U.S. flag

An official website of the United States government, Department of Justice.

Research, Science, and Policing

Date Published
October 18, 2018
Gary Cordner
Geoffrey Alpert

Expectations that policing will be evidence-based and scientific have increased significantly in recent years.[1] The logic behind this trend is undeniable – no government agency should use practices that are ineffective, and police in particular should adopt strategies, tactics, and policies that achieve the most good and cause the least harm.

The National Institute of Justice (NIJ) established its Law Enforcement Advancing Data and Science (LEADS) Agencies program to help law enforcement agencies meet these growing expectations.[2] The program’s objective is to help agencies become more effective through better use of data, analysis, research, and evidence. Science does not have all the answers, however. Police leaders have to balance research and data with experience and professional judgment.[3]

Overcoming Bad Habits

Designing police practices on a solid foundation of research makes sense for many reasons, especially if one considers the alternative. If policing is done a certain way because “we’ve always done it that way,” it is likely that changing times have rendered that approach ineffective, even if it was working well at one time. If practices are based on someone’s opinion about what works, there is a good chance that selective perception, limited personal experience, and bias (conscious or unconscious) are contaminating decisions that should promote the public good. If a police agency is satisfied to do things just because others do it, the agency is probably settling for mediocrity.

Limits of Science

But research and science have their limits, some technical and others more philosophical. For example, a randomized controlled trial (RCT), called by some the gold standard of research design, maximizes internal validity, which is the confidence we have in its findings. By its very nature, though, the extent to which the findings of a RCT are generalizable to other settings (external validity) is unknown. This helps explain why arrest for misdemeanor domestic assault was the most effective of three options for reducing subsequent assaults in Minneapolis, but not in other jurisdictions.[4]

It is also important to remember that science never “proves” anything. Rather, it tests theories (formal explanations about how something works) by confirming or disconfirming null hypotheses, a fancy way of saying that all scientific knowledge is tentative. Even principles and “facts” that are relatively well-established are periodically subjected to further testing, and often debunked. Research on eyewitness identification, for example, led to revised practices and model policies for conducting in-person and photo lineups.[5] Double-blind procedures are now the industry standard and sequential presentation seems to have advantages over simultaneous, but further studies are sure to challenge whatever becomes the new status quo.

A more philosophical issue arises because policing is a function of a government that is “of the people,” not “of science.” We choose to have guilt or innocence decided by a judge or jury, not a computer algorithm. Scientists might someday develop a brain scanning technology that can accurately detect deception, but whether and how police are allowed to use that technology will be determined by public opinion, politics, and judicial interpretation of the Fourth and Fifth Amendments, not by research.

Applying Science and Research

Several considerations are essential for the proper application of science and research in police administration. One is constant appreciation for the multi-dimensional “bottom line” of policing.[6] A study may determine that one strategy is more effective than another at reducing crime, but police must also consider its effects on fear of crime, public trust, efficient use of resources, and equitable use of force and authority, not to mention key values such as legality, transparency, and accountability. Researchers have the luxury of focusing their studies on one isolated outcome, but law enforcement executives have to juggle multiple outcomes, all of which matter.

A mistake that police leaders should avoid is over-interpreting a study’s results. For example, a widely accepted conclusion from 1980s response time studies was that rapid response did not matter. More accurately, however, the studies found that an immediate response to cold crimes had little benefit, whereas quick reaction to crimes in progress was actually quite productive.[7]

Two other factors that police should weigh when considering the implications of research are context and purpose. Several studies have found foot patrol to be effective,[8] but its applicability for the Wyoming Highway Patrol or even the typical suburban department might be limited. Hot spots patrol is regarded as an evidence-based crime reduction practice,[9] but if an agency is trying to reduce identity theft or acquaintance rape, some other approach is probably needed.

Striking a Balance

Not surprisingly, law enforcement needs to follow the middle way. Building and testing a scientific knowledge base for policing is a high priority that will pay huge dividends in increased effectiveness and better public service. At the same time, all concerned need to recognize that police policies and practices are inevitably influenced by law, values, politics, and public opinion. One of the responsibilities of police leaders is drawing on wisdom and experience to make their agencies as rational and scientific as possible, given the multitude of challenges and considerations that inevitably constrain real world decision-making.

Action Items

  • Stay abreast of CrimeSolutions,[10] the Campbell Collaboration,[11] the What Works Centre for Crime Reduction,[12] and other sources of evidence-based police practices.
  • Carefully assess relevant studies to determine their applicability to your jurisdiction.
  • Beware of confirmation bias, which is the tendency to value studies that confirm what you already believe and reject studies that challenge your beliefs.
  • When necessary, conduct studies in your own agency to answer key questions about what works best in your context.[13]
  • Don’t be afraid of tinkering and trial and error. Measuring performance, trying something a little different, and then measuring again is the path to continuous improvement. Most programs and practices don’t work perfectly right away, and even if they do, they will likely need tweaking over time as conditions change.[14]
  • Always remember the multi-dimensional bottom line of policing – a practice that achieves one outcome but has negative effects on others is a practice ripe for improvement.

About the Authors

Gary Cordner is a former police officer, police chief, professor, and CALEA commissioner. He is also past editor of the American Journal of Police and Police Quarterly. Along with Geoff Alpert, he is a Chief Research Advisor for the LEADS Agencies program.

Geoff Alpert is a professor of criminology at the University of South Carolina. As Chief Research Advisors, he and Gary Cordner help administer the LEADS Agencies program.

Date Published: October 18, 2018