Monday, January 25, 2021

Women had more favorable attitudes toward the police; those who were victims of crime or had more fear of crime had less satisfaction with the police

What matters in citizen satisfaction with police: A meta-analysis. Michelle A. Bolger, Daniel J. Lytle, P. Colin Bolger. Journal of Criminal Justice, Volume 72, January–February 2021, 101760. https://doi.org/10.1016/j.jcrimjus.2020.101760

Rolf Degen's take: Rolf Degen on Twitter: Meta-analysis: Women have more favorable attitudes toward the police than men.

Highlights

• We provide the first full meta-analysis of studies examining various factors predicting citizen satisfaction with police.

• We included 66 independent studies in our analysis.

• Random effects analyses revealed that race, age, gender, fear of crime, and victimization were statistically significant.

• Moderating analyses revealed Tthat inclusion of certain variables had impacts on other variable effect sizes.

• In sum, patterns across demographics are consistent with most prior research, but other key variables must be included.

• Furthermore, future research must work to standardize measurement of citizen satisfaction with police.

Abstract

Purpose: While there has been a sizeable amount of research on identifying the correlates of citizen satisfaction with police agencies, that research has not been synthesized to identify patterns across different studies. This study presents the results of a meta-analysis that assessed the predictive strength of the most commonly included correlates of satisfaction with police.

Methods: An exhaustive search for studies on satisfaction with police produced 66 studies eligible for inclusion in the meta-analysis. Random effects models were conducted along with moderating analyses.

Results: Findings revealed that gender, race, age, fear of crime, and victimization were statistically significant predictors of satisfaction with police. Moderating analyses revealed that certain variables, Hispanic ethnicity, socioeconomic status, and perceptions of crime, while not significant in the main effect size analysis, were significant in the moderator analysis.

Conclusions: It is important that future research establishes a more standardized form of measurement for satisfaction, with the consideration that confidence and trust may operate as distinct constructs. Additionally, it is imperative to move beyond investigating demographic factors alone and instead focus on variables related to procedural justice, performance theory, and neighborhood context.

Keywords: Satisfaction with policeTrustConfidenceMeta-analysis

5. Discussion

The results of this meta-analysis provide three insights. First, findings generally support previous studies regarding demographic effects. Second, findings suggest the need for inclusion of citizen experience and neighborhood perception variables such as fear of crime and victimization in studies examining citizen satisfaction. Third, findings highlight the need for more uniform and direct measurement of satisfaction with police.

As expected, gender, age, race, victimization, and fear of crime were statistically significant predictors of satisfaction. Consistent with most previous research, education and socioeconomic status were not statistically significant in either analysis. Surprisingly, prior police contact was not statistically significant despite being noted as a significant predictor in previous studies (Frank et al., 1996Frank et al., 2005Rosenbaum et al., 2005). This may be due, in part, to a relatively small number of studies which included the variable and potential interaction effects with other variables such as race, victimization, or fear of crime. Another potential explanation is that the measurement varies widely across studies. Perhaps prior police contact may not be influential, in some studies, measured as a binary prior police contact, but the quality and nature of the police-citizen interaction is important (Dai & Jiang, 2016Frank et al., 1996Rosenbaum et al., 2005Schafer et al., 2003Skogan, 2005Wells, 2007). A perceived negative interaction has a substantially greater impact on someone's global perceptions with the police, while a perceived positive interaction does not have the same impact (Dai & Jiang, 2016Schafer et al., 2003). It should be noted that the mean effect size for prior police contact was just outside the margin for statistical significance (p = .054). As such, future studies should include measures of prior police contact whenever possible in order to discern if it does make a difference in global perceptions of satisfaction with the police. Fear of crime, though relatively weak, was statistically significant in predicting satisfaction. Fear of crime and additional neighborhood perception variables, including physical and social disorder, should also be employed in studies surrounding satisfaction with police.

Another important finding is that some of the variables that were not significant in the general random-effects modeling mean effect size analysis were significant in the moderator analysis. Hispanic ethnicity and SES produced significant mean effect sizes when other correlates were included in the analysis. Hispanic ethnicity was related to police satisfaction when studies included measures of education, fear of crime, and victimization. SES was significant when race and prior police contact were measured. These findings could indicate that consistency in variables measured is an important consideration in studies of police satisfaction, and the model misspecification may mask important findings. It may be especially important to consider Hispanic ethnicity as a separate correlate from race given that Lytle (2014) found that Hispanic individuals were more likely to be arrested and Engel, Cherkauskas, Smith, Lytle, and Moore (2009) found that Hispanic individuals were more likely to be searched, but were significantly less likely to have contraband seized from the stop. As such, it is possible that Hispanic individuals may be more likely to have negative experiences with police.

Where the study occurs is generally not significant, which means that perceptions of police do not appear to vary based on whether the study is conducted in the United States or abroad or whether the environment is urban or rural. There were two exceptions to this from the moderator analysis. First, SES was significant in studies conducted outside of the United States, and Hispanic ethnicity was significant in studies that examined a mixture of environments relative to urban only studies and rural only studies.

Finally, a few measurement nuances were noted in the moderating analyses. Namely, gender produced greater effect sizes in OLS as opposed to probit and logistic regression, indicating that the level of measurement is important. Regarding the impact of SES, it was statistically significant when satisfaction was operationalized as trust only, but not with any other conceptualizations (confidence or general satisfaction). These findings suggest the need for more uniform measures of satisfaction with police. More uniformity in the conceptualization and level of measurement will aid in the inclusion of more studies in future reviews and will aid in our ability to make stronger conclusions about predictors of citizen satisfaction.

5.1. Limitations

There are important limitations to consider when interpreting our findings. First, there was wide variation in how the dependent variable was measured. This limits our ability to make stronger conclusions regarding the main effect sizes of various correlates with satisfaction. Indeed, moderating analyses suggested that there were differences in effect sizes based on conceptualization. We chose to include conceptualizations of trust, satisfaction, and multifaceted global measures of satisfaction to be as inclusive as possible. These varying measures are not necessarily invalid, but our analyses suggest that a more standardized measurement may be more useful, and as Cao (2015) suggested, satisfaction, confidence, and trust may indeed operate as distinct constructs. Second, there are other theoretically supported variables surrounding procedural justice theory and neighborhood factors such as concentrated disadvantage, physical and social disorder, and collective efficacy that were not included in our analyses. The absence of these variables was simply because too few studies included measures of these constructs to include in the analyses. We suggest that future research include measures of procedural justice, concentrated disadvantage, physical and social disorder, and collective efficacy whenever possible.

Finally, there were small sample sizes for some key variables we included, namely victimization, fear of crime, and prior police contact. The small sample size may have played a role in the lack of significance for prior police contact, and the fact that fail-safe N values for victimization and fear of crime were moderately stable. For instance, 25 of the 66 studies (37.88%) included victimization. Fear of crime was included in 21 of the 66 (31.81%). Even so, these variables retained statistical significance in random-effects models. Previous research has suggested that these key variables have predicted satisfaction above and beyond citizen characteristics, supporting theories surrounding performance theory and procedural justice. As such, it is imperative that future research includes these constructs when examining satisfaction with police.

The small sample size could be to blame for the lack of significance for prior police contact and Hispanic ethnicity. This is especially true for Hispanic ethnicity, given the significance of race. Of the 66 studies analyzed, only 15 (22.72%) examined Hispanic ethnicity, and 18 (27.27) included a measure of prior police contact. Future research should pursue attempts to measure these potentially relevant variables given the significance of analogous measures.

6. Conclusion

Despite our limitations, this meta-analysis provides important insights for research surrounding satisfaction with police. First, there is a consistent finding that, based on the studies analyzed, people that have been victims of crime are significantly less satisfied with police. This may indicate that police are doing a subpar job of serving the needs of crime victims. Moreover, this may be exacerbated among minority communities. In support of the majority of prior studies, race remains a statistically significant predictor of satisfaction with police. Based on the studies examined in this meta-analysis, police should make a concerted effort to improve the services they provide to these groups. For practitioners, as police departments across the United States struggle with community relations, particularly relations with African-American communities, these findings point to the importance of adding citizen experiences of victimization and perceptions of police contact as key measures for evaluating agency performance. While demographic factors should always be included in models predicting satisfaction, it is critical that other theoretically supported constructs such as fear of crime, victimization, and prior police contact are also included. It is important to note that the relationship between fear of crime and police satisfaction may be nuanced. Fear of crime was significantly related to police satisfaction in the main effect size analysis. However, based on the moderator analysis, when race, Hispanic ethnicity, or victimization were measured, fear of crime was not significant.

Citizen satisfaction with police has ripple effects regarding compliance with the law (Bolger & Walters, 2019Murphy et al., 2008Murphy et al., 2009Murphy & Cherney, 2012Sunshine & Tyler, 2003Tyler & Fagan, 2008). Additionally, research has indicated that citizen satisfaction impacts willingness to report a crime (Boateng, 2018Goudriaan et al., 2006Watkins, 2005). Given these two findings, increasing citizen satisfaction could improve policing efficiency in general. Based on the findings from this analysis, police have a significant problem with certain groups within society, non-white, younger individuals, those that fear crime, and victims of crime all expressed less satisfaction than their similarly situated counterparts. Moreover, the findings for race, age, and victimization were not conditioned on some other variable or a measurement issue. Considering the findings from this study, combined with those from other studies, it is reasonable to think that if police could improve relationships between those that fall into the dissatisfied categories, then they could increase compliance and citizens' willingness to report crimes. Police certainly need the community in this capacity, and increased cooperation could lead to a better ability to solve crimes (Braga & Dusseault, 2018Brunson & Wade, 2019Chaiken, Greenwood, & Petersilia, 1977Decker, 1996Regoeczi & Jarvis, 2013).

Second, there is consistent evidence that the effects of some theoretically relevant variables are conditional based on the presence of other covariates. Consistent with prior meta-analytic endeavors in research on policing literature (Bolger, 2015; Bolger & Lytle, 2018Lytle, 2014), model conceptualization and specification matters. The moderator analysis found differences among mean effect sizes based on the presence of covariates for gender, Hispanic ethnicity, SES, and perceptions of crime. It is essential for future studies that examine police satisfaction to include conditional variables in addition to those which tend to have statistically significant effects. Failure to include both could lead to misspecification of findings. Further, we recommend that police satisfaction literature should work to create more standardized measures as it is clear that there was a wide array of measurements for all variables involved.

In sum, as is the case with the police decision-making literature, a more precise measurement of the dependent variable and more inclusion of neighborhood variables and citizen encounter and experience variables must be employed in future studies. As nationally representative samples of communities and police departments are unlikely to be collected due to the cost and logistical challenges of such efforts, our understanding of what makes people more or less satisfied with policing services will be limited to localized samples unless researchers can agree on more unified measurement to facilitate more robust research synthesis efforts.

No comments:

Post a Comment