21
Citation: 33 Criminology 283 1995 Content downloaded/printed from HeinOnline (http://heinonline.org) Mon Aug 26 10:36:47 2013 -- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and Conditions of the license agreement available at http://heinonline.org/HOL/License -- The search text of this PDF is generated from uncorrected OCR text. -- To obtain permission to use this article beyond the scope of your HeinOnline license, please use: https://www.copyright.com/ccc/basicSearch.do? &operation=go&searchType=0 &lastSearch=simple&all=on&titleOrStdNo=0011-1384

The Impact of Sentencing Guidelines on Jail Incarceration in Minnesota*

Embed Size (px)

Citation preview

Citation: 33 Criminology 283 1995

Content downloaded/printed from HeinOnline (http://heinonline.org)Mon Aug 26 10:36:47 2013

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and Conditions of the license agreement available at http://heinonline.org/HOL/License

-- The search text of this PDF is generated from uncorrected OCR text.

-- To obtain permission to use this article beyond the scope of your HeinOnline license, please use:

https://www.copyright.com/ccc/basicSearch.do? &operation=go&searchType=0 &lastSearch=simple&all=on&titleOrStdNo=0011-1384

THE IMPACT OF SENTENCING GUIDELINESON JAIL INCARCERATION INMINNESOTA*

STEWART J. D'ALESSIOLISA STOLZENBERG

Indiana University-Purdue University Fort Wayne

Although available empirical evidence suggests that Minnesota'sDeterminate Sentencing Law has had little effect on prison incarcera-tion, it is still uncertain whether the sentencing guidelines affected jailuse. A few recent studies imply that the guidelines have had a positiveeffect on jail incarceration rates. Accounts have pointed to preexistingtrends, more severe sanctioning of repeat property offenders, and judi-cial concern with prison overcrowding as possible underlying causes ofthis observed increase. Using longitudinal data and an ARIMA studydesign, we investigate the validity of these competing explanations. Ourfindings show that the onset of the sentencing guidelines increased judi-cial use of the jail sanction beyond the effect of preexisting trends. Inaddition, the effect of mitigated dispositional departures from the noprison/prison outcome on jail use is salient only when prison popula-tion levels are high. This latter finding supports the thesis that judicialconcern with prison overcrowding motivated judges to circumvent theguidelines in order to shift the burden of incarcerating offenders fromthe state to the local level. The policy implications of these results fordeterminate sentencing reform are discussed.

Most research on the effects of Minnesota's determinate sentencingreform has focused primarily on whether the sentencing guidelinesincreased proportionality and equity in sanctioning criminal offenders.This research shows overwhelmingly that the guidelines not only reducedsentencing disparity but also maintained the prison population withinexisting capacity limits (Blumstein et al., 1983). Yet, as several socialscientists have noted, research conducted to date has been curiously silentabout the effect of the guidelines on jail incarceration (Parent, 1988;Tonry, 1987). Despite evidence that jail and prison populations are inti-mately linked (Mays and Thompson, 1988) and that jail incarceration rates

* An earlier version of this article was presented at the 1994 annual meeting of theAmerican Society of Criminology, Miami, Florida. We thank the anonymous reviewersfor their helpful comments and suggestions. We also are grateful to Sue Carter forproviding access to the data of the Minnesota Sentencing Guidelines Commission. Theopinions expressed are those of the authors and do not necessarily reflect the views ofthe Minnesota Sentencing Guidelines Commission.

CRIMINOLOGY VOLUME 33 NUMBER 2 1995 283

D'ALESSIO AND STOLZENBERG

increased markedly after the onset of the sentencing guidelines (Mietheand Moore, 1989; Minnesota Sentencing Guidelines Commission [MSGCJ,1991), scholars have not sought critically for possible connections betweenthe two. A review of the limited literature on this topic suggests threepossible alternative explanations. One explanation proposes that theobserved rise in Minnesota's jail incarceration rate resulted from the con-tinuation of a preexisting trend. A second explanation suggests that thesanctioning of property offenders, especially those with a criminal record(hereafter repeat property offenders), accounts for the observed increasein jail use. A third explanation, which draws largely from the insights ofRothman (1971, 1980) and Weber (1978), suggests that the organizationalimperative to maintain Minnesota's prison population within mandatedpopulation levels motivated judges to circumvent the guidelines in orderto transfer the burden of incarcerating convicted felony offenders from thestate to the local level.

Using longitudinal data and an ARIMA study design, we investigate thevalidity of these competing explanations. Regardless of which explanationis supported, the results of this study are important for gaining furtherinsight into the connection between determinate sentencing and jail incar-ceration. In addition, the findings will help determine the utility of consid-ering only the existing prison capacity as a policy constraint in undertakingdeterminate sentencing reform.

DETERMINATE SENTENCING ANDINCARCERATION

Recent research has documented a strong connection between determi-nate sentencing reform and rising prison populations in Illinois (Goodsteinand Hepburn, 1986), Connecticut (Hepburn and Goodstein, 1986), Florida(Bogan, 1990), Pennsylvania (Kramer and Lubitz, 1985), and the federalprison system (Mays, 1989). 1 Theorists attribute this increase followingthe inception of determinate sentencing to policymakers' failure to scaledown criminal penalties to reflect more accurately the time served inprison (von Hirsch, 1976). Within indeterminate sentencing structures,criminal sentences tend to be artificially inflated because judges expectthat incarcerated offenders will be paroled before completing their fullprison terms. With the implementation of determinate sentencing and theelimination of parole, however, the accumulation of "good time" credits istypically the only means of reducing an inmate's prison sentence. Accord-ingly, the Committee for the Study of Incarceration recommended that

1. Although California's prison population rose markedly after determinate sen-tencing was established, Cohen and Tonry (1983) reported that the increase resultedfrom a preexisting trend.

SENTENCING GUIDELINES IN MINNESOTA

policymakers reduce the severity of criminal penalties when undertakingdeterminate sentencing reform (von Hirsch, 1976). In practice, however,sanctions have not decreased in severity because many state legislaturesthat have implemented determinate sentencing have succumbed to publicpressure for stiffer criminal penalties (Cullen and Gilbert, 1982). Becausepolicymakers are unwilling to lessen the severity of criminal sanctions (i.e.,number of commitments to prison and average length of confinement) forfear of alienating the public, determinate sentencing is much more puni-tive than the reformers had intended.

Despite considerable evidence suggesting that determinate sentencingaccelerates prison growth, several empirical studies report that Minne-sota's sentencing guidelines did not increase prison populations (Hepburnand Goodstein, 1986; Knapp, 1982, 1984; Miethe and Moore, 1985).Indeed, as Mays (1989:192) notes, "the most notable exception to the gen-eral upward spiral of prison populations is the State of Minnesota." Acomprehensive study undertaken by the Minnesota Sentencing GuidelinesCommission (1984), for example, found that Minnesota's prison popula-tion remained within capacity limits during the first three years followingthe establishment of the sentencing guidelines. Similarly, Miethe andMoore (1989) reported that although the imprisonment rate rose in Min-nesota immediately after sentencing guidelines were established, thepostguideline imprisonment rate remained well below the preguidelinerate.

Minnesota's unique success is attributed to the MSGC's decision to bindsentencing practices to existing correctional resources (Knapp, 1982, 1984;Tonry, 1991, 1993; von Hirsch and Hanrahan, 1981). The MSGC theorizedthat if prisons became overcrowded, then prosecutors, judges, prisonadministrators, and others in the criminal justice system might take correc-tive action that potentially could conflict with the philosophical principlesof determinate sentencing (Knapp, 1982; von Hirsch, 1982). Humanitarianconcerns, appropriation problems, and possible legal challenges to theconstitutionality of the determinate sentencing law also were used to jus-tify linking determinate sentencing with available prison capacity (MSGC,1984).

The Minnesota state legislature also furnished the MSGC with theauthority to modify the guidelines when circumstances warranted suchaction. In 1982, when data showed that prison commitments and averagelength of confinement were increasing at a rate greater than originallyanticipated, the MSGC modified the guidelines to avert projected prisoncrowding (Miethe and Moore, 1989). These modifications included reduc-ing certain presumptive lengths of sentences, shortening mandatory prisonsentences by the amount of "good time" inmates earned, and amendingjail credit policies to award credit for time served in jail while on a split

D'ALESSIO AND STOLZENBERG

sentence if the probation later was revoked and the offender imprisoned(MSGC, 1984). These changes reduced the average prison term from 27months in 1982 to 22 months by early 1984. A projection analysis indi-cated that the policy changes would maintain Minnesota's prison popula-tion within existing prison capacity (Knapp, 1983).

Although many researchers have examined the impact of sentencingreform on prison populations, very little research explores the effect ofdeterminate sentencing on jail incarceration. This neglect is surprising,given that approximately 80% of the convicted offenders in Minnesotareceive a stayed sentence (MSGC, 1991). Further, unlike determinate sen-tencing systems in states such as Pennsylvania and Washington, Minne-sota's guidelines encompass only state imprisonment and make norecommendations about county jail sentences. Although the state legisla-ture gave the MSGC the authority to develop guidelines for nonprisonsentences, the Commission believed that developing guidelines for stayedsentences would encumber its work unduly (MSGC, 1984). Moreover, bygranting judges greater flexibility in disposing of cases that involved lessserious offenses, the Commission mollified political resistance to theguidelines by those deeply committed to rehabilitative ideals. Accord-ingly, Minnesota's guidelines afford judges an inordinate amount of discre-tionary latitude in determining type and severity of nonprison sentences(Kramer et al., 1989).

A small amount of empirical evidence also suggests, but does notdemonstrate conclusively, that jail rates rose markedly after the establish-ment of the Determinate Sentencing Law. The MSGC (1991), for exam-ple, reports that the rate of jail use for all convicted felons increased from35% in fiscal year 1978 to 50% in fiscal year 1983. Miethe and Moore(1989) also found that the use of jail as a condition of a stayed sentenceincreased sharply from the preguideline to the postguideline period inMinnesota, from 45% in 1978 to 66% in 1984.

Figure 1 depicts the monthly jail incarceration rate (i.e., the number ofconvicted felons who received a jail sentence per 100,000) and the impris-onment rate (i.e., the number of convicted felons who received a prisonsentence per 100,000) for Minnesota over time. The vertical lines repre-sent the establishment of the determinate sentencing law and the twomajor adjustments made to the guidelines by the MSGC. Although Figure1 shows that the sentencing guidelines had relatively little effect on Minne-sota's imprisonment rate, it provides strong visual evidence that the jailincarceration rate did increase markedly following the onset of the guide-lines.

If the jail incarceration rate in fact increased after the sentencing guide-lines were actuated, as suggested by Figure 1, what accounts for this

SENTENCING GUIDELINES IN MINNESOTA

Figure 1. Jail Incarceration and Imprisonment Rates per100,000

Rate per 100,00020 1 1

1 13 25 37 49 61 73 85 97 109 121 133 145 157

Month

NOTE: The preguideline (7/177-6130/78) and postguideline (9/1/80-12/31/92) serieswere joined to reflect a continuous series.

upward trend? Three lines of explanation have emerged. One explana-tion, advanced by the MSGC (1991), suggests that the increase in Minne-sota's jail incarceration rate resulted from a preexisting trend. Thisconclusion remains tentative, however, because prior research reliedexclusively on the one-group, pretest-posttest research design. Suchresearch designs are limited because they cannot disentangle discretechanges or effects associated with an intervention from the continuation ofa preexisting trend. In short, then, it is difficult to determine from theresults of previous studies whether the observed increase in jail ratesresulted from determinate sentencing or from a preexisting trend.

A second explanation is that the sanctioning of property offenders,especially repeat property offenders, accounts for the observed rise in jailuse. When the guidelines initially were developed, the sentencing grid wasconstructed so that first-time violent offenders would be sentenced moreharshly than repeat property offenders (MSGC, 1984). Many judgesobjected to this modification because they believed that the guidelinestreated repeat property offenders too leniently (Miethe and Moore, 1988).

D'ALESSIO AND STOLZENBERG

This reaction may have motivated judges to impose a jail sentence onrepeat property offenders, who normally would have received a prisonsentence before the guidelines, rather than releasing them on probation.

According to the third explanation, the organizational imperative tomaintain Minnesota's prison population within acceptable limits moti-vated judges to circumvent the guidelines (Tonry, 1987). The work ofRothman (1971, 1980) and Weber (1978) may help to provide a theoreticalframework for understanding this perspective. According to these theo-rists, the philosophical principles that provide the initial impetus and theunderlying rationale for an institutional reform frequently conflict with theimmediate, pragmatic objectives of the organization. Over time, theabstract philosophical principles that form the groundwork for a reformare supplanted by "policies and activities ... [that] maximize the rewardsand minimize the strains for the organization" (Chambliss and Seidman,1971:266). This "co-optation," which manifests itself through the processof reform, conflict, and displacement, is known as the "dialectic of con-science and convenience" (Rothman, 1980) and the "routinization of cha-risma" (Weber, 1978).

In Minnesota the organizational imperative to maintain the state'sprison population within acceptable limits reportedly has received morewidespread support than the philosophical goals associated with determi-nate sentencing. As Knapp (1984:189) readily notes, "the more immedi-ately pragmatic concern of the level of prison populations, however,currently appears to cause more compunction than the more philosophicalconcern with sentence proportionality and uniformity."

Recent empirical work by Stolzenberg and D'Alessio (1994) tends tosupport this view. In a longitudinal study of determinate sentencing inMinnesota, these authors found that the guidelines had an immediate andconsequential effect on reducing sentencing disparity, but that sentencinginequality in the no prison/prison judicial decision began to revert topreguideline levels as time passed. They theorized that this reversionresulted from judicial attempts to constrain the growth of Minnesota'sprison population. The results of this study, coupled with the generalobservation that dispositional departures from the guidelines are generallymitigating rather than aggravating (Miethe and Moore, 1985; MSGC,1991), suggest that researchers should examine "jail use patterns to dis-cover if the guidelines had indirect or displacement effects changing localconfinement practices" (Parent, 1988:188).

Guided by the theoretical explanations presented above, we broadenthe scope of previous inquiry by using an ARIMA study design and datacalibrated in monthly intervals to test the following three hypotheses:

1: The increase in the jail incarceration rate is due to a preexistingtrend. Thus, we expect a dummy-coded "intervention" variable,

288

SENTENCING GUIDELINES IN MINNESOTA

which measures the onset of the guidelines, to have little or no effecton jail rates.

2: With an increase in the rate of repeat property offenders who receivea nonprison sentence, the jail incarceration rate increases.

3: When prison populations are high, jail rates are increased by miti-gated dispositional departures from the no prison/prison sentencingdecision.

Our analysis also includes controls for several legal and extralegal factorsto attenuate the possibility of spurious results. To the best of our knowl-edge, no research conducted to date has explicitly evaluated the validity ofthese alternative explanations.

DATA, VARIABLES, AND METHOD OF ANALYSIS

DATA

The longitudinal data for this analysis are drawn from the Inter-Univer-sity Consortium for Political and Social Research (ICPSR) and from theMSGC.2 The preguideline data are for fiscal year 1978 (July 1, 1977through June 30, 1978) and account for approximately 50% of the con-victed felony offenders sentenced in the state for that year. The postguide-line data were available for all persons convicted of felony offensescommitted from May 1, 1980 through December 31, 1992.3 We use monthrather than year as our unit of analysis because monthly data are consid-ered superior to yearly data for interpreting change (McCleary and Hay,1980) and for reducing the confounding of history effects (Cook andCampbell, 1979). Analyzing monthly data also permits greater flexibility

2. The ICPSR data set used in this study is titled "Evaluation of Minnesota'sFelony Sentencing Guidelines, 1978-1984." Terance Miethe and Charles Moore haveconducted a series of studies using this data set, including an assessment of the effects ofMinnesota's sentencing guidelines on prosecutors' charging practices, plea negotiations,and sentencing decisions (Miethe, 1987; Miethe and Moore, 1985, 1989; Moore andMiethe, 1986).

3. In addition, we made two adjustments to the data. First, because jail use datawere unavailable for a 22-month period immediately preceding the onset of the sen-tencing guidelines, we followed Stolzenberg and D'Alessio's (1994) recommendation ofjoining the pre- and the post-guideline series to reflect a continuous series. The exclu-sion of these data should not be a serious limitation because it decreases the likelihoodof reactivity bias (Campbell and Stanley, 1963). That is, by excluding data immediatelyprior to the onset of the guidelines, we reduce the probability that our results would bevitiated by judges' prematurely modifying their sentencing behavior. Second, we elimi-nated the first four months of postguideline data from the study because the number offelony cases sentenced under the Determinate Sentencing Law was insufficient to per-mit any meaningful statistical analyses. The exclusion of these four months reduced thenumber of postguideline time periods to 148.

289

D'ALESSIO AND STOLZENBERG

in applying more sophisticated and more efficient statistical procedures,such as ARIMA, because of the increase in the number of observations.

VARIABLES

DEPENDENT VARIABLE: JAIL INCARCERATION RATE

The jail incarceration rate is operationalized as the total number of fel-ony offenders for whom the judge pronounced jail or workhouse time as acondition of a stayed sentence, divided by the state population and multi-plied by 100,000. We use the jail incarceration rate rather than the jailpopulation rate as our endogenous variable because jail population meas-ures are confounded by pretrial offenders, previously confined offenders,and jail release practices.

INDEPENDENT VARIABLES

Sentencing guidelines variables. We analyze the effect of the onset ofMinnesota's sentencing guidelines on jail use with a dummy variablecoded 0 for sentences passed before May 1980 and 1 for sentences passedthereafter. On the basis of Hypothesis 1, we expect the actuation of thesentencing guidelines to have little effect on jail rates.

We do not anticipate, however, that the impact of the guidelines will becaptured fully by the one intervention measuring its inception. The sen-tencing guidelines also were amended on two occasions; these modifica-tions may have changed or intensified the impact of the guidelines on jailuse. To analyze their effect, we added two dummy-coded variables to theanalysis. The first guideline modification variable is coded 0 for sentencespassed before November 1983 and 1 for sentences passed thereafter. Thisvariable captures the MSGC's attempt to adjust the guidelines to avertprojected prison overcrowding.

The second guideline modification variable models the MSGC's deci-sion to revise the guidelines so that existing correctional resources wouldcease to be a legitimate constraint on sentencing. The modificationsincreased the severity of prison sentences for drug offenses, sexual assault,and homicide. Although they reportedly increased Minnesota's prisonpopulation (Frase, 1993), the modifications' net effect on local jail usecould not be ascertained (MSGC, 1990). This variable was coded 0 forsentences passed before August 1989 and 1 for sentences passedthereafter.

Repeat property offender rate. This variable is defined as the monthlynumber of property offenders per 100,000 with a criminal history whoreceived a nonprison sentence. If judges sentenced repeat propertyoffenders to jail rather than releasing them on probation because of the

290

SENTENCING GUIDELINES IN MINNESOTA

perception that the guidelines treated such offenders too lightly, the coeffi-cient for repeat property offenders should be significant and in the positivedirection. Thus, as the rate of sentenced repeat property offendersincreases, the jail incarceration rate also should increase (Hypothesis 2).

Mitigated dispositional departure rate. This variable is operationalizedas the monthly number of mitigated dispositional departures from the noprison/prison sentencing decision per 100,000. Minnesota's presumptiveguidelines furnish judges with a recommended sentence range for the noprison/prison and the prison length decisions. Departures from the guide-lines are acceptable only for substantial and compelling reasons; judgesmust justify in writing their decision to impose a deviant sentence (MSGC,1984). Although prosecutors and defense attorneys have an opportunityto appeal departures from the guidelines, such appeals rarely are made(Parent, 1988). On the basis of the assumption that judges are more likelyto sentence mitigated dispositional departures from the no prison/prisonoutcome to jail than to impose probation or some other nonconfinementsanction, we expect to find a strong positive relationship between miti-gated departures and the jail incarceration rate.

Prison population rate. This variable is defined as the average monthlypopulation of offenders incarcerated in Minnesota's adult correctionalinstitutions per 100,000.

We also include an interaction term between mitigated dispositionaldepartures and the monthly prison population rate to assess the possibilitythat judges circumvented the guidelines to maintain Minnesota's prisonpopulation at or below available capacity levels (Hypothesis 3).4 To date,the impact of this interaction term on jail incarceration remains unex-plored. If the displacement thesis has any merit, we expect mitigated dis-positional departures from the no prison/prison sanction to increase jailrates when prison populations are high.

Although we are interested primarily in testing the three hypotheses, weincluded additional exogenous variables in the analysis to avoid basingconclusions on spurious or suppressed relationships. We selected thesevariables because of their potential associations with jail incarceration.

Legally mandated sentencing factors. We included two legally mandated

4. The variables that constituted the interaction terms were "centered," or takenas deviations from their respective means, before their inclusion in the maximum-likeli-hood equation. The centering of variables involved in creating multiplicative terms isrecommended to reduce multicollinearity, which can inflate estimates of standarderrors (Aiken and West, 1991).

D'ALESSIO AND STOLZENBERG

sentencing factors-offense seriousness and criminal history-in the anal-ysis to control for possible changes in the average severity levels of offend-ers who received nonprison sentences. We calculated monthly averagesfor both offense seriousness and criminal history of nonprison offendersusing severity scales developed by the MSGC (1984). Offense severity iscoded as a 10-point scale, ranging from 1 (less serious crimes, such as saleof a simulated controlled substance) to 10 (second-degree murder). Thecriminal history index, a composite measure of several prior record fac-tors, is coded as a 7-point scale ranging from 0 (no criminal history) to 6(extensive criminal history). Our analysis also included an interactionterm involving these two variables because Minnesota's sentencing guide-lines attach importance to criminal history in determining sentences forless serious offenses (MSGC, 1984).

Multiple conviction rate. This variable is operationalized as the monthlynumber of felony offenders per 100,000 who were convicted for multipleoffenses and received a nonprison sanction. We expect that judges punishnonprison offenders with multiple convictions more severely by sentencingthem to jail instead of releasing them on probation.

Nonprison sentenced offender rate. This variable is a measure of themonthly number of felony offenders receiving a nonprison sanction per100,000. The more offenders receiving nonprison sentences per month,the greater the potential effect on jail use.

Black offender rate. Although research shows that Minnesota's sentenc-ing guidelines reduced racial disparity in sentencing (Miethe and Moore,1985), African-Americans still continue to receive more aggravated sen-tencing departures (MSGC, 1984). To account for the possibility of racialdifferences in sentences to jail, we included in the analysis a control for themonthly rate of African-American offenders who received nonprisonsentences.

Unemployment rate. Previous research reports that incarceration ratesare associated positively with levels of unemployment, independent ofchanges in levels of crime (Chiricos and DeLone, 1992). According to thisline of reasoning, as an economic crisis deepens and as unemploymentrises, social control agencies increasingly perceive the unemployed aspotentially volatile (Jacobs, 1978). These agencies increase their use ofincarceration in response to the threat posed by labor surplus. Drawingfrom this research, we include in the analysis a control for the aggregateunemployment rate. We expect unemployment to increase judicial use ofthe jail sanction.

SENTENCING GUIDELINES IN MINNESOTA

METHOD OF ANALYSIS

The analysis is presented in two stages. In the first stage we constructthe univariate ARIMA model for the jail incarceration rate series. Thismodel, which is constructed through an iterative procedure, accounts forthe stochastic processes associated with a series (see McCleary and Hay,1980). In the second stage, to test the hypotheses, we incorporate theexogenous variables into the maximum-likelihood equation as fixedregressors.

We began by conducting a preliminary analysis to determine whetherthe jail incarceration rate series was stationary in variance. (A stationaryvariance is a necessary condition of ARIMA models.) An examination ofthe univariate distribution revealed skewness for this variable. In keepingwith conventional practices, we transformed the series by taking naturallogarithms.

In addition to a stable variance, another requisite condition of ARIMAis that the series be stationary in level. That is, the series should not trendor drift upward or downward over time. To determine whether the jailincarceration rate series was stationary in level, we used an "augmented"Dickey-Fuller test, which assesses whether a series has a unit root (Dickeyet al., 1986). This test showed that the series was stationary in level.

We also examined the series autocorrelation function (ACF) and partialautocorrelation function (PACF) for autoregressive (Y, = 1Y, + • • • +

, Y,-, + a,) and moving-average parameters (Y, = a, - 0 1a,-1 -. . . - Oqa-q).5The ACF measures the correlation between Y, and Y,-k for different lagvalues of k; the PACF depicts the correlations between the series errorterms (e, and etk), holding constant all other residuals for the kth lag. Anexamination of the ACFs and PACFs suggested an Ln (1,0,0) (1,1,0)12ARIMA model. A Box-Ljung Q statistic (Ljung and Box, 1978), whichtests the null hypothesis that a set of sample autocorrelations is associatedwith a random process, indicated that the residuals for this model wereuncorrelated (i.e., constituted "white noise").

RESULTS

Table 1 reports the multivariate maximum-likelihood ARIMA esti-mates.6 We used one-tailed tests at the .05 level of significance because

5. When month is selected as the unit of analysis, it is quite possible that the jailseries may show cyclical or periodic fluctuations. Because an examination of the ACFat lags of 12 months, 24 months, 36 months, 48 months, and 60 months indicated aseasonal process, we differenced the jail rate series seasonally.

6. We employed the ARIMA procedure in SPSS/PC+Trends (SPSS Inc., 1990),developed by Craig Ansley, to estimate the jail incarceration rate equation. SPSSTrends uses the traditional Box-Jenkins methodology to generate maximum-likelihood

D'ALESSIO AND STOLZENBERG

our hypotheses are directional. Five independent variables have signifi-cant effects on jail rates: the nonprison sentenced offender rate, the onsetof the guidelines, the 1989 guideline modifications, offense seriousness,and the interaction term that involves mitigated dispositional departuresand the prison population rate. As anticipated, the number of offendersreceiving nonprison sentences per month helps to explain patterns of jailuse in Minnesota: The more nonprison offenders sentenced, the greaterthe jail incarceration rate.

Contrary to predictions of the MSGC (1991), however, it appears thatthe establishment of the guidelines had a strong positive effect on jail use.Net of the effect of the other variables in the model, the onset of theguidelines increased jail incarceration by 26%.7 The major point of thisfinding is that it does not support the MSGC's contention that the rise injail use resulted from a preexisting trend. Rather, it seems that the incep-tion of the sentencing guidelines had both a positive and a substantiveeffect on the number of offenders sentenced to jail. 8

Although the effect for the guideline modifications made in 1983 is triv-ial in magnitude and is not statistically significant, it appears that the mod-ifications to the guidelines made in 1989 are important in explainingpatterns of jail use. We observed a significant 7% decline in jail use fol-lowing the 1989 guideline adjustments. This finding is somewhat surpris-ing given that the guideline modifications made in 1989 pertained only tooffenders sentenced to prison.

The coefficient for offense seriousness is also statistically significant. Asexpected, increases in jail incarceration are a partial function of increasesin offense severity levels. However, the coefficients for criminal history,multiple convictions, mitigated dispositional departures, prison popula-tion, and the interaction term between offense seriousness and criminalhistory are small and fall well below levels of statistical significance. Also,the coefficient for the aggregate unemployment rate is essentially 0 in themaximum-likelihood equation. Although aggregate unemployment rates

estimates for a standard linear regression model with an error structure that is not nor-mal. The autoregressive and moving-average terms are included in the maximum-likeli-hood equation as explanatory variables.

7. For ease of explication, we converted the intervention coefficient into a per-centage by using the formula (.229 - 1) x 100 = 26%.

8. However, as pointed out by an anonymous reviewer, a possibility exists thatmissing data (i.e., the 22-month period immediately before the guidelines and the 4-month period following the onset of the guidelines) might have affected our test ofHypothesis 1. In an effort to evaluate this possibility, we reanalyzed the data with aKalman filter, which uses a maximum-likelihood estimation algorithm to estimate val-ues for imbedded missing data (Harvey, 1990; SPSS Inc., 1990). The results from thereanalysis did not differ substantially from the results produced from our analysis of thejoined series.

SENTENCING GUIDELINES IN MINNESOTA

Table 1. Maximum-Likelihood Coefficients for JailIncarceration Rate Equation

AR(1)SAR(1)Sentencing Guidelines1983 Guideline Modifications1989 Guideline ModificationsRepeat Property OffendersMitigated Dispositional DeparturesPrison PopulationMitigated Dispositional Departures x Prison

PopulationOffense SeriousnessCriminal HistoryOffense Seriousness x Criminal HistoryMultiple ConvictionsNonprison Sentenced OffendersAfrican-American OffendersUnemployment

ConstantLog-LikelihoodAICSBC

Jail Incarceration

b SE(b)

.320* .082-. 520* .076

.229* .051

.035 .039-. 070* .033

.019 .014-. 001 .018

.067 .046

-.033*.021*.004.003.009.080*

-. 014-. 003

.015181.477

-328.955-278.002

.011

.012

.012

.007

.029

.007

.018.009

.016

* p < .05 (one-tailed test).

have been linked to levels of incarceration in theoretical arguments(Rusche and Kirchheimer, 1939) and by empirical evidence (Chiricos andDeLone, 1992; but see D'Alessio and Stolzenberg, 1995), our analysisfinds no support for the prediction that the unemployment rate is associ-ated with increased jail use. Similarly, we find no support for the conten-tion that African-American offenders are disproportionately sentenced toharsher sanctions (i.e., jail incarceration instead of receiving a probationsentence).

Contrary to predictions derived from Hypothesis 2, the proportion ofsentenced property offenders with a criminal history has no statisticallydiscernible effect on patterns of jail use. This finding suggests that

D'ALESSIO AND STOLZENBERG

although judges were dissatisfied with the guidelines regarding the leni-ency accorded repeat property offenders, their dissatisfaction did not moti-vate them to increase their use of the jail sanction for such offenders.9

Hypothesis 3 is clearly confirmed. Although mitigated dispositionaldepartures and the prison population rate have no statistically significantdirect effect on jail incarceration, the interaction term involving these vari-ables is negative and salient. Thus, the effect of mitigated dispositionaldepartures on jail use is significant only when the prison population ishigh. To facilitate further interpretation of the interaction effect, we con-structed a graph showing the rate of mitigated dispositional departuresthat resulted in a jail sentence for low, medium, and high prison popula-tion levels.

Figure 2 is intuitively appealing because it depicts the operation ofcausal mechanisms associated with displacement. During periods whenthe prison population was high (rates of 88 to 115 per 100,000), 55% of themitigated dispositional departures resulted in a jail sentence. This figure isnearly twice as great as that for medium prison population levels (rates of74 to 87 per 100,000), and three times as great as that for low prison popu-lation levels (rates of 61 to 73 per 100,000). These results strongly supportthe displacement argument: The organizational imperative to maintainMinnesota's prison population within acceptable limits motivated judgesto circumvent the guidelines by shifting the burden of incarceratingoffenders from the state to the local level.

POLICY CONSIDERATIONS

Past research on the relationship between determinate sentencing andincarceration is incomplete. A substantial amount of evidence indicatesthe success of the MSGC's strategy of linking determinate sentencing toexisting prison capacity in order to control prison growth, but previousstudies neglected the effect of the sentencing guidelines on jail use. Thefew existing studies report that the jail incarceration rate increased sub-stantially after the onset of the guidelines. Even so, the exact nature of theassociation between determinate sentencing and jail use remainedunexplored.

9. However, as suggested by an anonymous reviewer, one possible explanationfor our null finding is that judges are using aggravated dispositional departures toprison as a way of fulfilling their presumed desire to increase punitiveness towardrepeat property offenders. We conducted a supplemental analysis (not reported here)to determine whether repeat property offenders were more likely than other offendersto receive aggravated dispositional departures. The results of this analysis providedsome evidence that judges circumvented the guidelines to sentence repeat propertyoffenders to prison.

296

SENTENCING GUIDELINES IN MINNESOTA 297

Figure 2. Rate of Mitigated Dispositional Departures toJail at Low, Medium, and High PrisonPopulation Levels

Rate per 100,000

0 --- -80 (55%)

10 0 -- - - - - -

43 (30%)S 60- ----

40 - 21(15%)

3 20

0Low Medium High

Prison Population

NOTE: The total number of mitigated dispositional departures to jail is 144 per100,000.

The analysis conducted here explains jail incarceration under determi-nate sentencing in Minnesota more completely than in the past. Two cen-tral conclusions emerge. First, the data analysis contradicts the widelyheld belief that the rise in jail incarceration following the establishment ofdeterminate sentencing in Minnesota resulted from a preexisting trend.We have documented that the onset of guidelines increased jail use by26%. This finding demonstrates that it is unwise to generalize about thesuccess of Minnesota's guidelines in constraining judges' use of incarcera-tion solely on the basis of research on offenders sentenced to prison. Italso underscores the need to use longitudinal data sets with multipleobservation points to clarify the effect of determinate sentencing on jailuse.

To what extent can this finding be generalized to other states with deter-minate sentencing? Because so little research has accumulated thus far, adefinitive answer is not yet possible. Additional studies conducted inother states can show whether determinate sentencing reform "always"

D'ALESSIO AND STOLZENBERG

has a substantive and positive effect on jail incarceration. The more suchwork is done, the greater confidence we can place in the generalizability ofthis finding to different times and places.

The second interesting and important finding is that the effect of miti-gated dispositional departures on jail incarceration is more pronouncedwhen levels of prison population are high. Contrary to what the MSGCmight have desired, it appears that judicial attempts to constrain thegrowth of Minnesota's prison population contributed to the observedincrease in jail use. In Minnesota the organizational imperative to main-tain the prison population within acceptable limits seems to have sup-planted the more distant, more abstract goal of equity in criminalsentencing. Such a conclusion, as it relates to organizational theory,agrees in general with Rothman's and Weber's theoretical positions. Italso provides support for Feeley and Simon's (1992) thesis that the crimi-nal justice system is more concerned with system functioning than withpunishing or rehabilitating criminal offenders. Yet the lack of self-reportdata makes it impossible to evaluate directly whether judges increasedtheir use of mitigated dispositional departures to alleviate prison crowd-ing. In the absence of other explanations, however, and in view of themany independent variables considered here, the statistical significance ofthe interaction term seems to be readily interpretable as support for thedisplacement argument. This supposition is further supported by a signifi-cant reduction in jail use following the modifications made to the guide-lines in 1989, which severed the link between correctional resources andsentencing decisions.

The question which now must be addressed is, What are the policyimplications of our findings for determinate sentencing reform? Mostimportant, our findings suggest that sentencing reforms that place limitsonly on the growth of prison populations may result in substantialincreases in jail use and thereby may erode policymakers' efforts to pro-duce desirable long-term effects on "overall" incarcerated populations.Broader reform efforts that include constraints on jail use seem to be war-ranted. The recommendation that determinate sentencing be linked tolocal correctional resources is relevant for states that are consideringwhether to bind determinate sentencing reform to correctional resources(Knapp, 1993; Tonry, 1993). It also is relevant for Minnesota because theMSGC retains the authority to develop guidelines for nonprison sentences(MSGC, 1984). If the Sentencing Commission continues to rely on its1989 policy, one could expect a substantial reduction in jail incarcerationrates within the next few years, while Minnesota's prison population willprobably continue to grow unabated. Future research should investigatewhether demographic characteristics of the individual offender, such asrace or class, mediate departure decisions or sentences to jail when prison

SENTENCING GUIDELINES IN MINNESOTA 299

populations are high. Such analyses will help to determine whether prisonovercrowding or other organizational factors are diluting the philosophicalgoals of proportionality and equity in criminal sentencing.

REFERENCES

Aiken, Leona S. and Stephen G. West1991 Multiple Regression: Testing and Interpreting Interactions. Newbury

Park, Calif.: Sage.

Blumstein, Alfred, Jacqueline Cohen, Susan E. Martin, and Michael H. Tonry (eds.)1983 Research on Sentencing: The Search for Reform. Vol. 1. Washington,

D.C.: National Academy Press.

Bogan, Kathleen M.1990 Constructing felony sentencing guidelines in an already crowded state:

Oregon breaks new ground. Crime and Delinquency 36:467-487.

Campbell, Donald T. and Julian C. Stanley1963 Experimental and Quasi-Experimental Designs for Research. Boston:

Houghton Mifflin.

Chambliss, William J. and Robert J. Seidman1971 Law, Order and Power. Reading, Mass.: Addison-Wesley.

Chiricos, Theodore G. and Miriam A. DeLone1992 Labor surplus and punishment: A review and assessment of theory and

evidence. Social Problems 39:421-446.

Cohen, Jacqueline and Michael H. Tonry1983 Sentencing reform impacts. In Alfred Blumstein, Jacqueline Cohen,

Susan E. Martin, and Michael H. Tonry (eds.), Research on Sentencing:The Search for Reform. Vol. 2. Washington, D.C.: National AcademyPress.

Cook, Thomas D. and Donald T. Campbell1979 Quasi-Experimentation: Design and Analysis Issues for Field Settings.

Boston: Houghton Mifflin.

Cullen, Francis T. and Karen E. Gilbert1982 Reaffirming Rehabilitation. Cincinnati, Ohio: Anderson.

D'Alessio, Stewart J. and Lisa Stolzenberg1995 Unemployment and pretrial incarceration: Does the State use imprison-

ment to control labor surplus? American Sociological Review. Forthcom-ing.

Dickey, David A., William R. Bell, and Robert B. Miller1986 Unit roots in time series models: Tests and implications. American

Statistician 40:12-26.

Feeley, Malcolm M. and Jonathan Simon1992 The new penology: Notes on the emerging strategy of corrections and its

implications. Criminology 30:449-474.

Frase, Richard S.1993 Prison population growing under Minnesota guidelines. Overcrowded

Times 4:1ff.

300 D'ALESSIO AND STOLZENBERG

Goodstein, Lynne and John R. Hepburn1986 Determinate sentencing in Illinois: An assessment of its development and

implementation. Criminal Justice Policy Review 1:305-328.

Harvey, Andrew1990 Forecasting, Structural Time Series Models, and the Kalman Filter. New

York: Cambridge University Press.

Hepburn, John R. and Lynne Goodstein1986 Organizational imperatives and sentencing reform implementation: The

impact of prison practices and priorities on the attainment of theobjective of determinate sentencing. Crime and Delinquency 32:339-365.

Jacobs, David1978 Inequality and the legal order: An ecological test of the conflict model.

Social Problems 25:515-525.

Knapp, Kay A.1982 Impact of the Minnesota sentencing guidelines on sentencing practices.

Hamline Law Review 5:237-256.1983 The etiology of prison populations: Implications for prison population

projection methodology. Working Paper, Prison Overcrowding Project,Minnesota Sentencing Guidelines Commission, St. Paul.

1984 What sentencing reform in Minnesota has and has not accomplished.Judicature 68:181-189.

1993 Allocations of discretion and accountability within sentencing structures.Colorado Law Review 64:679-705.

Kramer, John H. and Robin L. Lubitz1985 Pennsylvania's sentencing reform: The impact of commission-established

guidelines. Crime and Delinquency 31:481-500.

Kramer, John H., Robin L. Lubitz, and Cynthia A. Kempinen1989 Sentencing guidelines: A quantitative comparison of sentencing policies

in Minnesota, Pennsylvania, and Washington. Justice Quarterly 6:565-587.

Ljung, G.M. and George E.P. Box1978 On a measure of lack of fit in time series models. Biometrika 65:297-303.

Mays, G. Larry1989 The impact of federal sentencing guidelines on jail and prison overcrowd-

ing and early release. In Dean J. Champion (ed.), The U.S. SentencingGuidelines: Implications for Criminal Justice. New York: Praeger.

Mays, G. Larry and Joel A. Thompson1988 Mayberry revisited: The characteristics and operations of America's small

jails. Justice Quarterly 5:421-440.

McCleary, Richard and Richard A. Hay, Jr.1980 Applied Time-Series Analysis for the Social Sciences. Beverly Hills,

Calif.: Sage.

Miethe, Terance D.1987 Charging and plea bargaining practices under determinate sentencing: An

investigation of the hydraulic displacement of discretion. Journal ofCriminal Law and Criminology 78:155-176.

SENTENCING GUIDELINES IN MINNESOTA 301

Miethe, Terance D. and Charles A. Moore1985 Socioeconomic disparities under determinate sentencing systems: A

comparison of preguideline and postguideline practices in Minnesota.Criminology 23:337-363.

1988 Officials' reactions to sentencing guidelines. Journal of Research inCrime and Delinquency 25:170-187.

1989 Sentencing Guidelines: Their Effect in Minnesota. Washington, D.C.:National Institute of Justice.

Minnesota Sentencing Guidelines Commission (MSGC)1984 The Impact of Minnesota's Sentencing Guidelines: Three Year Evalua-

tion. St. Paul: Minnesota Sentencing Guidelines Commission.1990 Report to the Legislature. St. Paul: Minnesota Sentencing Guidelines

Commission.1991 Summary of 1989 Sentencing Practices for Convicted Felons. St. Paul:

Minnesota Sentencing Guidelines Commission.

Moore, Charles A. and Terance D. Miethe1986 Regulated and unregulated sentencing decisions: An analysis of first-year

practices under Minnesota's felony sentencing guidelines. Law andSociety Review 20:254-277.

Parent, Dale G.1988 Structuring Criminal Sentences: The Evolution of Minnesota's Sentencing

Guidelines. Stoneham, Mass.: Butterworth.

Rothman, David J.1971 The Discovery of the Asylum: Social Order and Disorder in the New

Republic. Boston: Little, Brown.1980 Conscience and Convenience: The Asylum and Its Alternatives in

Progressive America. Boston: Little, Brown.

Rusche, Georg and Otto Kirchheimer1939 Punishment and Social Structure. New York: Columbia University Press.

SPSS Inc.1990 SPSS/PC+ Trends. Chicago: SPSS Inc.

Stolzenberg, Lisa and Stewart J. D'Alessio1994 Sentencing and unwarranted disparity: An empirical assessment of the

long-term impact of sentencing guidelines in Minnesota. Criminology32:301-310.

Tonry, Michael H.1987 Sentencing Reform Impacts. Washington, D.C.: National Institute of

Justice.1991 The politics and processes of sentencing commissions. Crime and

Delinquency 37:307-329.1993 Sentencing commissions and their guidelines. In Michael Tonry (ed.),

Crime and Justice: A Review of Research. Vol. 17. Chicago: Universityof Chicago Press.

von Hirsch, Andrew1976 Doing Justice: The Choice of Punishments. New York: Hill and Wang.1982 Constructing guidelines for sentencing: The critical choices for the

Minnesota Sentencing Guidelines Commission. Hamline Law Review5:164-215.

302 D'ALESSIO AND STOLZENBERG

von Hirsch, Andrew and Kathleen Hanrahan1981 Determinate penalty systems in America: An overview. Crime and

Delinquency 27:298-316.

Weber, Max1978 Economy and Society, ed. Guenther Roth and Claus Wittich. Vol. 2.

Berkeley: University of California Press.

Stewart J. D'Alessio is Assistant Professor in the School of Public and EnvironmentalAffairs at Indiana University-Purdue University Fort Wayne. He received his PhD incriminology from Florida State University in 1993. His current research examines therelationship between available capacity and levels of incarceration.

Lisa Stolzenberg is Assistant Professor and Criminal Justice Program Coordinator inthe School of Public and Environmental Affairs at Indiana University-Purdue Univer-sity Fort Wayne. She also received her PhD in criminology from Florida State Univer-sity in 1993. Her research interests include sentencing reform and criminal justicepolicy.