Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Sentencing Reforms and Their Effects Recent changes in sentencing laws and practices have variously affected judges, prosecutors, parole boards, and other officials in the criminal justice system. Many sentencing reforms have been directed only at one set of officials and have not addressed or attempted to anticipate shifts of discretion to other officials. For example, the California legislature eliminated parole release for most prisoners and established detailed statutory criteria for prison sentences but did nothing to control the discretion of prosecutors whose influence on sentencing through charg- ing and plea negotiation increased. Because the punishment process is complex, it is important if the effects of changes are to be recognized and understoodto look not only at processes that are formally and immediately affected by a change but also at earlier and later processes. Thus, an adequate assessment of the impact of sentence reforms requires that consideration be given to its implications for court procedures and plea-bargaining practices as well as to its apparent impact on sentences received by offenders. This chapter summarizes the findings of the literature on evaluations of the impact of sentencing reforms. A detailed review of that literature is contained in Cohen and Tonry (Volume II). We are primarily con- cerned with the effectiveness of sentencing reforms as a means of re- ducing disparities, altering sentence severity, and making decision mak- ing systematic. Consequently, we focus on how innovations affect what happens to defendants and how participants in the system have altered their behaviors in reaction to innovations. The innovations we consid- 184
Sentencing Reforms and Their Effects 185 ered were directed at the actions and decisions of prosecutors, judges, and parole authorities. More specifically, we review evaluations of ef- forts to abolish plea bargaining in three jurisdictions; mandatory min- imum sentencing laws in Michigan, Massachusetts, and New York; Cal- ifornia's Uniform Determinate Sentencing Law; presumptive and voluntary sentencing guidelines; and parole guidelines in four jurisdictions. This chapter is divided into four sections. The first section summarizes research findings on the extent of formal compliance with several sen- tencing innovations. The second section examines evidence of efforts by lawyers and judges to dispose of cases in ways inconsistent with apparently applicable rules and laws. The third section reviews evidence concerning the impact of sentencing innovations on sentence outcomes. Section four describes the major methodological shortcomings that char- acterize the evaluations. (The organization of this chapter requires that we discuss particular studies and innovations in more than one section. In order to minimize repetition, studies are described relatively fully when first mentioned and are thereafter described by cross-reference.) The corpus of sentencing reform impact evaluations is small, and most published reports suffer from serious methodological shortcomings. There has not yet been a sufficient number of well-executed evaluations to permit the panel to offer detailed conclusions about the effects of diverse sentencing innovations. However, the following four broad generali- zations emerged from many of the evaluations considered: 1. Formal compliance with the requirements of innovations has been widespread: assistant prosecutors have adhered to plea-bargaining bans and restrictions; parole board examiners have tended to set release dates that are consistent with applicable parole guidelines provisions; judges have tended to adhere to statutory sentencing standards, especially man- datory minimum sentence laws. Outside the parole context, however, compliance has often been formal rather than substantive. 2. Judges and lawyers have often substantially modified case-pro- cessing procedures in order to achieve dispositions of cases that were different from those specified in applicable rules or laws. Partial plea- bargaining bans and mandatory minimum sentencing laws appear es- pecially vulnerable to circumvention. 3. Parole and sentencing guidelines systems that have legal or ad- ministrative force and are subject to credible enforcement mechanisms have operated to reduce the extent of sentencing disparities. 4. Plea-bargaining bans and mandatory and determinate sentencing laws have produced modest changes in sentencing outcomes, particularly some increases in prison use. Typically, increases in severity have been
186 RESEARCH ON SENTENCING THE SEARCH FOR REFORM experienced by marginal offenders, who previously might or might not have received prison sentences. COMPLIANCE WITH SENTENCING REFORMS Whether officials comply with the formal requirements of sentencing innovations appears to depend on the legal authority of an innovation and whether it is subject to credible enforcement mechanisms. Plea- bargaining bans, mandatory minimum sentencing laws, Minnesota's pre- sumptive sentencing guidelines, California's determinate sentencing law, and parole guidelines commonly result in substantial formal compliance. Voluntary sentencing guidelines have not been shown to achieve high rates of formal compliance. Aso~T~oN OF PLEA BARGAINING There have been several efforts to "abolish" plea bargaining in full or in part. Some of these efforts have been evaluated: the Alaskan attorney general's 1975 ban on plea bargaining in that state (Rubinstein et al., 1980~; the actions of a county prosecutor in Michigan to abolish charge bargaining in drug trafficking cases (Church, 1976~; the Wayne County (Detroit) prosecutor's prohibition of bargaining in firearms cases subject to a mandatory 2-year sentence (Heumann and Loftin, 19794; and the restrictions placed on charge reductions in New York's mandatory sen- tencing laws for drug offenses (Joint Committee on New York Drug Law Evaluation, 1978~. These evaluations found that plea bargaining can be substantially controlled when the chief prosecutor wishes to do so, establishes internal review and management systems that effectively monitor assistant pros- ecutors' behavior, and wins the support of assistant prosecutors. When the ban is only partial (only charge bargaining is banned or only sen- tencing bargaining), judges and lawyers tend to shift to alternative bar- . . gaining systems. Alaska is the only jurisdiction to attempt the statewide elimination of plea bargaining in all its variant forms. On July 3, 1975, effective August IS, 1975, the attorney general of Alaska ordered state prose- cutors to desist from plea bargaining and sentence recommendations. Charge dismissals or reductions as inducements to guilty pleas were later forbidden, but unilateral charge dismissals for good-faith professional reasons were permitted. The Alaska Judicial Council evaluated the im- pact of the abolition in Anchorage, Fairbanks, and Juneau (Rubinstein et al., 1980~. Case record data were collected on case dispositions in
Sentencing Reforms and Their Effects 187 the 12-month periods before and after the ban, and interviews were conducted covering more extended periods. The credibility of the study's statistical analyses is doubtful, as are the conclusions deriving from the statistical data, but the rich interview data provide a firmer basis for most of the study's major conclusions. The study concluded that "plea bargaining as an institution was clearly curtailed" (Rubinstein et al., 1980:314. Sentence bargaining and prosecutorial sentence recommen- dations declined abruptly from 43.5 to 13.1 percent of all cases in the three jurisdictions. The interview data from judges, prosecutors, and defense attorneys supported the statistical indications that sentence bar- gaining had essentially ceased. The study concluded that charge bar- gaining also had substantially disappeared. An effort to eliminate prevailing charge-bargaining practices was in- itiated by the newly elected prosecutor in "Hampton" County, Michi- gan,~ in January 1973. He instituted a strict policy forbidding bargained charge reductions in drug sale cases and, at the same time, substantially tightened the standards under which drug prosecutions were authorized. Church (1976) collected information on drug sale warrants and dispo- sitions for the 12-month periods before and after January 1, 1973. These data were supplemented by data from interviews with judges, defense counsel, prosecutors, and the court administrator. Church concluded that guilty pleas to reduced charges fell from 81 percent of cases before the ban to 7.1 percent afterward but that charge bargaining was quickly replaced by sentence bargaining involving judges and defense lawyers. The Michigan Felony Firearms Statute created a new offense of pos- sessing a firearm while engaging in a felony and mandated a 2-year prison sentence, which could not be suspended or shortened by release on parole, to be served consecutively to the sentence imposed for the predicate (underlying) felony. Since the gun possession charge had to be separately charged, its applicability depended on the decisions of Michigan prosecutors. The law took effect on January 1, 1977, and was supplemented by the Wayne County prosecutor's ban on charge dis- missals of firearms charges pursuant to plea bargains. Since the charge determined the mandatory incremental sentence, prohibition of charge bargaining also accomplished a prohibition on sentence bargaining. Heu- mann and Loftin collected data from court records on cases disposed in the 6-month periods before and after the ban took effect and conducted interviews with lawyers and judges. They found that the prosecutor was ~ 'Hampton" County is a pseudonym used by the researcher to conceal the identity of the research site.
188 RESEARCH ON SENTENCING THE SEARCH FOR REFORM generally successful in obtaining formal compliance with the ban (Heu- mann and Loftin, 1979:402~. MANDATORY SENTENCING LAWS Numerous mandatory sentencing laws have been passed in recent years. Evaluations of three of the laws have been published (Beha, 1977; Heumann and Loftin, 1979; and Joint Committee on New York Drug Law Evaluation, 1978~. All three evaluations were largely concerned with the deterrent effects of the laws studied: case processing and dis- positions received subsidiary attention, and, accordingly, the data are sometimes unsatisfying and must be interpreted cautiously. There appear to have been few blatant refusals to impose the pre- scribed sentences on defendants convicted under the mandatory mini- mum sentencing laws. However, in each jurisdiction studied the per- centage of prosecutions resulting in convictions declined, which suggests that officials attempted to shelter some defendants from the law's effects. In the case of the Michigan felony gun law, for example, the proportion of offenders incarcerated after conviction in "other assault" cases (a category of assault cases of moderate severity) rose from 57 percent to 83 percent; however, the conviction rate declined by 20 percent (Heu- man and Loftin, 1979: Table 3~. Under New York's "Rockefeller" drug laws, which went into effect on September 1, 1973, severe mandatory prison sentences were pre- scribed for narcotics offenses at all levels, and selective statutory limits were placed on plea bargaining. The Joint Committee on New York Drug Law Evaluation (1978) found that the risk of incarceration after conviction increased substantially, from 34 percent in 1972 to 55 percent in 1976. However, the likelihood that a person arrested for a drug offense would be incarcerated remained the same because indictment and con- viction rates declined. Sentence lengths did increase substantially: the percentage of sentenced drug felons receiving minimum prison terms longer than 3 years increased from 3 to 22 percent. In Massachusetts the Bartley-Fox Amendment, effective April 1, 1975, required imposition of a 1-year mandatory minimum prison sentence, without suspension, furlough, or parole, for anyone convicted of car- rying an unlicensed firearm. Beha (1977) collected data on all prose- cutions for firearms crimes in the 6 months after the law took effect and for the corresponding 6 months of the preceding year. Only indirect evidence is available regarding compliance by lower court judges in imposing the 1-year minimum prison term for carrying a gun. The per- centage of cases proceeding to superior court either on an appeal, as a
Sentencing Reforms and Their Effects 189 trial de novo, or bound over directly from the lower court, increased from less than one-fifth of cases to more than one-half after implemen- tation of the new law, suggesting that lower court judges imposed more severe sentences in compliance with the law. However, these increases in severity were offset by sharp reductions in the number of cases avail- able for sentencing because dismissals and acquittals increased. DETERMINATE SENTENCING IN CALIFORNIA The original California Determinate Sentencing Law (DSL) took effect July 1, 1977, and was amended twice in 1978 to increase the severity of penalties for some offenses.2 The DSL prescribes three base terms for each crime (e.g., for robbery the terms are 2, 3, or 5 years). The middle term is the presumptive term to be imposed except in cases in which the judge concludes that mitigating or aggravating circumstances warrant the use of the lower or upper base terms. Seven major research projects have examined the postimplementation impact of determinate sentencing in California. As summarized in Table =1, these studies vary considerably in questions addressed, in the ju- risdiction levels and stages in case processing studied, and in the relative strengths or weaknesses of their evaluation designs. Together they pro- vide a comprehensive picture of the impact of determinate sentencing in California. The available evidence indicates a high degree of formal compliance with the requirements of DSL in California. Available evidence for fiscal 1977-1978 and for 1979 indicates that most offenders sentenced to prison in those years received the presumptive middle base term (61 percent in 1977-1978 and 54 percent in 19794. The use of the lower and upper base terms for mitigating and aggravating circumstances varied among offenses. Upper base terms were more likely to be imposed than lower base terms for crimes against persons, and lower base terms were more common than upper base terms for property and drug offenses. In more recent years there has been a marked increase in the proportion of convicted persons who receive the lower base term; possible explana- tions for this trend are discussed in Cohen and Tonry (Volume II). The law also prescribes additions to prison sentences ("enhance- ments") when particular aggravating circumstances, including weanon ___=, - --I 2 The crime types directly affected by the amendments were first-degree burglary, robbery, voluntary manslaughter, rape, crimes against children, and oral copulation. Both the middle and upper terms were increased for all these offenses except robbery, where only the upper term was increased.
190 - ~ ax ~ _' Ct ._ - ._ E Ct ._ ._ Ct . . ._ An a I o . Ct - Ct Lid Ct E i_ ._ Cal o ._ Ct ._ Ct ¢ U' . ~ Cal U: ~ Cal Cal I\ V ;^ ;^ ;^ ;^ C _ ~ _' ·4- o ~ ~ ~ ~ ~ U) V >~ ;~~ ~~ I'm ~ o ~ ~ ·:> ~ Cad ~ ;^ ;^ C=t ~ = ~ ·3 To ~ ~ oo ._ 3~ us ~ :^ ~ _ ;^ X~ VZ Z Z == c~ ~ u) ;^ ;^ ;^ A A ~ ~ ~ O ~ ~ U ~ ~ ~ ~ ~ 1) D Ct ._ ;> O a; c3' o~ ._ U: C~ o CD - ._ C5 C) 5 o ;^ C o~ C ._ ·3 o D e~ ._ C~ :> O cn Ct 3 o _ Ct ._ C_ "s: Ct C) C:
Sentencing Reforms and Their Effects 191 use, great bodily injury to the victim, large property loss, or prior prison terms, are charged and proved. Statewide, among persons committed to the Department of Corrections, the use of enhancements tends to be limited to weapon or firearm use, especially in robbery cases, for which the enhancement was charged in 90 percent of eligible cases and proved in 74 percent of charged cases. This contrasts sharply with enhancements for victim injury and prior prison, which were charged and proved in less than 25 percent of eligible cases among persons admitted to prison. The statutes impose no obligation on prosecutors to charge or prove facts that would support an enhancement. Charging patterns vary from county to county in California, and thus the imposition of enhancements also varies (see Casper et al., 1982; Utz, 19814. Contrary to the wide discretion exhibited by prosecutors in charging and proving enhancements, once enhancements were proved judges routinely added the prescribed additional years to the base sentence. Thus, with respect to the enhancements, there was considerable com- pliance by judges with the formal requirements of DSL (see Cohen and Tonry, Volume II:Table 7-17~. SENTENCING GUIDELINES As we discussed in Chapter 3, sentencing guidelines have taken many forms and approaches. A distinction is sometimes drawn between "de- scriptive" guidelines, which are based on statistical characterizations of past practice, and "prescriptive" guidelines, which are in the first in- stance the result of policy decisions about appropriate punishments. They are also distinguished by their respective legal authority. "Vol- untary" guidelines do not have legal authority: judges are not required or authoritatively encouraged to comply with voluntary guidelines, and defendants have no recourse against judges who fail to do so. "Pre- sumptive" sentencing guidelines, like Minnesota's, do have legal au- thority: judges are directed to impose sentences prescribed by the guide- lines unless "substantial and compelling" circumstances are present that justify departure from them, and defendants and prosecutors may seek review of the appropriateness of a departure. Several examinations of the construction and effects of voluntary/ descriptive guidelines have been conducted: Rich et al. (1981) assessed the construction and subsequent impact of judicially adopted guidelines in Denver and Philadelphia; Sparks et al. (1982) reviewed the construc- tion of the Massachusetts guidelines in depth and those of other juris- dictions in less detail; Cohen and Helland (1982) examined guidelines in Newark.
192 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM Formal compliance with voluntary/descriptive guidelines has appar- ently been limited in the jurisdictions studied. It is important to note that in this context we distinguish between "compliance" and "consis- tency." A sentence is compliant with guidelines when a judge has con- sciously considered the sentences indicated by the applicable guidelines and elected to impose a sentence from within the guideline range. A sentence may be consistent with guidelines even if a judge was unaware of their existence. Thus sentencing in a court could be consistent with guidelines but not compliant. Indeed, the original Denver guidelines were drafted with the intent that 8~85 percent of the cases in the construction sample would fall within ("be consistent with") the guide- lines. Rich et al. (1981) found that in Denver judicial decisions to incarcerate were consistent with guidelines in about 70 percent of cases, both before and after guideline implementation. Postimplementation consistency in terms of sentence length was lower, about 40 percent of those sentenced to prison. In Philadelphia and Denver, there was consistency on both the in/out and length decisions in only about one-half of all cases. Similar results were found in Newark (Cohen and Helland, 1982~. Lawyers and judges interviewed in Philadelphia and Denver indicated that few judges made significant efforts to comply with the guidelines. Thus there was little evidence of formal compliance and evidence of even less consis- tency than had been expected given the avowedly descriptive basis of the guidelines. As of early 1982, Minnesota was the only jurisdiction that had im- plemented sentencing guidelines that are both presumptive and pre- scriptive. The early indication from internal evaluations by the guidelines TABLE ~2 Percentage of Cases Sentenced Consistently With Minnesota Sentencing Guidelines Presumptive Outs77 Who Were Sentenced "Out77 Presumptive ``Ins77 Who Were Sentenced "Ink 1978 baseline cases 86 44 198~1981 sentences imposed 96 77 NOTE: The figures in this table were estimated from data provided by Knapp (1982). They are not precise because some cases that appear among the presumptive "outs77 are actually treated as presumptive "ins' under separate statutory provisions for mandatory sentencing.
Sentencing Reforms and Their Effects TABLE ~3 Percentage of Cases Sentenced Consistently With Presumptive Sentences Within Selected Cells of Minnesota Sentencing Guidelines 193 Percentage Actually Sentenced "Out" 1978 198~1981 Presumptive "Out" Cells Baseline Cases Sentences Imposed Offense 5, history 1 60.7 95.0 Offense 5, history 2 21.8 74.2 Offense 3, history 3 45.4 80.3 Percentage Actually Sentenced "In" 1978 198~1981 Presumptive "In" Cells Baseline Cases Sentences Imposed Offense 7, history 0 39.1 71.8 Offense 8, history 0 41.9 85.4 Offense 8, history 1 29.1 75.0 SOURCE: Knapp (19824. commission is that there is substantial formal compliance by judges in both decisions to incarcerate and decisions about sentence length. As shown in Table 3-1, the Minnesota guidelines are expressed in a grid format; cases falling in cells above the in/out line should receive "out" sentences (i.e., stayed state prison terms), and cases falling in cells below the line should receive state prison sentences. Table ~2 shows the percentages of cases in the commission's 1978 baseline sample that would have been sentenced consistently with the presumptive "in" and "out" sentences had the guidelines been in effect in 1978 and the percentages of cases sentenced consistently with the presumptive sentences under the guidelines in 198(}1981. For both "in" and "out" decisions there were marked shifts in sentences consistent with the guidelines. As Table ~3 reveals, these shifts in sentencing are often larger when individual cells in the guideline grid are examined than is apparent overall. The relatively low preguideline consistency with the guidelines illus- trates the extent to which the guidelines departed from previous sen- tencing practices in Minnesota. Since this is one of the few sentencing reforms that has resulted in substantial changes in the behavior of a state court system and its participants at least in the preliminary find- ings the panel encourages a longer-term and more extensive exami- nation of the nature and extent of compliance with the Minnesota sen- tencing guidelines.
194 RESEARCH ON SENTENCING THE SEARCH FOR REFORM PAROLE REFORMS Parole Abolition On May 1, 1976, Maine became the first state in modern times to abolish parole release and establish a determinate sentencing system in which the duration of prison terms could be calculated at the time of sen- tencing. Maine is not an ideal jurisdiction in which to assess the impact of parole abolition. The small numbers of criminal prosecutions and convictions in that state make meaningful statistical analyses of changes in sentences by offense type virtually impossible. However, two eval- uations of Maine's innovations have been completed. Kramer et al. (1978) assessed the impact of the change during its first 12 months; Anspach's (1981) report provides a content analysis of changes in Maine's substantive criminal law without data on the impact of the parole ab- olition. Both of the evaluations are fundamentally flawed, and neither pro- vides credible findings on the impact of Maine's abolition of parole. Plausible arguments have been made that parole release operates in important respects as a monitoring system for sentencing- evening out disparities in the lengths of prison terms and providing a device to relieve prison crowding; the Maine evaluations have not provided insights into these or other questions. It is to be hoped that other evaluations will be undertaken that are designed to test hypotheses about the compar- ative advantages of judicial and parole systems for determining the lengths of prison sentences. Parole Guidelines There have been three major evaluations of the operations of parole guidelines systems. Arthur D. Little, Inc., and Goldfarb, Singer, and Austern (1981, hereafter cited as ADL, 1981) examined the U.S. Parole Commission's parole guidelines system and state systems in Washington, Oregon, and Minnesota. Mueller and Sparks (1982) studied the oper- ation of the Oregon parole guidelines. In 1982 the General Accounting Office released a study on the operation of the federal parole guidelines system. Four primary questions have been studied: the extent to which parole guidelines are correctly applied in prison release decisions; the extent to which parole release decisions are consistent with apparently applicable guidelines; the extent to which parole guidelines serve to reduce disparities in punishment compared with parole release without guidelines and compared with the distribution of sentences imposed by
Sentencing Reforms and Their Effects 195 judges; and the effect of parole guidelines on the overall severity of prison sentences. Findings on the first two questions are reviewed here. Findings on the third and fourth questions are reviewed later in this chapter. Using parole guidelines, the Arthur D. Little and the General Ac- counting Office studies investigated consistency of decisions in two sen- ses. The first, which we refer to here as error proneness, concerns the consistency with which different decision makers apply the guidelines to individual cases. This was tested by having researchers or, in the General Accounting Office study, parole hearings examiners, calculate guidelines sentences for cases already decided and comparing those sentences to the ones actually imposed. Both the General Accounting Office and the Arthur D. Little studies of the U.S. Parole Commission's guidelines found serious error-prone- ness problems. Arthur D. Little researchers using a method in which two individuals separately evaluated each file, reconciled their decisions, and compared them with the actual case decisions were in agreement with the actual Parole Commission offense seriousness and salient factor score calculations in 61 percent of the cases studied (ADL, 1981b:49~. The General Accounting Office (1982:15-22) study found greater in- consistency, even when it had experienced parole examiners calculate guidelines sentences for 30 prisoners previously released. In the three states studied, Arthur D. Little researchers found wide variation in rates of error proneness. In Minnesota, from a sample of prisoners released in 1979, the researchers concluded that the parole board "applies parole decision guidelines in a highly consistent manner" (ADL, 1981d:97) . In Oregon, calculations were completely consistent with parole board calculations in 67 percent of the cases studied (ADL, 1981a:8~. In stark contrast, the complete agreement rate in Washington was only 13 percent (ADL, 1981c:24. The evaluators point out that their analyses may, for several reasons, overstate discordance. Nonetheless, for all but Minnesota's "simple and explicit" system, all of the guidelines systems appear highly subject to calculation errors, owing to various combinations of inherent complexity, poor quality-control procedures, insufficiently specific policy rules, and problems of missing and unreli- able data. Consistency in its second sense concerns the extent to which release dates are consistent with the apparently applicable guidelines (that is, the guideline term as determined by the examiner, which, as noted above, may be inaccurate). An important caveat must be noted: all parole guidelines systems authorize examiners to depart from the guide- lines in exceptional cases. Thus a release date not authorized by the
196 RESEARCH ON SENTENCING THE SEARCH FOR REFORM guidelines does not necessarily mean that it is not in compliance with the guidelines system. The discretionary "departure rates" under the U.S. parole guidelines have varied between 10 percent and 20 percent. Under the Minnesota guidelines, the overall discretionary departure rate in 1977-1979 was less than 10 percent (ADL, 1981d:40~. Compliance with Washington's first set of guidelines occurred in only about 30 per- cent of the cases, but those guidelines were later repealed and replaced with guidelines expressed in a different format; release dates were set within the new guidelines in 74 percent of cases (ADL, 1981c:14~. ADAPTIVE RESPONSES TO SENTENCING REFORMS Most sentencing innovations are designed to alter existing processes, procedures, or outcomes, and they generally originate outside the or- ganizational contexts in which decisions are made in individual cases. Legislatures, parole commissions, sentencing commissions, or chief prosecutors prescribe the new systems, but judges, assistant prosecutors, and parole hearing examiners must carry them out. There are numerous ways that officials can alter their behaviors to adapt to new procedures or rules that they believe to be inconvenient, impractical, or unwise. Most impact evaluations of sentencing innovations have identified ways in which the officials who make decisions have altered their operations in order to nullify new policies in some respects. ADAPTATIONS TO PLEA-BARGAINING BANS Courtroom personnel have personal and bureaucratic interests in the expeditious disposition of cases that they often believe are satisfied by plea bargaining. One might expect plea-bargaining bans to disrupt case processing by reducing guilty-plea rates and by increasing trial rates, case backlogs, and case-processing time. Or one might expect wide- spread efforts to circumvent such bans, particularly through adoption of forms of plea bargaining or consensual case disposition that have not been banned. The evidence is mixed. The evaluators of the Alaska plea-bargaining ban concluded that overt plea bargaining ceased to be an important factor in case processing in the jurisdictions studied and that implicit plea bargaining (in which a defendant is assured that guilty pleas will be rewarded by sentencing concessions) replaced overt bargaining only for some offenses. In Wayne County, "Hampton" County, and Black- hawk County, plea-bargaining bans produced shifts to forms of consen-
Sentencing Reforms and Their Effects 197 sual case disposition that were not banned (Church, 1976; Heumann and Loftin, 1979; lowa Law Review, 1975~. Alaska Sentence bargaining was the predominant form of bargaining prior to the plea-bargaining ban. For a brief period after the ban, charge bar- gaining increased in Fairbanks, but it ceased when it was prohibited by the state prosecutor there. Overall the ban appears to have been effec- tive (Rubinstein et al., 1980~. Case processing changed little: there was a slight tendency toward earlier dismissal of cases, but overall dismissal rates and guilty-plea rates were substantially unchanged. Trial rates increased, but the absolute numbers remained small. Court-processing times decreased. There was some evidence of changes in the handling of cases. Case screening tightened: the percentage of cases screened out increased from 10.0 to 12.9 percent in the year after the ban. Screening rejections of drug and morals felonies increased substantially; there may have been a tendency after the ban took effect to prosecute as misdemeanors cases that previously were handled as felonies. There was no increase in out- right dismissals: the overall dismissal rate prior to the ban (52.3 percent) was essentially unchanged after the ban (52.7 percent). Both the interviews and the statistical analyses indicated that sentence bargaining was not replaced by charge bargaining or by forms of implicit plea bargaining. And, contrary to expectations, the rate of guilty pleas to offenses originally charged declined only slightly: from 23.6 percent of the cases available after screening before the ban to 22.5 percent of such cases after the ban. "Hampton" County In "Hampton" County, where charge bargaining in drug cases was pro- hibited, the system adapted to the ban in ways that permitted consensual case dispositions to continue. First, sentence bargaining increased: roughly one-half the judges made some form of preplea sentence commitment in applicable casesa sizable behavioral shift given former practices and strong system norms against judicial participation in plea bargaining (Church, 1976:3874. Second, there was a substantial increase in the rate at which cases were dismissed outright. Under the plea-bargaining ban, nolle prosequi rates declined slightly from 15 to 10 percent. Judicial dismissal rates after the ban took effect increased from 19 percent for
198 RESEARCH ON SENTENCING THE SEARCH FOR REFORM 1972 warrants to 28 percent after the ban, as did "youthful trainee" convictions (which permit a sentence of probation) from 3 to 17 percent. Thus some drug sale cases that formerly were likely to result in a con- viction through a guilty plea to reduced charges before the ban were removed from the system following the ban. Wayne County, Michigan Although the county prosecutors filed and pursued gun law charges in conformance with the state law (which required a mandatory prison term), two types of adaptive mechanisms to avoid the law's impact and the simultaneous ban on plea bargaining greatly limited increases in sentence severity. First, especially for assault cases of moderate severity, "waiver trials" were used to avoid the mandatory 2-year sentence: some- times judges gave explicit prior indications that they would dismiss the gun charge at trial, often with the prosecutor's acquiescence; other times there were no explicit understandings, but judges acknowledged to re- searchers that they considered every possible defense and sought any available technical loophole. Second, researchers' interviews suggested that judges routinely nullified the 2-year mandatory sentence increment for a firearms offense by reducing the sentence otherwise imposed on the primary convicted felony offense by an offsetting 2 years (Heumann and Loftin, 1979~. MANDATORY SENTENCING LAWS One conventional hypothesis concerning mandatory sentencing laws is that lawyers and judges will dismiss charges, acquit defendants, and otherwise alter their practices in order to avoid imposing sentences they believe to be unduly harsh or otherwise inappropriate. Although the number of evaluations of the impact of mandatory sentencing laws is too small and their quality too uneven to permit confident generaliza- tions, the avoidance hypothesis appears to be confirmed by the few published evaluations. In Wayne County, Michigan, there is evidence from interviews and also from statistical analyses that efforts were made to prevent sym- pathetic defendants from being subject to mandatory imprisonment under the firearms law. Although there was little change in the disposition patterns for the most serious offense studied (armed robbery offenders continued to be imprisoned at high rates) or the least serious (few offenders were imprisoned for felonious assault, which often involved acquaintances), marked changes characterized the more ambiguous cat-
Sentencing Reforms and Their Effects 199 egory, "other assault." Early dismissal rates doubled, and there was an offsetting decline in the percentage of convictions (Heumann and Loftin, 1979~. Beha's (1977) evaluation of Massachusetts's Bartley-Fox law revealed few signs of widespread efforts to avoid compliance with the mandatory sentence provision in the lower courts. The only significant sign was a substantial increase from 16 to 36 percent of dispositions in the ac- quittal rate for defendants charged only with illegal carrying of a firearm. Beha only studied case dispositions in lower trial courts, from which convicted offenders could appeal to a trial de nova in the superior courts; he does not indicate whether circumvention of the mandatory sentence law occurred in the higher court. In the evaluation of the mandatory sentencing provisions of the Rock- efeller drug laws in New York (Joint Committee on New York Drug Law Evaluation, 1978), there are also indications that cases were screened more carefully at early stages of the process after the law took effect. The numbers of arrests for drug felonies in New York State declined substantially, as did indictment rates given arrest and conviction rates given indictment. Most of the mandatory sentencing law evaluations have been primarily concerned with the deterrent effects of the new laws, and the effects on case processing have received less attention. Neither Beha nor the Joint Committee conducted extensive interviews or used participant obser- vation methods. Cautiously phrased, our conclusion is that the evidence is not inconsistent with the avoidance hypothesis, especially when de- fendants have been charged with relatively less serious offenses. How- ever, it is important to note that there were high rates of formal com- pliance for cases that were not screened out. CALIFORNIA S DETERMINATE SENTENCING LAW Case processing in California changed little after implementation of DSL. The law left substantial discretion in charging and dismissing cases in the hands of prosecutors, whose processing and plea-bargaining ac- tivities apparently continued much as before. However, as in other Jurisdictions In WHICh sentencing stanctaros made sentences more pre- dictable, there was a tendency toward disposition of cases earlier in the process. Some observers expected a shift of sentencing power to the prosecutor (Alschuler, 1979~. However, it appears that the new law did not significantly alter power relations: in jurisdictions in which judges traditionally dominated sentencing, they continued to do so; where pros- ecutors traditionally dominated, this too continued. . . .. . . . · . . . . . .
200 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM Charging Prosecutors in the jurisdictions studied typically adhered to an explicit policy of full initial charging (although screening on the merits of the case was permitted) and used various administrative procedures to en- sure that assistant prosecutors complied. Both observation and interview data and the statistical analyses found little evidence of any major changes in initial charging for cases finally disposed of in superior court. For example, Utz's (1981) multivariate analysis of changes in initial charging for burglary cases in Alameda and Sacramento counties indicates that, after controlling for other attributes of the case, initial charging was not affected by DSL. Unfortunately, all analyses of charging are limited to cases finally disposed of in superior courts. It is possible to circumvent the deter- minate sentence provisions by initially charging cases as misdemeanors rather than felonies, in which instance they do not appear in superior court at all. From existing data it is impossible to determine if such changes have occurred. Plea Bargaining Utz (1981) and Casper et al. (1982), who between them studied five California counties, found little change in local plea-bargaining practices as a result of DSL. Jurisdictions that engaged in substantial bargaining before DSL incorporated explicit sentence length agreements into their bargaining practices after DSL; those jurisdictions with limited bar- gaining before DSL continued to refrain from bargaining after DSL. Controlling for crime type, there were no marked changes after DSL in the already high proportion of guilty pleas among convictions found in the five counties and for the state as a whole, although early guilty- plea rates (e.g., at initial court appearance) increased. As indicated in Figure ~1, without controlling for any variations in crime-type mix over time, a simple two-point comparison between 1976 before DSL and 1978 after DSL shows sharp increases in the proportion of early pleas entered at initial appearance among all guilty pleas in superior courts. Consideration of a longer time period before DSL implementation, however, reveals a long-term decline in the rate of early pleas from the late 1960s to 1976, making it unclear whether the increases in the early guilty-plea rate after DSL represent a real effect of the law on early guilty pleas or merely a random fluctuation in a cyclic phenomenon. Despite explicit prosecution policies in all five counties of "full en- forcement" of enhancement and probation ineligibility provisions, both
Sentencing Reforms and Their Effects 60 50 us 40 c: z a: ~ 6 O ~ to On ~ ~ TV O ~ - U] ~ _ ~ Z in 0 ~ 20 Z c) LL cr ~ U.l Z ~ UJ 10 O _ -it \ ~ / \ ; _ _ . a ·. \ 1 1 1 1 1968 1970 1972 1974 \ San Francisco,\ Santa Clarea ~ /j~ ~1 201 DS L I mplemsnted July 1, 1977 5a n Rernad; nca A \ 1 1 1 1 ~ ~ 1976 1 1977 1978 1979C 1980 YEAR FIGURE ~1 Trends in the timing of guilty pleas in California: percent of all superior court guilty pleas entered at initial appearance. SOURCES: a From Casper et al. (1982:Figure 13); b Derived from Lipson and Peterson (1980:Table 3~; c The rates for 1979 derived from data reported in California Department of Justice (1980~. Casper et al. (1982) and Utz (1981) report that the opportunities for prosecutors to drop such allegations played a significant part in plea negotiations. As expected, the allegations were used by prosecutors as bargaining chips, to be dropped in return for a guilty plea to the basic offense charge or an agreement on a prison sentence. SENTENCING GUIDELINES The Rich et al. (1981) evaluation of the initial voluntary sentencing guidelines systems attempted to study the effects of the guidelines on plea negotiations. Interview data from Philadelphia, Chicago, and Den- ver indicate that lawyers did not consider the guidelines to be important
202 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM and accordingly did not take them into account when negotiating plea agreements. Because Minnesota's presumptive sentencing guidelines have legal force arid prescribe narrow ranges from which prison sentence lengths must be selected, some guideline critics have suggested that opposing counsel would incorporate the guidelines into their plea negotiations. Since the applicable guideline range is based on conviction offenses, the outcomes of charge bargains would determine the applicable guideline sentence. Preliminary analyses by the Minnesota Sentencing Guidelines Commission found some evidence of changes in charge reduction pat- terns for cases in which aggravated robbery was the most serious initial charge. As evidenced in Table 4 4, the proportion of charge reductions from aggravated robbery to a lesser charge increased for defendants with low criminal history scoresfewer of these defendants were ac- tually convicted of aggravated robbery. Once again, there were appar- ently adjustments in case processing to avoid imposing the prescribed prison term when prison was not deemed appropriate by court person- nel. For defendants with high criminal history scores, on the other hand, the proportion of charge reductions declined, and more cases resulted in convictions on the original aggravated robbery charge. This pattern suggests that prosecutors and judges were operating to preserve dis- tinctions among cases on the basis of criminal history despite the explicit guidelines policy that prescribed prison terms for aggravated robbery cases. TABLE 4 4 Changes in Charge Reductions After Implementation of the Minnesota Sentencing Guidelines Criminal History Score Percentage of Cases Convicted of Aggravated Robbery When Aggravated Robbery Was the Most Serious Original Charge 1978 Cases Sentenced Under Baseline Cases Guidelines, 1980-1981 1° 0 59 49 1 75 60 2 64 66 3 54 70 4 58 70 SOURCE: Knapp (1982).
Sentencing Reforms and Their Effects 203 It was also anticipated by some that the guidelines would result in increases in the rate of cases going to trial. No such increase was observed during the first year after full implementation of the guidelines; the trial rate among felony convictions was 5 percent in 1978 and 4 percent among 5,500 cases disposed under the guidelines (Knapp, 1982~. However, in assessing the impact on trial rates it is important to also examine dis- position time. If disposition time increased, especially for trial cases, increases in trial rates might not be evident during the early implemen- tation period. This remains an issue for further exploration in the con- tinuing evaluation of the impact of the Minnesota guidelines. PAROLE GUIDELINES In most jurisdictions decisions under parole guidelines are not the result of an adversary process. Parole boards may responsively adapt policies to various pressures, but hearing examiners have few opportunities for adaptive responses. Although there has been wide variation in compli- ance with parole guidelines, this seems to result from ambiguities in the guidelines themselves rather than from systematic attempts to evade the guidelines. Two other forms of adaptive response to parole guidelines at sen- tencing have been hypothesized: sentence bargains in which the maxi- mum sentence imposed expires before the presumptive parole release date, thus avoiding the parole decision entirely; and judicial imposition of minimum sentences that expire later than the presumptive release date. There is considerable anecdotal evidence of such bargaining and sentencing by reference to parole guidelines, but no systematic studies are available. THE USE AND SEVERITY OF SANCTIONS Most sentencing innovations that affect the behavior of prosecutors and judges operate to make sentencing more predictable. If plea bargaining has been banned or regulated or if a case is subject to a mandatory sentence, statutory determinate sentencing, or presumptive sentencing guidelines, the parties can better predict the likely sentence than under indeterminate sentencing systems. For many defendants the increased predictability may affect plea negotiations, but it is unlikely to affect the nature of the sanction to be imposed. Offenders who have committed venial offenses are often unlikely to receive prison sentences. Offenders who have committed major violent crimes or who have extensive crim- inal records are likely to receive prison sentences whatever the sen-
204 RESEARCH ON SENTENCING THE SEARCH FOR REFORM fencing system. Generally, the innovations studied have not resulted in dramatic increases in the proportion of cases sentenced to prison for either venial or repetitive serious offenders. However, there have been increases in the lengths of prison terms imposed for an intermediate category of offender who might or might not have been imprisoned before an innovation. New sentencing standards could also substantially affect the sentences of marginal offenders. By definition these are ambiguous cases. New sentencing standards may resolve the ambiguity of the cases by directing that marginal offenders fitting a particular profile be imprisoned. Yet these are cases in which judges may often be loathe to impose prison sentences. It is hypothesized that at least two arguably undesirable out- comes may result. Judges and lawyers may circumvent applicable new standards when they appear to be too harsh in a particular marginal case, or they may apply them inappropriately, punishing marginal of- fenders more severel~with prison terms- than they want to. There is evidence to support both hypotheses: the findings on adaptive responses (discussed above) confirm the first hypothesis; other evidence (discussed below) suggests that sentencing outcomes do not appear to have been altered substantially except for marginal offenders, who often seem to receive harsher sentences. PLEA-BARGAINING BANS Bans on plea bargaining did not have a substantial overall impact on sentencing outcomes in any of the three jurisdictions in which evalua- tions are available; they did, however, affect the severity of sentences, especially for marginal offenders. In Alaska, although there were few marked changes in imprisonment rates, Rubinstein et al. (1980) conclude that there were some selected changes in sanction severity. Sentences did not become more severe where the-original charge was a violent felony or "high-risk"3 larceny, but drug cases experienced large in- creases in sentence severity as did "low-risk" burglary, larceny, and . . receiving cases. In Wayne County, Michigan, there was no substantial overall change in sentences for defendants processed by the court (including those dismissed and acquitted). The proportion of all defendants who received incarcerative sentences did not increase. There were, however, some 3 "High-risk" and "Iow-risk" characterizations were based on indicators of persistent . . . cr~m~na fry.
Sentencing Reforms and Their Effects 205 increases in the severity of prison terms imposed. The proportion of armed robbery defendants who received sentences of 5 or more years increased from 34 to 41 percent. The proportion of defendants who received sentences of at least the 2-year minimum increased by at least 50 percent, for "other assaults" (from 22 to 33 percent of defendants) and for felonious assaults (from 4 to 13 percent of defendants). MANDATORY MINIMUM SENTENCING LAWS In Wayne County, the percentage of convicted defendants who were incarcerated did not change markedly after passage of Michigan's fire- arms law (Heumann and Loftin, 1979~. However, the likelihood of incarceration after conviction did change significantly, from 57 to 82 percent for the marginal "other assault" offenders. Also, after the new law took effect, there was an increase in the length of sentences for imprisoned offenders. Of offenders imprisoned for felonious assault, the proportion sentenced to terms of 2 or more years increased from 30 to 71 percent. For offenders imprisoned for other assaults, the proportion receiving terms of 2 years rose from 59 to 81 percent after the law. There was little increase (from 87 to 93 percent) in use of 2-year terms for armed robbery offenders.4 Beha's (1977) data do not permit conclusions regarding changes in sanctions in Massachusetts. The substantial increase in the rate of ap- peals of lower court convictions suggests that the imposition of prison sentences increased substantially, but whether these sentences survived superior court processing is unknown. In New York State, the risk of incarceration for the small numbers of defendants who were convicted of drug offenses after passage of the Rockefeller drug laws increased significantly, but the steady decline in the number of drug felony convictions from 1972 to 1976 offset that development to yield a stable probability of incarceration given arrest. Overall and statewide, the proportion of persons arrested for a drug felony who were imprisoned remained essentially the same in 1972 (10.6 percent) and the first half of 1976 (11.6 percent). However, the severity of prison terms imposed on those New York drug offenders who were 4 Loftin and MeDowall (1981) report similar effects on a considerably expanded data set. Though they report no effect of the gun law on the expected time served by offenders charged with murder or armed robbery, the expected sentences for felonious assault and other assaults did increase more for cases involving guns. Similar results were found for the probability of prison among charged offenders.
206 RESEARCH ON SENTENCING THE SEARCH FOR REFORM sentenced to prison increased markedly. Under the old law (between 1972 and 1974) only 3 percent of sentenced drug felons received mini- mum sentences of more than 3 years. Under the new law, the use of long minimum sentences increased to 22 percent. Between September 1973 and June 1976, 1,777 offenders were sentenced to indeterminate lifetime prison terms, a sentence rarely imposed before the new drug law (Joint Committee on New York Drug Law Evaluation, 1978:99- 103~. CALIFORNIA'S DETERMINATE SENTENCING LAW Trends in Prison Use Prison use definitely increased after DSL, whether measured by the commitment rate to prison (commitments/population) or by the likeli- hood of a prison sentence after conviction in superior court. The in- creased use of prison has been accompanied by increased imprisonment of less serious, marginal offenders: persons convicted of less serious offenses constitute a larger proportion of persons sent to prison, and imposition of prison sentences has increased relative to jail sentences. However, these increases continue preexisting trends toward increased prison use in California and may not be an effect of DSL. Because of the greater certainty about lengths of prison terms, it was generally hypothesized that prison use would increase as a result of DSL. Under indeterminate sentencing laws, judges could not impose short prison sentences because all prison terms were for the statutory maximum sentence subject to earlier release at the parole agency's dis- cretion. It was widely believed that judges were hesitant to imprison persons convicted of less serious crimes for fear that they might be kept in prison unduly long. Under DSL's determinate provisions, a judge could impose a short sentence and know when the defendant would be released. Consistent with this hypothesis, most studies have found a definite increase in prison use, measured by commitment rates based on population and by rates based on superior court convictions. As indicated in Table ~5, the commitment rate for all offenses in- creased between 1976 and 1978 for the state as a whole and for individual jurisdictions within the state. Similar increases were generally found for the proportion of convicted offenders sentenced to prison in superior courts (see Table ~6), both across jurisdictions and for different kinds of offenses (except in Santa Clara County). When the observation period is extended to include multiple observations, however, several studies conclude that the increase in prison use after DSL is best viewed as a
Sentencing Reforms and Their Effects TABLE ~5 California Adult Prison Commitment Rate Commitments/100,000 Residents) Commitment Rate Jurisdiction 1976 (Before DSL) 1978 (After DSL) Males onlya State total 30.0 39.3 Counties Southern California 25.1 37.6 Los Angeles 27.9 39.1 9 other counties 22.5 35.9 San Francisco Bay 29.3 39.4 Alameda 25.0 46.0 San Francisco 50.2 83.7 7 others 26.5 37.1 Rest of state 37.8 44.8 10 Sacramento Valley 40.9 43.3 7 San Joaquin Valley 37.5 51.4 22 others 34.3 37.1 All adultsb State total 32.1 41.8 a Data from Lipson and Peterson (1980:Table 12~. The reported rates represent the number of males committed to state prisons per 100,000 total resident population (males and females). b Data from Brewer et al. (1980:Table 5~. The rates are total adult commitments (male and female) to state prisons per 100,000 total resident population. 207 continuation of a preexisting trend toward increased prison use in Cal- ifornia (Brewer et al., 1980; Casper et al., 1982; Ku, 1980; Lipson and Peterson, 1980), both for all offenses for the state as a whole (Figure ~2) and in San Bernardino and San Francisco counties (Casper et al., 1982~. Factors Contributing to Increased Prison Use Public and judicial attitudes toward criminals may simply have become more punitive in the late 1970s. However, several factors are also po- tentially important in accounting for the trend toward greater prison use in California. The Changing Impact of Probation Subsidies on Local Jurisdictions Brewer et al. (1980) note the contributing role of changes in the probation
208 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM subsidy program to counties. This program, initiated in 1965 to provide economic incentives for local jurisdictions to keep offenders under local supervision within their own communities, achieved this end through 1971. Because of growing dissatisfaction with local programs and es- calating costs of these programs, prison commitments began to rise again in the early 1970s. Under the provisions of the act, however, the amount of the probation subsidy was determined by the extent of prison use in a county; any increase in prison use relative to a baseline figure de- creased the probation subsidy the county received. Thus, as prison use began to increase, the amount of the probation subsidies received by counties decreased. With this decline in probation subsidies, prison com- mitments rose even higher, and the use of probation declined further. TABLE 4 6 Proportion of Convicted Offenders Sentenced to Prison in California Superior Courts Percentage to Prison Among Convictions Jurisdiction and Offense 1976 1978 (Before DSL) (After DSL) All offenses State totala 17.8 23.0 Counties Alamedab 14.2 23.2 Sacramentob 25.4 26.9 San BernadinoC 29.5 38.5 San FraneiscoC 25.0 31.5 Santa ClaraC 25.0 16.5 Burglary Alameda 17.8 42.5 Sacramento 23.0 21.3 San Bernadinoe 29.5 38.5 San Franciscoe 24.5 32.0 Santa Clarae 24.5 16.0 Robbery San Bernadinoe 65.0 63.0 San Franciscoe 44.0 49.5 Santa Clarae 59.5 57.0 a These data from the California Bureau of Criminal Statistics are reported in Brewer et al. (1980) and Lipson and Peterson (1980~. b Derived from data reported in Utz (1981:Appendix F). c Data from Casper et al. (1982:Figure 5~. ~ Data from Utz (1981:Table 39~. e Data from Casper et al. (1982:Figures 6 and 7~.
Sentencing Reforms and Their Effects 3 c' to o - cr: ~0 n z z 0 uJ tic < ~ 1Q a o 0 ~ 6 ~ 1 O o 1 0 It Z ~ > ~ Z UJ O C' o 1967 1969 1971 · · Percent to Prison Among Convictions ~ 0 Commitments/100,000 Population 20 a' / DS L I 60 _ i 0 - -' 1 1 1 o 973 1975 ~ 977 197 YEAR 209 50 40 of Z Z O UJ In I.L 0< Z Z UJ O 20 ~ A. 8g 1 o FIGURE =2 Prison use in California. SOURCES: a Data Mom Brewer et al. (1980:Table 5) and Lipson and Peterson (1980:Figure 2~; b Data from Brewer et al. (1980:Table 5~; c Data for 1979 were obtained from the California Department of Justice (1980~. Increased Seriousness Another factor in the increased use of prison is that the seriousness of cases sentenced in superior courts may have increased. This could have resulted from increases in judicial punitive- ness, increases in the seriousness of the cases that result in convictions, or changes in the distribution of cases between superior and municipal courts. Including elaborate controls for the seriousness of burglary cases dis- posed in superior courts in Alameda and Sacramento counties in 1976 (pre-DSL) and in 1978 (post-DSL), Utz's (1981:22-27) data indicate some increase in case seriousness between 1976 and 1978. Increases in prison sentences among those convicted in superior courts also may have resulted from changes in the pretrial filtering process that
210 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM affected the case mix in superior courts. In particular, a shift of less serious cases to municipal courts for final disposition would leave the superior courts with increasing proportions of more serious, prison- eligible cases. If so, the increase in prison sentences for superior court cases would be more apparent than real: the cases sentenced to prison remain essentially unchanged, while more cases of moderate seriousness are eliminated from the available pool of convictions. (However, this would not explain the increases in the prison commitment rate per population.) Major changes have occurred in the distribution of cases between superior and municipal courts. The proportion of total court dispositions for felony arrests that were handled in superior courts dropped dra- matically from 70.7 percent in 1968 to 30.4 percent in 1979. During this same period the proportion of felonies among superior court cases in- creased from 59.6 percent in 1968 to 89.1 percent in 1979. These changes result in part from statutory changes that permitted prosecutors to han- dle as misdemeanors or felonies certain offenses previously handled exclusively as felonies (Penal Code 17b(4~. Other changes similarly permitted judges to sentence certain cases as misdemeanors, even if they were filed as felonies (Penal Code 17b(5~. A variety of system changes could account for the recent concentration of felonies among superior court convictions and for recent increases in the prison sentence rates among these convictions. The increased rep- resentation of felonies at superior courts could have resulted from a shift in offending behavior to more Serious crimes, resulting in an in- crease in serious felonies at each stage of the process. Alternatively, there might have been a general shift to greater punitiveness manifested by prosecuting as felonies less serious offense incidents previously pros- ecuted as misdemeanors. These possible explanations for observed changes indicate the need to monitor and control for changes in the presentence filtering processes that affect the character of cases available for sentencing. Without such controls, changes in the way cases are filtered (which may be unrelated to a sentencing reform) could be mistakenly interpreted as changes in sentencing policy for "like" cases. Collecting data on control variables that reflect important aspects of the character of casesi.e., attributes that identify "like" cases for sentencing purposes increases the like- lihood of distinguishing sentence changes due to differences in the char- acter of cases available for sentencing from sentence changes due to real shifts in the sentencing policy for "like" cases. Demographic Shifts General demographic shifts, not mentioned in any of the studies, also may have contributed to the recent rise of prison
Sentencing Reforms and Their Effects 211 use in California. Both in the United States as a whole and in California, the anomalous pattern of a decrease in prison admission rates through the 1960s, during a period of rapidly rising crime rates, and an increase in prison admission rates in the 1970s, when crime rates increased much less, could be attributed to the changing demographic composition of the population (Blumstein et al., 1980~. In the 1960s the post-World War II baby-boom generation was moving into the high-crime ages, but as juveniles or first-offender adults these individuals were not likely to be sent to prison even if convicted. On the other hand, the increase in prison commitments in the 1970s occurred as a sizable portion of the baby-boom offenders became old enough to have developed adult crim- inal records. Based on population projections, these increases in prison commitments are likely to continue nationally at least until the end of the 1980s. In California the continuing high inmigration of persons ages 18 to 29 is likely to delay any reversal of the upward trend in prison commitments and prison populations in that state. Increased Punitiveness If the increases in prison use in California reflect a real shift toward increased punitiveness in the state, this increase should be reflected in increases in the proportion of persons committed to prison for less serious crimes and in increases in time served (con- trolling for offense seriousness). Data from both Ku (1980) and the California Bureau of Criminal Statistics on prison sentences do indicate a trend toward increased representation among prison commitments of the less serious offense of burglary. This apparent increase in burglars among prison commitments cannot be attributed to a shift in the seri- ousness mix for burglary offenders sent to prison (Utz, 1981~. Sparks (1981) finds that the greatest increases in prison use after conviction occur for less serious offenders, whether defined by offense type, prior criminal record, or custody status at the time of the offense. These changes served to narrow the differences in the likelihood of prison after conviction for cases of differing seriousness. In all cases, however, the pattern of increasing punitiveness for less serious cases began before DSL. Other studies find evidence of a slight increase in prison use relative to jail beginning in 1975, predating the implementation of DSL (Brewer, 1980; Casper et al., 1982~. This is consistent with a continuing increase in punitiveness as increasing portions of marginal cases are shifted from jail to prison. The weight of the evidence indicates no perceptible change in prison use as a result of DSL. The increases in prison use for superior court cases and associated shifts away from probation and jail sentences ev- ident after DSL implementation appear to be continuations of pre-
212 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM existing trends. These trends toward increased prison use are consistent with, and probably reflect the effects of some combination of increased punitiveness, general increases in the seriousness of cases handled at all levels of the criminal justice system, shifts of less serious cases from superior to municipal courts, and changes in the age structure of the population. Impact on Length and Disparity of Prison Terms Two issues are of central concern in considering the impact of DSL on prison terms: changes in the average severity of prison terms reflected in either increases or decreases in mean or median time served and changes in the variability or disparity in time served for similar cases. All the evidence points to a decrease in sentence lengths after DSL, but the post-DSL changes are part of a continuing trend that began before the law was implemented. There was also a tendency toward greater uniformity in sentences under DSL as sentence variation de- clined and the difference in sentences of men and women was essentially eliminated. Nevertheless, the range of sentences imposed for individual convicted offenses remained surprisingly broad. The impact of DSL on average prison terms was difficult to anticipate prior to implementation. The original base terms were selected to reflect recent time actually served under releasing policies of the California Adult Authority (the state parole board for adults). The good-time provisions under DSL, which allowed for a maximum of one-third off a sentence, the application of separate enhancements, and subsequent enactment of increased base terms all contributed to uncertainty in predictions about changes in average time served under DSL. A decrease in the variation or spread of prison terms was anticipated since an important goal of DSL was introduction of greater uniformity in sentences for offenders convicted of the same offense (Casper et al., 1982; Lipson and Peterson, 1980~. Length of Prison Terms Studies comparing the average length of terms under the old and new laws use both actual sentences imposed under DSL and adjusted DSL terms reflecting credits for jail time already served and/or good time off the sentence.5 These comparisons generally find decreases in mean or median time served under the new law, es- 5 Since most of the studies were undertaken in the first few years after ISLE imple- mentation, the number of individuals sentenced and subsequently released under DSL is
Sentencing Reforms and Their Effects 213 pecially when allowing for jail and maximum good time-discounts from the term imposed at sentencing. Brewer et al. (1980), for example, report that the mean time served for all offenses increased very slightly from the old to the new law (from 40.0 to 41.4 months) using the actual sentence imposed under DSL; but allowing for maximum good-time credits, the adjusted DSL mean time served is only 28.7 months. Sim- ilarly, for robbery, the mean time served from actual DSL sentences is higher than for actual time served before the law but is lower when adjusted for good time. For burglary, both the mean time served from actual DSL sentences and the mean from adjusted DSL sentences are lower after DSL than the mean time served found for prisoners released under indeterminate sentencing laws. Different post-DSL changes in time served for men and women were observed as the substantial gap between men's and women's terms was closed under DSL (Brewer et al., 1980~; see Table ~7. Greater uni- formity in time served by sex has been accomplished by DSL through increasing the terms of women imprisoned for offenses against persons, keeping women's terms for property offenses about the same, and re- ducing men's prison terms for all types of offenses. When the observation period is increased, the general decline in time served evident after DSL appears to be a continuation of a trend toward shorter terms that began several years before DSL implementation. Median prison terms between 1968 and 1975 for all offenses were con- sistently longer (about 36 months) than those in the preceding 23 years (about 26 months). From 1975 through 1978, however, the length of terms declined, falling to pre-1968 levels in 1978. The shorter DSL terms after discounting for jail and maximum good-time credits are fully con- sistent with this recent decline in time served. Variability in Prison Terms The statutory declaration that punishment is the primary purpose of imprisonment under DSL suggests that sen- tences for similar convicted offenses should receive similar sentences. In meeting this objective, reductions in the level of variation in prison terms for "like" offenses would be expected. Several of the studies explicitly addressing this issue report reductions in the variation or spread of prison terms after DSL when controlling for the offense of conviction, quite small. Information from the Department of Corrections indicates that, so far, with the admittedly limited experience with implementation of the early-release, good-time provisions, most prisoners have been released with maximum good time off their sentences (Brewer et al., 1980:1~15; Lipson and Peterson, 1980:25; Utz, 1981:150~.
214 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM TABLE i7 Changes in Length of Prison Terms by Sex Based on Statewide Data Mean Prison Terms, in Months Offense Time Served by Women as a Percent of Lime Served Men Women by Men All offenses Pre-DSL: 1972-1976 40.0 23.7 59 DSL adjusted:a 1977-1978 28.7 24.8 86 2nd degree burglary Pre-DSL: 1972-1976 30.0 19.5 65 DSL adjusted:a 1977-1978 18.4 16.0 87 Robbery Pre-DSL: 1972-1976 44.8 26.7 60 DSL adjusted:a 1977-1978 35.7 29.6 83 Assault with deadly weapon Pre-DSL: 1972-1976 40.9 22.3 55 DSL adjusted:a 1977-1978 32.9 34.7 105 a Adjusted to reflect maximum allowable good-time credits. SOURCE: Derived from data in Brewer et al. (1980:Tables 7 and 8~. particularly when discounted DSL terms are used. For men, for example, the comparisons of DSL terms actually imposed with time served before DSL in Brewer et al. (l980:Table 7) indicate that the standard deviation decreased from 20 to 50 percent for five of the seven crime types with increases in means (Brewer et al., 1980~. However, Casper et al. (1982:Table 17) note that the range of DSL sentences actually imposed in robbery and burglary cases is quite broad, due principally to the use of enhancements and consecutive terms on multiple charges in sentences for offenders convicted of the same crime type. In sum, despite the magnitude of the change in sentencing procedures under DSL, there is no compelling evidence of substantial changes in sentence outcomes attributable to DSL. While prison use increased and time served decreased after the new sentencing law, both changes rep- resented continuations of trends that began several years before deter- minate sentencing was implemented. SENTENCING GUIDELINES When the impact of various voluntary sentencing guidelines has been examined, there is little or no evidence of changes in sentencing prac-
Sentencing Reforms and Their Effects 215 tices. In particular, the relative use of different sanctions and the length of prison terms imposed have remained unchanged, and there has been little effect on the extent of variation in sentences imposed on like cases, as classified by the guidelines. This absence of substantial differences in sentences can be attributed largely to the nature of the guidelines themselves. In addition to being voluntary in their implementation, the guidelines were largely "descriptive," articulating past sentencing practices without intending to substantially alter them. As a result, very little change from past practices was expected from even strict compliance with the guide- lines. Instead, the guidelines were intended to provide judges with a description of prevailing practices in their jurisdiction, to be used as a standard in their own sentencing decisions and to serve as a basis for possible reconsideration of those practices in an iterative process of description, evaluation, and modification of the guidelines. The Minnesota guidelines represent a complete departure from this model of voluntary/descriptive guidelines. The Minnesota guidelines are presumptive, having the force of a legislative requirement, and the pre- scribed sentences represent a deliberate departure from past sentencing practices. On the basis of early in-house data, Minnesota's presumptive guidelines appear to have significantly altered sentencing patterns in that state. The Minnesota guidelines included an explicit policy choice to increase the use of prison for serious offenses against persons by offenders with limited criminal histories while decreasing the use of prison for property offenders, regardless of their prior criminal history. Consistent with the guidelines, the proportion of total commitments to state prisons rep- resented by individuals convicted of person offenses increased from 32 to 46 percent. There was no increase in the proportion of convictions for person offenses; cases with presumptive prison sentences represented about 13 percent of convictions before (1978) and after (198~1981) guidelines implementation. Table ~8 provides further evidence of the effectiveness of the guidelines in shifting prison sentences from property to person offenses. The proportion of low-history offenders convicted of serious offenses who were sentenced to prison increased sharply, from 45 to 77 percent after implementation of the guidelines, while the proportion of high-history offenders convicted of the least serious fel- onies who were sentenced to prison decreased from 53 to 16 percent (Knapp, 1982; Minnesota Sentencing Guidelines Commission, 1982~. Another explicit choice articulated in the Minnesota guidelines was in the direction of uniform sentences, in particular that sentences should be neutral with respect to the race, sex, and socioeconomic status of
216 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM the defendant. One indicator of the success of the guidelines in achieving more uniform sentencing is the rate of departure of sentences from the guidelines for different demographic groups. The data in Table =9 indicate that variations in sentences remained after implementation of the guidelines. The total in/out departure rate was reduced from 19.4 to 6.2 percent after guidelines were implemented, and similar reductions in departures were found for all demographic groups. Nevertheless, minority, male, and unemployed offenders continued to experience higher rates of departures from the presumptive sentences, and these depar- tures tended to be in the direction of more severe sentences: presumptive '~outs" who were in fact sentenced to the state prison. The mix of cases differed sharply among different demographic groups: cases of convicted whites, females, and employed offenders were more likely to involve low-seriousness offenses and low criminal history scores. Departure rates were also generally lower for these less serious cases; the typical reasons for departures related to the extent of injury to victims- conditions that do not apply in low-seriousness property of- fenses. These differences in the distribution of cases could affect com- parisons of departure rates across demographic groups. As a minimum control for the potential influence of differences in the distribution of cases, departure rates were estimated separately among presumptive "ins" and presumptive "outs." As indicated in Table ~10, the differ- ences across race and sex remain after minimally controlling for case distribution and the differences between employed and unemployed offenders are increased. TABLE ~8 Shift in Prison Sentences From Property to Persons Offenses Under Minnesota Sentencing Guidelines Percent of Cases Sentenced to State Prisons Cases Sentenced 1978 Under Offense Criminal Baseline Guidelines, 198 Severity Level History Score Cases 1981 VII, VIII, IX 0,1 45 77 (high) (low) I, II 3, 4, 5 53 16 (low) (high) SOURCE: Data from Minnesota Sentencing Guidelines Commission (1982). Also available from Knapp (1982).
Sentencing Reforms and Their Effects TABLE ~9 In/Out Departure Rates for Cases Sentenced Under Minnesota Sentencing Guidelines in 198~1981 217 Percentage of Percentage of Severe Departures Lenient Departures ~ Among Total Among Total Percentage of (Presumptive (Presumptive Demographic Departures, "Outs" Who Were "Ins" Who Were Group All Cases Sentenced "In") Sentenced "Out") Total 6.2 3.1 3.1 Race Whites 5.2 2.6 2.7 Blacks 9.6 4.9 4.7 Native Americans 12.4 7.5 4.9 Sex Males 6.5 Not reported Not reported Females 3.1 Not reported Not reported Employment Employed 3.4 0.2 3.2 Unemployed 8.9 5.0 3.9 SOURCE: Data from Minnesota Sentencing Guidelines Commission (1982~. The actual departure rates were compared with an independent as- sessment of justified departures by the guidelines commission, and sim- ilar differences by race and sex were found. Based on this analysis, it appears that differences in case seriousness account for a large part of the differences found in departure rates by sex and raced The rate of severe departures for blacks relative to whites is, however, somewhat higher than expected and remains a matter of concern. PAROLE GUIDELINES Departure Rates As we noted earlier, evaluations of parole guidelines have noted de- parture rates that have varied from under 10 percent (Minnesota in 1979) to 68 percent (Washington in 1976) (ADL, 1981c,d). Unfortu- nately, departure rates alone are not informative their salience de- pends on several factors, including the specificity of policy guidance, the width of guidelines ranges, whether the applicable guideline has 6 No independent assessment is available for the unemployment variable.
218 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM TABLE ~10 Departure Rates Among Presumptive "Ins" and Presumptive "Outs" (Under Minnesota Sentencing Guidelines) Percentage Sentenced Percentage Sentenced Demographic "In" Among Presumptive "Out" Among Presumptive Group "Outs"a "Ins" Race Whites 3.1 15.4 Blacks 6.8 10.7 Native Americans 9.5 17.2 Sex Male 4.0 14.2 Female 1.6 25.9 Employmentb Employed 0.2 46.4 Unemployed 6.3 18.8 a Severity level VI offenses are excluded from the presumptive "outs" because some of these offenses are in fact presumptive "ins" under the terms of separate mandatory sentencing laws. b The departure rates by employment status are estimated from data on departure rates and the distribution of cases for different categories of offenders available from the Minnesota Sentencing Guidelines Commission. The figures estimated here are approxi- mations based on estimates of both the number of departures and the total number of cases in each category. They include severity level VI offenses among presumptive "outs." SOURCE: Data from Minnesota Sentencing Guidelines Commission (1982~. been correctly identified, and clarity about the character of a departure. To elaborate this last point, we repeat our earlier distinction between consistency and compliance: a decision may be inconsistent with guide- lines but compliant if the case is one which the developers intended the offender to receive an aggravated or mitigated sentence outside a normal range; conversely, a decision may be consistent but noncompliant if the case is one that should have been handled by a departure but was not. Consequently, data on departure rates are necessarily ambiguous unless full and detailed contextual information is available. Changes in Disparity and Severity All of the studies we reviewed that assessed the impact of parole guide- lines on disparity found evidence that the guidelines reduced sentencing disparities. For Oregon, Mueller and Sparks (1982:2~21, 36) concluded that, controlling for offense seriousness and using the parole board's
Sentencing Reforms and Their Effects 219 offender scoring system, the variability of prison terms in that state was less in 1976 and 1978, under guidelines, than it had been in 1974, before guidelines were implemented. The Arthur D. Little study of the impact of the U.S. parole guidelines on disparity compared actual times served by prisoners convicted of robbery and selected property offenses who were released in 1970 (preguidelines) and- 1979 (postguidelines) and found "measurably less dispersion in the distribution of actual time served" for the 1979 releases that could not be explained by reduced variability in sentences imposed by judges (ADL, 1981e:3~. Finally, for Minnesota, Arthur D. Little found that for persons convicted of aggra- vated robbery, "offenders released in 1979 under the guidelines system tended to serve more nearly the same amount of time . . . when stratified into subgroups based upon prior history" than did aggravated robbery prisoners who were released in 1974, before guidelines (ADL, 1981d:63~. Thus it appears that well-managed parole guidelines systems can operate to reduce sentence disparity among persons imprisoned. Only one study has addressed the question of whether the overall severity of prison sentences served increased with the implementation of parole guidelines. Mueller and Sparks (1982:20) concluded that in Oregon between 1974, before guidelines were implemented, and 1978, when guidelines had been in effect for several years, there was "an overall increase in severity of terms." They cautioned, however, as we do in regard to the California evaluations, that one cannot conclude that "the guidelines caused the observed changes" (Mueller and Sparks, 1982: 1). CONCLUSIONS SUBSTANTIVE FINDINGS Most of the studies we reviewed reported that formal compliance with the procedural requirements of reforms has been achieved. Prosecutors have refrained from bargaining, judges have imposed the mandated sentences on convicted offenders, and parole boards have released pris- oners according to guideline requirements. But this behavioral change often represented compliance more in form than substance. When par- ticipants considered the new rules inappropriate, they routinely at- tempted to circumvent the procedural changes by filtering out those cases they believed should not be subject to those rules. With respect to sentence outcomes, the impact of the sentencing reforms has been modest. There has been some increase in prison use, some increase in
220 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM sentence severity for marginal cases, and some decrease in disparity. These changes have varied by type of reform and jurisdiction. Formal Compliance The mechanisms for achieving compliance with new sentencing rules differ. Plea-bargaining and parole reforms have been successfully im- plemented through administrative orders. The three evaluations of plea- bargaining bans indicate that when prosecutors sought to abolish plea bargaining in general or in a particular form and were serious about it they were able to do so. Similarly, the evaluations of parole guidelines indicate that parole examiners attempt to follow the guidelines of parole boards. High compliance of prosecutors and parole boards with sentencing reforms, when it occurs, is likely to be the result of both administrative controls within the relevant agencies and favorable responses to the sentencing innovations by participants. Prosecutors working in systems that have prohibited plea bargaining tend to prefer the new regime (although defense lawyers dislike it). For prosecutors, the shift from haggling over sentences and charges to expending greater effort to de- velop cases for trial enhances their self-image and demands more "professional" behavior. The support for parole guidelines sometimes expressed by hearing examiners and parole board members, too, was not surprising, since the guidelines represent the policy of the board that initiated their development, can be changed by the board when guidelines and practice tend to diverge, and relieve individual members of some of the difficulties of decision making. Judges, who traditionally operate as independent agents relatively free of administratively imposed changes and organizational controls, have complied with new sentencing provisions only when changes have been mandated by statute, as in the instances of mandatory and deter- minate sentencing laws and statutory sentencing guidelines. In the ju- risdictions studied, voluntary sentencing guidelines have produced no measurable judicial compliance. Adaptive Responses There is much evidence of adaptive responses by officials who alter case- processing methods in order to circumvent new rules and procedures for some categories of offenders. Increased case screening or other early disposition of cases effectively avoids application of sentencing laws. Voluntary sentencing guidelines have had no discernible impact on ju- dicial behavior or court processing; they have simply been ignored.
Sentencing Reforms and Their Effects 221 Sentencing Outcomes, Disparity, and Marginal Cases Modest changes in sentencing outcomes, particularly in the use of prison sentences, followed implementation of sentencing innovations. In- creases in sentence severity were most typically found in those marginal cases for which imprisonment had been most uncertain prior to the innovations. Both high-seriousness cases, for which imprisonment had been and continued to be likely, and low-seriousness cases, which were unlikely to be given a prison term, experienced little change. Concerns that defendants with minor records or those accused of minor offenses would become enmeshed in the rigidity of the new scheme were expressed by critics of plea-bargaining bans and mandatory sen- tencing laws. The evidence from virtually every study indicates that these apprehensions were well founded: marginal offenders who did not ben- efit from early filtering decisions have been subject to harsher sentences. Ironically, while the severity of those sanctions that were imposed for certain offense types sometimes increased, the rate of persons impris- oned and the likelihood of imprisonment declined. In New York, for example, about 11 percent of those arrested on drug charges were im- prisoned in both 1972 and the first half of 1976, but the number of drug arrests was much smaller in 1976 so that there were fewer prison sen- tences imposed overall. One of the goals of the sentencing reforms was the reduction of disparity in time served by like offenders with like cases. Several studies present some evidence suggesting that reduction in sentence variation did result from DSL in California, Minnesota's sentencing guidelines, and parole guidelines in several jurisdictions. However, this does not address variations in case processing before the sentencing stage and their effects. For example, two cases in which the underlying offense behavior is the same might still result in different charges at conviction and thus still be sentenced differently as a result. METHODOLOGICAL CONCERNS Our review of impact evaluations suggests the need to address a number of key methodological issues in subsequent evaluations of criminal jus- tice reforms generally and of sentence reforms particularly. The Need for Extended Observation Periods Several of the evaluations reviewed here involved simple two-point, proreform and postreform designs that are inadequate for a number of reasons. First, such designs do not permit distinguishing discrete changes
222 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM or effects associated with a reform from the continuation of preexisting trends. Multiple observations of outcomes before implementation of a change are highly desirable, as indicated by several studies of DSL in California; those extended observations were crucial to the conclusion that introduction of determinate sentencing in California resulted in no substantial changes in sentencing outcomes in that state. Ideally, the postreform observation period should also extend for multiple observations after the reform, because case processing often takes months or even years. Thus a sentencing reform that is to apply to all cases involving offenses committed after January 1 of a year may not be applied to any substantial number of cases until well into the second year after the reform is implemented. If the reform increases case-processing time, there will be further delay before full impact. To the extent that cases disposed early differ significantly from cases that take longer to resolve, early evaluations of effects are likely to be biased and may indicate opposite effects from later evaluations. The possibility of delayed impact was strongly suggested both by the dramatic increase in early guilty-plea rates in California and under the New York drug laws, where median disposition times increased and conviction and imprisonment rates for drug felonies fell immediately after the law went into effect. A gradual increase of sanction rates in New York by 1976 and a drop in early guilty pleas in the second year under DSL in California suggest the need for more data points. To avoid possible spurious findings of effects arising from delays, evaluations should routinely include measures of case-processing times and changes in work loads and backlogs. These variables are important not only as direct indicators of impact, but also for identifying necessary follow-up periods after a reform. The Need for Outcome Measures at All Levels of Case Processing Evaluations are often limited to aspects of the process directly affected by a reform and fail to address processing at earlier or later stages in the criminal justice system. This narrowness of focus fails to acknowl- edge the complexity of criminal case processing and the many oppor- tunities for the exercise of discretion that it affords. While in a literal sense criminal sentences are limited to the sanctions imposed by the court on convicted offenders, the character of these sentence outcomes is substantially influenced by factors determining which cases are actually available for sentencing. If those cases least likely to end in a prison sentence if convicted are
Sentencing Reforms and Their Effects red = 223 weeded out by changes in screening, charging, case dismissal, or shifting final disposition from higher to lower courts, the cases that reach the higher courts will be increasingly restricted to the more likely prison cases. An observed increase in prison use among higher court convictions then might be more apparent than real, because it derives at least in nart from a chance in the mix of cases at the higher court rather than from a real change in policy to extend prison use to cases previously sentenced to nonprison outcomes. The importance of changes in the filtering process is clearest in the evaluations of the New York drug laws (Joint Committee on New York Drug Law Evaluation, 1978) and the mandatory sentencing law for firearms violations in Detroit, Michigan (Heumann and Loftin, 1979~. In both jurisdictions prison use for convicted offenders increased dra- matically after the reform, but conviction rates fell and fewer cases entered the system; consequently, there was virtually no change in the proportion of all cases entering the system that resulted in imprisonment. The considerable opportunities for filtering cases before they reach the sentencing stage cannot be ignored. The need to adequately address the effects of changes in filtering is a central lesson from our review. The Need for Adequate Controls for Changes in Case Attributes General changes in the character of cases particularly changes in the seriousness of cases are related to but certainly not limited to the filtering process. Case attributes relevant to sentencing outcomes might also be affected by general changes in offending patterns and demo- graphic changes in the offender population. Failure to control for any resulting changes in ease attributes before and after a reform can seri- ously jeopardize the validity of conclusions about the impact of that reform on case outcomes at various stages, particularly sentencing out- comes. The Need for Qualitative Analysis of System Functioning Many evaluations are limited entirely to statistical analysis of abstracted case-processing data, often available from centralized automated data systems. While such analyses can provide aggregate average character- izations of case processing for large numbers of cases, they seldom provide adequate data to understand the ways courtroom participants alter their behavior to cope with the changes. The complexity of the sentencing process strongly indicates that statistical research based on
224 RESEARCH ON SENTENCING: THE SEARCH FOR REFORM official records cannot adequately address the system impact of legal changes. Although statistical analyses are an important component of evaluation research, they must be augmented by extensive use of par- ticipant observation, systematic interviewing, and other qualitative methods.
