Debates about racial inequalities in the criminal justice system have ramped up in the last several years. This has been largely driven by discussions about racially-biased police violence, but many also speculate about disparities in how the police treat crime victims. For example, one common belief is that the thoroughness of police investigations varies based on victim and officer race. But a recent study published in The Journal of Law and Economics (subscription required for access) suggests otherwise, at least in the context of residential burglary.
The study was conducted in 2016 in Tucson, Arizona by Rebecca Goldstein of UC Berkeley. That year, the Tucson Police Department (TPD) investigated approximately 35,000 crimes (4,000 of which were violent). With only 870 sworn officers to share the workload, not all investigations would receive the same level of attention, resulting in variation in investigative quality. To better understand patterns and correlates of this variation, Goldstein studied granular data from a sample of residential burglaries.
Methods
Goldstein sought to explore whether victim, officer, or neighborhood demographics had an impact on the investigative thoroughness of residential burglary investigations. Data was collected from TPD case files for 2,763 residential burglaries that occurred in Tucson in 2016. These files contained information about the initial incident, victim demographics, and notes regarding actions taken during the investigation. Officer demographics were obtained from employee records. Data on neighborhood demographics (at the census block-level) were pulled from the U.S. Census and linked to the addresses of each burglary.
Investigative thoroughness was coded using the contents of each case file. There were three main dependent variables, including officers’ time spent on the scene (generated from timestamps), whether the police dusted for fingerprints, and whether a detective was eventually assigned to the case. These measures were used because they are widely applicable to most burglary cases, whereas other potential measures of thoroughness (e.g., DNA collection, witness interviews) are not.
To estimate the causal impacts of victim and officer race on investigative thoroughness, regression models were used. The benefit of using regression models is that Goldstein was able to control for numerous factors that could be related to the outcomes (e.g., various case characteristics, as well as other fixed effects such as hour, month, and neighborhood), which helped eliminate alternative explanations for the findings. The main dependent variables included three indicators of investigative thoroughness: 1) officer time spent on the scene, 2) whether police dusted for fingerprints, and 3) whether a detective was assigned to the case. The secondary dependent variable was case clearance, though the analysis for this outcome was limited (see below for more details).
Sample Characteristics
51.1% of burglary victims were White, while 27% were Hispanic, and 5.1% were Black. These breakdowns are similar but not identical to the race/ethnicity distribution of Tucson’s overall population, which is approximately 47.7% White, 42% Hispanic, and 5% Black. For some reason, Hispanics are under-represented in the share of burglaries while Whites are slightly over-represented. In terms of neighborhood-level demographics (per the U.S. Census), the distribution of burglaries was fairly representative of Tucson’s overall population. 45.8% of burglaries occurred in areas comprised of mostly White people, while 42.8% occurred in Hispanic neighborhoods, and only 4.2% occurred in Black neighborhoods. When it comes to officers who responded to burglaries, 57.4% were White, 31.6% were Hispanic, and 3.9% were Black.
Burglaries were more concentrated in poorer areas of the city, with 29.8% of burglaries occurring in high-poverty neighborhoods. Coincidentally, they were often clustered within denser parts of the city, with 31.1% occurring in apartment buildings. The former is not surprising, considering that crime rates are always higher in poorer areas. The latter finding is probably because short distances between homes allows greater opportunities for burglaries to occur. In other words, it behooves criminals to learn the way into a housing complex with a large number of units. This mechanism would be consistent with past research on repeat and near-repeat burglaries, which is a phenomenon where subsequent residential burglaries occur within 1-2 blocks and within 1-2 weeks of an initial burglary.
Results
Victim and officer demographics had no impact on investigative thoroughness nor case clearance. Rather, forced entry was actually the most salient predictor of investigative thoroughness, which was significant for all three measures. Goldstein also incorporated different interaction terms into the models to explore whether there were any snyergistic effects between different couplings of officer/victim demographics and forced entry. Only two interaction terms were statistically significant. When examining the combined impact of forced entry, all white officers, and all white victims (via a three-way interaction term), Goldstein found that these cases were slightly less likely to have detectives assigned. Police also spent less time at scenes with all white victims.
Forced Entry
Of the cases examined, 64.5% (n=1,782) were classified as forced entry, meaning that the perpetrator entered the residence through a locked door or window (either by breaking it, prying it open, or using a tool to disable the lock). Forced-entry burglaries leave more physical evidence and are therefore somewhat easier to investigate. Similarly, when there is potential physical evidence to link a suspect to a crime, the case is viewed as more “solvable.” Cases that are “more solvable” tend to be prioritized over “less solvable” ones (especially for high-volume crimes like burglary) in order to maximize clearance rates. Thus, these cases are more likely to have detectives assigned.
Forced entry significantly boosted all three measures of investigative thoroughness, even when controlling for other factors. Fixed effects (i.e., census block group, date, and hour) helped explained some of the variation in investigative thoroughness, but forced entry was the strongest predictor, remaining significant even when additional factors were included in the model. Conditional on fixed effects, officers spent about 19% more time (14 additional minutes) at the scene of forced-entry burglaries relative to unforced-entry burglaries (averages of 88 and 74 minutes, respectively). Second, police were 67% more likely to collect fingerprints in forced-entry cases (45%) than unforced-entry cases (27%). Third, forced-entry cases were 49% more likely to have a detective assigned to the case (37% vs. 25%, respectively).
Investigative thoroughness also significantly increased the possibility that a case would be cleared. This suggests that forced entry had an indirect impact on case clearance via its impact on investigative thoroughness. A direct link between forced entry and case clearances could not be established, though, because forced entry and investigative thoroughness were too highly correlated. In other words, forced entry and investigative thoroughness were both predictors of clearance but their impacts were confounded with each other and could not be disentangled. Nonetheless, out of the 2,763 cases, a total of ~282 (about 10.2%) were eventually cleared.
Neighborhood
An interesting wrinkle is that the probability of forced entry varied significantly by neighborhood. Simply put, unforced-entry cases occurred more often in less-advantaged residences and neighborhoods. High-poverty neighborhoods, apartment complexes, and neighborhoods with a high renter share (more than two-thirds) were overrepresented among unforced-entry cases and underrepresented among forced-entry cases. These differences were all statistically significant. While the exact reasons for this are unclear, burglary investigators were able to speculate some potential reasons.
24% of forced-entry burglaries occurred in high-poverty neighborhoods, compared with 29% of unforced-entry cases. The most probable explanation for this is that people living in high-poverty areas are less likely to be homeowners. Therefore, their homes are maintained by property owners who have less incentive to install proper security measures than most homeowners would normally have. Consequently, these homes might be more run-down and lack proper security measures. This theory is also supported by the finding that 36% of forced-entry cases occurred in neighborhoods with higher proportions of renters (i.e., more than two-thirds), compared with 44% of unforced-entry cases.
Relatedly, 28% of forced-entry burglaries occurred in apartment complexes, compared with 44% of unforced-entry burglaries. This isn’t entirely surprising considering the finding above about neighborhoods with higher shares of renters. Additionally, TPD burglary officials speculate that burglars take advantage of multifamily residences because they can easily (and quickly) case many homes for unlocked doors and windows.
Discussion
The results are inconsistent with prior research showing that police are less likely to clear a homicide case when the victim is Black or Hispanic than when the victim is White. Perhaps though, this is due to a lack of cooperation with homicide investigators in communities where relations between the police and the community are strained, as some scholars have suggested. This debate highlights an important question: are the factors that largely determine clearance rates mostly outside of investigators’ control? If so, then clearance is not an accurate reflection of thoroughness. Relatedly, this would mean that disparities in clearance rates are probably not inherently driven by racial discrimination or malicious intent.
An advantage of the present study is that investigative thoroughness was measured directly using TPD records, without relying on clearance outcomes as a proxy. Importantly, this is different from a lot of the past research that measures thoroughness based whether a case was cleared. This might explain the inconsistent findings of this study, i.e., the result that the thoroughness of police investigations did not vary by victim or officer race, suggesting no evidence of racial bias. Many scholars, policymakers, and the media argue that the police fail to take crimes seriously when racial minorities are the victims. Finding that the police do not discriminate against racial minorities when they are residential burglary victims, then, represents an informative null result.
Conclusion
Many people claim that police do not investigate crimes as seriously when racial minorities are the victims. Goldstein’s study on residential burglaries found otherwise, though. There were no impacts of victim or officer race on investigative thoroughness, but forced entry emerged as a significant predictor of thoroughness. In other words, investigative thoroughness was similar across victims of residential burglary, conditional on whether the burglary involved a force or unforced entry. Through the impact on investigative thoroughness, forced entry likely had an indirect impact on case clearances, but this could not be confirmed in the current study.
However, the probability of forced entry varied significantly by neighborhood and across types of residence, such that unforced-entry cases occurred more often in high-poverty neighborhoods, neighborhoods with more renters, and apartment complexes. This resulted in some inadvertant disparities in investigative quality, but this was unintentional and driven by differences in forced entry.
These findings show that it’s possible that investigative thoroughness (and maybe clearance rates) are driven, at least in part, by factors outside of investigators’ control. This would mean that disparities in investigations do not necessarily indicate discrimination or malicious intent.
The post Misconceptions about racial disparities in police investigations: Results from Tucson appeared first on Crime & Consequences.