Skip to main content
NCBI home page
Psychiatry, Psychology, and Law logoLink to Psychiatry, Psychology, and Law
. 2020 Dec 7;28(5):623–644. doi: 10.1080/13218719.2020.1821826

Actor–observer asymmetry in perceptions of parole board release decisions

Logan A Yelderman a,✉, Timothy I Lawrence b, Courtney E Lyons c, Alicia DeVault d
PMCID: PMC9103362  PMID: 35571597

Abstract

In the current study, the actor–observer effect is tested with both mock parole board members and the public evaluating the responsibility of parole board members for a decision resulting in a parolee reoffending and committing a murder. Participants (two samples with a combined N = 1317) were randomly assigned to act as a mock parole board member and make a decision (which ended in the parolee reoffending) or as a member of the public who read a story about the same parole decision and outcome. Findings suggest that the traditional actor–observer asymmetry emerged across blame and responsibility concepts, emotion and moral judgments. Overall, the public held harsher judgments than the mock parole board members. Implications regarding self-enhancement, methodology and attribution theory are discussed.

Key words: actor–observer, attributions, blame, decision-making, emotion, parole, responsibility

Parole has been a part of corrections reform since the early twentieth century, and it shares many goals with a more rehabilitative criminal justice approach seen gaining momentum recently (Cullen, 2017; Petersilia, 2000). Perceptions that prison sentences were too harsh and did not allow any room for rehabilitation spurred support for a more lenient approach to parole, and models of releasing less violent inmates that have shown good behavior were adopted (Clear & Cole, 1997). In general, there are two types of parole decisions; discretionary and mandatory parole (Maruschak & Bonczar, 2013). Discretionary parole involves a decision process through which parole board members review an inmate’s files and decide whether to grant or deny supervised release (parole). There is no required action given to the board, but the action is left up to the discretion of the board. In contrast, mandatory parole is a decision process by which inmates are released under specific circumstances after they have served a certain period of time in prison and do not have major violations or other factors barring their release (Hughes et al., 2001). Under mandatory release guidelines, the parole board is often obligated to release inmates to supervised release, usually based on determinate sentencing and good behavior, unless some major factor prevents such action (Abadinsky, 2012). Recently, discretionary parole has been used at a higher rate than mandatory parole, creating a trend over the past several years (Kaeble, 2018; Maruschak & Bonczar, 2013).

In the United States, there are currently about 875,000 inmates on parole, with over 450,000 inmates being released on parole in 2016 (Kaeble, 2018). With an increase in discretionary parole decisions and a large parole population, the parole board likely faces scrutiny for the outcomes of its decisions and is held accountable by the public for ensuring safety and economy (e.g. Abadinsky, 2012; Mackenzie, 2001; National Parole Resource Center, 2012; Paparozzi & Guy, 2009). Parole board members’ perspectives and the public’s perspective might differ as they relate to how parole decisions are perceived and the accountability of the boards’ decisions. This divergence is important because legislation applicable to parole board decisions is intended to incorporate both perspectives, and any dialogue about parole should involve an attempt at mutual understanding. However, it is possible that the divergence in perspectives is due to experience and observational standpoint. The purpose of the current study is to apply an attribution framework to better understand the difference between perspectives of actors (parole board members) and observers (the public) by using a mock parole decision-making paradigm and comparing both parole board and public perspectives on the decision process and outcome with a focus on blame and responsibility for negative outcomes.

Attributions

The extent to which two events are linked is dependent on numerous factors including, but not limited to, nominal association and frequency of co-occurrence (Heider, 1944; Kelley & Michela, 1980). Differences in causal explanations linking two events based on perspective is called the actor–observer effect (Jones & Nisbett, 1972). Specifically, the actors (those displaying the behavior) are more likely to attribute their own behaviors to situational or environmental factors, while the observers (those watching the actor’s behavior) are more likely to attribute the actor’s behavior to internal characteristics or dispositions (Jones & Nisbett, 1972).

There is support for the actor–observer effect across different types of decisions, explanations and roles (Goldschmied & Hochuli, 2014; Hansen et al., 2001; Malle & Knobe, 1997; Nisbett et al., 1973; Wallace & Hinsz, 2009). One explanation for the difference in types of attributions is that actors and observers differ in their knowledge of event details and in their assessments of the actors’ control over the outcome. Specifically, actors would have more access to both personal and situational information and perceive themselves as having less control than observers (Harvey et al., 1975; Jones & Nisbett, 1972; Malle & Pearce, 2001; Nadelhoffer & Feltz, 2008). Applied to parole decision-making, a parole board would have more access to risk assessment information and criminal history information, which are both incorporated into release decisions, whereas the public would not have this information. It is also possible that the parole board would admit less control over what the inmates do after release than would the public. Similarly, parole boards might consider themselves to be restrained by a process, whereas the public might believe the parole board has more free will and discretion than it actually has.

Though attributional differences between actors and observers are well supported and extend across populations and methods (e.g. Johnston & Lee, 2005; Krueger et al., 1996), these differences might be dependent upon other moderating factors, such as perceived similarities (Arkin et al., 1978; Green & McClearn, 2010; McKillip & Posavac, 1975), attributional complexity (Wilson, Levine, Cruz, & Rao, 1997), emotion (Finney & Helm, 1982; Hennessy & Jakubowski, 2007) or other individual differences (Robins et al., 1996). Actors tend to make favorable attributions about their own morality depending on the outcome, suggesting a self-serving or protective evaluation (Mitchell, 1985). However, the actor–observer effect appears to only be consistent with original theory when the outcomes are negative (Eisen, 1979; Lowe & Ansen, 1976) and might even be in the opposite direction for positive outcomes (Malle, 2006).

While there is support for the actor–observer effect, there are also several studies that failed to find such differences in perspective (Lewis, 1995; Malle et al., 2007; Miller & Norman, 1975; Newman, 1978; O’Connor, 1988; Vázquez et al., 2017). However, in Malle’s (2006) meta-analysis, traditional actor–observer effects are more likely when using hypothetical events, negative outcomes, between-subjects designs and a Western sample. Because the current study uses a between-subjects design and uses a U.S. sample with attribution ratings following a hypothetical negative event, it is quite probable that traditional actor–observer asymmetry will emerge. It is possible that this asymmetry is due to the parole board’s informational advantage over the public and motivation to engage in self-enhancing judgments; however, other explanations might also exist. This asymmetry could lead to the parole board considering negative outcomes to be flukes or unrepresentative of their typical decisions (see Campbell & Sedikides, 1999; Kelley & Michela, 1980). Thus, a parole board will likely hold themselves less responsible for parolee recidivism than the public in order to preserve self-image and self-concept.

In parole decisions, parole boards might be perceived as a causal actor, but they might also be viewed as a responsible or blameworthy party associated with the offender. Although responsibility and blameworthiness have roots in causal attributions (Gailey & Falk, 2008; Shaver, 1985; Shultz & Schleifer, 1983), they can also be applied to non-causal actors (Brank et al., 2011; Shultz et al., 1987). In both crime and accident attributions, responsibility increases with the severity of the outcome (Gebotys & Dasgupta, 1987; Walster, 1966). When individuals are seen as less responsible, the desire to assign punishment tends to wane (e.g. Cullen et al., 1985). Attributing responsibility is intimately linked with one’s role; when a person’s role is associated with their action or decision, the outcome is more strongly tied to judgments of responsibility (Hart, 1968). Roles often obligate responsibility and accountability for one’s actions and decisions, implying that an actor will be evaluated and asked to justify their decision (Frink & Klimoski, 2004; Holdorf & Greenwald, 2018; Lerner & Tetlock, 1999). When responsibility involves a decision in which the decision-maker’s role and obligation are to provide safety, a moral concept, decisions are also judged on moral grounds (Fincham & Jaspars, 1980). In models of responsibility and blame, aspects of morality and moral responsibility, including moral emotions, are often included (Hart, 1968; Shaver, 1985; Tangney et al., 2007; Weiner, 1995). Because parole boards exercise discretion, provide supervision plans for parolees, and make decisions that weigh both public safety and economy, their decision outcomes are tied to their roles. If parole boards release inmates who then commit crimes shortly after release, the parole board might be held responsible, blamed for the crime, and held accountable for the negative outcome because of its role and status.

Similar to judgments of responsibility and risk, judgments of blame include evaluations of intentionality, foreseeability, moral wrongness, causality, knowledge of possible outcomes, severity of consequences and other aspects of mental states (Cushman, 2008; Feinberg, 1968; Guglielmo, 2015; Lagnado & Channon, 2008; Malle et al., 2014). Moreover, blame is intimately tied to emotion, especially in cases of self-blame; thus, emotions are likely to play a role in blame judgments (Sheikh & McNamara, 2014). Regarding the actor–observer effect, increases in responsibility attributions are associated with increases in blame (Fincham & Jaspars, 1979). Therefore, if observers are more likely to make internal attributions, they are also likely to assign more blame to the actor. This means the public will likely view parole boards as more blameworthy and responsible as a result of increased attributional evaluations and emotions. Overall, parole boards and the public likely differ in the extent to which they consider the parole board blameworthy and responsible for parolees’ crimes committed shortly after release.

The assessment of risk is also key in making decisions with potential negative outcomes. Each parole release decision has an element of risk since the decision is based on an inmate’s potential level of risk posed to the community in terms of reoffending. Actors and observers differ in acceptable levels of risk allowed in a decision. Specifically, actors tend to make riskier decisions than observers, possibly because of the differences in available information between the two perspectives (Fernandez-Duque & Wifall, 2007; Horswill & McKenna, 1999; Nuijten et al., 2018). Therefore, parole boards are more likely to make a risky decision about releasing a parolee than the public. It is also possible that the public is harsher toward the parole board because they view each release as posing an increasing, potentially unnecessary, risk to the community. This is quite clearly evidenced by the increased incorporation of victim statements.

Victim impact statements have been infused into sentencing and parole decisions over the past several decades and involve the aspect of victims’ rights and restorative justice (Rhine et al., 2017). The idea behind victim impact statements is that it allows victims to provide input to the parole authorities about how they have been personally affected by the inmate’s crimes with an emphasis on long-term effects (Council of State Government/American Probation and Parole Association, n.d.). Most states have legislation providing guidance on how to facilitate victim statements, but the evidence on how these statements impact parole release decisions is mixed. Some researchers demonstrate clear effects from victims’ statements (see Caplan, 2007, for review; Morgan & Smith, 2005) while others suggest no effects or potentially even misguided release decisions (Caplan, 2010; Roberts, 2009). Though victim impact statements’ influence in decisions are likely jurisdiction-specific and potentially case-specific, the ability of victims to provide input plays an important role in the parole landscape and related legislation (see Vîlcică, 2016, 2018). Victims are sometimes viewed as acting as representatives of the community harmed by a particular criminal act (see Rosebury, 2011). Additionally, parole decisions influence corrections policies (Rhine et al., 2018), and with this logic, victim impact statements can indirectly shape corrections policies through their impact on parole decisions. Moreover, victims and victims’ groups might not only make statements to the parole board but also assist in reentry efforts for the offenders (Herman & Wasserman, 2001). Therefore, the representativeness of victims as spokespeople for the community increases the importance of understanding community sentiment toward failed parole decisions.

One case illuminating immediate recourse after parole boards released an inmate who then committed a violent crime occurred in Pennsylvania when a recently released inmate shot four police officers, killing one of them. This prompted the Pennsylvania governor to place a moratorium on the parole board as a result of increased pressure from both the public media and political administration (Vîlcică, 2016). Thus, it is safe to acknowledge that public attitudes can translate into swift policy changes. This is particularly true with politicized and sensational stories about parolee crimes. Therefore, inmate risk must be considered a priority because the public views risk as a priority. Parole boards are tasked with evaluating inmates on risk levels, and when they release an inmate who then commits a violent crime, it is possible the public feels as if the parole board should have seen it coming. In these cases, the public is likely harsher on the parole board than the parole board members themselves because the public likely perceives the outcome as an immediate threat as well as avoidable and foreseeable. The current research proposes that this difference in attributions and emotions could be explained by the actor–observer effect.

Cognitive dissonance

Though it is possible that the parole board might have more information as to why they made a specific decision and use that as a way to explain their decisions, thus resulting in lesser internal attributions, it is also possible that parole board members are motivated to justify decisions that have negative outcomes and cause inner turmoil (McGrath, 2017). Festinger’s (1957) cognitive dissonance theory suggests that when a person’s behavior and attitudes contradict, they tend to find ways to alleviate the negative feeling that results. One possible way this might occur is through self-affirmation processes by which people think positively of themselves or behave in ways that restore their sense of self-worth (Steele & Liu, 1983). In this case, parole boards might think positively of themselves or their actions to reduce dissonance, which would result in decreased feelings of moral wrongfulness and blame. It is also possible that the simple denial of responsibility might help the parole board members alleviate dissonance (Gosling et al., 2006), and it is possible that both denial of responsibility and self-affirmation occur simultaneously after parole board members release an inmate who then commits murder. This would allow the parole board’s sense of decisional competence and authority to remain intact. Therefore, it could be dissonance that drives down parole boards’ attributions and emotions related to responsibility and blame for releasing inmates who then commit a violent offense, and this reduction causes the difference between the parole board and public in such judgments.

Hypotheses

Based on the preceding literature, the current study examines the following two hypotheses.

  1. Participants in the public condition will report greater attributions of causality, responsibility and blame to the parole board than the mock parole board condition and lesser attributions of moral rightness, excuses and justifications.

  2. Participants in the public condition will report greater negative emotions and anticipated emotions than participants in the mock parole board condition.

Overview

The current study examined actor and observer differences in attributions and emotions toward a parole board after releasing an inmate who committed a homicide within two years of release. The analysis included two samples, a student sample and an Amazon Mechanical Turk (MTurk) sample. An analysis of variance (ANOVA) was used to assess mean differences across the two groups for each variable. All participants answered questions about accountability, responsibility, blame and emotions related to parole release decisions from either the perspective of a parole board member or a member of the public.

Method

Samples

The current study employed a two-sample design with the goal of establishing results that replicate across samples. Both samples completed identical procedures and measures, and data were collected between 2015 and 2019. An a priori power analysis suggested that a sample of 352 was needed to detect a small to moderate effect with the following criteria: two groups, F = 0.15, α = .05, power (1−β) = .80. Both student and MTurk samples were sufficient in power. The MTurk sample was much larger than the minimum sample size because of available funding for such data collection. Similarly, the student sample was larger than the minimum sample size because two student samples were collected with the intention of meeting sufficiency, but they did not independently meet the minimum sample size requirement so they were combined (see Table 1 for demographics).

Table 1.

Demographics for both student and MTurk samples.

Demographic variable Student sample
%		 MTurk sample
%
Male 37.2 49.7
Female 62.8 50.3
White/Caucasian 43.1 77.9
Black/African-American 36.3 8.2
Asian-Pacific Islander 5.5 7.2
Hispanic/Latino 11.9 4.6
Other race 3.1 1.9
College educated — 71.7
High school educated — 27.7
Less than high school educated — 6
Protestant/Christian 38.7 24.2
Catholic 20.6 18.1
Atheist 5.3 17.8
Agnostic 6.0 17.4
Believe in God but have no particular affiliation 21.4 12.8
Other religion 8.0 9.5
Democrat 48.8 44.6
Republican 31.7 33.0
Independent 16.9 17.8
Other political affiliation 2.7 2.8
Liberal 30.9 49.6
Moderate 51.0 28.9
Conservative 18.1 21.4
Age (M/Mdn/SD), years 22/20/5.74 35/32/11.6
Open in a new tab

Student sample

A total of 515 students from two universities were included in the study (i.e. one western university, n = 323, and one southern historically black college/university, n = 192). Students were recruited through an online recruitment system (SONA) and were awarded one research credit for participation. Student data from both universities were combined into one general ‘student’ sample. School and race differences were assessed and modeled as moderators to determine whether any major variance existed in group differences based on school or race. Race did not moderate any group difference. School moderated group differences for two outcome variables (intentionality and anticipated regret), but the effects were consistent with general group differences (public means were larger than parole means) and only varied in magnitude (ΔMdifference = −.29 and .66, respectively) so the two student samples were collapsed together to improve power. Overall, the student sample was 62.8% female1; 43.1% White/Caucasian (36.3% Black/African-American, 11.9% Hispanic/Latino, 5.5% Asian-Pacific Islander, and 3.1% other); 48.8% Democrat (31.7% Independent, 16.9% Republican, and 2.7% other); 51% moderate (30.9% liberal and 18.1% conservative); 38.7% Protestant/Christian (21.4% believe in God but have no particular affiliation, 20.6% Catholic, 6.0% agnostic, 5.3% atheist, 8% other); and were 22 years old on average (Mdn = 20, SD = 5.74).2

MTurk sample

A total of 802 MTurk participants were recruited on Amazon’s Mechanical Turk and were paid $2.65 for participation.3 MTurk participants were required to be at least 18 years old and U.S. citizens. Overall, MTurk participants were 50.3% female; 77.9% White/Caucasian (8.2% Black/African-American, 7.2% Asian-Pacific Islander, 4.6% Hispanic/Latino, and 1.9% other); 71.7% at least college educated (27.7 high school educated and .6% less than high school educated); 24.2% Protestant/Christian (18.1% Catholic, 17.8% atheist, 17.4% agnostic, 12.8% believe in God but have no particular affiliation, and 9.5% other); 44.6% Democrat (33% Republican, 17.8% Independent, and 2.8% other); 49.6% liberal (28.9% moderate and 21.4% conservative); and 35 years old on average (Mdn = 32, SD = 11.61).

Procedure

All participants received identical procedures. After agreeing to participate, participants were randomly assigned to either the mock parole condition or the public condition. In the mock parole condition, participants acted as a parole board member and made a decision to release an inmate. These decisions were made using the seventh member paradigm in which the participant had the final vote on parole release to break a tie among the rest of the parole board members (see Appendix). This paradigm is based on the ‘9th justice paradigm’ developed by Finkel and Duff (1991) in which participants acted as the deciding justice to indicate their support for various legal actions. All inmates were low risk to increase the chance that the participants would choose to release the inmate and feel actively agentic in the release decision. Parole board members viewed risk assessment details, which were developed and adapted from observed parole hearings and various parole risk assessment instruments (see Appendix). Participants who decided to release the inmate were told that the inmate was released as a result of their vote. They then read a news story about the inmate they just released, finding out that the inmate murdered a young girl after being released.4 Mock parole board members were then asked questions about their own role in the outcome of the crime. Questions included rating their responsibility, blame and emotions regarding the crime and their role. This constituted the actor perspective. Lastly, they completed demographics. Participants who chose not to release the inmate were excluded from analyses.

In the public condition, participants read the same news story as the one that the parole condition read and completed questions about their perceptions of the parole board members’ role in the outcome of the crime. These questions were identical to the parole board members’ questions but were asked from the observer perspective. They also completed demographics at the end. For both conditions, the perception measures were a subset from a larger questionnaire; however, the questions in this study occurred first and were not subject to order effects.

Measures

Condition

The primary grouping variable was dichotomous. Participants were in either the parole board condition or the public condition. In the student sample, 48.6% of participants were in the parole condition, and in the MTurk sample, 40.6% were in the parole condition.

Blame

Blame was measured by a single item asking the extent to which the parole board was to blame for the death of the girl on a 5-point Likert scale from 1 (not at all to blame) to 5 (completely to blame). This item was developed for the purposes of this study.

Responsibility

Responsibility was measured by a single item asking the extent to which the parole board members are responsible for the girl’s death on a 5-point Likert scale from 1 (not at all responsible) to 5 (completely responsible; adapted from Gailey & Falk, 2008).

Accountability

Accountability was measured by a single item asking the extent to which the parole board is accountable for the girl’s death on a Likert scale from 1 (not at all accountable) to 5 (completely accountable). This item was developed for the purposes of this study.

Answerability

Answerability was measured by a single item asking the extent to which the parole board should have to answer for the girl’s death on a Likert scale from 1 (should not have to answer at all) to 5 (should definitely have to answer). This item was developed for the purposes of this study.

Preventability

Preventability was measured by a single item asking the extent to which the parole board was able to prevent the girl’s death on a Likert scale from 1 (could not have prevented at all) to 5 (could have completely prevented; adapted from Gailey & Falk, 2008).

Control

Control was measured by a single item asking the extent to which the parole board was able to control the parolee’s actions on a Likert scale from 1 (not able to control at all) to 5 (completely able to control). This item was developed for the purposes of this study.

Causality

Causality was measured using a five-item index (Gailey & Falk, 2008). These items included questions about the parole board being at fault for the girl’s death and being able to avoid the girl’s death. These items were measured on a 5-point Likert scale with response options specific to the particular wording of the question. For example, response options for the fault question were 1 (not at all at fault) to 5 (completely at fault). The scale had adequate reliability for both student and MTurk samples, α = .77 and α = .84, respectively.

Knowledge

Knowledge was measured using a seven-item index (Gailey & Falk, 2008). These items included questions about the parole board’s ability to foresee the outcome and the extent to which they were aware of the potential consequences of releasing the inmate. Items were measured on a 5-point Likert scale with response options specific to the particular wording of the question. For example, response options for the foreseeability question were 1 (not at all able to foresee) to 5 (completely able to foresee). The scale had adequate reliability for both student and MTurk samples, α = .70 and α = .74, respectively.

Coercion

Coercion was measured using a four-item index (Gailey & Falk, 2008). These items included questions about the extent to which the parole board’s decision was influenced by someone or something else. These items were measured on a 5-point Likert-type scale with response options specific to the particular wording of the question. For example, response options for the direct coercion question were 1 (not at all coerced) to 5 (completely coerced). The scale had poor reliability for both student and MTurk samples, α = .58 and α = .69, respectively.

Intentionality

Intentionality was measured using a three-item index (Gailey & Falk, 2008). These items included questions about the extent to which the parole board intended for the girl to be harmed. These items were measured on a 5-point Likert scale with response options specific to the particular wording of the question. For example, response options for the direct intentionality question were 1 (not at all) to 5 (definitely). The scale had adequate reliability for both student and MTurk samples, α = .82 and α = .80, respectively. It is worth noting that both samples reported very low intentionality scores overall.

Morality

Morality was measured using a three-item index about the extent to which the parole board’s decision was morally right (Gailey & Falk, 2008). These items were measured on a 5-point Likert-type scale with response options specific to the particular wording of the question. For example, response options for the moral question were 1 (not acting morally at all) to 5 (completely acting morally). The scale had somewhat adequate reliability for both student and MTurk samples, α = .69 and α = .82, respectively.

Excuse

Excuse was measured by a single item asking the extent to which the parole board’s decision should be excused on a 5-point Likert scale from 1 (not excused at all) to 5 (completely excused). This item was developed for the purposes of this study.

Justification

Justification was measured by a single item asking the extent to which the parole board’s decision to release the inmate was justified on a 5-point Likert scale from 1 (not justified at all) to 5 (completely justified). This item was developed for the purposes of this study.

Anger, guilt and regret

Anger, guilt and regret were measured by single items asking the extent to which anger or guilt was felt about the crime against the victim either as the parole board or at the parole board on a 5-point Likert-type scale from 1 (not angry/guilty/regretful at all) to 5 (extremely angry/guilty/regretful). These items were developed for the purposes of this study.

Anticipated anger, guilt and regret

Anticipated anger, guilt and regret were measured by single items asking the extent to which participants would feel a certain emotion if they were to make a decision in the future that ended with releasing an inmate who then committed a violent crime shortly after release. For the public, this question captured the level of emotion that the public thought the parole board members should expect to feel if they were to make such a decision, reflecting affective forecasting for the parole board. The comparison made here provides additional insight into the emotion expectations of divergent perceptions on how parole boards should feel if their decisions have poor outcomes. These items were measured on a 5-point Likert-type scale with the same response options as anger, guilt and regret items. These items were developed for the purposes of this study.

Demographics

Demographics were collected by asking participants to self-report their age, race, religion, gender, highest completed level of education, political affiliation and political orientation.

Results

To assess mock parole board members’ and public attributions and emotions toward parole boards after releasing parolees who reoffend, a one-way ANOVA was used to test mean differences across the two groups for each of the dependent variables. Initially we assessed and reported each dependent variable separately; however, a reviewer suggested using factor analysis of the dependent variables (DVs) to reduce the total number of analyses, decrease family-wise error and simplify the DVs. Therefore, all 18 DVs were factor analyzed to reduce the number of tests in the final analyses. Three factor analyses were conducted. First, the student sample and the MTurk samples were assessed independently, and then they were combined and assessed. Initial factor analysis across all three tests suggested correlated factors (i.e. at least one correlation above r = .3) so direct oblimin rotations and Kaiser normalizations were used. In addition, one of the goals of this analysis was to reduce the total number of DVs for simplification purposes and decrease Type 1 error, so principal component analysis was used to create the smallest emergent factor structure. In all three factor analyses, factors with an eigenvalue above 1.0 were retained, and factor loadings above .5 were considered to load onto that particular factor. In cases in which items cross-loaded, and the loadings were both under .5, the highest two loadings were assessed and interpreted within the theoretical constructs from prior literature. All three analyses resulted in a four-factor structure with the same four factors emerging.

Factor 1, identified as ‘blame and responsibility’, included loadings above .5 for blame, responsibility, preventability, accountability and answerability (weak loading for student only sample) across all three analyses. Causality and knowledge loaded onto Factor 1 in two of the three analyses (not in the student only sample). Also, guilt loaded onto Factor 1 in the student only sample, and anger loaded onto Factor 1 in the student only sample and cross-loaded onto Factors 1 and 4 in the combined sample. Factor 2, identified as ‘motivation’, included controllability (weak loading in MTurk only sample), coercion and intentionality across all three samples. Causality and knowledge loaded onto Factor 2 in the student sample, though knowledge did not load above .5. Factor 3, identified as ‘moral exemption’, included morality, excuse and justification across all three analyses. Factor 4, identified as ‘emotionality’, included regret, anticipated guilt and anticipated regret in all three analyses. Guilt loaded onto Factor 4 in two of the three analyses (not in the student only sample), and anger loaded onto Factor 4 in the MTurk only sample (see Table 2 for all factor loadings).

Table 2.

Factor analysis for both samples independently and combined.

Factor Student sample
 MTurk sample
 Combined samples
1 2 3 4 1 2 3 4 1 2 3 4
Blame .765 −.062 −.100 .010 .871 .025 .063 .135 .851 −.002 −.003 .034
Responsibility .771 −.064 −.013 −.041 .887 .007 .017 .051 .893 −.052 −.014 −.044
Accountability .843 −.109 −.037 −.132 .801 .130 −.048 .065 .854 .065 −.020 −.025
Answerability .402 − .394 −.264 .064 .540 .094 −.385 .022 .502 .180 −.335 .053
Preventability .556 −.236 −.004 .112 .661 −.033 −.139 .147 .659 .008 −.115 .113
Causality .254 − .614 −.107 −.046 .857 −.008 −.083 .060 .705 .111 −.089 .065
Knowledge .004 − .491 −.009 .355 .470 −.033 − .397 −.196 .321 .162 − .276 .012
Control .191 − .614 −.107 −.046 .332 .440 −.238 −.092 .289 .524 −.154 −.037
Intention .117 − .705 .103 −.271 .189 .781 .230 −.115 .157 .794 .232 −.119
Coercion −.177 − .764 −.163 .062 −.189 .874 −.122 .092 −.172 .855 −.085 .103
Morality .009 .193 .759 −.116 −.005 −.094 .807 −.149 .000 −.114 .794 −.131
Excuse −.039 −.012 .841 .079 −.109 .024 .863 −.100 −.097 .058 .840 .062
Justification −.047 −.156 .839 −.008 .005 .024 .863 −.100 −.012 .092 .873 −.021
Anger .719 .186 −.015 .215 .479 .036 −.005 .487 .562 −.061 .037 .402
Guilt .534 .148 −.046 .480 .415 .005 −.019 .604 .435 −.053 .001 .578
Regret .272 .200 .127 .687 .120 −.047 .080 .845 .127 −.085 .141 .837
Ant. Guilt .150 −.013 −.106 .779 .113 .049 −.203 .715 .108 .039 −.145 .753
Ant. Regret −.149 −.154 −.103 .870 −.109 −.020 −.155 .867 −.166 .065 −.117 .892
Open in a new tab

Note: All factor structures are based on principal component analysis and pattern matrices using direct oblimin rotations with Kaiser normalization. Largest loadings are bolded for each item. For items that cross load, second highest loadings are bolded and italicized. Factor 1 = Blame and Responsibility; Factor 2 = Motivation; Factor 3 = Moral Exemption; Factor 4 = Emotionality. Ant. = anticipated.

When assessing the factor structures across the three analyses, theory guided decisions regarding how to categorize factors with inconsistent loadings. First, causality and knowledge were included in Factor 1 because they are theoretical components of blame and responsibility models. For anger and guilt, they were combined with the other emotions in Factor 4. The final factor structure including four factors, which were blame and responsibility (i.e. blame, responsibility, causality preventability, accountability, answerability and knowledge), motivation (i.e. control, coercion and intention), moral exemption (excuse, justification and morality), and emotionality (i.e. anger, guilt, regret, anticipated guilt and anticipated regret). These four factors are used as the DVs for all final analyses in both samples.

MTurk sample results

In the MTurk sample, all of the outcome variables were normally distributed except motivation, which was slightly positively skewed (1.26). Motivation was assessed with and without a transformation. The transformation did not change the outcome of the analysis or nature of the mean difference, and the untransformed variable is used in reported analyses. Reliability coefficients for the four outcomes were α = .93 (blame and responsibility), α = .57 (motivation), α = .88 (moral exemption) and α = .89 (emotionality). Because the reliability for motivation is so low, results should be interpreted with hardly any confidence. No outliers were removed from the dataset. Missing data for any one variable entered into the model were less than 1.5% (largest n = 5) of all data available for each variable; thus, missing data were not considered problematic, listwise deletion was used, and missing data patterns were not assessed.

Mock parole board (n = 326) and public (n = 476) conditions were compared across responses on blame and responsibility, motivation, moral exemption and emotionality (see Table 3 for F values and effect sizes). All significance values are based on the Welch robust test of equality of means (Tomarken & Serlin, 1986). Across all four outcomes, the public made harsher judgments about the parole board than did the mock parole board members. The public reported more blame and responsibility, more culpable motivation, a stronger perceived emotional response and less moral exemption (ps < .001). Demographics, prior arrest and prior work in law enforcement were assessed as moderators; however, no significant moderation emerged.

Table 3.

Means, standard deviations and ANOVA results.

  Student sample
 MTurk sample
Public
M (SD)		 Parole
M (SD)		 F p η2 Public
M (SD)		 Parole
M (SD)		 F p η2
Blame/responsibility 2.96 (0.74) 2.71 (0.76) 13.75 <.001 .03 3.04 (0.94) 2.57 (0.92) 48.41 <.001 .06
Motivation 2.15 (0.74) 1.64 (0.58) 75.10 <.001 .13 1.84 (0.64) 1.59 (0.63) 301.95 <.001 .27
Moral exemption 2.85 (0.83) 3.50 (0.82) 79.16 <.001 .13 2.76 (0.98) 3.92 (0.83) 29.15 <.001 .03
Emotionality 3.75 (0.83) 3.50 (1.06) 9.06 .003 .02 4.02 (0.94) 3.62 (1.17) 29.01 <.001 .03
Open in a new tab

Note: All comparisons with homogeneity of variance violations used the Welch corrected robust test of equality of means asymptotic F distribution statistic and adjusted p value. ANOVA = analysis of variance.

Student sample results

In the student sample, all of the outcome variables were normally distributed. Reliability coefficients for the four outcomes were α = .84 (blame and responsibility), α = .65 (motivation), α = .77 (moral exemption) and α = .85 (emotionality). Because the reliability for motivation is so low, results should be interpreted with little confidence. No outliers were removed from the dataset. Missing data for any one variable entered into the model were less than 1.5% (largest n = 6) of all data available for each variable; thus, missing data were not considered problematic, listwise deletion was used, and missing data patterns were not assessed. Three participants did not have condition level data and were removed from the original N = 515.

Parole board (n = 249) and public conditions (n = 263) were compared across responses on blame and responsibility, motivation, moral exemption and emotionality (see Table 3 for F values and effect sizes). All significance values are based on the Welch robust test of equality of means (Tomarken & Serlin, 1986). Across all four outcomes, the public made harsher judgments about the parole board than did the mock parole board members. The public reported more blame and responsibility, more culpable motivation, a stronger perceived emotional response and less moral exemption (ps < .01). Demographics, prior arrest and prior work in law enforcement were assessed as moderators; however, no significant moderation emerged. Overall, both the MTurk and student samples showed consistent actor–observer asymmetry across all four dependent variables (see Figure 1).

Figure 1.

Figure 1.

Open in a new tab

Mean differences between conditions across both samples by outcome.

Discussion

When parole boards make decisions, there is an inherent risk regarding whether or not the person they release is ready to reenter society peacefully or whether they will reoffend. In some cases, parolees commit crimes shortly after release, and parole boards might receive negative attention because the inmate’s release was ultimately up to the board’s discretion. The results of this study support the notion that the public might hold parole boards more accountable for the outcomes of their decisions than parole boards do, at least in part. Though the public attributed the negative outcome of the parolee’s crimes to the parole board more than the mock parole board members did, many of the means on these outcomes were below the midpoint, which suggests that partial blame and accountability might be attributed to the parole board but the public does not completely blame the parole board. Therefore, responsibility and blame are likely attributed to other actors/factors as well.

Overall the hypotheses were supported for each variable across both samples; however, each hypothesis is addressed in order. Hypothesis 1 was supported across both samples. Both the student and MTurk samples were consistently asymmetrical in their actor–observer perceptions across attributions of blame and responsibility, though the effect was twice as large in the MTurk sample. The effect sizes were not large, but indicate that parole boards might be held more accountable by the public than themselves. For example, in the Martinez v. California (1980) case, the victim’s survivors held the parole board directly responsible, but the parole board and United States Supreme Court found otherwise. Similarly, the mock parole board members saw their behavior as more morally right and that they should be exempt from any responsibility, compared to the public. This was consistent across both samples, but the effect was much larger in the student sample. Lastly, both samples did not attribute nefarious motives to the mock parole board members as a result of the parolee’s murder, but this was even truer for the mock parole board members. However, due to such low reliability ratings for the motivation measurement, these findings are uninterpretable.

Hypothesis 2 was also supported. Both samples consistently showed a traditional actor–observer asymmetry in negative emotions attributed to the parole board. The public responded to the news story with more emotion than the mock parole board members did. Negative emotions are consistent with appraisals of causality and responsibility for negative outcomes and also blame attributions (Weiner, 1995). These discrepancies in effect sizes, and also means, between student and MTurk samples might be due to the demographics of the two groups, as previous research has shown that student samples do not always model actual parole boards (Lindsey & Miller, 2011), while the MTurk sample might have approached the backgrounds of actual parole boards a little better (though still quite different). Student samples tend to be more lenient in mock criminal justice decision-making settings (Lindsey & Miller, 2011), which could explain effect size differences. Further exploration of this difference would be important for understanding the possible implications of this research.

One explanation of these differences might be attributed to differences in perceived risk. A parolee’s crime threatens a community because it presents a risk of endangerment. However, it might not pose the same threat to the parole board, or, even if the parole board members and the parolee live in the same community, its members were part of the decision and might perceive the release as involving an acceptable level of risk. In contrast, the public might think in absolutes, such that the public perceives that a release should be an absolute absence of risk. Previous research has shown that perceiving risk as controllable is associated with more internal attributions, while perceiving risk as uncontrollable is more associated with external attributions of causality (Rickard, 2014). This could be translated to the current findings, where the public perceiving risk as a factor that must be controlled are more likely to assign causal blame to the parole board, whereas the parole board is more likely to understand that some risk cannot be controlled and assign less blame to themselves. Further research would need to address these possible explanations in the current context.

Another explanation could be that the parole board attributes less blame, responsibility and emotion to themselves as a way to buffer the threat to their decision-making ability by reducing cognitive dissonance. The parole board members want to feel confident in their own abilities, and a ‘failure’ such as releasing a parolee only for them to reoffend could create a sense of cognitive dissonance, or stress, as a result of conflicting information about oneself (Festinger, 1957). Dulled emotions and reduced internal attributions might be a coping strategy or a way to buffer negative feelings regarding their professional, and possibly even personal, identities, therefore reducing cognitive dissonance. This coping process could be necessary to avoid the impact of previous failed decisions on subsequent parole decisions. This might also be a way to keep one’s professional image intact and reduce self-doubt.

It is also possible that the mock parole board members’ lower emotion scores can be explained by dissatisfaction with the process. They might not report as much of an emotional response or expect an emotional reaction because the process to them is more nuanced or potentially less revealing of their individual influence. Moreover, the public might have had the entire parole board’s contribution in mind when attributing blame and reporting emotion, whereas the mock parole board members considered their individual contributions. However, additional evidence is needed to explain why parole boards might perceive themselves less responsible and report less emotion for the parolees’ actions.

If the public attributes parolees’ recidivism to the board, then corrective actions might be pursued by the public through legislative changes, reducing parole board discretion. In this case, the public might favor mandatory parole or longer sentence length preceding parole eligibility. However, mandatory parole would remove the decision-making process in which a parole board considers relevant factors to assessing the risk of releasing an inmate to parole. The current research is necessary to define and explain differences in views between the public and parole board. A discussion of each perspective is key to understanding the efficacy of parole as a method of reducing recidivism.

Implications of differences in attributions between the public and parole board members could be applied to various legal and policy-related settings. Policy reform could be affected by public opinion towards the success or failure of discretionary parole decisions as compared to mandatory parole outcomes. These opinions might be conveyed through legislative petition (such as in Martinez v. California), through outspoken victims at parole hearings or through public media pressure (as in the Pennsylvania case). Incorporating the strength and influence of victims and victim groups, as well as the media, will likely continue to shape parole decisions either directly through legislative changes or indirectly by affecting the parole board members themselves. Additionally, disclosure of relevant risk assessment information to the public following a parole release decision could lead to fewer discrepancies between the public and parole board. Many risk assessments used by state parole boards might not be publicly available, and some researchers suggest that risk assessments are inaccurate or biased (e.g. van Eijk, 2017). If the public believes that risk assessments are some ephemeral and abstract tool that unknowingly impacts sentencing, parole release and reentry, then disclosing the details to the public might enhance public relations, increase the public’s perceptions of credibility for the decision-makers and improve confidence in the criminal justice and corrections systems. This affect might then reduce the asymmetry found in the current study. Knowing whether and how published risk information impacts public attitudes and thoughts about parole decisions might become quite critical in future research.

Limitations and future directions

The current study assessed differences in perceptions of parole decisions; however, because of the simulated nature of the study, there are several limitations. First, the parole decisions were mock parole decisions and suffered from reduced consequentiality and verisimilitude. However, prior literature suggests that simulations can lead to effective role playing and mimic actual decisions (Kerr et al., 1979). In this case, mimicking actual decision-making would be difficult as parole boards are unique in training and caseload, but using a simulated decision likely bolstered consequentiality and verisimilitude over speculation or imagined scenarios (see Diamond, 1997). Second, though the sample characteristics were similar to typical student and community samples, they differed in two respects. The current student sample was much more racially diverse than most student samples (assessing race as a moderator), which likely improved and added to the current literature using student samples. Also the community sample was an online community sample, which might differ from traditional community samples, so the literature assessing student and community samples in legal decision-making might differ from our research conclusions (see Bornstein 1999; Bornstein et al., 2017).

Third, risk perception confounded perspective in this study. The mock parole board members were able to factor in the information about the parolee’s initial risk assessment, indicating low risk, in their overall judgments. However, the public was only able to determine and speculate about risk after reading about the parolee committing a murder. Therefore, risk perception is confounded with condition and the exact explanation of the parole versus public perspective differences might be better explained as differences in perceived risk. Future research should vary risk perceptions for both parole and public perspectives to assess this potential explanation.

Lastly, it is quite possible that parole boards conceptualize their decisions differently than the current mock parole boards as a result of formal training. Because the current sample did not receive formal parole decision-making training, these differences were not accounted for in this study. Although formal training might be difficult to simulate in mock parole research, a training component could be introduced in order to account for simulated training effects.

Future studies could benefit from assessing the make-up of constructs such as blame and accountability in parole board settings, acceptable levels of risk associated with parole decisions and understanding the difference in available information between actors and observers in this setting. An interesting extension of this study could be to examine the actor–observer bias associated with parole decisions that do not result in the parolee reoffending within 2 years. Finally, future research could benefit from asking real parole board members and members of the public about their attributions following actual parole release cases to increase the external validity of the claims generated by the findings of studies such as this one.

Conclusion

The public and parole boards have different perspectives when assessing the cause of parolees’ behavior. The current research analyzed these divergent perspectives according to the actor and observer framework of attributions. Findings support the applicability of the actor–observer effect to this setting. When a discretionary parole decision is made and results in the parolee reoffending within 2 years, different types of attributions are made by the public as compared to the parole board. The public, or observer, places more blame and control on the parole board, while the parole board attributes more justification and understanding of risk on themselves. Clarification of risk assessment criteria between the parole board and public might reduce the discrepancy in attributions resulting from future failed decisions. The current study presents a new perspective to understand asymmetries between public and parole board evaluations of discretionary parole decisions, and is key in understanding the stakes of parole decision-making in the United States.

Ethical standards

Declaration of conflicts of interest

Logan Yelderman has declared no conflict of interest

Timothy Lawrence has declared no conflict of interest

Courtney Lyons has declared no conflict of interest

Alicia DeVault has declared no conflict of interest

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee (University of Nevada, Reno and Prairie View A&M University Institutional Review Boards, IRBs) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

This study was declared exempt by both the University of Nevada, Reno IRB, and Prairie View A&M University IRB and did not include formal written consent documentation.

Appendix.

Parole condition

You have been elected to serve on a parole board to help make parole release decisions that are best for the community, the inmate and the prison. Please read the following information carefully.

[Offender’s name] is eligible to be released on parole. You are a member of a parole board in charge of making the release decision for [Offender’s name] that is best for the safety of the public, the rights of the individual and the prison. The parole board that you are serving on consists of seven members. The other six members have already made their release decision. Three of them elected to release [Offender’s name], and three of them elected not to release [Offender’s name]. You have the deciding vote whether to release [Offender’s name] or not.

Name: [Offender’s name]

Sex: Male

Race: Caucasian

Education: High school graduate

Age at time of incarceration: 40

Current age: 45

Prison sentence: 7 years

Time spent in prison: 5 years

Reason for imprisonment: Aggravated assault

Volunteer work within the prison: Maximum allowed

Signs/symptoms of mental illness: None

Social/family support: High

Prior arrests: None

Risk of physical harm to self: Very low

Risk of physical harm to others: Very low

Potential risk to the community: Very low

Risk of dangerousness: Low

Public condition

Instructions: You will read a news story about an event. It is important and necessary that you read the news story in its entirety so that you receive all of the information. You will then be asked several questions about the news story. At the end you will be asked about general beliefs and your demographics.

News story [text only]

[Offender Name] RAPES AND MURDERS NEIGHBOR’S DAUGHTER AFTER BEING PAROLLED FOR ONLY [time]

[Offender Name], a Caucasian male, has been arrested for raping and murdering his neighbor’s daughter. He was a felon who had been released from prison [time] prior to this crime. Neighbor [Neighbor’s name] says, ‘He was arrested for violence before, I can’t believe he committed this awful crime only [time] after being released from prison. Now a sweet and innocent child suffered a terrible death’. Police say that [offender’s name] brutally raped his neighbor’s daughter, [victim name] (age 6), and then strangled her to death in front of her two older brothers. Police also say that her brothers might have been able to help save her had they been a couple of years older. [Victim]’s mom was at work at the time of the crime. Her kids had called her yelling hysterically on the phone. She called the police and rushed home to find her two traumatized boys (ages 11 and 13) screaming and crying, and the police examining her daughter’s dead body. [Offender’s name] had already left the scene by the time the police arrived.

Funding Statement

This work was supported by University of Nevada, Reno - Graduate Student Association, Outstanding Graduate Student Scholarship Award; Research, Travel, and Materials Grant Program.

Footnotes

1

Gender estimates were based only on 312 participants because of computational error and missing data.

2

All percentages are valid percentages. They might not equal zero due to rounding.

3

Completion times showed this to be consistent with around a $5–$7 an hour rate.

4

An experimental manipulation of 2 weeks, 2 months or 2 years between release and the time of the crime was employed but yielded no differences across the conditions or as a moderator so they were collapsed together.

References

  1. Abadinsky, H. (2012). Probation and parole: Theory and practice. Pearson. [Google Scholar]
  2. Arkin, R. M., Gabrenya, W. K., Jr., & McGarvey, B. (1978). The role of social perspective in perceiving the causes of success and failure 1. Journal of Personality, 46(4), 762–777. 10.1111/j.1467-6494.1978.tb00196.x [DOI] [Google Scholar]
  3. Bornstein, B. (1999). The ecological validity of jury simulations: Is the jury still out? Law and Human Behavior, 23(1), 75–91. 10.1023/A:1022326807441 [DOI] [Google Scholar]
  4. Bornstein, B., Golding, J. M., Neuschatz, J., Kimbrough, C. D., Reed, K., Magyarics, C., & Luecht, K. (2017). Mock juror sampling issues in jury simulation research: A meta-analysis. Law and Human Behavior, 41(1), 13–28. 10.1037/lhb0000223 [DOI] [PubMed] [Google Scholar]
  5. Brank, E. M., Greene, E., & Hochevar, K. (2011). Holding parents responsible: Is vicarious responsibility the public’s answer to juvenile crime? Psychology, Public Policy, and Law, 17(4), 507–529. 10.1037/a0024566 [DOI] [Google Scholar]
  6. Campbell, W. K., & Sedikides, C. (1999). Self-threat magnifies the self-serving bias: A meta-analytic integration. Review of General Psychology, 3(1), 23–43. 10.1037/1089-2680.3.1.23 [DOI] [Google Scholar]
  7. Caplan, J. M. (2010). Parole release decisions: Impact of victim input on a representative sample of inmates. Journal of Criminal Justice, 38(3), 291–300. 10.1016/j.jcrimjus.2010.02.012 [DOI] [Google Scholar]
  8. Caplan, J. M. (2007). What factors affect parole: A review of empirical research. Federal Probation, 71(1), 16–19. [Google Scholar]
  9. Clear, T., & Cole, G. (1997). American corrections. Wadsworth Publishing. [Google Scholar]
  10. Council of State Government/American Probation and Parole Association (n.d.). Fact sheet 5: Promising victim related practices in probation and parole. https://www.appa-net.org/
  11. Cullen, F. T. (2017). Correctional rehabilitation. Academy for Justice, 4, 235–260. [Google Scholar]
  12. Cullen, F. T., Clark, G. A., Cullen, J. B., & Mathers, R. A. (1985). Attribution, salience, and attitudes toward criminal sanctioning. Criminal Justice and Behavior, 12(3), 305–331. 10.1177/0093854885012003003 [DOI] [Google Scholar]
  13. Cushman, F. (2008). Crime and punishment: Distinguishing the roles of causal and intentional analyses in moral judgment. Cognition, 108(2), 353–380. 10.1016/j.cognition.2008.03.006 [DOI] [PubMed] [Google Scholar]
  14. Diamond, S. S. (1997). Illuminations and shadows from jury simulations. Law and Human Behavior, 21(5), 561–571. 10.1023/A:1024831908377 [DOI] [Google Scholar]
  15. Eisen, S. V. (1979). Actor–observer differences in information inference and causal attribution. Journal of Personality and Social Psychology, 37(2), 261–272. 10.1037/0022-3514.37.2.261 [DOI] [Google Scholar]
  16. Feinberg, J. (1968). Action and responsibility. In White A. R. (Ed.), The philosophy of action (pp. 95–119). Oxford University Press. [Google Scholar]
  17. Fernandez-Duque, D., & Wifall, T. (2007). Actor/observer asymmetry in risky decision making. Judgment and Decision Making, 2(1), 1–8. 10.1037/t04601-000 [DOI] [Google Scholar]
  18. Festinger, L. (1957). A theory of cognitive dissonance. Stanford University Press. [Google Scholar]
  19. Fincham, F. D., & Jaspars, J. M. (1980). Attribution of responsibility: From man the scientist to man as lawyer. In Advances in experimental social psychology (Vol. 13, pp. 81–138). Academic Press. 10.1016/S0065-2601(08)60131-8 [DOI] [Google Scholar]
  20. Fincham, F. D., & Jaspars, J. (1979). Attribution of responsibility to the self and other in children and adults. Journal of Personality and Social Psychology, 37(9), 1589–1602. 10.1037/0022-3514.37.9.1589 [DOI] [Google Scholar]
  21. Finkel, N. J., & Duff, K. B. (1991). Felony-murder and community sentiment: Testing the Supreme Court’s assertions. Law and Human Behavior, 15(4), 405–429. 10.1007/BF02074079 [DOI] [Google Scholar]
  22. Finney, P. D., & Helm, B. (1982). The actor-observer relationship: The effect of actors’ and observers’ responsibility and emotion attributions. The Journal of Social Psychology, 117(2), 219–225. 10.1080/00224545.1982.9713430 [DOI] [Google Scholar]
  23. Frink, D. D., & Klimoski, R. J. (2004). Advancing accountability theory and practice: Introduction to the human resource management review special edition. Human Resource Management Review, 14(1), 1–17. 10.1016/j.hrmr.2004.02.001 [DOI] [Google Scholar]
  24. Gailey, J. A., & Falk, R. F. (2008). Attribution of responsibility as a multidimensional concept. Sociological Spectrum, 28(6), 659–680. 10.1080/02732170802342958 [DOI] [Google Scholar]
  25. Gebotys, R. J., & Dasgupta, B. (1987). Attribution of responsibility and crime seriousness. The Journal of Psychology, 121(6), 607–613. 10.1080/00223980.1987.9712690 [DOI] [PubMed] [Google Scholar]
  26. Goldschmied, N., & Hochuli, R. (2014). Do fans and officials think alike when assessing the home-field advantage? Support for actor–observer asymmetry in the National Football League. Basic and Applied Social Psychology, 36(2), 155–163. 10.1080/01973533.2014.882258 [DOI] [Google Scholar]
  27. Gosling, P., Denizeau, M., & Oberlé, D. (2006). Denial of responsibility: A new mode of dissonance reduction. Journal of Personality and Social Psychology, 90(5), 722–733. 10.1037/0022-3514.90.5.722 [DOI] [PubMed] [Google Scholar]
  28. Green, T. D., & McClearn, D. G. (2010). The actor-observer effect as a function of performance outcome and nationality of other. Social Behavior and Personality: An International Journal, 38(10), 1335–1344. 10.2224/sbp.2010.38.10.1335 [DOI] [Google Scholar]
  29. Guglielmo, S. (2015). Moral judgment as information processing: An integrative review. Frontiers in Psychology, 6, 1637. 10.3389/fpsyg.2015.01637 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hansen, E. M., Kimble, C. E., & Biers, D. W. (2001). Actors and observers: Divergent attributions of constrained unfriendly behavior. Social Behavior and Personality: An International Journal, 29(1), 87–104. 10.2224/sbp.2001.29.1.87 [DOI] [Google Scholar]
  31. Hart, H. L. A. (1968). Punishment and responsibility. Oxford University Press. [Google Scholar]
  32. Harvey, J. H., Harris, B., & Barnes, R. D. (1975). Actor-observer differences in the perceptions of responsibility and freedom. Journal of Personality and Social Psychology, 32(1), 22–28. 10.1037/h0076852 [DOI] [Google Scholar]
  33. Heider, F. (1944). Social perception and phenomenal causality. Psychological Review, 51(6), 358–374. 10.1037/h0055425 [DOI] [Google Scholar]
  34. Hennessy, D. A., & Jakubowski, R. (2007). The impact of visual perspective and anger on the actor-observer bias among automobile drivers. Traffic Injury Prevention, 8(2), 115–122. 10.1080/15389580601048937 [DOI] [PubMed] [Google Scholar]
  35. Herman, S., & Wasserman, C. (2001). A role for victims in offender reentry. Crime & Delinquency, 47(3), 428–445. 10.1177/0011128701047003008 [DOI] [Google Scholar]
  36. Holdorf, W. E., & Greenwald, J. M. (2018). Toward a taxonomy and unified construct of responsibility. Personality and Individual Differences, 132, 115–125. 10.1016/j.paid.2018.05.028 [DOI] [Google Scholar]
  37. Horswill, M. S., & McKenna, F. P. (1999). The effect of perceived control on risk taking. Journal of Applied Social Psychology, 29(2), 377–391. 10.1111/j.1559-1816.1999.tb01392 [DOI] [Google Scholar]
  38. Hughes, T. A., Wilson, D. J., Beck, A. J. (2001). Trends in state parole, 1990–2000. Bureau of Justice Statistics. http://www.bjs.gov
  39. Johnston, C., & Lee, C. M. (2005). Children’s attributions for their own versus others’ behavior: Influence of actor versus observer differences. Journal of Applied Developmental Psychology, 26(3), 314–328. 10.1016/j.appdev.2005.02.005 [DOI] [Google Scholar]
  40. Jones, E. E., & Nisbett, R. E. (1972). The actor and the observer: Divergent perceptions of the causes of behavior. In Jones E. E., Kanouse D., Kelley H. H., Nisbett R. E., Valins S., & Weiner B. (Eds.), Attribution: Perceiving the causes of behavior (pp. 79–94). General Learning Press. [Google Scholar]
  41. Kaeble, D. (2018). Probation and parole in the United States, 2016. Bureau of Justice Statistics. http://www.bjs.gov
  42. Kelley, H. H., & Michela, J. L. (1980). Attribution theory and research. Annual Review of Psychology, 31, 457–501. 10.1146/annurev.ps.31.020180.002325 [DOI] [PubMed] [Google Scholar]
  43. Kerr, N. L., Nerenz, D. R., & Herrick, D. (1979). Role playing and the study of jury behavior. Sociological Methods & Research, 7(3), 337–355. 10.1177/004912417900700305 [DOI] [Google Scholar]
  44. Krueger, J., Ham, J. J., & Linford, K. M. (1996). Perceptions of behavioral consistency: Are people aware of the actor-observer effect? Psychological Science, 7(5), 259–264. 10.1111/j.1467-9280.1996.tb00371.x [DOI] [Google Scholar]
  45. Lagnado, D. A., & Channon, S. (2008). Judgments of cause and blame: The effects of intentionality and foreseeability. Cognition, 108(3), 754–770. 10.1016/j.cognition.2008.06.009 [DOI] [PubMed] [Google Scholar]
  46. Lerner, J. S., & Tetlock, P. E. (1999). Accounting for the effects of accountability. Psychological Bulletin, 125(2), 255–275. 10.1037/0033-2909.125.2.255 [DOI] [PubMed] [Google Scholar]
  47. Lewis, P. T. (1995). A naturalistic test of two fundamental propositions: Correspondence bias and the actor‐observer hypothesis. Journal of Personality, 63(1), 87–111. 10.1111/j.1467-6494.1995.tb00803.x [DOI] [Google Scholar]
  48. Lindsey, S. C., & Miller, M. K. (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. 10.1080/13218719.2011.625619 [DOI] [Google Scholar]
  49. Lowe, C. A., & Ansen, R. D. (1976). Motivational influence of behavioral desirability on actors’ and observers’ attributions. Social Behavior and Personality: An International Journal, 4(1), 17–25. 10.2224/sbp.1976.4.1.17 [DOI] [Google Scholar]
  50. Mackenzie, D. L. (2001). Sentencing and corrections in the 21st century: Setting the stage for the future. National Institute of Justice. [Google Scholar]
  51. Malle, B., Guglielmo, S., & Monroe, A. (2014). A theory of blame. Psychological Inquiry, 25(2), 147–186. 10.1080/1047840X.2014.877340 [DOI] [Google Scholar]
  52. Malle, B. F. (2006). The actor-observer asymmetry in attribution: A (surprising) meta-analysis. Psychological Bulletin, 132(6), 895–919. 10.1037/0033-2909.132.6.895 [DOI] [PubMed] [Google Scholar]
  53. Malle, B. F., & Knobe, J. (1997). Which behaviors do people explain? A basic actor–observer asymmetry. Journal of Personality and Social Psychology, 72(2), 288–304. 10.1037/0022-3514.72.2.288 [DOI] [Google Scholar]
  54. Malle, B. F., Knobe, J. M., & Nelson, S. E. (2007). Actor-observer asymmetries in explanations of behavior: New answers to an old question. Journal of Personality and Social Psychology, 93(4), 491–514. 10.1037/0022-3514.93.4.491 [DOI] [PubMed] [Google Scholar]
  55. Malle, B. F., & Pearce, G. E. (2001). Attention to behavioral events during interaction: Two actor-observer gaps and three attempts to close them. Journal of Personality and Social Psychology, 81(2), 278–294. 10.1037/0022-3514.81.2.278 [DOI] [PubMed] [Google Scholar]
  56. Martinez v. California, 444 US 277 (1980). [Google Scholar]
  57. Maruschak, L. M., & Bonczar, T. P. (2013). Probation and parole in the United States, 2012. Bureau of Justice Statistics. http://www.bjs.gov
  58. McGrath, A. (2017). Dealing with dissonance: A review of cognitive dissonance reduction. Social and Personality Psychology Compass, 11(12), e12362. 10.1111/spc3.12362 [DOI] [Google Scholar]
  59. McKillip, J., & Posavac, E. J. (1975). Judgments of responsibility for an accident. Journal of Personality, 43(2), 248–265. 10.1111/j.1467-6494.1975.tb00705.x [DOI] [Google Scholar]
  60. Miller, D. T., & Norman, S. A. (1975). Actor-observer differences in perceptions of effective control. Journal of Personality and Social Psychology, 31(3), 503–515. 10.1037/h0076485 [DOI] [Google Scholar]
  61. Mitchell, T. E. (1985). Actor-observer differences in attributions to morality. The Journal of Social Psychology, 125(4), 475–477. 10.1080/00224545.1985.9713526 [DOI] [Google Scholar]
  62. Morgan, K., & Smith, B. L. (2005). Victims, punishment, and parole: The effect of victim participation on parole hearings. Criminology & Public Policy, 4(2), 333–360. 10.1111/j.1745-9133.2005.00025.x [DOI] [Google Scholar]
  63. Nadelhoffer, T., & Feltz, A. (2008). The actor–observer bias and moral intuitions: Adding fuel to Sinnott-Armstrong’s fire. Neuroethics, 1(2), 133–144. 10.1007/s12152-008-9015-7 [DOI] [Google Scholar]
  64. National Parole Resource Center (2012, June) Vignettes from the national parole resource center learning sites. Bureau of Justice Assistance. http://nationalparoleresourcecenter.org/
  65. Newman, A. (1978). Actor-observer differences in perception of self-control. The Journal of Social Psychology, 105(2), 199–204. 10.1080/00224545.1978.9924115 [DOI] [Google Scholar]
  66. Nisbett, R. E., Caputo, C., Legant, P., & Marecek, J. (1973). Behavior as seen by the actor and as seen by the observer. Journal of Personality and Social Psychology, 27(2), 154–164. 10.1037/h0034779 [DOI] [Google Scholar]
  67. Nuijten, A., Keil, M., van der Pijl, G., & Commandeur, H. (2018). IT managers’ vs. IT auditors’ perceptions of risks: An actor–observer asymmetry perspective. Information & Management, 55(1), 80–93. 10.1016/j.im.2017.04.002 [DOI] [Google Scholar]
  68. O’Connor, B. P. (1988). Actors and observers: Are their descriptions of behavior indications of their causal attributions? Social Behavior and Personality: An International Journal, 16(2), 207–213. 10.2224/sbp.1988.16.2.207 [DOI] [Google Scholar]
  69. Paparozzi, M. A., & Guy, R. (2009). The giant that never woke: Parole authorities as the lynchpin to evidence-based practice and prisoner re-entry. Journal of Contemporary Criminal Justice, 25(4), 397–411. 10.1177/1043986209344561 [DOI] [Google Scholar]
  70. Petersilia, J. (2000). Parole and prisoner reentry in the United States. Perspectives, 24, 32–46. [Google Scholar]
  71. Rhine, E. E., Mitchell, K., Reitz, K. R. (2018). Levers of change in parole release and revocation. Robina Institute of Criminal Law and Criminal Justice. www.apaintl.org
  72. Rhine, E. E., Petersilia, J., & Reitz, K. R. (2017). The future of parole release. Crime and Justice, 46(1), 279–338. 10.1086/688616 [DOI] [Google Scholar]
  73. Rickard, L. N. (2014). Perception of risk and the attribution of responsibility for accidents. Risk Analysis : An Official Publication of the Society for Risk Analysis, 34(3), 514–528. 10.1111/risa.12118 [DOI] [PubMed] [Google Scholar]
  74. Roberts, J. V. (2009). Listening to the crime victim: Evaluating victim input at sentencing and parole. Crime and Justice, 38(1), 347–412. 10.1086/599203 [DOI] [Google Scholar]
  75. Robins, R. W., Spranca, M. D., & Mendelsohn, G. A. (1996). The actor-observer effect revisited: Effects of individual differences and repeated social interactions on actor and observer attributions. Journal of Personality and Social Psychology, 71(2), 375–389. 10.1037/0022-3514.71.2.375 [DOI] [PubMed] [Google Scholar]
  76. Rosebury, B. (2011). The political logic of victim impact statements. Criminal Justice Ethics, 30(1), 39–67. 10.1080/0731129X.2011.559061 [DOI] [Google Scholar]
  77. Shaver, K. G. (1985). The attribution of blame: Causality, responsibility, and blameworthiness. Springer. [Google Scholar]
  78. Sheikh, S., & McNamara, M. E. (2014). Insights from self-blame and victim blaming. Psychological Inquiry, 25(2), 241–244. 10.1080/1047840X.2014.904138 [DOI] [Google Scholar]
  79. Shultz, T. R., Jaggi, C., & Schleifer, M. (1987). Assigning vicarious responsibility. European Journal of Social Psychology, 17(3), 377–380. 10.1002/ejsp.2420170314 [DOI] [Google Scholar]
  80. Shultz, T. R., & Schleifer, M. (1983). Towards a refinement of attribution concepts. In Jaspars J., Fincham F. D., & Hewstone M. (Eds.), Attribution theory and research: Conceptual, developmental and social dimensions (pp. 37–62). Academic Press. [Google Scholar]
  81. Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-affirmation. Journal of Personality and Social Psychology, 45(1), 5–19. 10.1037/0022-3514.45.1.5 [DOI] [Google Scholar]
  82. Tangney, J. P., Stuewig, J., & Mashek, D. J. (2007). What’s moral about the self-conscious emotions? In Tracy J. L., Robins R. W., Tangney J. P., (Eds.), The self-conscious emotions. (pp. 21–37). The Guilford Press. [Google Scholar]
  83. Tomarken, A. J., & Serlin, R. C. (1986). Comparison of ANOVA alternatives under variance heterogeneity and specific noncentrality structures. Psychological Bulletin, 99(1), 90–99. 10.1037/0033-2909.99.1.90 [DOI] [Google Scholar]
  84. van Eijk, G. (2017). Socioeconomic marginality in sentencing: The built-in bias in risk-assessment tools and the reproduction of social inequality. Punishment & Society, 19(4), 463–481. 10.1177/1462474516666282 [DOI] [Google Scholar]
  85. Vázquez, J. J., Panadero, S., & Zúñiga, C. (2017). Actors, observers, and causal attributions of homelessness: Differences in attribution for the causes of homelessness among domiciled and homeless people in Madrid (Spain). The American Journal of Orthopsychiatry, 87(1), 15–22. 10.1037/ort0000130 [DOI] [PubMed] [Google Scholar]
  86. Vîlcică, E. R. (2016). Studying parole in the “spotlight”: Lessons from a large American jurisdiction. European Journal on Criminal Policy and Research, 22(1), 61–88. 10.1007/s10610-015-9284-8 [DOI] [Google Scholar]
  87. Vîlcică, E. R. (2018). Revisiting parole decision making: Testing for the punitive hypothesis in a large U.S. jurisdiction. International Journal of Offender Therapy and Comparative Criminology, 62(5), 1357–1383. 10.1177/0306624X16668512 [DOI] [PubMed] [Google Scholar]
  88. Wallace, D. M., & Hinsz, V. B. (2009). Group members as actors and observers in attributions of responsibility for group performance. Small Group Research, 40(1), 52–71. 10.1177/1046496408326576 [DOI] [Google Scholar]
  89. Walster, E. (1966). Assignment of responsibility for an accident. Journal of Personality and Social Psychology, 3(1), 73–79. 10.1037/h0022733 [DOI] [PubMed] [Google Scholar]
  90. Weiner, B. (1995). Judgments of responsibility: A foundation for a theory of social conduct. Guilford. [Google Scholar]
  91. Wilson, S. R., Levine, K. J., Cruz, M. G., & Rao, N. (1997). Attribution complexity and actor-observer bias. Journal of Social Behavior and Personality, 12(3), 709–726. [Google Scholar]

Articles from Psychiatry, Psychology, and Law are provided here courtesy of Taylor & Francis

RESOURCES