Introduction

Available data (aggregated at the national level) suggest that individuals serve less than one-half of their imposed sentence on average. The Bureau of Justice Statistics (BJS) calculated the actual amount of time that individuals released from State prisons during calendar year 1994 served. The BJS examination revealed that such individuals served 41% of their imposed sentence on average (Langan & Brown, 1997:9). This percentage varied by crime category. Persons convicted of violent offenses (excluding life or death sentences) served 46% of their sentence on average, whereas persons convicted of property offenses served 41% of their sentence on average, and persons convicted of drug offenses served 36% of their sentence on average (Langan & Brown, 1997:9).

The present study will collect data on the actual amount of time that persons sentenced in Maryland circuit courts serve. To that end, samples of persons convicted of person, drug, and property offenses and sentenced in Maryland circuit courts during calendar year 1993 were selected and followed through the prison system.

Sample Selection

The primary reasons for selecting calendar year 1993 were two-fold. First, calendar year 1993 was selected to ensure that a sufficient number of individuals would have completed their sentence by the time of data collection (September, 1997). Second, calendar year 1993 was selected to ensure that release practices in place during the course of the study (which clearly affect time-to-serve) would be similar to present-day release practices. However, it should be noted that legislation affecting parole eligibility (Article 41, Section 4-516) was modified after the samples were sentenced. The modification effective October 1, 1994 requires persons convicted of violent offenses (as well as burglary and daytime housebreaking) to serve at least 50% of their sentence before they are considered eligible for parole.

Due to the relatively short time frame of the study, sample selection was limited to individuals who had received sentences of greater than 1 year and less than 10 years of incarceration. Individuals with sentences of 12 months or less were excluded from participation because it is most likely that they would have served their sentence in a local jail, precluding the collection of Department of Public Safety and Correctional Services (DPSCS) data. Individuals with sentences of 120 months or more were also excluded from the sample because it is likely that they would still have been incarcerated at the time of data collection. As a consequence, data collected as part of the present study will not generalize to persons who received short sentences to be served in local jails (i.e., 1 year or less) or to persons who received sentences exceeding 10 years.

Time-to-serve will be estimated separately for these two groups. Generally speaking, persons sentenced to serve less than 1 year in a local facility serve a high proportion of the total sentence (e.g., 70% to 75%). Such individuals typically serve the judicially imposed sentence less good time credit. Since they are not eligible for parole release unless they have been sentenced to at least six months, parole release is uncommon (given that the remaining time-to-serve is relatively short).

Due to legislative and procedural changes governing the parole process, time-to-serve for persons sentenced to 10 or more years has been more variable. The Department of Public Safety and Correctional Services (1996) estimated that persons convicted of violent crimes (643B offenses) who were recently released served 60% of their sentence on average. The percentage of time served was slightly higher (67.8%) for those offenders convicted of rape or sex offenses. The average sentence length for 643B offenses was 13 years.

During calendar year 1993, 6,196 individuals convicted of a single count and sentenced in circuit courts received a sentence involving some period of incarceration (e.g., partial suspension, split-sentence, incarceration only). Nearly one-half (48.3%) of those persons sentenced during calendar year 1993 received sentences involving a period of incarceration of 12 months or less. Another 8% received sentences of 10 or more years. Roughly 44% (N=2,706 persons) received terms of incarceration between 1 and 10 years and were therefore eligible for participation in the study. Roughly one-half of the eligible individuals (N=2,706) had been convicted of drug offenses (53.3%), followed by person offenses (31.5%) and property offenses (15.2%).

Because a separate sentencing guidelines matrix is used for person, property, and drug offenses and because time-to-serve seems to vary by offense category, a sample of individuals convicted of each offense category was selected. A simple random sample of 20% of the individuals who fell within each cell of the relevant sentencing guidelines matrix (i.e., person, drug, property) was then selected. The final samples consisted of the following: (1) a sample of N=185 individuals convicted of person offenses; (2) a sample of N=305 individuals convicted of drug offenses; and (3) a sample of N=102 individuals convicted of property offenses.

Procedure

Since time-served data for the target sentence may be confounded with time-served resulting from other convictions (e.g., prior or subsequent convictions or parole/probation revocations), the DPSCS also indicated whether individuals were required to serve additional time stemming from prior or subsequent convictions and/or parole or probation revocations.

Results

The percentage of the total sentence actually served was calculated for each individual (e.g., 12 months served out of a total sentence of 18 months=0.67 or 67%). The individual percentages were then averaged across each sample.

Time-served was also examined as a function of the following factors: (1) Apure (target sentence only) versus Anonpure sentence (target sentence plus additional time); (2) release type (parole, mandatory release, court release); (3) offense score (a measure of the seriousness of the offense used in the sentencing guidelines); and (4) offender score (a measure of criminal history used in the sentencing guidelines).

Person Offenses

Individuals were most frequently convicted of the following person crimes: (1) robbery (23%); (2) robbery with a deadly weapon (16%); and (3) common law assault (13%). Another 17% of the sample had been convicted of other forms of assault (e.g., assault with intent to maim, murder, rape, or prevent arrest) or other forms of robbery (e.g., attempted, accessory, or conspiracy).

Complete information was available for 123 individuals (67.6%). Time-served data was not collected from the remainder of the sample for the following reasons: (1) twenty-nine (15.9%) individuals had been incarcerated in a local jail (even though the actual sentence exceeded 1 year); (2) twenty-six (14.3%) individuals were still incarcerated at the time of data collection; and (3) four (2.2%) individuals could not be found in the database. The average sentence length of the individuals who had been incarcerated in a local jail was 20 months and the average sentence length of the individuals who were still incarcerated was 175 months (including additional time).

Roughly 37% of the sample had received additional time stemming from prior or subsequent convictions or parole/probation revocations. The remaining 63% had not received additional time. These cases were considered Apure cases.

Individuals with complete case information (n=123) served 55% of their imposed sentence on average (see Table 1 and Appendix). The proportion of the total sentence served did not vary by whether the sentence length was Apure (i.e., additional time was not incurred) or Anonpure (i.e., additional time stemming from a prior or subsequent conviction was added to the target sentence). Apure and Anonpure cases served 55% and 56% of the imposed term, respectively.

Time-served also varied slightly depending on the seriousness of the offense and the prior record of the offender (see Table 1). Offense score was dichotomized into Alow@ and Ahigh@ seriousness. One-half of the sample (50.4%) consisted of individuals with Alow@ offense seriousness scores. The other half of the sample (49.6%) consisted of individuals with Ahigh@ offense seriousness scores. Similarly, the offender score (or prior record summary) was dichotomized into Alow@ and Ahigh@ scores. Roughly one-half of the sample had Alow@ offender scores (52.8%) and the other half of the sample had Ahigh@ offender scores (47.2%).

The results of the offense and offender seriousness examination revealed that persons with Alow@ offense seriousness scores served 53% of their sentence on average whereas persons with Ahigh@ offense seriousness scores served 57% of their sentence on average. Similarly, persons with Alow@ offender scores served 54% of their sentence on average and persons with Ahigh@ offender scores served 57% of their sentence on average.

Time-served was also examined as a function of sentence length. The total sample was subdivided into the following three subsamples based on sentence length: (1) short (13 to 41 months); (2) medium (42 to 60 months); and (3) long (61 months and greater). Approximately, 36% of the sample had short sentences, 44% had medium sentences, and 20% had long sentences. The analysis suggested that individuals with short and medium sentences served roughly the same percentage of their total sentence (58% and 56%, respectively), whereas individuals with long sentences served a slightly smaller percentage (50%).

The difference time-served by sentence length (short, medium, or long) seems to stem in part from the fact that individuals with long sentences are more likely to be released on parole. Individuals released on parole tend to serve a smaller proportion of their sentence than persons released by means of mandatory release. Slightly over one-half of the individuals with long sentences, for example, were released on parole. In comparison, 25% of the persons with short sentences were released on parole and 30% of the persons with medium sentences were released on parole.

Property Offenses

Time-served data was available for slightly over one-half of the sample (52.9%). Twenty individuals (19.6%) served their time in a local jail and 15 (14.7%) could not be located in the database. Another 12 (11.8%) were still incarcerated at the end of the data collection period. The sentence of one individual had been stayed.

Thirty-one (57.4%) of the complete cases were Apure@ cases (i.e., the target sentence was not confounded with additional time). The remaining cases (n=23) had additional time added to the target sentence as a result of a subsequent or prior conviction or probation/parole revocation. Examination of time-served using the total sample revealed that individuals served 55% of the imposed sentence on average (see Table 2 and Appendix). When Apure@ cases are distinguished from Anonpure@ cases, the Apure@ subsample served 53% of the imposed term and the Anonpure@ served 58% of the imposed term.

Time-served was also examined as a function of offense and offender score (see Table 2). The total sample was divided into individuals with Alow@ and Ahigh@ offense scores and Alow@ and Ahigh@ offender scores. Individuals with Alow@ offense scores served 53% of the imposed term and individuals with Ahigh@ offense scores served 61% of the imposed term. Similarly, individuals with Alow@ offender scores served 51% of the imposed term and individuals with Ahigh@ offender scores served 57% of the imposed term.

When time-served was examined as a function of sentence length, individuals with short sentences (13-41 months) served 58% of their sentence (see Table 2). Individuals with medium sentences (42-60) served 49% of their sentence and individuals with long sentences (greater than 61 months) served 57% of their sentence.

Drug Offenses

Complete information was available for approximately 70% of the cases. Time-served data was unavailable for 30% of the sample for the following reasons: (1) fifteen percent (15%) of the sample had been incarcerated in a local jail; (2) six percent (6%) could not be located; and (3) ten percent (10%) had not yet been released. Sixty percent (60%) of the complete cases were deemed Apure@ cases.

Drug offenders served 50% of the imposed sentence on average (see Table 3 and Appendix). Drug offenders deemed Apure@ cases served a smaller percentage of their sentence (48%) than Anonpure@ cases (53%).

Time-served did not vary substantially by offense or offender score (see Table 3). Persons with low offense seriousness scores served 49% of the imposed term while persons with high offense seriousness scores served 50% of the imposed term. Likewise, persons with low offender scores served 48% while persons with high offender scores served 52% of the imposed term.

Similarly, time-served did not vary much by total sentence length (see Table 3). Drug offenders with short sentences (13-41 months) served 52% of their sentence on average, whereas drug offenders with medium sentences (42-60 months) and long sentences (greater 61 months) served 49% and 48% of their sentence, respectively.

Summary

Time-served data were available for approximately two-thirds of the sample. Data for remaining one-third were unavailable for two primary reasons: (1) sample members had been incarcerated in local jails rather than the state prison system (even though sentences were greater than 1 year); and (2) sample members were still incarcerated at the end of the data collection period. In addition, a small percentage of persons could not be located.

Missing data-values (i.e., cases where complete case information was unavailable) present a problem to the analysis (i.e., potentially biased results) if the cases that are missing differ systematically from cases that are not missing, and any such differences are related to the construct of interest, time-to-serve. Since the intent of the study is to focus exclusively on persons who served sentences of less than ten years in prison (as opposed to jail), the fact that information was unavailable for persons who served time in jail should not be problematic. The results of the present study will only generalize to individuals who served time in prison. Similarly, results of the study were only intended to generalize to individuals required to serve a term of less than 10 years. Among those individuals who were still incarcerated at the time of data collection and therefore lacked complete case information, approximately 51% (across the three samples combined) were required to serve 10 or more years (including additional time). The results of the present study revealed that drug offenders served 50% of the imposed prison term on average. Both person and property offenders served 55% of the imposed prison term on average. Examination of Apure@ cases (target sentence only) suggested that such individuals served roughly the same percentage of time as Anonpure@ cases (target sentence plus additional time), although Anonpure@ property and drug offenders tended to serve a slightly greater percentage of the imposed term. Due to the modification of parole eligibility legislation (Article 41, Section 4-516) which requires person and select property offenders to serve 50% of their sentence before being considered eligible for parole, time-to-serve for person and select property offenses may increase in the future.

Persons in the present study (individuals serving between 1 and 10 years) served a slightly smaller percentage of the imposed term than persons convicted of violent offenses (643B offenses) (average sentence length of 13 years). The DPSCS estimated that persons convicted of serious, violent offenses served 60% of their sentence on average.

Release type varied substantially across the samples. Approximately, 56% of the drug offender sample, for example, were released on parole as compared to 33% of the person offender and 17% of the property offender samples. Court release was surprisingly common. Among the three samples combined, roughly 9% of the offenders were released by means of court release.

The release mechanism (parole, mandatory release, or court release) was the major determinant of time-served. Persons released by the court served the smallest portion of their total sentence (roughly one-quarter of the sentence). Persons released on parole served between 40% (drug offenses) and 48% (person offenses) of their total sentence, whereas persons released via mandatory release served between 64% (person offenses) and 71% (drug offenses) of their total sentence. The method of release appeared to be related to offender score since persons (among the three samples combined) with low offender scores (0-2) were more likely to be released on parole, thereby serving a smaller percentage of their total sentence.

Implications for Sentencing Guidelines

Court release seemed to produce the most noticeable disparity. Sentenced offenders may be released by the court in several ways. One form of court release empowers judges to reconsider sentences within a 90-day window. Court release may also result from review by a judicial Review Panel or from a defendant's appeal. The OBSCIS I data base does not distinguish between specific types of court release.

Due to the court release process, criminal sentences may appear to be consistent with the sentencing guidelines, while in reality defendants receive a substantially different sentence. Note, however, that the use of court release does not appear to be common throughout the state of Maryland. Nearly one-half of the court releases in this sample took place in 2 counties (Prince George's and St. Mary's counties).

In summary, although structured sentencing may reduce disparity at the front-end, it is also important to consider the effect of discretionary decision making at the back-end since such discretion materially influences the severity of a given sentence. To that end, an understanding of the factors which influence court release and parole decision making in the state of Maryland is essential. What percentage of offenders are released by means of court release? To what extent is parole release based on objective measures of recidivism risk? To what extent is parole release used to rectify jurisdictional disparity in sentencing? In order to reduce unwarranted disparity in both the judicially imposed sentence and the percentage of time-served, both structured sentencing and structured parole decision making processes would seem necessary.

Implications for a Truth-in-Sentencing Policy

Clearly, in order to implement a truth-in-sentencing policy, the percent of the judicially imposed sentence that individuals serve must be increased. However, since time-to-serve under a truth-in-sentencing policy (e.g., 85%) often departs dramatically from time-to-serve under the present system, a critical question becomes whether to maintain or reduce currently imposed sentence lengths. If currently imposed sentence lengths are maintained (and time-to-serve is increased), the severity of a given sentence will obviously increase. However, if currently imposed sentence lengths are reduced (and time-to-serve is increased), the severity of a given sentence will remain relatively constant. The overall level of punishment associated with a particular crime will therefore not fluctuate over time.

In order to assess the potential impact of truth-in-sentencing policies on the prison system, accurate estimates of time-served are essential. Clearly, the effect of truth-in-sentencing on the prison system depends on the definition of truth-in-sentencing that is adopted. What percentage of an imposed sentence would offenders have to serve to constitute truth-in-sentencing? What types of offenses would fall under the rubric of a truth-in-sentencing policy?

Consider the following illustration of the impact of an 85% truth-in-sentencing policy using the present sample as an example. In order to meet a hypothetical 85% truth-in-sentencing mandate, the current sample (individuals required to serve between 1 and 10 years) would have to serve an additional 35% of the original sentence on average. The total sample of 387 individuals served 795 years (or 9,540 bed-months) out of a total of 1,559 possible years (or 18,708 bed-months). Under 85% truth-in-sentencing, those same individuals would be required to serve an additional 530 years for a grand total of 1,325 years (or 15,900 bed-months). Even for a small sample such as this, an 85% truth-in-sentencing policy would appear to have a significant impact on prison bed space needs.

A more restrictive truth-in-sentencing policy which only applies to a small number of serious, violent crimes may also be implemented (e.g., murder, rape, armed robbery, etc.). The impact on the prison system may not be as dramatic as an across the board truth-in-sentencing policy. Nevertheless, even if such inmates already serve a high percentage of their imposed sentence (due to the nature of the crime), an 85% truth-in-sentencing policy would still appear to have a nontrivial impact on the prison population since prison populations tend to be driven primarily by persons serving longer sentences (Young & Brown, 1992). For example, a 10% increase in time-to-serve from 75% to 85% for a twenty-year sentence amounts to an increase in time-served of 2 years. Using the DPSCS time-served estimate of 60% for serious, violent crimes (i.e., 643B offenses), a 25% increase in time-served would translate to an additional 5 years for the same hypothetical twenty-year sentence.

During calendar year 1996, the DPSCS reported that 544 persons sentenced for serious, violent offenses (643B offenses) were committed to state facilities with sentences exceeding 10 years (not including 82 life sentences) (DPSCS, 1996). If each new intake was required to serve an additional 25%, the intake group together would serve a minimum of 1,360 additional years (assuming sentence length was only 10 years). Over the long run, the increase in time served would accumulate as additional cohorts of inmates sentenced for violent crimes entered the system. The DPSCS (1996) projected that by the year 2015, an 85% truth-in-sentencing policy for 643B violent offenses would require 4,192 additional beds (not including beds required for parole eligible lifers).

Thus absent new prison construction, it seems clear that in order to implement truth-in-sentencing policies (either for a subset of serious violent offenses or for all felony offenses), current sentence lengths as recommended by the voluntary sentencing guidelines would have to be reduced. For example, if truth-in-sentencing is expected to apply to the full-range of offenses, one possible strategy would be to reduce the imposed sentence length by 30% for all offenses. Given the political constraints surrounding the reduction of sentence lengths for serious, violent offenses (even if the severity of the punishment or time-to-serve remains the same), an alternative strategy may be to reduce sentence lengths for all offenses except serious, violent offenses. Such a strategy would effectively increase the severity of punishment for persons convicted of this class of crime only. According to DPSCS estimates, however, the adoption of the latter strategy would still greatly increase the demand for prison bedspace.

In conclusion, the principal objectives and likely consequences of a truth-in-sentencing policy must be carefully considered. Is the principal goal of a truth-in-sentencing policy to close the gap between imposed sentence length and time-to-serve such that judges, defendants, victims, and the public at large would know at the time of sentencing how long an offender would be expected to serve? Is the truth-in-sentencing policy intended to maintain the overall severity of the sanctioning system or is it intended to increase the level of punishment for all offenses or for a subset of offenses? Are resources available for the additional prison construction such a policy would require?

References

Langan, P.A. (1997, July). Felony Sentences in the United States, 1994. Bureau of Justice Statistics Bulletin. Washington, D.C.: U.S. Department of Justice, Office of Justice Programs.

Lynch, J.P. & W.J. Sabol. (1997, August). Did Getting Tough on Crime Pay? The Urban Institute Crime Policy Report. Washington, D.C.: The Urban Institute.

Young, W. & M. Brown. (1996). Cross-national Comparisons of Imprisonment. In M. Tonry (Ed.) Crime and Justice; A Review of Research (vol. 17). Chicago: The University of Chicago Press.

Appendix