Parole practises in Lithuania: Factors predicting court decisions

Abstract

This article examines factors that predict parole decisions in Lithuanian courts. The study state has a two-stage discretionary parole system where applicants are first evaluated through a parole board hearing, and the board’s decision is then reviewed in court. The study sample included 360 court verdicts from various court institutions. Intergroup comparisons suggest that parole boards tend to grant parole more often than courts. The results of regression analysis suggest that courts weigh heavily on the decision made by the parole board as well as the number of misconduct reports, time left to serve and previous parole or probation violations.

Keywords

Parole conditional release decision-making Lithuania regression analysis

Introduction

The legal context of parole applications

In Lithuania, parole is a form of conditional release from custody that may be applied in the case of a determinate sentence after some proportion of the sentence has been served. The study state currently has a two-stage parole application system where the applicant may be granted parole on discretionary conditions. The option of automatic parole has been established recently; however, discretionary parole remains the most prevalent framework. The current parole application system was developed in 2012 after the initiation of the Probation reform. The reform involved significant updates to the Code for the Execution of Criminal Penalties (CECP) and the adoption of the Probation Act. The main purpose of the reform was to address prison overcrowding problems and ensure smoother social integration of probationers and parolees.¹

Until 2012, the main parole condition was a duty to serve a certain part of the sentence based on the length of the original sentence (CECP, 2002). The decision-makers were given no specific guidelines or criteria to rely on while dealing with parole cases. Hence, courts exercised a great deal of discretion in terms of parole decisions.

After 1 July 2012, the legal terms of parole application were redefined to be more specific. Following this, inmates continued to be required to serve a certain part of the sentence depending on the original length.² Additional criteria included (1) the completion of measures listed in the correctional plan, (2) level of criminal risk³ and (3) the inmate’s institutional conduct (Law on changes of the CECP, 2012). The correctional plan had to be based on the results of the criminal risk assessment and address the specific criminogenic needs of each inmate. This instrument was developed to ensure the routinisation of correctional measures and facilitate the monitoring of the rehabilitation process. Thus, from then on, the parole decision has depended on the level of criminal risk, institutional conduct and the inmate’s efforts to adhere to the correctional plan.

However, the list of criteria was not finite as ‘other significant circumstances’ could also be considered (Law on changes of the CECP, 2012). Although the new framework was more specific, the legal terms of parole still lacked clarity. It was not clear what level of criminal risk and which aspects of institutional conduct indicated eligibility for conditional release.

Another significant change made in 2012 was the establishment of parole boards.⁴ Before 2012, local courts dealt with all parole cases. Parole boards emerged as an additional parole decision-making authority charged with considering social research reports for each applicant and providing individualised recommendations on parole conditions. Local courts were then given the power to make parole decisions based on parole board recommendations. In addition, the 2012 revisions permit only board members to meet the applicants in person, while written forms are reviewed for parole cases in court.⁵ These changes added complexity to the parole application system, as inmates were now required to pass two parole stages instead of one.

In addition, intensive supervision (IS) using electronic monitoring⁶ was introduced in 2012, which allows inmate who agree to its terms to apply for parole 6 months prior to the term set in the CECP. Every inmate eligible for parole could apply for this option regardless of the sentence length or crime committed (Law on changes of the CECP, 2012). IS was a novel system in Lithuania and created a new category of parolees. The introduction of IS required decision-makers to consider a new parole option and adjust their decision-making strategies to a new system.

Further changes to the parole system occurred on 1 September 2015, resulting in the liberalisation of parole terms. These revisions required inmates to satisfy three basic requirements in order to obtain parole: (1) the obligation to comply with (rather than complete) the measures included in the correctional plan, (2) being classified as low-risk and (3) the progress of risk reduction, indicating that the inmate shall not reoffend. In addition, inmates who agreed to IS could apply for parole 9 months (instead of 6 months) prior to the term set in the CECP (Law on changes of the CECP, 2015).

Due to these updates, institutional conduct and ‘other significant circumstances’ became irrelevant, while the progress of risk reduction became the key criterion for parole application. Unfortunately, it was not specified how the progress should be measured, as no corresponding guidelines were provided to decision-makers.

Further substantial changes to parole application in the CECP came into force on 1 June 2020. Since then, some parole cases⁷ have been assigned to boards and some inmates were granted automatic parole along with IS. Nevertheless, a significant number of inmates have to pass the same two parole stages. Furthermore, the criteria for parole application include only the level of criminal risk and the progress of risk reduction (Law on changes of the CECP, 2019). At this stage, it is difficult to tell whether the latest changes in the parole system will improve parole rates in Lithuania. However, regardless of the complexity of the new system, the criteria for parole application remain rather obscure.

Practical issues of parole application

The establishment of a new parole system in 2012 led to a significant drop in the number of parolees. In 2008, parolees made up approximately 51% of all inmates released from prison. However, in 2012, this number dropped to approximately 37% (Report of the general unit of the Prison Department, 2009, 2013). In 2017 and 2018, the percentage of parolees dropped to 27% and 20% of all inmates released from prison, respectively (Report of the general unit of the Prison Department, 2018a, 2018b, 2018c; Report of the task planning unit of the Prison Department, 2019a, 2019b, 2019c). This meant that a substantial number of inmates stayed in prison rather than being supervised and assisted in re-entry into society.

In 2015, the Prison Department conducted an in-depth analysis of parole practises in the state and found that boards received 8782 parole applications from 1 July 2012 to 1 January 2015. While boards granted parole in 4886 cases (nearly 56%), courts granted parole in only 2769 cases (31%). It was concluded that courts tended to take a more reserved stance on parole application, as they granted parole less often than boards (Report of the supervision unit of the Prison Department, 2016).

It has also been observed that, in Lithuania, parole trends vary depending on the correctional facility. Offenders with extensive criminal history are placed in special correctional facilities such as the Alytus correctional house. This group of inmates usually has a greater number of prior convictions and a higher level of criminal risk, and may, therefore, face more severe integration difficulties. Meanwhile, a proportion of those serving their first prison sentence are placed in remand prisons, where inmates are employed as serving staff. These inmates usually get employed in custody, manifest fewer substance abuse problems and have fewer misconduct reports. The comparison of parole application rates in various correctional facilities revealed substantial differences between institutions. Inmates serving sentences in remand prisons were granted parole more often (45%) than inmates with extensive criminal history (23%; Report of the supervision unit of the Prison Department, 2016).

As the percentage of parolees has been gradually decreasing, local scholars assumed that the decrease may be due to stereotypes escalating in local media. The media tend to portray parolees as dangerous recidivists posing a serious threat to society. Scholars from abroad also recognise that the only time the society hears anything about parolees is when they commit a new crime (Ball, 2011). However, data indicate that, from 1998 through 2013, the percentage of parolees in Lithuania who were returned to prison for parole violations or committing a new offence was between 7% and 13% (Sakalauskas, 2013). Hence, reoffending could be considered quite uncommon.

In addition, local parole trends appear pessimistic in the European context. According to correctional statistics from Nordic countries, from 2009 to 2013, the percentage of parolees among probationers in Denmark was between 17% and 19%, Finland −29% and 36%, Iceland −45% and 50%, Norway −14% and 16%, and Sweden −30% and 31% (Kristoffersen, 2014). During this period, the percentage of parolees among probationers in Lithuania decreased to between 11% and 17%. In 2018, the percentage of parolees among probationers in Lithuania dropped to 8%, which is below the European average (Aebi et al., 2019). These numbers show that despite the legal developments, parole remains an uncommon form of community sanction in Lithuania.

A high rate of parole denial results in most inmates being released at the end of their sentence without any assistance or follow-up. Furthermore, local parole trends may negatively affect inmates’ motivation to rehabilitate and seek parole. To date, research has not sought to examine the inmates’ perspective on parole decision-making in Lithuania. In 2000, a group of foreign authors investigated inmates who had been denied parole and revealed that the factors inmates believed to affect parole decisions differed significantly from those that parole boards considered relevant (West-Smith et al., 2000). Thus, after parole denial, inmates may feel that their attempts to rehabilitate had been pointless and question whether parole is worth the effort. High rates of denial may cause parole to lose its main advantage, that is, the possibility to motivate inmates toward positive behavioural changes.

Factors associated with parole decision-making

Although low parole rates may hinder gradual offender integration, parole decision-making is unquestionably complex. Therefore, some research has focused on factors affecting parole decision-making, with some studies identifying historical risk factors as influential. Parole boards may regard inmates’ criminal history, institutional conduct, program non-compliance, number of previous incarcerations, age of the first offence, previous escape attempts and parole violations (Houser et al., 2019; Nuffield, 1982; Warner, 1923). Other studies show that parole decisions may depend on the severity of the crime committed, crime type, violent behaviour in custody, lack or delay of attending relevant programs in prison, the number of previous convictions, incarceration length and history of mental illness (Caplan, 2007; Caplan and Kinnevy, 2010; Feder, 1994; Lindsey and Miller, 2011; Morgan and Smith, 2005; Proctor, 1999). In other words, historical information may be of great importance when it comes to parole decision-making.

Other studies have shown that other characteristics are also important when it comes to parole decisions, including inmate’s age, intellectual abilities, officer’s recommendation, participation in educational programs, prospects of employment after release, marital status and the number of dependents (Heinz et al., 1976; Houser et al., 2019). Another study has shown that, while parole boards emphasise factors such as the nature of the current offence, institutional conduct and education, their decisions may also depend on the level of criminal risk as well as the race and ethnicity of the inmate (Huebner and Bynum, 2008). A study on parole of men incarcerated for sexual offences showed that decision-makers heavily weigh the seriousness of the offence, institutional conduct, and parole readiness scores; however, victim and offender age are also significant factors (Huebner and Bynum, 2006). Therefore, it seems that parole decision-making may be affected by various legal and extralegal factors, while decision-makers may sometimes unconsciously rely on particular stereotypes about certain applicant groups.

Due to undeniable parole application difficulties, Lithuanian scholars have attempted to discover their possible underlying causes. One study published in 2017 suggests that the main motives for parole denial in court were inadequate time served, the severity of the crime committed, the presence of some criminogenic factors and personality traits. In addition, it was concluded that judges often disagreed regarding the progress of risk reduction due to various interpretations of this factor (Vosyliute, 2017).

Further, a qualitative study regarding parole decision-making in Lithuania found that board members tend to focus on the CECP criteria, while some additional factors, such as institutional conduct, the nature of the crime committed and other characteristics, are also influential. Similarly, judges indicated that institutional conduct, criminal history, personality traits and time spent in custody are also important. In addition, judges believed that parolees should be classified as low-risk, while compliance with the plan was not given high importance, as it was considered a formal requirement rather than the indicator of successful rehabilitation (Michailovic and Jarutiene, 2017).

Although there have been several attempts to uncover the factors that could be associated with parole decisions, no proper quantitative study has ever been conducted in Lithuania. Therefore, the main purpose of this study is to identify particular variables that predict parole decisions in Lithuanian courts.

Materials and methods

Research data

The data for this study were collected from the LITEKO open-access database, which contains all court verdict records enunciated in the state.⁸ The search period covered 4 years (from 1 September 2015 to 1 November 2019), which correspond to the period of validity of the latest parole application system. The study state contains six district courts,⁹ which have recently been dealing with parole applications from eight correctional facilities. Since applicants may appeal unsatisfactory district court verdicts, there are also three regional courts dealing with parole cases.¹⁰ Accordingly, parole verdicts from all nine court institutions were selected for this study. The sample size was determined by the number of independent variables (at least 20 cases per independent variable) and included 360 verdict records.

After data selection, all applicants’ characteristics relevant to the study were encoded using IBM SPSS Statistics 21.

Variables used in the study

Dependent variable. The dependent variable in this study was court decision (parole granted or denied) for each applicant. As the measure of the dependent variable was dichotomous, logistic regression analysis was chosen as the most suitable statistical method.

Independent variables. Independent variables fell into three categories: (1) statutory, (2) legal/extralegal and (3) procedural. A more detailed description of the independent variables is provided in Table 1.

Table 1.

Independent variables used in the study.

Variables	Categories
Statutory variables
Compliance with the plan	0 = does not comply
Compliance with the plan	1 = does comply
Level of criminal risk	OASys score (0–168 points)
OASys score shift	OASys score shift since last assessment (negative values indicating the drop of OASys score, 0 value indicating no shift, positive values indicating increase of OASys score)
Application for IS	0 = did not apply
Application for IS	1 = did apply
Legal/extralegal variables
Number of previous convictions	Total number of previous convictions
Crime severity	1 = minor
	2 = moderate severity
	3 = severe
	4 = extremely severe
Nature of crime committed	1 = violent
	2 = property
	3 = drug-related
	4 = economic
	5 = sexual
	6 = other
Institutional conducta) Valid + invalid misconduct reportsb) Rewards	a) Total number of valid and invalid misconduct reports
	b) Total number of rewards granted in custody
Remaining part of the sentence	Total number of months left to serve
Previous parole/probation violations	0 = no
Previous parole/probation violations	1 = yes
Plea in court	1 = not guilty
	2 = no contest
	3 = guilty
Compliance with restitution order	0 = does not comply
Compliance with restitution order	1 = does comply
Sex	0 = male
Sex	1 = female
Procedural variables
Board’s decision	0 = parole denied
Board’s decision	1 = parole granted
Court instance type	0 = district court
Court instance type	1 = regional court

Statutory variables were based on the parole conditions determined in the CECP, while procedural variables were based on the current parole application system in the state. Legal/extralegal variables were drawn from previous studies that revealed possible predictors of parole decision-making. Some extralegal variables such as ethnicity, race, age of the first offence, marital status and education, were not included, as they were not available in the records drawn for the study.

During the study, two independent variables (i.e. Plea in court and Compliance with restitution order) were excluded from the analysis, as the number of missing values exceeded 60%.

Data analysis

IBM SPSS Statistics 21 was used for statistical data analysis. A multiple imputation procedure was used to deal with missing values. The chi-square test was used for intergroup comparisons. Before applying logistic regression, a correlational analysis of independent variables was performed to avoid multicollinearity issues. As some of the variables used a scale and some were categorical, Spearman correlation was utilised. Finally, logistic regression models were developed to identify the variables that predict parole decisions.

Results

General findings

Detailed characteristics are provided in Table 2. Among the 360 parole applicants included in the sample, 74% were denied parole, while 26% were granted parole in court. Of the total, 78% were male and 22% were female.¹¹ The majority of applicants were convicted for severe or extremely severe crimes such as aggravated assault, murder and drug trafficking. The results indicated that parole boards take a less rigorous stance toward the applicants, as the parole approval rate was notably higher compared to courts, with 72% of applicants granted parole during parole board hearing.

Table 2.

Sample characteristics.

	Parole denied (N = 266; 74%)	Parole granted (N = 94; 26%)	Total sample (N = 360)
Statutory characteristics
Compliance with the plan	Does comply (245; 92%)	Does comply (94; 100%)	Does comply (339; 94%)
Compliance with the plan	Does not comply (21; 8%)	Does not comply (0; 0%)	Does not comply (21; 6%)
Level of criminal risk	Low (78; 29%)	Low (49; 52%)	Low (127; 35%)
	Moderate (181; 68%)	Moderate (45; 48%)	Moderate (226; 63%)
	High (7; 3%)	High (0; 0%)	High (7; 2%)
OASys score shift	M = −5; SD = 7.7	M = −6.4; SD = 7.1	M = −5.4; SD = 7.6
Application for IS	Did apply (75; 28%)	Did apply (28; 30%)	Did apply (103; 29%)
Application for IS	Did not apply (191; 72%)	Did not apply (66; 70%)	Did not apply (257; 71%)
Legal/extralegal characteristics
Sex	Male (202; 76%)	Male (78; 83%)	Male (280; 78%)
Sex	Female (64; 24%)	Female (16; 17%)	Female (80; 22%)
Crime severity	Minor (27; 10%)	Minor (13; 14%)	Minor (40; 11%)
	Moderate severity (61; 23%)	Moderate severity (23; 25%)	Moderate severity (84; 23%)
	Severe (83; 31%)	Severe (21; 22%)	Severe (104; 29%)
	Extremely severe (95; 36%)	Extremely severe (37; 39%)	Extremely severe (132; 37%)
Board’s decision	Parole granted (170; 64%)	Parole granted (89; 95%)	Parole granted (259; 72%)
Board’s decision	Parole denied (96; 36%)	Parole denied (5; 5%)	Parole denied (101; 28%)
Previous parole/probation violations	No (75; 28%)	No (45; 48%)	No (120; 33%)
Previous parole/probation violations	Yes (191; 72%)	Yes (49; 52%)	Yes (240; 67%)
Number of previous convictions	M = 5.3; SD = 3	M = 4.4; SD = 3	M = 5.1; SD = 3
Number of misconduct reports	M = 2.7; SD = 4.5	M = 1.2; SD = 2.3	M = 2.3; SD = 4.1
Number of rewards	M = 3.5; SD = 3.8	M = 4.3; SD = 4.5	M = 3.7; SD = 4
Months left to serve	M = 24.9; SD = 15.9	M = 22.5; SD = 15.8	M = 24.3; SD = 15.9

Intergroup comparisons

Intergroup comparisons showed that the differences between court and parole board practise were statistically significant (χ² = 31.1; p < 0.001). This indicates that board members tend to trust parole applicants more than judges.

A comparison of criminal risk level groups showed that the majority of those who were granted parole were classified as low-risk. Furthermore, the percentage of moderate-risk inmates granted parole was smaller compared to those who were not. Finally, there were no high-risk inmates among those who were granted parole. The chi-square test showed that these differences were statistically significant (χ² = 17.2; p < 0.001).

The analyses also revealed some differences between district and regional court practise. Regional courts granted parole less often (13% of applicants) than district courts (33% of applicants; χ² = 14.3; p < 0.001). Thus, in Lithuania, appeal courts seem to demonstrate less confidence in parole-seeking inmates compared to trial courts.

There were no statistically significant differences between those who applied for IS and those who did not. In addition, inmates granted parole did not significantly differ from those who were denied parole in terms of crime severity. The distribution of those who committed minor, moderate severity, severe or extremely severe crime was similar between the two groups.

Intergroup comparisons showed that the majority of those who were denied parole (72%) had a previous history of parole or probation violation. Furthermore, the group of unsuccessful applicants included a higher percentage of those who had a previous violation history, compared to the group of successful applicants (χ² = 21.3; p < 0.001).

No statistically significant differences were found between male and female applicants. In addition, applicants who had been sentenced for violent, property, economic, sexual or other types of crime had a success rate similar to those who had not been sentenced for these types of crime.

Furthermore, the results showed that there were fewer successful applicants among inmates sentenced for drug-related crime (χ² = 3.8; p = 0.05). In other words, the results have shown that those who had been sentenced for drug-related crime were less likely to be paroled.

Correlational analysis

In order to avoid multicollinearity problems in the logistic regression model, all independent variables underwent correlational analysis. The results are presented in Table 3.

Table 3.

The results of correlational analysis¹³.

	Compliance with the plan	Level of criminal risk	Application for IS	Previous convictions	Crime severity	Violent	Property	Drug related	Economic	Sexual	Other	Misconduct reports	Rewards	Months left to serve	Parole/probation violations	Board’s decision	Court institution	Sex	OASys score shift
Compliance with the plan	1
Level of criminal risk	−0.111*	1
Application for IS	−0.078	−0.044	1
Previous convictions	−0.047	0.468**	−0.08
Crime severity	−0.03	−0.483**	0	−0.431**	1
Violent	−0.05	−0.01	−0.12*	−0.02	0.214**	1
Property	−0.05	0.306**	0.05	0.315**	−0.422**	−0.02	1
Drug related	−0.01	−0.09	0.158**	−0.15**	0.209**	−0.476**	−0.332**	1
Economic	−0.03	−0.177**	0.05	−0.147**	0.09	−0.249**	−0.167**	0.232**	1
Sexual	−0.04	0.03	−0.05	−0.002	0.02	0.09	0.02	−0.09	−0.03	1
Other	−0.121*	0.03	0.148**	0.03	−0.08	−0.131*	0.123*	−0.03	0.04	0.05	1
Misconduct reports	−0.193**	0.03	0.1	−0.01	0.287**	0.126*	−0.02	0.09	0.05	−0.004	0.01	1
Rewards	0.08	−0.285**	0.02	−0.190**	0.426**	0.161**	−0.139*	0.09	0.184**	0.01	−0.112*	0.296**	1
Months left to serve	0.05	−0.416**	0.210**	−0.391**	0.582**	0.05	−0.257**	0.251**	0.158**	−0.01	−0.06	0.2**	0.411**	1
Parole/probation violations	−0.01	0.403**	−0.08	0.439**	−0.349**	−0.03	0.06	−0.08	−0.03	0.001	−0.01	−0.04	−0.208*	−0.32**	1
Board’s decision	0.293**	−0.304**	−0.03	−0.191**	0.1	0.05	−0.1	0.02	0.07	−0.07	−0.08	−0.109*	0.155**	0.1	−0.136*	1
Court institution	−0.05	−0.08	0.02	−0.06	0.133*	0.05	−0.06	0.05	0.01	−0.02	−0.05	0.03	0.06	0.111*	−0.04	−0.004	1
Sex	0.08	−0.109*	−0.07	−0.222**	0.1	−0.170**	−0.143**	0.128**	−0.130**	−0.07	−0.125*	−0.232**	−0.25**	0.124*	0.08	0.1	0.189**	1
OASys score shift	−0.04	0	0.06	−0.04	0.08	−0.142	−0.07	0.120	−0.02	−0.117	−0.02	0.06	−0.09	0.07	−0.02	−0.06	0.221*	0.191*	1

The results did not reveal any serious multicollinearity issues. Only two independent variables (number of months left to serve and crime severity) appeared to be strongly correlated (r = 0.582; p = 0.01). A strong association between these two variables is reasonable, as inmates sentenced for more severe crimes were likely to receive longer sentences and, as a result, have longer unserved sentence portions. As these two variables could provide redundant information, the Crime severity variable was excluded from further analysis.

Logistic regression model

During logistic regression analysis, three regression models were developed (Table 4). Model 1 included only statutory variables. The analysis showed that the only significant predictor of court decision was the inmate’s level of criminal risk based on the OASys score. As expected, higher OASys scores indicated lower chances of being paroled. This may mean that judges tend to be more forgiving with inmates who were classified as low-risk. Although the Omnibus and Hosmer–Lemeshow tests showed that the model was appropriate, it was able to account for only 6–9% of the variance (Table 4).

Table 4.

The results of logistic regression analysis.

	Model 1			Model 2			Model 3
Independent variables	Exp (B)	B	p	Exp (B)	B	p	Exp (B)	B	p
Statutory variables
Compliance with the plan	522,283,257	20	1.000
Level of criminal risk (OASys score)	0.981	−0.019	0.002	0.982	−0.018	0.029	0.988	−0.012	0.19
Application for IS	1.119	0.112	0.68	1.317	0.275	0.37	1.299	0.261	0.43
OASys score shift	0.974	−0.026	0.15	0.983	−0.017	0.42	1.003	0.003	0.98
Legal/extralegal variables
Number of previous convictions				0.939	−0.063	0.32	0.957	−0.044	0.52
Violent crime				0.777	−0.252	0.47	0.787	−0.240	0.54
Property crime				0.958	−0.043	0.89	0.893	−0.113	0.75
Drug-related crime				0.530	−0.635	0.11	0.490	−0.714	0.09
Economic crime				0.733	−0.311	0.65	0.656	−0.422	0.57
Sexual crime				0.446	−0.807	0.47	0.648	−0.434	0.72
Other type of crime				1.057	0.055	0.87	1.165	0.153	0.67
Number of misconduct reports				0.858	−0.154	0.01	0.872	−0.136	0.03
Number of rewards				1.065	0.063	0.1	1.054	0.053	0.19
Number of months left to serve				0.970	−0.031	0.01	0.976	−0.025	0.04
Previous parole/probation violations				0.471	−0.752	0.05	0.443	−0.815	0.05
Sex				0.643	−0.442	0.27	0.747	−0.291	0.50
Procedural variables
Court instance type							0.297	−1.214	0.001
Parole board’s decision							7.842	2.060	0.0001
Cox–Snell and Nagelkerke R-squared	6–9%			21–33%			29–46%

Further, Model 2 included statutory¹² as well as legal and extralegal variables and revealed some additional significant predictors (Table 4). Inmate’s level of criminal risk as well as the number of misconduct reports, number of months left to serve, and the history of previous parole or probation violations were found to be significant. The results suggest that higher OASys score, greater number of misconduct reports, a longer unserved sentence, and previous parole or probation violation history predicted a lower chance of being paroled. The Omnibus and Hosmer–Lemeshow test confirmed that Model 2 was appropriate, while Cox–Snell and Nagelkerke R-squared test showed that the fit of the model improved substantially, explaining 21–33% of the variance.

Finally, Model 3 was developed, including statutory, legal, extralegal and procedural variables. Both procedural variables contributed substantially to the model and were found to be the most significant predictors of parole decisions in court. In Model 3, the OASys score was not a significant predictor. The parole board’s decision was the most significant variable for parole decision in court. These findings indicate that judges rely heavily on the board’s opinion. In addition, the court instance type was significant for parole decision-making. This indicates that parole cases considered in appeal courts are less likely to be resolved favourably. Other predictors, including the number of misconduct reports, time left to serve, and previous history of parole or probation violations, were also significant. The Omnibus and Hosmer–Lemeshow test confirmed that Model 3 was appropriate, while Cox–Snell and Nagelkerke R-squared showed that it explained 29–46% of the variance. Thus, Model 3 was the most beneficial of all three models.

Discussion

This study aimed to identify predictors of parole decision-making in court. The results showed that parole boards tend to recommend parole more often than courts. This corresponds to the findings of the local Prison Department (2016). The current two-stage parole model does not seem to be beneficial to inmates, as a large percentage of applicants do not succeed in getting through the second parole stage. The latest changes in the CECP may partially solve this problem, as some parole cases are being assigned solely to parole boards. However, some applicants still have to receive approval from both decision-making authorities and will presumably face the same difficulties.

This study revealed other interesting insights. Overall, half of the inmates granted parole were classified as low-risk, while those denied parole were mostly classified as moderate-risk. These findings are consistent previous research indicating judges’ beliefs that parolees should mostly be classified as low-risk (Michailovic and Jarutiene, 2017). Although this criterion is included in the CECP, the inmate’s progress with risk reduction should also be considered. Finally, most inmates in this study sample cannot be classified as low-risk due to multiple previous convictions, parole or probation violations, or having been convicted of severe or extremely severe crime. As such, the requirement of a low-risk score may be unreasonable, and it is probably worth reconsidering this particular criterion.

Finally, intergroup comparisons showed that those convicted for drug-related crimes were less likely to be granted parole in court. It should be noted that some previous studies abroad have revealed the opposite tendencies (Feder, 1994; Houser et al., 2019; Huebner and Bynum, 2008). This tendency may be due to a rather strict criminal policy regarding drug-related crime in Lithuania. Since 2017, all types of drug-related crimes, except drug use, have been criminalised in the state. An analysis of punitive practise for drug-related crime showed that from 2009 through 2012, in most cases, the perpetrators were given custodial sentences (Venckeviciene, 2013). Furthermore, in Lithuania, drug distribution is considered a severe crime regardless of the amount distributed, so these custodial sentences range from 2 to 8 years for the distribution of small amounts (Lankauskas, 2013). Thus, the legislation, as well as punitive practise regarding drug-related crime, is rather strict. It may be assumed that the general attitude toward drug-related crime in this state affects parole decisions.

The logistic regression model included statutory, legal, extralegal and procedural variables and showed that procedural variables were the strongest predictors of court decisions. The results of this study suggest that judges tend to weigh heavily on decisions made by parole boards. A favourable decision by the board predicted a greater chance of being paroled in court. Furthermore, the results showed that parole cases that were considered in trial courts had a greater chance of being successful. It was assumed that the decision-making process in appeal courts might be informed by the so-called “framing effect”, a cognitive heuristic where the decision is influenced by the way information is presented. In most cases, applicants were denied parole either during board hearings or in the trial court. The former decisions may serve as a reference point and shift appeal courts’ decisions toward parole denial. Generally, the significance of the procedural variables implied that the structure of the parole application system contributes to the decrease in the parole rate in the state.

Some legal variables were also found to be significant in terms of parole decision in court. The number of misconduct reports as well as time left to serve were significant predictors of the court decision. A greater number of misconduct reports and longer part of the sentence unserved decreased the chances of earlier release. In addition, inmates with previous history of parole or probation violation were less likely to be paroled. The results suggest that parole decision-making in court is mostly affected by negative historical information about applicants. These findings are consistent with the results of previous studies regarding parole decision-making (Caplan, 2007; Houser et al., 2019; Huebner and Bynum, 2008; Lindsey and Miller, 2011; Morgan and Smith, 2005; Nuffield, 1982; Proctor, 1999; Warner, 1923). Unlike some previous studies (Heinz et al., 1976; Huebner and Bynum, 2008; Michailovic and Jarutiene, 2017), this study showed that the level of criminal risk appears to be insignificant in terms of parole decision-making. This may indicate that judges tend to focus on the risk factors they believe to be significant in terms of recidivism and public safety. The criminal risk assessment framework might still seem novel and complicated and, therefore, is often disregarded during the decision-making process.

It should be noted that all variables that were significant predictors of the court decision are taken into account during the criminal risk assessment. OASys covers such characteristics as previous parole or probation violations and misconduct reports. Therefore, the results suggest that some historical risk factors are repeatedly evaluated in court, potentially unreasonably amplifying perceived criminal risk. This issue has been discussed in previous research, which argued that frequent application of criminal risk assessment for legal decision-making may lead to the accumulation of bias; thus, such practise, in fact, becomes exclusionary to some individuals (Van Eijk, 2020). Instead of providing an informative basis for parole decision-making, the results of risk assessment may become an obstacle to conditional release. It has also been argued that parole decisions should not be informed by risk assessment, as it is difficult to measure dynamic risk factors in custody and predict the course of integration after release. Another issue may be due to the way risk assessment results are communicated to decision-makers. Although the results provide information on the probability of recidivism, decision-makers receive no comments regarding the confidence interval of risk estimates (McGarraugh, 2012). To avoid these issues, judges should receive proper training regarding risk assessment instruments, their exact structure, and the possible level of error inherent in the presented results. Furthermore, the focus on risk should be reconsidered, and additional evidence-based criteria could be relied upon to avoid exclusionary practises.

The apparent insignificance of the statutory criteria set in the CECP might be due to the absence of clear parole application guidelines in the state. As such, decision-makers should be provided with an advisory decision recommendation matrix or framework, which would help categorise inmates in terms of their eligibility for parole. Such tools are widely used in the US (Ford, 2012; Georgia Board of Pardons and Paroles Annual Report, 2019; Gutierrez, 2019; Serin and Gobeil, 2014; Wardrop and Serin, 2019) and may include components such as the level of criminal risk, institutional/community behaviour, release plan, and other relevant information. In addition, as the progress of risk reduction remains a significant criterion after the latest changes in CECP, it should be specified how the progress should be measured.

This study had some obvious limitations concerning the content of verdict records uploaded to the LITEKO database. In most cases, no definite information regarding the sociodemographic characteristics of applicants was available; therefore, variables such as age, education, marital status, race and ethnicity were not included in the model. As such, it was not possible to explore the role of race and ethnicity, which has been addressed in previous studies (Huebner and Bynum, 2008). In addition, data did not include any information regarding the rehabilitation process in custody (employment, educational and correctional program completion) and particular criminogenic needs of the applicants. Finally, this study did not include any information regarding applicants’ mental health status, which is also considered a significant factor in parole application (Feder, 1994; Houser et al., 2019). The inclusion of such data would have allowed the development of a more detailed model and identification of additional predictors of parole decisions. Access to applicants’ case files would help solve these problems; therefore, such an option should be considered in future studies.

Conclusions

Parole decision-making in Lithuanian courts highly depends on procedural and historical risk variables, while statutory criteria provided in the CECP do not seem to be significant. The decision made by the parole board and court institution type appears to be strong predictors of parole decision-making in court. These results suggest that the structure of the parole application system has a great impact on the possibility of conditional release. Previous parole or probation violation history, number of misconduct reports and the number of months left to serve also predict the final court decision. This may lead to an excessive evaluation of particular risk factors and inaccurate perception of the actual risk.

Such tendencies may be a result of a rather obscure definition of the legal terms set in the CECP and the absence of clear parole decision-making guidelines in the study state. To facilitate the decision-making process and reduce the level of disparity between similar parole cases, decision-makers should be provided with an advisory matrix or framework. In addition, receiving proper training regarding criminal risk assessment tools and their structure and confidence intervals could help judges improve their decision-making strategies. Finally, it is worth reconsidering the legal demand to classify inmates as low-risk, as most applicants cannot meet this requirement.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Liubove Jarutiene

Notes

Author biography

Liubovė Jarutienė is a researcher at the Law institute of Lithuania. She has received MA degree (Psychology of Law) from Vilnius University and is currently a doctorate student in Vilnius University. Ms Jarutienė‘s areas of scientific interest include prison conditions, offender rehabilitation and parole decision-making. Liubovė Jarutienė is a co-author of a study, ‚Conditions of Imprisonment and Premisses for Social Integration of Prisoners’ published in 2019. Ms Jarutienė also has practical experience of working in Lithuanian correctional system. Recently, Liubove Jarutiene participated in several scientific project funded by Research Council of Lithuania that were aimed at disclosing the specific punitive practices applied to women and the prevalence of stalking in Lithuania.

References

Aebi

Hashimoto

Tiago

(2019) Probation and prisons in Europe, 2018: Key findings of the SPACE reports. Available at: http://wp.unil.ch/space/files/2019/05/Key-Findings_SPACE-II_190520-1.pdf (accessed 17 August 2020).

Ball

(2011) Normative elements of parole risk. Stanford Law and Policy Review 22: 395–412.

Caplan

(2007) What factors affect parole: A review of empirical research. Federal Probation 71(1): 16–23.

Caplan

Kinnevy

(2010) National surveys of state paroling authorities: Models of service delivery. Federal Probation 74(1): 34–42.

Code for the execution of criminal penalties of the Republic of Lithuania (2002) Government Bulletin No. 73–3084. Available at: https://e-seimas.lrs.lt/portal/legalAct/lt/TAD/TAIS.171368/GcdFNpzgun (accessed 10 October 2020).

Feder

(1994) Psychiatric hospitalization history and parole decisions. Law and Human Behavior 18(4): 395–410.

Ford

(2012) Overview: Colorado state board of parole administrative release guideline instrument. Available at: https://cdpsdocs.state.co.us/ors/docs/Risks/SUMMARY-PBRGI_2012-11-01.pdf (accessed 24 April 2020).

Georgia board of pardons and paroles annual report FY 2019 (2019). Available at: https://www.tdcj.texas.gov/bpp/publications/PG%20AR%202018.pdf (accessed 24 April 2020).

Gutierrez

(2019) The Texas Board of Pardons and Paroles Parole Guidelines Annual Report FY 2018. Available at: https://www.tdcj.texas.gov/bpp/publications/PG%20AR%202018.pdf (accessed 24 April 2020).

10.

Heinz

Senderowitz

, et al. (1976) Sentencing by parole board: An evaluation. Journal of Criminal Law and Criminology 67(1): 1–31.

11.

Houser

Vilcica

Saum

, et al. (2019) Mental health risk factors and parole decisions: Does inmate mental health status affect who gets released. International Journal of Environmental Research and Public Health 16(16): 2950.

12.

Huebner

Bynum

(2006) An analysis of parole decision making using a sample of sex offenders: A focal concerns perspective. Criminology 44(4): 961–991.

13.

Huebner

Bynum

(2008) The role of race and ethnicity in parole decisions. Criminology 46(4): 907–938.

14.

Kristoffersen

(2014) Correctional Statistics of Denmark, Finland, Iceland, Norway and Sweden 2009−2013. Oslo: Correctional Service of Norway Staff Academy.

15.

Lankauskas

(2013) Drug control policy and legal regimes in Europe and Lithuania. Legal Issues 2(80): 81–117.

16.

Law on changes of the 18, 19, 66, 90, 91, 126, 138, 140, 152, 154, 157, 158, 159, 164, 176 Articles and invalidation of the 127, 160, 161, 162, 163, 179 Articles of Code for the execution of criminal penalties of the Republic of Lithuania (2012) Government Bulletin No. 4-110.

17.

Law on changes of the Code for the execution of criminal penalties of the Republic of Lithuania (2015) Register of Legal Acts No. 11069.

18.

Law on changes of the Code for the execution of criminal penalties of the Republic of Lithuania (2019) Register of Legal Acts No. 11180.

19.

Lindsey

Miller

(2011) Discretionary release decisions of actual and mock parole board members: Implications for community sentiment and parole decision-making research. Psychiatry, Psychology and Law 18(4): 498–516.

20.

McGarraugh

(2012) Up or out: Why sufficiently reliable statistical risk assessment is appropriate at sentencing and inappropriate at parole. Minnesota Law Review 97: 1079–1113.

21.

Michailovic

Jarutiene

(2017) Factors that influence parole boards’ and judges’ decisions on parole application in Lithuania. Baltic Journal of Law and Politics 10(1): 230–259.

22.

Morgan

Smith

(2005) Parole release decisions revisited: An analysis of parole release decisions for violent inmates in a Southeastern state. Journal of Criminal Justice 33: 277–287.

23.

Nuffield

(1982) Parole decision-making in Canada: Research towards decision guidelines. Available at: https://www.ncjrs.gov/pdffiles1/Digitization/126179NCJRS.pdf (accessed 24 April 2020).

24.

Proctor

(1999) The “new parole”: An analysis of parole board decision making as function of eligibility. Journal of Crime and Justice 22(2): 193–217.

25.

Report of the general unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the number, structure and change of prison population in year 2008 (2009). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

26.

Report of the general unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the number, structure and change of prison population in year 2012 (2013). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

27.

Report of the supervision unit of the Prison Department under the ministry of justice of the Republic of Lithuania on parole process and parole boards’ and courts’ motives of parole denial in year 2015 (2016). No. LV948.

28.

Report of the general unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the number, structure and change of prison population in year 2017 (2018a). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

29.

Report of the general unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the operation of social rehabilitation services in year 2017 (2018b). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

30.

Report of the general unit of the Prison Department under the ministry of justice of the Republic of Lithuania on working with persons supervised by district probation services in year 2017 (2018c). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

31.

Report of the task planning unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the operation of social rehabilitation services in year 2018 (2019a). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

32.

Report of the task planning unit of the Prison Department under the ministry of justice of the Republic of Lithuania on the number, structure and change of prison population in year 2018 (2019b). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

33.

Report of the task planning unit of the Prison Department under the ministry of justice of the Republic of Lithuania on working with persons supervised by district probation services in year 2018 (2019c). Available at: http://www.kaldep.lt/lt/kalejimu-departamentas/administracine_informacija/ataskaitos/metu.html (accessed 4 April 2020).

34.

Sakalauskas

(2013) Parole application after the enactment of Probation act: Theory and practice. Legal Issues 4(82): 5–39.

35.

Serin

Gobeil

(2014) Analysis of the Use of the Structured Decisionmaking Framework in Three States. U.S. Department of Justice, National Institute of Corrections.

36.

van Eijk

(2020) Inclusion and exclusion through risk-based justice: Analysing combinations of risk assessment from pretrial detention to release. The British Journal of Criminology 4(60): 1080–1097.

37.

Venckeviciene

(2013) A Criminological Analysis of Liability for Crimes Relating Disposal of Narcotic and Psychotropic Substances: Social Demographic Portrait of Convicts. Available at: http://teise.org/wp-content/uploads/2016/07/Studija-nuteistuju-soc-dem-portretas.pdf (accessed 24 April 2020).

38.

Vosyliute

(2017) Parole as a premise of integration: Issues of court decisions. In Sakalauskas

, Sentencing and the Execution of Sentences: An International Comparison. Trends and Perspectives in Lithuania. Vilnius: Zara: 205–232.

39.

Wardrop

Serin

(2019) Evaluating the structured parole decision making framework in three U.S. states. The American Journal of Forensic Psychology 37(2): 5–30.

40.

Warner

(1923) Factors determining parole from Massachusetts reformatory: The report of the director of committee on criminal records and statistics. Journal of the American Institute of Criminal Law and Criminology 14(2): 172–207.

41.

West-Smith

Pogrebin

Poole

(2000) Denial of parole: An inmate perspective. Federal Probation 64(2): 3–10.