A year after Michael Brown was shot and killed in Ferguson, Missouri, the public continues to focus attention on American policing like never before. The sustained scrutiny and calls for change—fueled largely by egregious police encounters captured on video and then widely shared on social media—have already generated a series of reforms. New York will use special prosecutors to investigate some officer-involved homicides, body cameras are being adopted by agencies around the country, and the concept of “guardian policing” is gaining traction among police executives, instructors, and front-line officers. These and other reforms are expected to contribute to better police-community relations, as well as increase the public trust and police legitimacy that communities deserve and that effective law enforcement relies on.

But it won’t be enough.

Police violence is at the core of the tension between police agencies and the communities they serve, but the U.S. simply doesn’t have the information it needs to make informed policy decisions or provide officers with thorough, evidence-based training. Consider just a few of the questions that could help develop better training, equipment, and policy: How often do officers successfully deescalate tense encounters, avoiding violence? When officers resort to force, are there predictable factors that can help them properly calibrate the amount of force in any given situation? Are there common elements that increase the chances of officers shooting unarmed suspects? How often and why are tactical units, such as SWAT teams, used? Do tactical teams encounter more or less resistance when controlling for other factors? After public protests, do civilians resist more (or less) and do officers use more (or less) force? Does the reason for the protest matter? If there is some effect, how long does it last?

The answers are unknown. Sure, there are a plethora of anecdotes about individual encounters, but war stories cannot be the foundation of important policy decisions. All too often, such anecdotes are unrepresentative or inaccurate. Consider a relatively benign example. It is often considered common knowledge within law enforcement that the full moon brings with it an increase in crime and psychological problems, making it an especially demanding time for officers. As a result, training officers and veterans tell rookies to be especially alert, while some agencies have been known to increase the number of on-duty officers to deal with the so-called “lunar effect.” But the common knowledge is wrong. Statistical studies have shown that the lunar effect is a myth; people don’t commit more crime or act any more bizarrely than usual. So those extra resources are wasted responding to something that does not exist.

When the public, U.S. elected officials, the courts, and even law-enforcement agencies have an inaccurate understanding of the police environment, they fail to make well-informed policy choices. They react to the world that they think exists, rather than the reality. But taking things for granted can skew the resulting rules and policies. Worse, it can have dire consequences. This is painfully evident in the criminal-justice context, where defendants have been convicted on the basis of deeply flawed forensic techniques such as bite-mark analysis, hair and fiber comparison, toolmark analysis, and ballistics. In a 2009 report, the National Academy of Sciences strongly criticized these and other techniques, pointing out that “many forensic tests … have never been exposed to stringent scientific scrutiny” because “researching their limitations and foundations was never a top priority.” The fact that hundreds of convicted inmates have been exonerated by DNA—one of the few exceptions to the hokum that saturates traditional forensic sciences—has proven other investigative techniques to be less reliable than frequently assumed, including eyewitness testimony, photographic line-ups, and interrogations.

The use of force is no exception—like forensics and investigative procedures, the use of force has been both plagued by an appalling lack of thorough, empirically grounded scientific review, and tainted by junk science. A recent New York Times article presents one example in the form of William Lewinski and his company, the Force Science Institute. (Two of us, Jeff Noble and Geoff Aplert, have worked with and against Lewinski on several cases.) Lewinski raises interesting and important questions about police behavior and reaction time.* The questions he asks are important enough to deserve answers, but those answers should be developed by applying human-factors engineering, cognitive psychology, kinesiology, and other methodologies that Lewinski—a psychologist—would seem unqualified to offer.

Without the appropriate methodology and rigor, use-of-force research is left in the hands of an entrepreneur borrowing concepts from areas of sophisticated science and applying them in questionable ways by referring to data developed in disputed studies. But those concepts sound good to the non-scientific layperson, judge, and juror, which may explain why officers pay $1,500 each to attend a Force Science Certification Course. It’s also how Lewinski is able to peddle his dubious conclusions as an expert witness, testifying almost exclusively on behalf of officers.

Like everyone else, officers deserve a robust defense, and rules should account for the fact that officers can and will make reasonable mistakes in use-of-force situations. In a similar vein, expectations must be tempered by a realistic understanding of the job that officers do. However, when an officer’s unreasonable and egregious actions go not just unchallenged, but endorsed by someone dressed in the guise of scientific validity, it casts a long shadow on the legitimate actions taken by good officers.

This is not the only available option. The U.S. could change course by identifying and promoting industry-best practices. This can be achieved by requiring all police agencies to release annual use-of-force summaries and to provide state and federal authorities with detailed information. In addition, change could be built on a more accurate and comprehensive evidence-based understanding of use-of-force situations, with research conducted by scientific teams combining multiple areas of expertise. And to achieve that, state and federal governments could make funding for such research a top priority. By ensuring informed policy positions, the U.S. would better protect the police and the public, mitigate the costs of uncertainty, and identify and meaningfully reduce failures and abuses that undermine public trust in policing.

* This article originally stated that Lewinski's ideas on police behavior and reaction times had not been subject to academically rigorous evaluation, nor tested through the peer-review process. In fact, Lewinski has published more than a dozen peer-reviewed articles and presented at academic conferences. We regret the error.