Is the police "legitimacy crisis" legitimate?

I can't tell what "police legitimacy" means or if it matters.

Over at National Review, Charles Fain Lehman has a good article criticizing activists and academics who have decided to pin the rise in crime on a loss of “police legitimacy.” Lehman gives the laundry list of people pushing this theory in the wake of the rise in murder and summarizes the “legitimacy crisis” argument:

Law professor John Pfaff has argued that violent crime “is the product of anger at police forces that kill far too many Black men, as well as at the remarkably violent, riotous way the police responded to last summer’s protests.” Criminal-justice-policy analyst Jonathan Blanks advanced a similar notion last summer, writing that the protests show that the police have lost public trust — a loss linked to rising crime. Thomas Abt, a senior fellow at the Council on Criminal Justice, has explicitly argued that reduced trust has made violence more common.

Each of these arguments invokes the “legitimacy hypothesis,” which holds that the primary reason people follow the law is not that they fear punishment but that they believe in its “legitimacy” — its deservingness of respect and compliance. In this view, police actions, such as killing innocent people, that reduce their legitimacy in the public eye weaken the rule of law and therefore lead to more crime.

Lehman notes the problems with legitimacy theory, chief among them the fact that the quantitative evidence — as seen in polling and 911 calls — doesn’t support it very strongly. People’s perceptions about police, as reported in polling, are stable over time. And because very few people have any actual interaction with police, “many Americans’ judgments about the police’s legitimacy are a by-product of a handful of rare but emotionally powerful stories.” To the extent police legitimacy is a problem, Lehman says, it’s partly because anti-police politicians have intentionally de-legitimized police.

I think Lehman is basically right about this, and I try not to write much about things that other people already have addressed. But I have questions about the basic concept of police legitimacy and the mechanisms by which it supposedly works that I’ve never really found a good answer to. It’s possible this is because I’m a dumb ex-cop and not a criminologist or economist so if you have good answers to these questions, please email me. I usually feel pretty strongly about the opinions I share, but here, I’d say my epistemic confidence is low.

I. What makes a police tactic legitimate?

At first, the answer to this question seems obvious: if police tactics are widely disliked by the public, police will be less effective. This makes intuitive sense. If you have a lot of brutal or corrupt cops going around stealing from people and beating them, people won’t feel inclined to call 911 or provide useful information. They might physically assault police and flee from arrests or detentions, and controversial incidents might explode into riots when they otherwise would not have. I’ll call this “legitimacy-as-popularity” — when people like and trust the police, the police are legitimate.

But we immediately run into conceptual problems here, because police legitimacy cannot be based on popularity without creating troubling issues. Let’s imagine the San Francisco Police Department adopts an unofficial policy of harassing and arresting anyone and everyone who comes within city limits to hold a pro-Trump rally. And let’s imagine 60% of San Francisco residents approve of this policy because they fear Trump supporters rallying in San Francisco will engage in violence.

Is it legitimate for the police in San Francisco to illegally arrest people just for holding a rally? That can’t be right. People have a right to protest. But if legitimacy is just a function of popularity, the San Francisco police become more legitimate here by arresting protesters just for protesting.1 In fact, they risk losing legitimacy if they don’t arrest the protesters, because a majority of people who reside in San Francisco will feel the police department can’t be relied on to protect them against violent people.

Another conceptual problem: if legitimacy stems from public approval, then saying something police are doing is “illegitimate” becomes a circular argument. By criticizing the practice, you make it less popular and delegitimize it. Nothing stops you from endorsing it and making it more legitimate instead. Police could theoretically legitimize any tactic with an effective propaganda campaign.

So perhaps legitimacy isn’t about popularity — perhaps it’s about respect for legal rights. This solves my San Francisco hypothetical; everyone has a First Amendment right to protest which police are obligated to respect. Perhaps by following the law and equally enforcing the same schema of legal rights for everyone without favor or bias, police gain legitimacy. Call this “legitimacy-as-legal-rights.”

Upon further examination, I think this also falls apart. For one thing, I don’t believe most people even know what their legal rights are — many times I detained someone based on reasonable suspicion, and they would demand to know what my probable cause was. Then they’d get angry when I told them I didn’t need any because they weren’t under arrest. Like most people, they did not have a firm understanding of Fourth Amendment jurisprudence on this topic.

For critics, the legality of a police tactic is not sufficient to make it legitimate. Let’s look at the example of pretext stops. A pretext stop happens when police pull over a car for a minor traffic violation in order to investigate some other crime. Critics of police hate pretext stops and frequently call them illegitimate. Jonathan Blanks has written an entire law review article about pretext stops called: “Thin Blue Lies: How Pretextual Stops Undermine Police Legitimacy.”

But pretext stops show legitimacy isn’t about respect for legal rights at all because they are, according to the Supreme Court, entirely legal and constitutional. And they always have been. So why does Blanks think they’re illegitimate? He grudgingly concedes pretext stops are legal, and then falls back on popularity:

…white and black drivers generally felt the traffic safety stops were legitimate because they knew they were pulled over for speeding and were most often treated in a way they viewed was fair. However, when the stop was for a minor infraction and led to the officer asking prying questions and requesting to search the vehicle, the stops engendered hostility and resentment among all races, but particularly among African Americans and Latinos—who were stopped much more often for investigatory purposes—whether or not the officer was polite and respectful.

Inherent in this argument is the implication that if drivers believed pretext stops were legitimate and widely approved of them, pretext stops would be legitimate and fine for police to do. But now we’re just back to legitimacy-as-popularity. What if a majority of drivers across all racial groups approved of pretext stops?2 Then pretext stops would be both popular and legal. Yet I suspect that Blanks would still say pretext stops are illegitimate. Why?


One possibility is that there’s some kind of “natural rights” argument in play here — people have a natural right to be free from traffic stops for no good reason, and such a stop inherently offends all free people. But we know this isn’t correct, because virtually every other developed Anglophone democracy permits police to stop a car for literally no reason at all.3 Police in these countries can just pull cars over to see if the driver has a valid license or force them to take a breath test— which makes sense, given that driving is a dangerous and highly regulated activity that requires a government license. As far as I can tell people in other countries have no issue with totally suspicionless traffic stops.

A second possibility is that pretext stops are illegitimate because they seem “tricky” — there’s something about saying “I’m stopping you for X” when you’re really interested in Y that’s just inherently dishonest (“blue lies”). I’m not sure you can actually separate this from legitimacy-as-popularity, and I’m also not sure why “I’m stopping you because I can” would be seen as legitimate when this is not. But if it’s really just “trickery” causing the legitimacy problem, we can resolve that issue by changing the law to allow for totally suspicionless traffic stops. No pretext is necessary if the police don’t need a reason to stop cars.

I think what’s actually happening here is that people are just using “X undermines legitimacy” as shorthand for “I dislike X.” Blanks is a libertarian; he would be against pretext stops even if drivers across all races overwhelmingly approved of them. He would not be happy if the United States allowed for suspicionless traffic stops even if most people thought they were legitimate. He generally believes we should expand individual rights. That’s fine. Living in a liberal democracy means balancing competing demands for liberty and security. But this has nothing to do with “police legitimacy.” It’s just a policy preference.

I sometimes see the word “legitimacy” used in even more incoherent contexts. Last year, Professors David Pyrooz, Justin Nix, and Scott Wolfe wrote an op-ed about how a police “legitimacy crisis” brought on by George Floyd’s murder was leading to de-policing and violent crime in Denver. I agree that de-policing is a big problem leading to more crime. But I’m not sure what legitimacy has to do with it. It seems like the authors threw the word “legitimacy” into the editorial so they could sound the alarm about de-policing without being accused of “blaming” protesters for the rise in crime.

All this to say: I still don’t know how to tell what makes a police tactic “legitimate” or “illegitimate.” It seems like popularity has something to do with it, and legal rights have something to do with it. Perhaps I’m an idiot for not getting this, but conceptually, legitimacy seems incoherent.

II. Does police behavior even affect legitimacy?

Blanks has also argued that even if police are not responsible for declining police legitimacy, it’s the responsibility of police departments to fix it. Lehman is skeptical, noting that claims about a collapse in police “legitimacy” don’t seem to be backed up by a lot of data. For example, in the aftermath of George Floyd’s murder, the number of 911 calls4 to police didn’t really change at all.

For me, seeing this data creates an even broader concern. Does anything police do — individually or as an institution — even affect citizen perceptions of legitimacy? To me, these graphs strongly suggest that basically nothing police do changes perceived legitimacy. If Derek Chauvin can murder a dude on video in Minneapolis, be seen doing it on every Twitter feed and TV channel in the country, and people in Minneapolis keep cooperating with police just as often as they did beforehand? I think that’s devastating for legitimacy theory.

This finding is something of a trend. Lehman references studies from Michael Zoorob and Roland Fryer which also find that high-profile instances of police misconduct have no lasting effect on citizen reporting to 911.

Even if high-profile media events involving police misconduct have no meaningful impact on police legitimacy, perhaps the low-profile, day-to-day way that police officers treat citizens does. But here too, I am skeptical. Check out this chart:

I don’t know anything about Lincoln PD, but what’s interesting to me about this data is how almost nothing changes. There is, across over twenty years of data, a very slight upward trend in citizens reporting officers treated them fairly. The gap in racial perceptions has never changed. Are police in Lincoln doing everything the same way they were two decades ago? I doubt it. Think about how much policing changed in general during this time period. New case law, the invention of the Taser, body cameras, cell phone cameras, the 1032 program, better information technology, social media, etc. Apparently, none of it changed how people believed Lincoln police were treating them — no big improvement, no big decline.

I also spend a lot of time thinking about Gallup’s annual “confidence in institutions” poll. Every year, Gallup asks people how much confidence they have in a variety of public and private institutions, including police:

Again, we see a remarkable amount of stability here. But there is a slight decline in confidence, hitting an all-time low of 48% in 2020. At first glance, this makes sense for obvious reasons because of Floyd’s murder and the ensuing riots.

But in objective terms, American police today are much better than police in the 1990s. Police in the 1990s used far more deadly force and were much less accountable. Violent crime was higher. There’s been an all-around dramatic improvement in policing since those days — both uses of force and crime have declined. So police in America today are better than they’ve ever been, and yet are apparently still losing ground when it comes to legitimacy. Once again, it seems the actual quality of policing isn’t having any effect on perceived legitimacy.

A tangential point: slight decline aside, American police remain one of the most trusted institutions in the country. Americans trust the police more than any other national institution except for the military and small business.

Is legitimacy not a problem for these other institutions? We don’t hear much about the collapse in trust of public schools, or about a “media legitimacy crisis.”


III. Is legitimacy really more effective than deterrence?

The key idea of legitimacy theory is that fair policing increases legitimacy which in turn reduces crime. Sometimes those who are writing about police legitimacy assert that legitimacy is an even more effective crime reduction tool than deterrence. Blanks, for example, cites a paper by Tom Tyler “noting that the results of one empirical experiment that showed the influence of legitimacy was about five times greater on compliance than deterrence.”

Taken to the extreme, the idea that legitimacy matters more than deterrence is obviously untrue. The Chinese police presumably have very little legitimacy in Hong Kong, for example. And yet the rioting and protesting that was happening there a year ago are over and the Chinese government definitely still runs things. They have deterred people just by arresting them and using force. So clearly, if you just go hog-wild on deterrence and punishment, legitimacy is totally irrelevant.

Of course, we’d rather live in a democracy and not a totalitarian state, so we don’t want to do what the Chinese do. But that seems like a political choice rather than a necessity. We’d prefer not to coerce because we like freedom, so we try to have a set of laws and rules that are legitimate and fair in the eyes of the people who live here. And we aren’t willing to use the kind of violence the Chinese are, so it’s probably true that without legitimacy we’d have anarchy. But that’s not the same thing as saying that coercion doesn’t work. It definitely works. We just don’t want to rely on it.

And I think there’s another problem here, which is that at some point the absence of effective coercion undermines police legitimacy. We see this extreme in countries with very weak systems of law enforcement, like Somalia. But I also saw this all the time on a much smaller scale as a cop. I’d go talk to some shopkeeper and ask him what was going on. He’d tell me about some guy coming in and stealing from his store every single day. “Why not call and report it?” I’d ask. “Well, I used to, but every time I did nothing happened because it took you an hour to get here.” Without enough coercion — arrests, stops, etc. — law enforcement also loses legitimacy.

Assuming that how police operate actually has an impact on whether they are perceived as legitimate, it seems what we really want in policing is a balance of coercion and legitimacy-inspired cooperation. We want people to buy into the law and comply voluntarily, but enough coercion to deal with the minority for whom that won’t suffice. Too much reliance on either, and the system breaks.

So are we relying too much on coercion, or too much on cooperation? At this point, many police departments have adopted models of procedural justice policing aimed at improving legitimacy, so there are now a fair number of studies on how that affects crime. They aren’t super promising. A 2013 DOJ review of the research concludes:

Despite most individual studies showing a null effect for reoffending, the meta-analysis showed that the interventions overall resulted in a decrease in reoffending that was marginally significant at the .05 [p = 0.053] level. The weighted mean g for the 26 evaluations combined was −0.07 using the random effects model (see Table 2), and the 95 percent confidence interval included zero at the very upper limit (lower limit = −0.14, upper limit = 0.00).

The DOJ puts a positive spin on this, but “there’s a good chance legitimacy-enhancing policing tactics have no effect on crime at all” doesn’t strike me as a super persuasive result. RAND Corporation agrees:

…there are insufficient data on whether legitimacy policing efforts lead directly to improvements on residents’ perceived legitimacy of the police. Similarly, there are insufficient data on whether these efforts lead to improved legal compliance and less crime.

We can contrast this mixed-at-best evidence with the overwhelming body of evidence suggesting that proactive policing strategies and hiring more cops are successful at deterring crime. When Tom Tyler says that relying on deterrence to maintain social order is “expensive, inefficient, and psychologically naive” he’s running up against a huge body of economic literature repeatedly finding real-world results like “each dollar we spend on police reduces the cost of crime by $1.63.” There probably is a point at which coercion becomes inefficient, but I don’t think we’re there yet.

Closing thoughts

I suspect some arguments about “legitimacy” — especially those that associate it with less policing — are pushing philosophical preferences under the guise of crime reduction. But I also suspect some arguments about “police legitimacy” are driven by a desire to show that police doing good things — like being polite during traffic stops — will also reduce crime. A DOJ paper for police departments, for example, gives an anecdote about a suspect who was going to shoot an officer, but decided not to because the officer had always treated him with “general mutual human respect.”

It would be great if treating suspects with mutual human respect reduces crime and improved officer safety. But even if it doesn’t, it’s a good thing that police that should do anyway. And as the DOJ paper notes — it’s what good cops have always done.


One can imagine even more problematic issues with “legitimacy-as-popularity” in cases involving racial and sexual minorities that a majority wants to see persecuted.


This is probably actually true. Among drivers who were actually searched or arrested by police during a traffic stop in 2015, 70% of white drivers, 68% of black drivers, and 62% of Hispanic drivers said they thought the police behaved properly. (2015 PPCS, Table 14)


See: Canada (“The police can stop cars at any time…”); the UK (“A police officer can legally stop any vehicle at any time…”); Australia (“…since the introduction of the random breath test, a police officer can pull you over anytime they’d like…”); and New Zealand (“The police can signal you to stop whenever you’re driving…”). I suspect the rule is the same in most other developed European countries as well.


Having failed to resolve what makes something legitimate, one might wonder if this is even a useful metric. But if legitimacy theory hinges on “legitimate” policies increasing public cooperation with police, 911 calls from citizens to police seem like a decent measure, even if we can’t resolve the question of how legitimacy is generated.