Introduction

Interest in the process-based model of regulation has increased dramatically in the past decade. Much of the growing body of research uses survey data to test whether subjective fairness judgments of police process influence supportive values, and whether the latter affects general cooperation (see Tyler 2003, p. 284, for a graphical depiction). The weight of the empirical evidence largely supports this two-part argument (Mazerolle et al. 2013a). Although such research is frequently referenced in police policy discussions, critics caution that subjective procedural justice judgments may not accurately reflect actual police behavior (Nagin and Telep 2017).

The process-based model also posits that police action consistent with the principles of procedural justice—respectful treatment, participation opportunity, neutral decision making, and motive-based trust—directly influence various outcomes that are of interest to the police, such as the willingness to accept authorities’ decisions (Tyler 2003). What is more, the salience of procedural justice during encounters is said to exceed the favorability of outcomes (Tyler and Huo 2002). To date, general population survey studies of the main effect of procedural justice judgements are relatively limited (Nix et al. 2015; Reisig 2007). Systematic investigations of the “process versus outcome hypothesis” using data from individuals who had recent contact with legal authorities are even sparser. This state of affairs is unfortunate because it is in the best interest of the police that encounters unfold in an orderly and cooperative fashion, even when issuing a citation is necessary.

Although research has used samples of individuals with recent contact with the police to test the main effects of procedural justice (Penner et al. 2014; Skogan 2005), the process versus outcome hypothesis is rarely subjected to empirical scrutiny. But such tests do exist. Paternoster et al. (1997) not only found that individuals who were arrested for spousal assault who judged the police in their case favorably in terms of procedural justice were significantly less likely to reoffend, but they also observed that the procedural justice effect was stronger than whether the participant received a favorable outcome (i.e., shorter term of detention). Tyler and Huo (2002), who sampled individuals with recent contact with the police or courts, report that procedural justice judgments significantly influence trust, acceptance of decisions, and satisfaction with the decision maker. Importantly, the procedural justice effects are stronger than outcome favorability. Although studies using samples of individuals with recent criminal justice experience likely benefit from more reliable data about the quality of police action (Reisig 2002), such studies also rely on subjective fairness judgments.

Methodologies that do not rely on subjective judgments of police action have also been used to test process-based model hypotheses. A small number of studies use systematic social observation (SSO) data. This method involves trained observers transcribing detailed accounts of transactions between officers and individuals according to a predetermined protocol. Arguably, this methodological approach effectively captures objective accounts of police behavior. Mastrofski et al. (1996) found that citizens who were treated disrespectfully by the police were less likely to comply with police directives. The effect of respectful police behavior was null. McCluskey et al. (1999) found that both respectful and disrespectful police behavior shape citizens’ compliance. Finally, Dai et al. (2011) found that police disrespect had no appreciable effect on compliance. At best, extant SSO procedural justice research is incomplete: the main effects of procedural justice elements vary across studies and researchers have yet to consider the importance of process relative to favorable outcomes.

Recent field experiments have assessed the influence of procedurally just police action. The Queensland Community Engagement Trial (QCET) involved the use of a script infused with procedural justice principles during random breath test traffic stops. Drivers receiving the experimental intervention were more satisfied with the police than drivers receiving a standard breath test (Mazerolle et al. 2013b; see also Sahin et al. 2016). The results from the Scottish Community Engagement Trial (ScotCET) showed that the experimental treatment did not produce the desired effect (MacQueen and Bradford 2015). Field trials testing the impact of procedurally just policing are relatively new, though the evidence is encouraging. Field experiments considering the relative importance of procedural justice by including a second stimuli regarding the resolution of police encounters, such as variation in the dosage of police sanctions, would help fill a critical void in the process-based model literature.

A handful of non-field experiments have examined how citizens perceive scenarios of hypothetical police encounters in terms of procedural justice. Maguire et al. (2016) used videos depicting traffic stops and found that individuals who observed procedurally unjust treatment were less willing to cooperate with police, had a lower sense of obligation to obey, and reported less trust and confidence in police. Outcomes for those observing procedurally just stops were much more positive (see also Lowrey et al. 2016). Barkworth and Murphy (2015) used written traffic stop scenarios. They found that participants who received the procedurally just encounter reported lower negative affect scores and had a greater propensity for compliance relative to those who received the procedural injustice condition. As with field trials, the primary interest of researchers conducting non-field experiments has been the main effect of procedural justice. The importance of fair processes relative to the favorability of outcomes has taken a back seat, leaving an important empirical question open.

Current focus

This study tests the effects of procedural injustice on encounter-specific outcome variables by using full factorial vignette survey data from a university-based sample. Participants were randomly given one of two scenarios—one involving a traffic stop and the other depicting a noise complaint. Both scenarios included two experimental conditions: (1) police behavior consistent with procedural injustice and (2) police issuing a citation. Ordinal regression models were estimated to test the magnitudes of the procedural injustice and police citation effects. The results will demonstrate the potential negative impact police processes can have on encounters with the public.

Methods

Data

This study uses data from self-administered surveys, distributed to undergraduate criminology and criminal justice (CCJ) students at Arizona State University (ASU). Surveys were administered during the 2015–16 academic year. Students from 13 different CCJ courses were surveyed, nine of which were introductory classes that are open to all ASU students (n = 449). Four classes were upper-level courses that were only available to CCJ majors (n = 147). The survey took 20 minutes to complete. The research protocol was approved by ASU’s institutional review board. A total of 596 of the 605 surveys were returned (participation rate = 98.5%). There was very little missing data (125 cells of the total of 4172 used in the analyses, or less than 3% of cells). Missing values were handled using similar response pattern imputation (or hot deck nearest neighbor method), which works well with Likert-type survey data (Jönsson and Wohlin 2006). Two cases with persistently missing values were omitted.

Sample

The sample was 57.3% female and 42.7% male. A little less than half of the sample identified as White (45.2%), 38.0% Latino, 6.2% African American, 4.2% Asian, 3.5% Native American, and 2.9% identified as “other” minority. A majority of participants were between the ages of 18 and 19 years old (55.5%), 12.6% were 20 years old, and 31.9% were 21 years old or more. When compared to the university student population, the sample was more racially diverse, younger, and included a higher proportion of females.

Design

Participants received survey booklets containing a variety of closed-ended items and a single hypothetical scenario (either traffic stop or noise complaint; see Appendix). The scenario was located near the end of the instrument. Presenting participants with only one vignette reduces carelessness when completing the instrument (Stolte 1994). The scenarios were short, concise, and true-to-life for many college students. Indeed, when asked how clearly they could imagine the story, most participants said “very clearly” or “somewhat clearly” (96.2% for traffic stop and 97.3% for noise complaint). Each survey altered the experimental stimulus and control condition for two different variables (i.e., procedural injustice and police citation), making for a total of four different possible versions of each scenario. Surveys were randomly pre-sorted and distributed by row in each classroom. After reading the scenario, participants were asked questions related to the hypothetical police encounter. Some of these items are used as outcome variables and others for manipulation checks.Footnote 1

Measures

The study uses four dependent variables. Encounter satisfaction is a single item that asked, “How satisfied are you with the way the police officer handled the situation?” This item featured a closed-ended scale (1 = very dissatisfied to 4 = very satisfied). Handle differently is a single item that read, “In a similar situation in the future, you would like to see the police handle the situation differently.” The response set ranged from strongly disagree (coded 1) to strongly agree (coded 4). Immediate compliance is a single-item variable that asked, “How likely would you be to do what the officer asked you to do?” A closed-ended response was used (1 = very unlikely to 4 = very likely). Decision acceptance is operationalized using an item that read, “How willing would you be to accept the officer’s decision?” A Likert-type response set was featured (1 = very unwilling to 4 = very willing; see Table 1 for summary statistics).

Table 1 Summary statistics for dependent variables
Full size table

The two independent variables capture experimental conditions. One condition features a police officer behaving in a manner that is not consistent with the principles of procedural justice. In both scenarios, officers use profanity (disrespectful treatment), make judgmental statements (fail to remain neutral), and ask rhetorical questions in a condescending manner (fail to provide participation opportunity). Another element of procedural justice, motive-based trust, is not well represented in the manipulation. Procedural injustice reflects whether participants were administered the experimental condition (1 = yes, 0 = no). The second variable, police citation, indicates whether the participant responded to a hypothetical encounter where the police issued a citation (1 = yes, 0 = no).

Analyses and results

Manipulation checks

To determine whether the injustice manipulations had the desired effects, one-way analysis of variance (ANOVA) models were estimated. After reading the scenarios and responding to the items for the outcome measures, participants were presented three items asking them to judge the behavior of the police officer in the scenario they received (i.e., “The officer treated you with respect”, “The officer treated you politely”, and “The officer showed concern for your rights”). Response sets ranged from strongly disagree (coded 1) to strongly agree (coded 4). Table 2 shows the mean scores for the evaluative measures across groups. The results demonstrate that individuals exposed to the injustice condition rate the police in their scenario more negatively in terms of procedural justice relative to those who received the control condition. These findings confirm that the injustice manipulations had the desired effect. Because the police citation manipulation was an objective action, checks similar to those employed for the injustice manipulation were not performed. Instead, the sign and significance of the effect of the police citation manipulation will offer evidence as to whether this condition worked as intended.

Table 2 One-way analysis of variance (ANOVA) models testing procedural injustice manipulations
Full size table

Multivariate regression models

Given the nature of the dependent variables, the ordinal regression model was used. The parallel lines assumption was met for the eight models presented in Table 3. Each model contains two variables that reflect the experimental conditions. Concerning the main effects of procedural injustice, the results are consistent in terms of sign and significance both within and across the two scenarios. In other words, participants who received the procedural injustice manipulation were less satisfied with how the police resolved the encounter, more likely to wish the police handle such situations differently, less likely to follow police directives, and less willing to accept police decisions. The effect sizes were also similar across scenarios. With regard to the outcome of encounters, the pattern of estimates indicates that the citation manipulation worked as intended. A negative effect was observed for encounter satisfaction, immediate compliance, and decision acceptance. And, as expected, these same participants were more likely to report that the police should handle these types of situations differently. All told, the evidence shows that both process and outcome were important predictors.

Table 3 Ordinal regression models
Full size table

When it comes to the process versus outcome hypothesis, the evidence is somewhat mixed. For example, the regression estimates are clearly stronger for procedural injustice relative to police citation in the handle situation differently and immediate compliance models for both scenarios. However, when it comes to encounter satisfaction, the relative effect of procedural injustice is as expected in the traffic stop subsample, but fails to provide support among individuals responding to the noise complaint scenario. Finally, the effect of police citation was considerably stronger in the decision acceptance models, suggesting that the process-based model may underestimate the salience of citations among college students who may be hit comparatively hard by even modest economic sanctions.

Discussion

The primary objective of this study was to test the process versus outcome hypothesis. The findings are largely supportive, with a couple of exceptions. Such findings suggest that the police can serve their cause well during public encounters by acting in accordance with the principles of procedural justice, even in situations when they issue citations. The contrary evidence regarding the salience of citations was unexpected. Until the findings are replicated using independent samples, caution must be exercised in calling for theoretical reform. Nevertheless, the potential theoretical questions are intriguing to consider. Is it possible that this aspect of the theory is age graded? Are younger people who are less financially secure more sensitive to economic sanctions? Do older adults who are similarly vulnerable in economic terms accept citation decisions more readily? More research is clearly needed.

Researchers should also keep in mind that vignette methodologies can be used to investigate other questions, such as the role of procedural injustice in other types of police encounters, such as voluntary contacts. Vignette research can also help assess whether procedural justice is more salient than distributive justice in police–citizen encounters. These are but two potential areas of future inquiry.

The practical significance of the findings is straightforward: police process matters. More specifically, unfair treatment by the police negatively shapes how their encounters with the public unfold. Coupled with SSO, experimental, and survey-based research, the evidence suggests that adhering to procedural justice principles during encounters pays dividends to police officers in the short and long run. Accordingly, extant research indicates that the continued development of academy and in-service procedural justice training could promote beneficial outcomes, though such reforms, like many past efforts to modify policing, will likely be met with some level of resistance.

This study is not without limitations. For example, vignette research is limited by the fact that the situations are hypothetical. It is unclear whether participants would make the same decisions and feel similarly if the situation had actually occurred. To address this concern, police encounters familiar to college students were used, and the language used in the procedural injustice manipulations was based on the authors’ personal observations and second-hand accounts of police–student encounters. Another concern is the use of college students. It is well known that younger individuals view the police differently than older people. The results should be viewed as tentative until they are replicated using more representative samples. A third concern relates to the experimental conditions. The injustice stimuli did not capture all of the procedural justice elements. As such, it is impossible to conclude that the absence of such elements did not contribute to the unexpected findings regarding the salience of citation outcome on decision acceptance. Despite the limitations, the evidence reported here contributes to the body of research on the utility of the process-based model in a policing context.

Coupled with other experimental and vignette studies, this study marks another attempt to move process-based research beyond subjective fairness judgments. At this time, the evidence is clear: the way the police behave during public encounters matters. Police processes consistent with the tenets of procedural justice offer promise in ways that should be appealing to police professionals of all stripes, from the mayoral-appointed chief to the newbie officer walking the beat.