- Participants will be able to explain traffic stop data collection programs, traffic stop data, benchmark data, and benchmark surveys.
- Participants will be able to list at least four critical elements of a comprehensive data collection program.
- Participants will be able to identify the advantages and disadvantages associated with the use of various external benchmark data.
- Participants will be able to list the key components utilized when conducting similarly situated officer analysis using internal benchmarks.
In Commonwealth v. Lora, the Supreme Judicial Court of Massachusetts concluded (1) " that evidence of racial profiling is relevant in determining whether a traffic stop is the product of selective enforcement violative of the equal protection guarantee of the Massachusetts Declaration of Rights" and (2) "that statistical evidence demonstrating disparate treatment of persons based on their race may be offered to meet the defendant's burden to present sufficient evidence of impermissible discrimination so as to shift the burden to the Commonwealth to provide a race neutral explanation for such a stop". Further, the Court pointed to the way that data should be collected and analyzed. It is with this ruling as a backdrop that we turn to a discussion of data collection and analysis.
I. Data That Should Be Collected
The SJC pointed to New Jersey v. Soto as a model for data collection in a case that alleges racial profiling. In that case data about all traffic stops by a police department, not only those that ended in a citation, were collected. Collecting all stops is a better model to follow than just citations because it gives a complete picture of police activity and does not leave any ambiguity concerning the race/ethnicity of the motorists who were stopped but not cited. For traffic stops the data that should be collected are:(1) the date, time, exact location of the stop;(2) the infraction that caused the officer to make the stop;(3) the race/ethnicity, gender, age of the driver; (4) post-stop activity, which should include any request to exit the vehicle, searches, legal authority and reason for the search, as well as the outcome of the search; and (5) whether the stop resulted in a citation.
A. Advantages of collecting stop versus citation data.
- Data from all stops assures that all police activity that results in a stop is available for analysis. When all stops are collected, whether the department is found to be targeting one or more minorities or not, both police and citizens know that every stop was included.
- Collecting citation data alone is more convenient for police, particularly in departments that do not require warning tickets for motorists a police officer decides not to cite. However, previous research has found that as many as two thirds of stops are not cited.
B. Disadvantages of collecting stop or citation data.
- Collecting all stops takes more time and involves more work and possibly more danger for officers. When a policy of collecting all stops is instituted there has been a temporary decrease in the number of stops that police officers make.
- When only citations are collected, if the date shows race-neutral activity, the analysis of the data is open to the charge that the unrecorded stops were heavily minority citizens.
II. Data Should Be Collected With The Utmost Care And Accuracy
Data can be recorded on a paper form, as in a citation, or electronically via a data terminal or hand held device. As virtually every stop includes the driver's license, the gender and age of the driver is obtained from the license. Race/ethnicity should be the officer's perception of that variable, as it is sometimes inflammatory to ask a driver who has been stopped for that information. Likewise, in a situation where a driver's license is not available, the officer's perception of age and gender should be used.
III. Search Data Should Include Only High Discretion Searches
Any search that is mandated by the department,such as searches incident to arrest and inventory searches, should not be included in the analysis. Consent searches are the most discretionary, followed by "Terry pat/frisks" and probable cause searches.
IV. Data Analysis for External Benchmarks
A. Once the stop (citation) data has been collected it must be considered in light of the race/ethnicity, gender and age of drivers in traffic (or those drivers violating a law). There are several benchmarks that have been utilized.
- Unadjusted (or minimally adjusted) census data have been utilized most often and are a very poor benchmark against which to measure the stops. Census data sometimes underestimates and sometimes overestimates the race/ethnicity, age and gender of the traffic. The difference is sometimes slight and sometimes quite large. Census data should not be used as a benchmark.
- The Driving Population Estimate was utilized as a benchmark for urban suburban areas and observational data were collected for citations issued on highways by the State Police in the Massachusetts study of 2004. The Driving Population Estimate of the driving population was developed by adjusting census data utilizing certain assumptions (e.g., "push" of drivers out of a jurisdiction and the draw of drivers to jurisdictions). Many municipalities believed that the Driving Populations Estimates for their locality were inaccurate. It is presently unclear what position the SJC would take on the use of this estimate as a benchmark.
- Observational benchmarks of the driving population are developed by scientifically sampling traffic on roadways and observing the driving population. This means that traffic is observed in specific areas or on specific roadways at randomly selected times and days. The racial/ethnic make-up of the traffic (or those violating a traffic law) is compared to stops on those roadways or at those specific areas. This type of observation has been utilized in court cases in New Jersey, Maryland and Arizona and has been endorsed by the Massachusetts Supreme Judicial Court as a reasonable method for a benchmark.
B. Whichever benchmark is used, it is important that stops (citations) should be compared to the racial/ethnic, gender and age composition of the traffic which an officer sees as he/she is deciding to make a stop. From this start, the analysis can be conducted.
- Officers' stops are compared to the traffic in the location where the stop was made. Officers who work in an area that has a high concentration of minority motorists are not compared to traffic which does not have this high concentration. Similarly, the stops of officers who select motorists to stop from a traffic stream that is highly non-minority are compared to that traffic stream.
- Analyses should be conducted in specific locations comparing the benchmarks to the stop (citation) data for those areas. The specific analyses used should be carefully employed to assure that they are correctly used.
V. Data Analysis For Searches Is Not As Straightforward As It May Seem
While some have argued that a simple comparison of the race/ethnicity of those searched can serve as the benchmark for searches, that explanation is too facile. First there are the considerations of which type of searches should be analyzed. In many police departments there are mandatory searches. Generally speaking these are searches incident to arrest and inventory searches. As these searches are not discretionary they should either not be analyzed or analyzed in a special category. Searches that are discretionary are far more helpful in determining whether officers are targeting a specific group. Additionally, there are at least four special considerations for analyzing searches, and possibly more given the specifics of the police department in question.
- Where was the search conducted and what were the demographics of traffic in that location?
- How many officers were located in the area of the search?
- Was the officer serving in a special unit at the time of the search?
- Did the officer know whether the motorist was on parole or probation?
The specifics of the situation should be taken into account and a multiple regression analysis including the above variables is often best.
VI. Data Analysis for Internal Benchmarks
Internal benchmarks are utilized to compare the stops (citations) of officers to other officers who are similarly situated in the department.
A. This type of analysis is very helpful in determining which officers, if any, are "outliers" in terms of stopping more or fewer minorities than other officers in the department. NOTE: Outliers are individuals who lie at the extremes (either higher or lower) of the normal distribution. What this means practically in our context is that these individuals stop either many more or many fewer minorities.
- This analysis must be undertaken with great care to assure that officers are similarly situated with regard to the composition of the traffic stream from which they draw drivers to stop.
- This must include, at the least, the same patrol locations, the same shifts, and the same days of patrol, as all of these variables may influence the race/ethnicity, gender and age of motorists.
- Once similarly situated officers are determined, then the comparisons can be made and the data utilized by the department and the officers to evaluate those officers. Possibly these data are most useful in the context of an early warning system.
B. Internal benchmarks by themselves cannot determine if a department is targeting minorities as the comparison is officer to officer within the department. However, comparing similarly situated officers on their stops of minority motorists should be considered as an important part of an early warning system.
Downloadable Reference Materials
- Practitioners Guide file size 3MB
- Soto Case Statistical Analysis
- San Antonio Report, particularly pp. 42-47 for a discussion of search variables.
- Commonwealth V. Lora