It should also be noted that the offense-specific prosecution-rate data above (as well as the conviction-rate data below) is only roughly comparable to the victim-survey and FBI data presented previously.)

Final dispositions: conviction rates. Nationwide data on these issues are, again, only available for certain jurisdictions, offenses, and years. Table 5 shows, by arrest offense, the percentages of those charged with selected felonies who were convicted (including conviction on lesser charges) in the nation's seventy-five largest counties.

Guilty pleas and plea bargaining. There are no national data on the practice of plea bargaining per se; information is limited to the prevalence of guilty pleas, and, once again, is further limited to certain jurisdictions, offenses, and years. Implicit in a defendant's decision to plead guilty is the assumption that the sentence will be more lenient than it would have been had the defendant Table 4 SOURCE: Hart and Reaves, p. 2, Table 1; Uniform Crime Reports data available from the Inter-University Consortium for Political and Social Research (ICPSR) web site (http://www.icpsr.umich.edu/). been convicted at trial, but the percent of plea agreements that are explicitly negotiated is unknown. The available data on felony dispositions in the nation's seventy-five largest counties show that, in 1996, 93 percent of felony charges resulting in a conviction (either felony or misdemeanor) were the result of a guilty plea, while only 7 percent were the result of a trial. Table 6 shows that guilty pleas are less likely when the stakes are highest. Murder convictions are almost as likely to result from a trial (47 percent of convictions) as from a guilty plea. By contrast, drug possession convictions were the product of a guilty plea in 97 percent of the cases.

Impact of method of disposition on case processing time. Table 7 shows the median time in days between arrest and sentencing, controlling for the manner of case disposition—trial ( jury or bench) versus guilty plea.

It is evident from Table 7 that, across conviction types, guilty pleas and bench trials are associated with much shorter case processing times than are jury trials. For example, regarding cases in which the most serious conviction offense was murder, those disposed with bench trials (191 days) took only half as long as those with jury trials (377 days), and two-thirds as long as cases disposed with a guilty plea. For other offenses, median days to disposition by plea were roughly the same as for disposition by bench trial, with time to disposition for jury trials markedly longer.

The slower case processing associated with jury trials is partly due to court backlogs, but another contributing factor is that many jury trials Table 5 SOURCE: Hart and Reaves, p. 24, Table 23. occur only after the parties have failed to reach a plea agreement—often after lengthy negotiations and attempts to "wear down" the other side. It is therefore quite possible that, if plea bargaining were curtailed and cases were either pled as charged or set for trial, average disposition times might actually decrease. Indeed, this is what happened in the only state (Alaska) that has ever attempted to sharply limit plea bargaining (Zimring and Frase, pp. 678–679).

Sentencing outcomes. Once again, national data on sentencing in the United States is limited to certain jurisdictions, offenses, and years. Table 8 shows the distribution of prison, jail, and probation sentences for selected felonies, by most serious conviction offense, in the seventy-five largest counties; it also reports the median sentence length for prison and jail sentences.

Table 8 shows that the total percent incarcerated varies greatly across conviction offense charges. For example, all of those convicted for murder received a prison sentence, whereas roughly one-third of those convicted on theft, drug possession, and weapons charges received probation. Among those who are incarcerated, the likelihood of getting a prison sentence as opposed to a jail sentence varied too. Those convicted of robbery were over three times more likely to receive a prison sentence than a jail sentence (71 percent versus 19 percent), whereas those convicted of drug possession were almost twice as likely to receive a jail sentence as they were to get a prison sentence.

Table 6 SOURCE: Hart and Reaves, p. 24, Table 23.

In light of the high proportions of custodial sentences reported in Table 8, it should be noted that many other Western countries appear to make much less use of custodial sentences, particularly for nonviolent crimes (Tonry and Frase, 2001; Frase, 2001).

Table 8 also shows that maximum sentence lengths are by far the longest for murder (median of 360 months), followed by rape (96 months). By comparison, those arrested for theft or drug possession had a median maximum sentence of only 24 months. Sentence lengths for those receiving a sentence of jail are markedly shorter since, in most states, jail sentences can be no longer than one year.

Actual time served in prison is almost always less than the maximum sentence imposed (usually much less). In 1996 it was expected that those convicted of violent felonies would serve 51 percent of their sentence (e.g., murder, 50 percent). In comparison, it was expected that those who were convicted of felony property offenses, as well as drug trafficking, would serve 42 percent of their sentences (Brown et al., p. 4, Table 4).

At the end of 1999, 3,507 prisoners were being held under sentence of death; 98 prisoners, in twenty states, were executed during that year (Snell, pp. 1, 7).

Case volumes at various stages (by offense, over time, and across jurisdictions). Table 9 shows the number of cases nationwide at several different stages in the criminal court process for six felony offenses that have the greatest comparability across reporting series. These numbers are aggregates, and do not represent individual cases as they flow through the system. Yet, they are the best available national data showing case volume (and therefore, case attrition) at several different points in the criminal court process. For example, in 1996 the police arrested more than 322,000 adults for drug trafficking (which is almost always a felony). In that year more than 212,000 adults were convicted of drug trafficking, and about 84,000 received a prison sentence. The last three columns in Table 9 show that in 1996, for every 100 adult arrests for drug trafficking, there were 66 felony convictions, and 48 custodial sentences (26 sentences to prison, and 22 to jail).

The screening decisions carried out at each successive stage of the criminal process result in dramatic reductions in case volume, as cases move through the system. For example, among the four felony offenses in Table 9 for which there is information on the number of offenses committed (robbery, aggravated assault, burglary, and motor vehicle theft), for every 1,000 felonies committed 444 were reported by the police, 71 adults were arrested, 18 adults were convicted, 13 adults were incarcerated, and 9 adults were imprisoned. Although data are not available, it seems likely that case attrition is at least as great for less serious crimes. In 1996, for example, there were about 12.3 million adult arrests in the United States (FBI, 1997, pp. 214, 224), but there were only about one million felony convictions in state criminal courts in that year (Brown et al., p. 2, Table 1). On the other hand, the attrition data above reflect cases, rather than offenders; since many offenders commit more than one offense, and thus are likely to be eventually caught for one of their crimes, it is likely that the proportion of offenders arrested, prosecuted, convicted, and imprisoned is somewhat higher than the numbers above would suggest.

Number of individuals in prisons and jails (overall and by state). As of midyear 1999, 1,254,600 persons were held in federal and state prisons, and 606,000 federal and state prisoners were held in local jails (Beck, 2000a, p.1). These inmate counts represented a national average of 460 adults per 100,000 residents incarcerated in state and federal prisons, and 222 per 100,000 incarcerated in local jails. However, as Table 10 shows, these national averages belie great variation across states. Southern states such as Texas (726 state prisoners per 100,000) and Oklahoma (625 per 100,000) imprison at much higher rates than some Midwestern and Northeastern states Table 7 SOURCE: Brown et al., p. 8, Table 11. (Minnesota, 121 per 100,000; Maine, 131 per 100,000).

Jail rates also vary greatly across jurisdictions. In 1999 Louisiana jailed at a rate (585 jail prisoners per 100,000) that was six and a half times higher than Maine's rate (89 per 100,000). Moreover, the relative use of prison and jail varied greatly. For example, whereas Missouri's prison rate was 3.6 times its jail rate, the ratio of prison use to jail use was much lower in most other states; indeed, two states (Louisiana and Tennessee) had substantially higher jail rates than prison rates. These dramatic variations in the rate and type of incarceration are only partially explained by variations in crime rates and criminal caseloads.

Number of individuals on probation and parole. At the end of 1999 there were approximately 3,773,600 adults on probation and 712,700 on parole (Bureau of Justice Statistics, 2000 (press release), p. 1). These caseloads represented an aggregate rate of 1,848 probationers per 100,000 residents, and 317 persons on parole per 100,000.

Trends in the number of individuals under criminal justice control. The last quarter of the twentieth century witnessed a sharp increase in all forms of correctional supervision—both custodial (prison and jail) and noncustodial (probation and parole). Figure 1 provides a graphic illustration of these increases (some figures have been scaled up or down, to facilitate trend comparisons). Figure 1 shows that, from 1977 through 1999, the number of adults on state or federal probation increased from 816,525 to 3,773,624—an increase of 362 percent. Over the same period, the number of adults on parole increased at a slightly lower rate (from 173,632 to 712,713, or 310 percent). State and federal year-end prison populations increased from 291,667 to 1,263,226 (up 333 percent), while average daily jail populations increased from an estimated 155,200 in 1977, to 607,978 in 1999 (up 292 percent). As with the state-to-state variations shown in Table 10, only some of these dramatic increases can be attributed to rising crime rates—as shown in Figure 1, adult arrests increased by only 50 percent during the same time period.

Appeals. It is estimated that about 122,000 criminal appeals were filed in 1996. More than two-thirds of these were mandatory appeals—cases that the appellate courts must hear as a matter of a defendant's right. The balance were discretionary petitions that appellate courts agreed to hear (National Center for State Courts, 1997, pp. 71, 74).

Criminal justice budgets. In 1997 total direct criminal justice expenditures (in billions) at federal, state, county, and city levels of government were, respectively: $20.5, $42.4, $31.6, and $35.3 (Bureau of Justice Statistics web site). Federal government expenditures had grown the most since 1982 (4.8 times higher in 1997), reflecting the dramatic increase in federal prison populations (which rose even faster than state prison populations during this period). State and county expenditures were about four times higher in 1997 while city expenditures (mostly for police) were about three times higher. When examined by component, the direct expenditures in 1997 for policing, the courts, and corrections, respectively, were: $57.8, $28.5, and $43.5 Table 8 NOTE: Sentence medians are the medians of maximum sentences. One percent of defendants in both burglary and drug possession cases, and 2 percent of defendants in weapons cases, were fined only. SOURCE: Hart and Reaves, pp. 30-33, Tables 30-32. billion. Correctional expenditures (mostly to operate prisons and jails) grew the most (4.8 times higher than they were in 1982); police and court expenditures were 3.0 and 3.7 times higher, respectively.