Sentiment Analysis: The Missing Link in Police Performance Management Systems

In the eyes of many, the profession of policing is facing a crisis of trust and confidence in what has come to be called the “Post-Ferguson” era. This crisis illustrates a vital lesson for local governments and their police agencies. To build public trust, each police agency must operate as an open system, using feedback as a learning loop for constant performance improvement, and become more responsive to public needs and mindful of the impact of policing efforts.

Measuring success in policing is complicated. What exactly is law enforcement’s role in a free society? By which performance measures is police work most appropriately measured?

Some believe police exist primarily to reduce crime by arresting offenders as the entry point for the criminal justice system. The heavy reliance on arrest as the primary strategy creates a “Catch 22”—the more the police focus on the location and frequency of serious crimes, the more their enforcement efforts fall upon certain communities, fueling perceptions of biased policing. On the other hand, one could argue it would be immoral for police to react to allegations or public criticism by not using the tools permitted by law to intervene in on behalf of those living in crime-plagued neighborhoods.

Since police do not have the capability to impact the root cause issues of poverty—lack of employment opportunities; dearth of mental health or addiction services, mass incarceration, and the like—the paradigm of police as “law enforcement” can entrap agencies. The more police drive down crime through arrest and incarceration and measure their performance solely by those outcomes, the more collateral damage can be done to public trust and confidence in those communities they are trying hardest to protect.

Ironically, the application of the very innovations used to increase law enforcement’s effectiveness in fighting crime—CompStat, broken-windows policing, hot spot policing, intelligence-led policing and so forth—appear to be damaging the police organizations’ relationships with many of those they serve. The problem is not the tools per se; these innovations have undoubtedly contributed significantly to the very low crime rates In many areas.

However, while decreasing crime is a good thing, it is not the only thing.

Policing services are best applied with a far broader view of the role of police. Law enforcement would be well served to follow other professions, like medicine, financing, and education, and take an “outcomes-based” perspective for measuring their performance—evaluating police services based upon how those services impact upon the communities an agency serves.

A broad role for was enforcement was reportedly espoused by Sir Robert Peel with the creation of the first modern police force in 1820s; reiterated by Herman Goldstein in his seminal work, Policing a Free Society, in 1977; and reinforced almost uninterrupted under the banner of community-oriented policing throughout the 1980s, to include the 2016 President’s Task Force on 21st Century Policing.1 The common thread throughout this history is the charge that police are most effective when they work with the communities, exist to reduce crime, fear, and disorder and to create communities safe for normal civic life to occur. Police help solve community problems. The question is, how does this vision become a reality?

It has been often said, “What gets measured gets done.” Therefore, for the police to work collaboratively with the public, the responsible use of data to identify emerging trends, and to measure the effectiveness of interventions is vital to continuing to enhance the effectiveness of crime reduction efforts and a broad array of police services.

As importantly, since police effectiveness is dependent upon the cooperation of the public, police must be perceived as being legitimate agents for justice. Law enforcement personnel and organizations must be viewed as being fair and unbiased in their decision-making; as treating people with dignity and respect; and as being judicious in the use of coercive authority and physical force. Note the subjective nature of these metrics—how people feel about safety in their neighborhoods and public spaces, what they feel to be the issues most adversely impacting on their quality of life, and how they feel about the quality of police services they receive are all measures of social sentiment.

As one considers the broad range of services police provide and law enforcement’s important role in creating safe and just communities, it makes sense to explore performance metrics that better measure the broader array of services: Are police effective in creating safe public spaces? How successful are the police at identifying and responding to emerging public order trends? Is law enforcement engaging with other community stakeholders to ensure important quality-of-life problems are addressed? These are but a few examples of measuring the outcomes of policing efforts, rather than simply measuring numerical outputs such as arrests, citations, and calls for service.

Sentiment Analysis: What and Why?

For the purposes of this discussion, “sentiment analysis” refers to the process of gathering and analyzing available data in order for decision makers to have a well-informed understanding of the issues of greatest concern within the various communities being served.

The very best kind of data is, of course, the understanding and insights that are the by-product of trust-based relationships with as large and diverse a spectrum of the public as possible. These relationships enable a constant feedback loop to allow decision makers to refine their understanding by asking probing questions. Further, when trust is genuine, these relationships become vehicles for collaboration in solving the problems identified.

Community surveys and feedback from community outreach efforts are the next best type of sentiment data. These methods provide insights about the issues and concerns from members of the public who may be less satisfied or less trustful of police and government than those with whom leaders regularly consult. By learning concerns from those less engaged, officials can carefully examine police activities to see if there are opportunities to improve outcomes and enhance trust. Knowledge about community frustrations is vital. Frustration, especially when it arises from perceived injustices and unfulfilled needs, are common precursors to unrest and disorder.

The truth is, however, that there is a level of trust involved in someone simply completing a survey. Those who distrust government and police and feel alienated from the larger community, often will not participate at all. Many people have no confidence that their efforts will yield positive outcomes, and some may even fear their participation may be used against them.

When alienation is high, people do not talk to government or participate in its processes—but they do talk about it.

Through the analysis of publicly available sources of data, such as social media postings, it is often possible to begin to identify those issues of greatest concern to those less engaged through other means—to hear those voices who are often overlooked. There is great opportunity to be gained by learning to understand those concerns and be responsive to them, and great peril in ignoring those concerns.

Sentiment analysis is about far more than measuring how people feel about police. Sentiment analysis is about understanding the underlying community narratives. Sentiment analysis seeks to identify the unfulfilled needs, underlying fears, and resultant frustrations that impact how people feel about their public spaces, neighborhoods, and the police and government officials who serve them.

By understanding those narratives, officials have a chance to earn trust by seeking to help fulfill those needs and resolve the related fears and frustrations. Having gleaned insights into those issues of greatest concern, officials have the opportunity to demonstrate caring and responsiveness and to earn a measure of trust in the process.

Public trust and confidence is earned when the police are both effective and act with integrity. Measuring performance outcomes can help accomplish both goals. When a police agency holds itself accountable for the outcomes of their policing efforts, by taking the time to measure those outcomes, and by adapting police activities to further improve service, the agency earns vital public trust and confidence.

Components of Sentiment Analysis

When attempting to measure the outcomes of policing efforts—how those efforts have impacted on how the public feels about their quality of life or their trust and confidence in the police—there are three key components of social sentiment: quality of life, quality of services, and perceptions of procedural justice (legitimacy).

Quality of Life

Community members are concerned about crime and disorder and a host of other non-criminal conditions that impact their sense of safety, security, and well-being in their neighborhoods and public spaces.

Feedback on crime and its related disorder is of most direct concern to the police. When considering public sentiment about crime and disorder, public concerns will typically fall in one of three categories—incidents, trends, and actors.

First, community members often have concerns about specific crime incidents or events. When something bad happens in the community, it can change how the community members feel. The incident may have undermined their sense of security and safety in their homes or when going out in public spaces. As a result of the incident, they feel less safe and are therefore less likely to engage in normal civic activities.

Community members also often know who is driving the crime and disorder in their neighborhood, and what those drivers are up to—sometimes far more so than the police. By becoming aware of sentiment around trends, police can establish an “early warning system” to enable them to respond to emerging trends in a timely manner, before the emerging trend becomes an entrenched crime hot spot.

In some cases, community members feel safe and comfortable enough to report information on local criminal elements to the police; often, they do not. However, in the age of social media, people will frequently vent about those groups or individuals who are causing troubles. When these vents are posted on publicly available sites, which is often the case, social sentiment analysis can yield tips that might serve to assist police investigators address patterns of criminal conduct.

Community concerns that are not directly related to crime are often still first reported to the police, since law enforcement is the only government service that will respond to concerns 24/7, 365 days a year.

Quality-of-life issues reported to police can include noise; loitering; drug activity, addiction, and disease; traffic and parking problems; graffiti; prostitution; property disrepair; and blight or infestation. For many of these community concerns, police have a vital role to play beyond enforcement, which might not be the best solution to mitigate or eliminate many problems in the longer term. Police are representatives of government, so when an issue of community concern is raised, police have an important problem-solving role to play. Their role is not to take ownership or solve the problem alone, but to identify and mobilize the relevant stakeholders. When called upon to assist with issues outside of their immediate job descriptions, police have the opportunity to earn public appreciation and trust.

Quality of Service

Measuring the outcomes of police efforts also includes obtaining feedback on how people feel about the quality of the services they receive. Do members of the public believe their local police to be effective at their jobs? Do they feel their police are responsive to their needs?

At times, the flash point of tension between police and communities is created by differing expectations of the police’s role.

When crime rates become the bottom line of policing, many law enforcement organizations will believe themselves to be highly effective when they have reduced crime rates and calls for service. Police often use data on reported crime and calls for service to show reductions as means of arguing their effectiveness and use measures of police activity—arrests, citations, calls for service handled—to argue their responsiveness. However, while there are many reasons why not all crimes or quality-of-life concerns get reported to police, a lack of public trust and confidence is one important variable.

At the same time police are arguing their effectiveness, members in those communities will argue that the police are ineffective, citing persistent, ongoing crime and disorder trends within their neighborhoods—quality-of-life concerns that have gone unaddressed. For police to be optimally effective, there must be measures of sentiment.

For example, if a city has a problem of retaliatory, gang-related violence that impacts a particular community, police will typically develop a response plan that might include high-visibility police presence, targeted investigative work, covert surveillance for criminal conduct, and even collaborative engagement with other stakeholders to interrupt the violence and arrest those involved. Typically, these interventions will drive down reported crime, although sometimes only in the short term. Sentiment analysis provides a feedback tool to allow officials to gain a perspective on the outcomes for the impacted communities. Did the intervention make people feel safer? Did police action reduce public fear? Has the intervention increased confidence in the police? Or, the reverse—has the intervention had unintended consequences? Do community members feel the police response was heavy handed or biased?

Until law enforcement agencies measure the outcomes of their interventions from the perspectives of those impacted, it’s impossible to know if they really are effective at creating safer communities.

Perceptions of Procedural Justice

The third component of sentiment analysis is perceptions of police legitimacy and procedural justice. In other words, do people feel the police treat them with dignity and respect, in a fair and unbiased manner? Do they feel that the police are judicious in their use of coercive authority (including physical force)?

Thanks to the work of researchers like Tom Tyler, Tracey Meares, and others, there has been much discussion about police legitimacy and procedural justice in recent years. In short, studies have shown that where the answers to the preceding questions are all positive—when police provide people with the chance to explain their perspectives (voice), treat people well (respect), make decisions in an unbiased manner (fair) and are honest and transparent in their actions (trustworthy)—police will be viewed as treating people with procedural justice. People will be more likely to view police acting in procedurally just ways as being legitimate agents of government. When police are viewed as legitimate and acting in procedurally just ways, research has shown that people are more likely to cooperate with the police and, even more importantly, are more likely to obey the laws absent police presence. This correlation makes public perceptions of police legitimacy pivotally important.2

Since increasing voluntary compliance with the law and increasing public cooperation with police are crucial to creating safer communities, measuring public perceptions of police legitimacy becomes critically important. Are officers treating people with respect? Are they fair and unbiased in their decision-making? Are they giving people voice by offering community members the chance to say their piece? Are officers perceived to be honest and trustworthy? These are all highly subjective, highly variable from individual-to-individual and community-to-community, but are all outcomes of community-police interactions developed over time. It is incumbent upon police agencies to find ways to get this feedback from their communities. Once collected, these are data that must be analyzed and incorporated into law enforcement performance management systems.


Improving policing elevates the police and serves the communities need for safety and justice. As Peel reportedly said, “The police are the community and the community are the police.”

Police performance management systems, and the use of data to direct policing efforts have created huge advances for the profession. Programs like CompStat, evidenced-based policing, and intelligence-led policing have permitted law enforcement to dramatically improve their effectiveness in reducing reported crime.

Reducing crime is a good thing, but it’s not the only thing.

By incorporating sentiment analysis, it is possible for police personnel and government officials to further measure the outcomes of their work by evaluating their effectiveness from the perspective of those impacted. By measuring the outcomes of policing efforts, creating a feedback loop, and using this feedback to refine their work to improve outcomes, law enforcement organizations can improve both their effectiveness and their responsiveness to public concerns, earning public trust and confidence in the process.



1 Law Enforcement Action Partnership, “Sir Robert Peel’s Policing Principles”; Herman Goldstein, Policing a Free Society (Cambridge, MA: Ballinger Publishing Company, 1977); Final Report of the President’s Task Force on 21st Century Policing (Washington, DC: Office of Community Oriented Policing Services, 2015).

2 See, for example, Tom R. Tyler, “Procedural Justice, Legitimacy, and the Effective Rule of Law,” Crime and Justice 30 (2003): 283–357; Tracey Meares, “Policing and Procedural Justice: Shaping Citizens’ Identities to Increase Democratic Participation,” Northwestern University Law Review 111, no. 6 (2017): 1525–1535.