IMPROVE YOUR DATA
How to improve policing data
Communities have different kinds of access to data on their police department, such as how many stops officers make. Some departments don’t yet collect data in a useful way. For example, they might use paper records which cannot be easily analyzed. Or they might not include information about the racial group of the person they stopped. Other departments do collect data, but the data are missing a lot of important information and don’t meet CPE’s requirements to produce results for the Justice Navigator.
Data quality issues can happen when departments do not set complete, clear, and enforced standards for when officers should record their actions. Departments might also be limited by the forms and electronic systems they have in place for collecting data. These systems can be expensive and difficult to change. The many factors that contribute to data quality mean that there are a lot of ways to improve data on policing and public safety.
Below are key questions any community or department can ask to improve their public safety data, no matter what stage they are at.
1
Does the department have a complete policy on data collection?
Communities can advocate for comprehensive, transparent data collection to measure when police stop or use force on members of the public. Departments and policymakers should seek communities’ opinions on data collection by creating clear and easy-to-access avenues for input, such as through advisory boards or meetings with elected officials.
Complete data collection means that officers record each interaction they have with a member of the public where the person is likely to believe they are not free to leave. Officers should record specific information about that interaction that can help answer questions about public safety. The Justice Navigator has a complete checklist of what specific information officers should record about each of these stop or use of force incidents.
The following key components should be recorded in every record about a stop or use of force incident:
- Information about the person stopped. Officers should record their initial perception of the racial group, age, and gender (e.g., man, woman, non-binary) of the person being stopped or subjected to force. Demographic data based on perception may differ from self-reported demographic data, so analyses comparing these data to Census demographics using population benchmarking may underestimate the size of disparities overall. However, using the officer’s perception is broadly supported in social science research as the best way to assess disparities and potential bias. Using the officer’s perception also eliminates the burden on community members of being asked to share their racial background or gender identity. Communities that are concerned about police treatment of other groups of people, such as people with disabilities, should ask departments to collect information about these factors as well.
- Information about where the stop or incident happened. Details should include the time, date, and exact location.
- Information about what happened to the person during the stop. This should include the outcome of the stop (for example, the person received a citation or was arrested), details about any searches (the type of search, the reason for the search, whether or not the person consented to the search, and the search outcome), and details about any force (all types of force used; any de-escalation techniques used; any resistance encountered; and whether the person was injured, taken to the hospital or hospitalized, or died).
The following key components should be recorded every time officers respond to a call for service:
- Information about the call reason. Data should include information about the origin of each call for service, including who made the call, the initial call type, and the final call type.
- Information about where and when the call happened. Details should include the dispatch and arrival times, close time, date, and exact location.
- Information about what happened during or as a result of the police response to the call for service. Officers should record the outcome of each call.
Many communities are redesigning the systems that deliver public safety and creating alternatives to police response. These systems might include civilian-led crisis response teams or non-armed traffic enforcement agencies. Any public safety system should collect complete data, with these same standards, and in a comparable format to police data. This will allow communities to monitor whether alternative systems are achieving their goals of safety and equity.
2
Do officers understand how and when to record their actions?
Clear requirements for data collection are not enough to produce quality data. They need to be supported by good practices. Departments should train all officers on collecting data. Regular training should include: what information to collect, when to collect it, how to enter data into the computer system, and how the data will be used.
Officers should understand that they need to record any interactions immediately. If any paper forms are used, officers should know how to transmit them electronically at a later time so that they match exactly. Officers should also be required to record an option for each question in a data collection system. In other words, they should not be able to leave any option blank, but rather record “unknown” or “N/A,” so that it is clear when data is missing rather than when something simply didn’t happen.
3
Are officers held accountable for recording complete data?
Supervisors should review records to make sure that officers are completing them properly. This should happen regularly–for example, at the end of every shift. Departments should have clear procedures in place to check for data quality. A computer system can help with this checking, for example, by notifying supervisors when an officer has not submitted a stop data form by the end of their shift. Any errors discovered should be addressed immediately, and patterns of errors should be addressed systematically, including through re-training, policy changes, or disciplinary measures to prevent the same issue from happening again.
4
Has the department invested in the right data collection technology?
The computer systems used by police departments to record officer behavior are usually known as a Record Management System (RMS). The right RMS can make data much easier to collect and analyze. There is often a cost to update or upgrade RMS, but certain changes to these systems can have a big pay-off in data quality.
Communities can ask their department for the following qualities in an existing or new RMS system:
- Make important data fields required as forced-choice questions. The officer’s perception of the racial group of the person should always be entered in the record of any stop or use of force incident. This allows for measuring racial disparities in enforcement. Officers should not have an option to record “unknown” or “other” (without a corresponding description of what “other” means) in the field for the perceived racial group.
- Limit the use of free-text fields. Free-text fields allow an officer to type any response into the system, rather than choosing from a limited number of dropdown options. Because free-text fields are more difficult to analyze and create the opportunity for human error and typos, any time the system is collecting information with a limited number of possible answers, these categories should be listed in a dropdown menu.
- Block or flag data entry errors. Some computer systems can be set up to automatically detect and flag when data is describing an impossible situation (for example, a date of 1/32/22).
- Allow for easily improving the reporting process. Any RMS should allow for changing the options for report entry (for example, adding or removing value fields) without needing advanced computer science skills. If a lot of officers are typing out the same response in an “other” free-form category, for example it would be useful to add that entry as a standard option in the dropdown menu to minimize the possibility of typing errors.
- Invest in flexible and user-friendly reporting systems. If five people were stopped while walking at the same time by the same officer, the labor of recording details of each person stopped may discourage officers from complete recording. The RMS can make recording such incidents easier by having an option to automatically duplicate the location, time, and date for one person stopped to other people under the same incident number, and requiring officers to complete the remaining fields. Additionally, departmental data can be used in combination with many other types of data to drive more informed decisions about where communities should invest their resources. For example, calls for service data (such as 911 or 311 calls) combined with departmental stop data can answer a number of public safety questions. This combined data can show what types of activity the police are being asked to do, what types of activity officers are starting themselves, and how much of officers' time is spent on each of these types of activities. To combine this data, all data collection systems should make it possible to tell which stops are related to which calls for service. This can be accomplished either by using the same unique identifier (or number) to label the same specific incident, or if systems use different identifiers, including a column in the stop dataset that shows the calls for service incident number associated with each stop incident.
- Use geocoding technology for recording addresses. Geocoding takes a text-based description of a location, such as an address, and matches it to geographic coordinates. This information can help communities understand where police activity is concentrated. Including this technology in an RMS can help make address entries for incident locations more accurate. For example, the system can incorporate a free geolocating tool to automatically fill in a geocodable location when an officer enters an approximate or incorrect address.
What happens after improving data?
It is worth noting that departments who already publish data and then make improvements to their data collection may see a large increase in the number of total stops or use of force incidents recorded. For example, a department may only collect data on vehicle stops when a citation or arrest occurs. A more comprehensive data collection policy would require officers to also record vehicle stops where no arrest or citation is made, and therefore would increase the total recorded number of stops the department makes. This leap in numbers may look concerning. It is important for communities and departments alike to understand that these new numbers likely are closer to the number of stops that took place before the new data collection practices, rather than indicating an actual large increase in stops. In other words, good data practices can uncover the truth of what is already happening.
Resources
Putting policing data to work
Why policing data matters to safety and equity, and how the Justice Navigator helps make policing data useful.
What if my department doesn't have a Justice Navigator assessment?
If your department doesn’t have a publicly available assessment, you can still use policing data to understand disparities and drive the reimagination of public safety.