Joel M. Caplan
Rutgers University
School of Criminal Justice

Susan C. Kinnevy
University of Pennsylvania
School of Social Policy and Practice

THE CONCEPT OF PAROLE has been around for more than a century in the United States and the goals and activities of parole agencies have evolved over time as social and political environments changed. Throughout the history of parole, calls for reform and new parole models have been voiced by prominent criminologists (i.e., Petersilia, 1999; Caplan, 2006; Criminological Research Associates, 1974); however, no attempts have been made to identify existing commonalities among state paroling authorities (i.e. parole boards). They are often nonchalantly categorized by authors of published articles and reports using binary descriptors that meet their literary needs. For example, two categorizations commonly referenced in the literature are discretionary parole, by which paroling authorities decide releases for eligible inmates on a case-by-case basis, and mandatory parole, by which judges or statutes define parole release as a function of an inmate’s sentence (e.g., Ireland & Prause, 2005; Griset, 1995; Shade, 1982). Indeterminate and determinate sentencing structures have also been used repeatedly to describe paroling authorities (Pew Charitable Trusts, 2007; Ashford & Winston, 1993; Bottomley, 1990). Other less common descriptors include casework and surveillance models (e.g., West & Seiter, 2005), summary parole and regular parole models (Star, 1979), and part-time parole boards and full-time parole boards (Conley & Zimmerman, 1982; Hart, 1978).

This exploratory study assumed that paroling authorities (PAs) are more complex in terms of their structures and operations than their traditional references, which only describe one aspect of parole and are not exhaustive. The limited attention given to parole models may be attributable to the fact that reasonable people can disagree over which label best characterizes a particular agency or jurisdiction (Tonry, 1999b). However, this explanation does not account for the absence of models that aggregate shared attributes of PAs among different states rather than debate experts’ preferences. A census of contemporary paroling authority (PA) attributes must be identified first before new models can be developed or existing models improved. Heeding calls for reform of parole is difficult when the system of independent agencies is greatly unknown in the broader context of the nation.

This paper presents a national snapshot of parole in the U.S. that was constructed from national survey data of state paroling authorities. Models depict the most common characteristics among all state parole jurisdictions. A Geographic Information System (GIS) was used to map regional differences and further illuminate variance in the models. The principle research question of this study was straightforward: What are the different types of state parole models currently operating in the United States? The answer, discussed here, was long overdue and is greatly needed to better understand the structural and operational characteristics of parole in America.

Data Sources and Methods
A national survey was designed and implemented in consultation with the APAI that aimed to identify characteristics common to all paroling authorities (PAs), to capture differences among PAs, and to solicit information on regulations pertaining to release, supervision, and revocation decision-making activities. The survey was sent to the administrative heads of 67 PAs that were members of APAI as of November 2007, including all 50 states. The administrative head (i.e. chairperson, director) was asked to complete the survey himself or herself, or to designate an appropriate representative to do so in his or her place. Respondents were asked to consult whatever resources were available to them to ensure that the answers provided were accurate and up-to-date. The survey was administered online via Survey Monkey, a web-based survey tool. It provided respondents with a user-friendly interface to complete the survey and it streamlined the collection and management of response data. Forty-seven (47) of the 50 states that were asked to respond did so; the states that did not respond were California, Indiana, and Mississippi. Data from the 2008 APAI survey formed the basis for modeling.

Originally, the APAI dataset contained 108 categories of variables for each state in the U.S. All data were recoded into dichotomous variables, yielding 577 variables. Only those variables with sufficient variability were used for subsequent analyses. Sufficient variability was defined as more than 25 percent and less than 75 percent agreement among the states. For example, 44 out of 47 states permit victim input at parole hearings. Because there was little difference among states with regard to this attribute, it was excluded from a model. As shown in Table 1, remaining variables were categorized into "structural" or "operational" variables, with operational variables sub-categorized as “pre-release,” “post-release: supervision,” and “post-release: revocation.” Structural variables included measures of PA composition, PA affiliation and jurisdiction, PA statutory regulations and authority to impact incarceration lengths. Five multilevel crosstabs were computed using all variables within each (sub)category, respectively. One multilevel crosstab included all “structural” variables; another included all “operational” variables combined; another included all “operational pre-release” variables; another included all “operational post-release: supervision” variables; and another included all “operational post-release: revocation” variables. As shown in Tables 2 through 5, the greatest overlap of variables within each crosstab became a model. Models represent the most common characteristics of paroling authorities in the U.S.—for each (sub)category. A geographic information system (GIS) was then used to assess spatial distribution of the models.

Table 1: Variables Included within each Category and Sub-Category

STRUCTURAL	OPERATIONAL
Mixed sentencing structure (both indeterminate and determinate)	Pre-Release
All full-time board members	Uses actuarial instruments to decide release
Has discretionary authority to release inmates	Time off credits are available
Has no discretionary power to set minimum time incarcerated (left up to courts/statutes)	Program completion is required prior to release
Has discretionary authority to terminate sentences prior to max	Board member voting for release works within a panel structure*
Independent/autonomous agency*	Minimum number of votes needed for parole approval*
Has jurisdiction over state inmates only*	Post-Release: Supervision
Has jurisdiction over adult inmates only*	Parole board has full authority over supervision of parolees in the community
	Board has discretionary authority to set conditions of parole
	Board uses actuarial instruments to set conditions
	Board uses actuarial instruments to set security levels of supervision
	Post-Release: Revocation
	Has discretionary authority to revoke parole
	Does not use actuarial instruments for revocation decisions
	Case hearing officers have discretionary authority to make final revocation decisions*
*excluded from final analysis due to either limited or extreme variability.

Results

Overview of Parole in the U.S.: Key Findings from the APAI Survey

Most paroling authorities (PAs) consist of members appointed by the governor and who serve an average of five years. They are most often independent agencies or affiliated with the Department of Corrections. A majority of state PAs have the authority to make final release decisions and make those decisions within a mixed determinate and indeterminate sentencing structure. Over half the PAs require interviews with parole-eligible offenders prior to release, with most interviews conducted in-person by a panel of PA members. A minimum of three panel members and three votes are needed to decide release.

The top three sources of input considered by PAs in their release decision-making process are from the victim, the offender’s family and the district attorney. Other factors that impact most heavily on the decision to release are crime severity, crime type, and offender criminal history, respectively. The most frequently cited factor in delayed release is a delay in program completion. Program completion is a prerequisite for release in most states; almost all PAs report that they do not have enough available programs. Most states do give time off credits (TOC), the most common one being statutory good time.

More than half of PAs have full authority over supervision and most have the power to set conditions of supervision for all their offenders across crime categories. More than half the PAs also have the authority to terminate supervision prior to maximum sentence for all offenders across crime categories. The most often cited responses to violations of supervision are outpatient and inpatient treatment programs, electronic monitoring, and house arrest. Most PAs can approve motions to revoke parole and over half can issue arrest warrants. Almost all PAs have the authority to manage or adjudicate violations, although only 75 percent can set the time to serve for revocation.

Over 90 percent of PAs can revoke supervision for all offenders across crime categories. Most PAs include both revocation options that return offenders to prison with or without treatment and non-revocation options that place offenders in intermediate sanctions or community-based facilities. Management of community-based facilities usually resides with the state’s correctional authority. With regard to instruments used to guide the parole process, the most commonly cited are Static-99, LSI-R, and instruments developed in-house. The only instruments that are routinely validated are those developed in-house.

The most easily produced and regularly published statistic by PAs is the number of offenders paroled in a given calendar year. Other statistics seem to be difficult to produce, apparently because the PAs are not always the entity that manages statistics. Only 29 PAs provided recidivism rates, with averages ranging from 25.1 percent calculated for one year to 4.28 percent calculated for over three years. The offender population used to calculate rates varied too much to report a pattern. The events used to calculate recidivism were generally those that resulted in incarceration. Only 19 PAs reported having secure facilities that can be used in place of incarceration.

Overall, the survey was successful in gathering a great deal of information about the policies and practices of domestic paroling authorities. Full findings from the survey are published in a report entitled “Findings from the APAI International Survey of Releasing Authorities,” and can be downloaded at http://www.apaintl.org/documents/surveys/2008e.pdf.

Paroling Authority Models

Nine states share common structural attributes. As shown in Table 2 and Figure 1, the structural model comprises states that share the following characteristics: The paroling authorities (PAs) are composed of all full-time members; their sentencing structure is mixed (both determinate and indeterminate); they have discretionary power to release; they have no authority to set minimum time (this function is left up to the courts or statutes); they have the authority to terminate the maximum sentence. There does not appear to be a significant spatial cluster of states that share structural attributes, although most of the states are in the eastern and mid-western part of the country.

Twenty-four states share common pre-release characteristics. As shown in Table 3 and Figure 2, the pre-release operational model comprises states that share the following characteristics: Program completion is required prior to release; time-off credits are available; PAs use risk assessment instruments in release decisions. There does not appear to be significance in the spatial pattern, although the plurality of states is in the south and mid-west.

Thirteen states share common post-release characteristics with regard to supervision. As shown in Table 4 and Figure 3, the post-release operational model for supervision comprises states that share the following characteristics: Their PAs have full authority over parolee supervision; their PAs have the authority to set conditions of parole; they use risk assessment instruments to set conditions and levels of parole. The spatial pattern shows some clusters in the east, south and mid-west.

Thirty-one states share the most common post-release characteristics with regard to supervision. As shown in Table 5 and Figure 4, the post-release operational model for revocation comprises states that share the following characteristics: Their PAs have authority to revoke supervision and they do not use risk assessment instruments to decide revocation. There does not appear to be any obvious spatial clusters, as most states fit this model.

The combined operational model comprises state PAs that share all the operational commonalities: Program completion is required prior to release; time off credits are available; they use risk assessment instruments in release decisions; they have full authority over parolee supervision; they have the authority to set conditions of parole; they use risk assessment instruments to set conditions and levels of parole; they have authority to revoke supervision; they do not use risk assessment instruments to decide revocation. Only five states share these common characteristics and spatially cluster in the southern part of the United States, as shown in Figure 5. Only one state—Tennessee--has all operational and structural characteristics discussed here.

Table 2. Model 1 – Structure

Authority to terminate prior to max	Has power to set minimum time	Authority to release offenders	Board membership			Mixed sentencing structure
Yes	Yes	Yes	All full-time	State	AL	--
					GA	Yes
				Total		1
			All part-time	State	OK	--
				Total		--
			Mixed	State	HI	--
					IA	--
					UT	--
				Total		--
	No	Yes	All full-time	State	FL	Yes
					IL	Yes
					MI	Yes
					MN	Yes
					MO	--
					NB	Yes
					NJ	Yes
					NY	Yes
					OH	Yes
					TN	Yes
					WVA	--
				Total		9
			All part-time	State	AK	--
					ID	Yes
					NH	--
					SD	--
					VT	--
					WY	--
				Total		1
			Mixed	State	CT	Yes
					KY	--
					WA	Yes
				Total		2
No	Yes	Yes	All full-time	State	TX	--
				Total		--
			All part-time	State	MT	--
				Total		--
No	No	Yes	All full-time	State	AR	Yes
					LA	--
					MA	Yes
					MD	--
					OR	Yes
					PA	--
					SC	Yes
				Total		4
			Mixed	State	DE	Yes
					NV	--
					RI	--
				Total		1
		No	All part-time	State	NM	--
				Total		--
			Mixed	State	NC	Yes
				Total		1

Table 3. Model 2 – Operational, Pre-Release

Time off credits available?	Program completion required?			Use parole decision making instruments?
				Yes	No
Yes	Yes	State	AK	1	0
			AL	1	0
			AR	1	0
			AZ	0	1
			CO	1	0
			DE	1	0
			FL	1	0
			GA	1	0
			IA	1	0
			KS	0	1
			KY	1	0
			LA	1	0
			MA	1	0
			MD	1	0
			ME	0	1
			NB	1	0
			ND	1	0
			NJ	1	0
			OK	1	0
			RI	1	0
			SC	1	0
			SD	1	0
			TN	1	0
			TX	1	0
			WA	1	0
			WI	0	1
			WVA	1	0
			WY	1	0
			Total	24	4
	No	State	IL	1	0
			NV	1	0
			NY	1	0
			OR	0	1
			VA	0	1
			Total	3	2
No	Yes	State	CT	1	0
			HI	0	1
			ID	1	0
			MI	1	0
			MO	1	0
			MT	1	0
			NH	0	1
			OH	1	0
			PA	1	0
			UT	1	0
			VT	1	0
			Total	9	2

 

Table 4. Model 3 – Operational, Post-Release Supervision

Assessment instruments to set levels?	Assessment instruments used to set conditions?	Set conditions of parole?			Full authority over supervision of parolees?
					Yes	No
Yes	Yes	Yes	State	AL	1	0
				AR	1	0
				GA	1	0
				KY	1	0
				LA	1	0
				MA	1	0
				ME	1	0
				MI	0	1
				MN	0	1
				MO	1	0
				MT	0	1
				NB	0	1
				NJ	1	0
				OK	0	1
				PA	1	0
				SD	1	0
				TN	1	0
				WVA	0	1
				WY	1	0
				Total	13	6
	No	Yes	State	HI	1	0
				NH	1	0
				OH	0	1
				Total	2	1
No	Yes	Yes	State	IA		1
				Total		1
	No	Yes	State	FL	0	1
				NC	0	1
				NM	1	0
				NV	1	0
				NY	1	0
				OR	1	0
				RI	0	1
				SC	0	1
				VT	0	1
				Total	4	5

 

Table 5. Model 4 – Operational, Post-Release Revocation

Risk assessment for revocation decisions?			Authority to revoke supervision?
			Yes	No
Yes	State	CT	1
		FL	1
		IA	1
		KY	1
		MA	1
		ND	1
		NJ	1
		PA	1
		SD	1
		UT	1
		VT	1
		WY	1
		Total	12
No	State	AK	1	0
		AL	1	0
		AR	1	0
		AZ	1	0
		CO	1	0
		DE	1	0
		GA	1	0
		HI	1	0
		ID	1	0
		IL	1	0
		KS	1	0
		LA	1	0
		MD	1	0
		MI	1	0
		MN	1	0
		MO	1	0
		MT	1	0
		NB	1	0
		NC	1	0
		NH	1	0
		NM	0	1
		NV	1	0
		NY	1	0
		OH	1	0
		OR	1	0
		RI	1	0
		SC	1	0
		TN	1	0
		TX	1	0
		VA	1	0
		WA	1	0
		WVA	1	0
		Total	31	1

Figure 5

Discussion and Conclusion

Results suggest that state paroling authorities are more complex than previously documented. For instance, the structural model indicates that categorizing parole into “determinate” and “indeterminate” is not valid because most states operate under a mixed sentencing structure. Within this structure, references to discretionary and mandatory release are more complex as well, in that parole boards also have the power to terminate maximum sentence, but no authority to set the minimum time incarcerated.

The paroling authority models constructed for this study highlight the variability among the states’ approaches to release. Although each state may operate in a consistent way on a daily basis, there is much disparity in the way that incarcerated offenders are considered for and supervised on parole across the country. Some non-shared operational characteristics have important implications for practice. For instance, the fact that 31 states do not use risk assessments for their revocation decisions indicates a need for further standardization in this area to ensure that prisoners across the country are similarly treated and assessed for their risk to the public in a reliable and evidence-based manner. Furthermore, the fact that program completion prior to release is required in 24 states and, as survey results indicate, there are not enough programs, is also alarming. When factors that inmates believe affect release decision are different from the factors that paroling authorities actually consider, or when they are limited by the resources that are made available, inmates will be confused and angry and will be less likely to conform to requirements for institutional control (West-Smith et al., 2000). “Each parole case that is deferred or set back becomes another story, duly embellished,” wrote West-Smith et al. (2000, p. 9), “that makes its rounds throughout the prison population, fueling suspicion, resentment, and fear of an unbridled discretionary system of power, control, and punishment.”

The Model Penal Code was developed in the 1950s for the American Law Institute, a professional organization of lawyers, judges, and law professors (Robinson & Dubber, 1999). The Code’s purpose was “to stimulate and assist legislatures in making a major effort to appraise the content of the penal law by a contemporary reasoned judgment—the prohibitions it lays down, the excuses it admits, the sanctions it employs, and the range of the authority that it distributes and confers” (American Law Institute, n.d., para 1). Prosecutors, psychiatrists, mental health specialists, judges, academic scholars, and leading corrections professionals wrote the Code’s sentencing and corrections provisions (Tonry, 1999b). According to the first official draft of the Model Penal Code, adopted at the 1962 Annual Meeting of The American Law Institute, the first three general purposes for the sentencing and treatment of offenders was: a) to prevent the commission of offenses; b) to promote the correction and rehabilitation of offenders; and, b) to safeguard offenders against excessive, disproportionate, or arbitrary punishment (Robbinson & Dubber, 1999; Tonry, 1999b). The Code stated that prisoners should be eligible for parole release on completion of their minimum sentences and it created a presumption that prisoners would be released when they first became eligible.

The Model Penal Code was never adopted in toto by any of the 50 U.S. states, the District of Columbia, or the federal government. The few similarities among the paroling authorities reviewed in this study may be in large part due to the Model Penal Code’s influence (Robinson & Dubber, 1999). More apparent, however, is that after nearly 60 years, and well into a new century, the Code has had little impact on producing a cohesive system of parole in the U.S. There remains no standard approach to parole across jurisdictions today (Tonry, 1999b; Kinnevy & Caplan, 2008). This highlights the trend that parole systems have been moving farther away from a unifying system-wide model that was common throughout much of parole’s early history in the U.S.

It remains unclear whether a system of disparate paroling authorities is the most appropriate framework for the United States. On the one hand, it permits the individualization of parole at the state level—which is consistent with the federated system of government in the U.S. and the historically common principle to assess risk and parole release on a case-by-case basis. Arguably, this principle can apply to states’ environmental, social, political, and economic contexts as well as to an offender’s personal and criminal attributes. On the other hand, dissimilar paroling authorities permit and perpetuate unequal release and reentry outcomes for similarly-situated inmates across state lines. Findings from this study suggest that twenty-first century paroling authorities are complex systems that cannot be labeled according to only one of their attributes, as is commonly done. More accurate labels should take into account common structural and operational characteristics of paroling authorities.

___________________________________________________________________________________
* This study was paid for, in part, by a grant from the JEHT Foundation. Special thanks to the Association of Paroling Authorities International for professional advice and significant contributions to the data collection efforts.

back to top

References | Endnotes