Skip to main content

Advertisement

Log in

Policing persons in behavioral crises: an experimental test of bystander perceptions of procedural justice

  • Published:
Journal of Experimental Criminology Aims and scope Submit manuscript

Abstract

Objectives

Policing is in the midst of a legitimacy crisis. Procedural justice is a primary avenue for police reform, including when police officers interact with vulnerable populations. The purpose of the current study was to evaluate the nuanced circumstances in which the public may endorse police interactions with persons in crisis as more or less procedurally just.

Methods

We recruited a nationally representative sample of 569 Americans and a diverse sample of 809 undergraduates. Using factorial survey vignettes, we assessed bystander perceptions of procedural justice to encounters between officers and a person suffering a behavioral crisis, which varied in officer tactics, use of force, and the cause of crisis.

Results

Officers were perceived as more procedurally just when they employed tactics consistent with Crisis Intervention Team (CIT) training. Use of force reduced perceptions of procedural justice, but use of force by CIT officers was perceived as more procedurally just than conventional officers’ actions, regardless of use of force. Conventional treatment and use of force were considered less procedurally just when the person’s crisis was due to mental illness compared with substance use.

Conclusions

The current findings suggest bystanders did not uniformly endorse use of force by police but were more tolerant of force when officers used CIT-informed tactics and when a person’s crisis was due to substance use. Use of force against persons with mental illness was viewed as procedurally unjust, perhaps reflecting the public’s increasing sensitivity to this population and a growing dissatisfaction with police involvement as the often standard response to persons in crises.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

  1. The ongoing debates regarding the operationalizations of procedural justice and legitimacy, and the causal relationship between them, are beyond the scope of the current study. See Gau (2011, 2014), Huq et al. (2017), and Nagin and Telep (2017, 2020) for discussion of these issues.

  2. Some researchers have used arrest reports, surveys of both the public and the police, and direct observations to conservatively estimate the frequency of nonlethal force (Garner et al. 2018; Hickman et al. 2008). These methods suggest that the police engaged in approximately 337,590 use of force incidents in 2012 (Garner et al. 2018), translating to 10.8 incidents per 10,000 residents, 45.4 incidents per 100 officers, and 19.6 incidents per 100 arrests for violent offenses, with large variations depending on the size and agency type.

  3. A police officer’s decision to use force is often not purely discretionary. Many times, this choice is made within the confines of departmental policies and situational exigencies, which create a set of contingencies that can both guide and restrain officer behavior. Nevertheless, for the purposes of our experimental design, we treat the use of force as a decision that is either made or not made within a given situation.

  4. Qualtrics uses double-opt-in market research panel providers to recruit participants. Identity is verified prior to participation through various methods, including IP address and digital fingerprinting. Participants are randomly selected to participant if they qualify and are compensated based on their agreement with Qualtrics (e.g., gift cards, charitable donations). Qualtrics panels are better able to replicate the demographics of the US population than either Amazon’s Mechanical Turk (MTurk) or SurveyMonkey (see Graham et al. 2020; Heen et al. 2014).

  5. The results disaggregated by sample are as follows: for the student sample, 9%—“too much,” 33%—“about right,” and 59%—“too little”; for the national online sample, 17%—“too much,” 46%—“about right,” and 37%—“too little.”

  6. Data and materials, including vignettes, are available via OSF: https://osf.io/u6s58/?view_only=e5df63119cf44647bac5445115756e82

  7. Participants assigned to the mental health crisis condition were more likely to correctly report Mike’s schizophrenia as the reason 911 was called (93.2% vs. 6.8% for those assigned to the substance use crisis condition), while those assigned to the substance use crisis condition were more likely to correctly report that 911 was called because Mike smoked meth (91.9% vs. 5.1% assigned to the mental health crisis condition), χ2(2, N = 1377) = 1048.61, p < .001, φ = .87. Participants assigned to the CIT tactics condition were more likely to correctly report that the responding officers received CIT training (92.8% vs. 27.3% for those assigned to the conventional condition), while those assigned to the conventional tactics condition were more likely to correctly report that the responding officers had not received any specialized training (72.7% vs. 7.2% assigned to the CIT tactic condition), χ2(2, N = 1376) = 616.99, p < .001, φ = .67. In addition, participants believed the amount of force used by the officers on Mike was more appropriate when they were assigned to the no force condition (vs. force), t(1375) = 13.58, p < .001, d = .73, 95% CI (.62, .84).

  8. In addition to demographics, all participants answered several other questions. Participants indicated their political affiliation (1 = very liberal, 5 = very conservative). They also indicated whether any close friends or family has struggled with drug abuse or mental illness, ever been arrested, been employed in law enforcement, or served in the military (all coded 1 = yes). Participants were also asked if they have ever been employed in law enforcement or served in the military (1 = yes; see Table 1). While the addition of control variables unnecessarily in models can introduce bias (Berk et al. 2013; L. D. Robinson and Jewell 1991), we conducted a series of ordinary least square (OLS) regressions to estimate models with the control variables. These are available in Appendix B and replicate the ANOVA results presented in text.

  9. The manner in which police officers treat persons in behavioral crises cannot be entirely examined independently of the disparity in how police treat White Americans and Black Americans. Indeed, psychosis may constitute an important indirect avenue between race and police (mis)use of force in the USA. Poorer Americans (relative to wealthier ones) and Black and brown Americans (relative to White ones) have less access to healthcare, poorer treatment when they do have access, and they disproportionately bear the social and developmental consequences of poor health (Braveman and Barclay 2009; Lantz et al. 1998; Williams et al. 1997).

References

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Angela M. Jones.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix A

Table 3 Question wording and factor loadings for situational procedural justice

The items composing the perceived procedural justice scale loaded on a single factor and explained 67.30% of the variance.

Appendix B

Sensitivity analyses with control variables

To examine the potential effects of the control variables, we also ran a series of OLS regression models (see Tables 4 and 5). In model 1, where procedural justice was regressed on the control variables, respondent age (β = .10, p = .002) and political orientation (conservatism; β = .09, p = .002) were positively correlated with perceptions of procedural justice, while the respondent reporting they had known someone who has struggled with mental health issues was negatively associated with perceptions of procedural justice (β = −.12, p = .008). Still, these characteristics explained little variation in perceptions of procedural justice in this vignette (R2 = .033). Model 2 describes the main effects of the experimental manipulations on perceived procedural justice, net of the control variables. Again, consistent with hypotheses 1, 2, and 3, each of the experimental manipulations had a statistically significant effect (p < .001) on procedural justice. In Table 5, models 3 through 5 reproduce the statistical interactions produced using ANOVA, net of the control variables. These again show support for hypotheses 4 and 5. Thus, the OLS results are substantively equivalent to the ANOVA results.

Table 4 OLS regressions predicting perceptions of procedural justice (N = 1358)
Table 5 OLS regression of manipulations and interactions on perceptions of procedural justice (N = 1358)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Jones, A.M., Vaughan, A.D., Roche, S.P. et al. Policing persons in behavioral crises: an experimental test of bystander perceptions of procedural justice. J Exp Criminol 18, 581–605 (2022). https://doi.org/10.1007/s11292-021-09462-1

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11292-021-09462-1

Keywords

Navigation