Last week, USA Today published a thorough review of police records in Baltimore from January 2014 through May of 2017. Here’s what the paper found:

Just before a wave of violence turned Baltimore into the nation’s deadliest big city, a curious thing happened to its police force: officers suddenly seemed to stop noticing crime.
Police officers reported seeing fewer drug dealers on street corners. They encountered fewer people who had open arrest warrants.
Police questioned fewer people on the street. They stopped fewer cars.
In the space of just a few days in spring 2015 — as Baltimore faced a wave of rioting after Freddie Gray, a black man, died from injuries he suffered in the back of a police van — officers in nearly every part of the city appeared to turn a blind eye to everyday violations. They still answered calls for help. But the number of potential violations they reported seeing themselves dropped by nearly half. It has largely stayed that way ever since.

Before the spring of 2015, Baltimore cops were involved in about 15,000 “police actions” per month. It’s been between 7,500 and 10,000 ever since — and that period roughly corresponds to the period in which violent crime in Baltimore has soared.

The report has generated a lot of discussion about what we ought to expect from police officers, the relationship between proactive policing and crime, and the so-called Ferguson Effect. So it’s worth breaking down what it does — and doesn’t — tell us.

What does the “Ferguson Effect” actually mean?

By one definition, the “Ferguson Effect” is a description of exactly what happened in Baltimore: Cops either stop or severely cut back on proactive policing in response to protests, indictments of other cops, viral cellphone videos and other public criticism. The second way the term is used is in reference to a rise in crime that is supposedly connected to de-policing. And there was also a spike in crime in Baltimore over about the same period.

But there’s a third way the term has come to be used — a sort of “alternative Ferguson Effect”: After high-profile incidents of alleged police brutality, people in marginalized communities are less likely to call, provide tips to and cooperate with the police. This is especially true in cities that already have a history of tension between those communities and the police.

To assess what happened in Baltimore, let’s start with the first way the term is used. The USA Today report makes pretty clear that Baltimore police officers dramatically scaled back what you might call proactive policing. They still responded to calls but were far less likely to stop people on the street, to look for suspicious activity and to conduct drug sweeps.

There are several explanations for why such incidents might make the ordinary street cop reluctant to do his or her job. The first is a fear that some justified action the officer takes might be captured on video, be mischaracterized or taken out of context, and then go viral, turning him into a public pariah. The second is that these high-profile incidents heighten public resentment of the police more generally, making these officers’ everyday interactions more hostile, and therefore more dangerous. And the third is the perception that more scrutiny of police increases the possibility of legal liability — whether through criminal indictments or lawsuits.

All of these fears are misguided. It may seem like these viral videos are saturating the Internet, and it’s true that cellphone video has brought more transparency and, to a lesser degree, more accountability to policing. But for the ordinary officer, the odds of starring in one of these videos are pretty slim. A Google News search of “cell phone video” and “police” turns up about 20 stories about alleged misconduct caught on video over the past week. About half of those are updates on the Eric Garner story, a four-year-old incident. The rest cover six separate incidents.

There are somewhere between 600,000 and 800,000 sworn police officers in the U.S. (the last official count was in 2012, and put the figure at about 750,000). So the odds of any given cop — even a rogue cop — getting caught in video footage that goes viral is pretty much a rounding error. Of course, many (or perhaps most, depending on your perspective) of these videos really do depict misconduct. Of the six incidents that turn up on Google New from the past week, one was a Chicago cop who boasted about “killing motherf*****s.” Another involved officers confiscating and turning off someone’s cellphone during a confrontation. Two others were about a Chicago cop caught on video while engaging in two separate bar fights while off-duty.

In other words, the odds of a given good cop getting caught in video footage that has been selectively edited or taken out of context to portray a false narrative are even slimmer. This isn’t to say it has never happened. But given the number of cops in the United States, the odds of such an injustice happening to any one of them are certainly nowhere near high enough to justify altering the way they do their jobs.

What about legal liability? In 2015, the Baltimore police union explicitly argued that the indictments of the officers involved in Freddie Gray’s death would make other officers in the city afraid to do their jobs. But criminal culpability for police officers is exceedingly rare, and convictions are even rarer. As for lawsuits, the qualified immunity given to cops means most cases don’t even get to a jury. When they do, juries rarely come down against law enforcement, and even when that happens, most officers are indemnified by the city they work for against any damages.

There also just isn’t much evidence that viral videos, protest and criticism of police shootings makes cops’ jobs more dangerous. While killings of police officers on the job fluctuate year to year, the general trend over the past 20 to 25 years is that policing has become dramatically safer. Over the very period in which we’ve seen the proliferation of cellphone video and the ability to amplify those videos via social media, killings of police officers and assaults on police officers have declined significantly.

Even looking at the past half dozen or so years, while there have certainly been a few instances of people specifically targeting cops because they’re cops, such as the mass killing of five officers in Dallas in 2016, the aggregate data doesn’t show even a correlation between the onset of protests and police criticism and the killing of police officers. The current swell of protest from groups such as Black Lives Matter began in the second half of 2014 during the protests in Ferguson, Mo., then gained momentum at the tail end of that year and well into 2015. Yet 2015 saw a drop in killings of police officers, and was the second safest year for police officers on record. Moreover, the targeted killings of police officers have been ambushes — either surprise attacks on cops from afar or by sneaking up on them, or attacks after the perpetrator calls 911 calling in a complaint about a crime such as domestic abuse. Neither method would be prevented by a reduction in proactive policing.

Of course, when it comes to issues such as crime and safety, much of the public is often misinformed. Police officers are people too. If most police officers feel like the job is getting more dangerous, or that it’s just a matter of time before they’re unfairly depicted in a cellphone video or unjustly charged, it doesn’t really matter what the statistics say.

This is where the pundits and law enforcement officials who push these narratives — people who should know better — seem especially cynical. It’s always been odd that people who argue for the Ferguson Effect do so in defense of law-enforcement officers. The argument here is that if we criticize bad cops, if we bring some transparency to policing, if we start to hold abusive, racist and dangerous cops more accountable, then the good cops will stop doing their jobs. It paints police as either incredibly sensitive or vindictive. Neither is a good look.

And the people pushing these views know (or at least should know) that policing is getting safer. Yet they claim that the job getting more dangerous because it’s politically advantageous for them to do so. They say this while believing that if cops think the job is getting more dangerous, they will stop policing proactively. They also believe that a reduction in proactive policing will cause an increase in violent crime. It’s hard to escape the conclusion that the people making this argument are sacrificing lives to gain political points.

Fortunately, the link between de-policing and homicide isn’t nearly so clear cut.

So does the USA Today study prove that the Ferguson Effect is real?

While the USA Today article did find a decline in police activity that corresponded with a rise in crime, there’s good reason to be skeptical that one caused the other. If you look at the line graph comparing police actions and shootings that accompanied the report, it’s true that there are stretches over the 41-month period that the paper analyzed in which police actions decline dramatically while shootings go up. But there are also stretches were police actions increase along with shootings. And there are others where both decrease from previous months.

We can also look to other cities. In 2013, when a federal judge required New York police to dramatically scale back the use of “stop and frisk,” there were widespread predictions that crime would soar across the city. Instead, New York has seen some of its lowest crime rates on record. And when the NYPD engaged in what appeared to be an organized “slowdown” in late 2014, crime declined.

In contrast to New York, Attorney General Jeff Sessions and others have pointed to Chicago, where the murder rate appears to have spiked shortly after the city entered into an agreement with the ACLU that required police officers to document and record such stops. Sessions cited a paper that claimed to prove this statistically and dubbed it the “ACLU Effect.” But as we documented here at The Watch, critics have poked holes in that paper and pointed to other possible causes for the rise in murders, including fluctuating funding for anti-violence intervention groups, that seem to be much more closely tied to the city’s murder rate.

So what else might have contributed to the spike in murders in Baltimore? One possibility is the alternative Ferguson Effect noted earlier. The general idea is that in places with a long history of police misconduct, abuse, unnecessary shootings and racial profiling, there’s a corresponding mistrust among police and the areas they serve, particularly in the minority communities that get the brunt end of such abuses. I’ve noted before the work of sociologist and historian Randolph Roth, who has argued that historically, violent crime has gone up when there’s widespread mistrust in government legitimacy. If that’s true, one could see how the problem would be compounded in communities where there’s ample distrust of the one government entity with whom people most often interact.

And there’s some data to back this up. A 2016 study found that 911 calls from black neighborhoods in Milwaukee dropped 20 percent after the high profile police beating of a black man. Murders in the city then soared to rates that surpassed similar stretches in both prior and subsequent years.

In fact, many of the cities where crime has risen recently have also been the subject of well-documented reports depicting law-enforcement cultures rife with abuses such as racial profiling, unjustified shootings and predatory policing, usually with little community outreach, and little to no accountability for even the most egregious and repeat offenders. The Justice Department reports on police practices in Chicago, Cleveland and St. Louis County, Mo., were particularly damning. It isn’t difficult to see why residents of these cities, especially those in high crime areas, might come to fear the police more than they fear the criminals.

We also see evidence for this in the data from sanctuary cities. Studies have repeatedly shown that such cities have lower crime rates (in fact, they’re especially safer for white people). This is likely because when a significant portion of the population fears or distrusts the police, they’re less likely to report crimes, provide tips or cooperate with police investigations.

It isn’t difficult to see why Baltimore’s black and brown residents might be mistrustful of the police. The city was the subject of a damning DOJ report about police practices there. The city also has a long history of aggressive stop and frisk policies, and “clearing” streets and street corners with mass arrests for low-level crimes such as loitering.

“Law and order” types sharply criticized the DOJ’s Baltimore report when it came out. But subsequent news has bolstered its credibility. The city was just rocked by the trial of officers in the Gun Trace Task Force, one of the most elite police units in the city. The officers who accepted plea bargains from federal prosecutors testified to incredibly brazen crimes, including stealing from witnesses and suspects, planting evidence and coverups. The city’s head of internal affairs — the division charged with holding bad cops accountable — was also implicated, as was a deputy commissioner. The scandal tainted thousands of cases going back a decade. Since then, there have also been multiple incidents in which body camera footage has appeared to show officers planting drug evidence. Earlier this year, another officer was caught lying in court. Prosecutors continued to use him, anyway.

One other point: If it is indeed true that Baltimore cops have responded to criticism by refusing to do their jobs, and have done so while under the impression that their work slowdown will result in more Baltimore residents — innocent and otherwise — being murdered, robbed and otherwise victimized by crime. Doesn’t that suggest a rather low regard for the people they’re supposed to be protecting? Doesn’t it suggest that maybe Baltimore residents ought to be distrustful of the police?

So does this prove that the alternate Ferguson Effect caused Baltimore’s spike in murders?

Not at all. It’s just an alternate theory that ought to make us question the dominant theory. One frustrating thing about these two theories is that the data support both. Let’s say there’s a high-profile incident of alleged police brutality, that police in that city respond to criticism by de-policing and that the crime rate subsequently goes up. It may have gone up because of the de-policing. But it may also have gone up because the incident further eroded trust between the police and marginalized communities.

If someone were to do a study of Baltimore such as the one in Milwaukee, and if that study were to find that 911 calls to police from black neighborhoods dropped dramatically over the period that crime went up, that would suggest that the alternate Ferguson Effect could explain at least part of the increase in violence. But every city is unique, and crime trends are driven by a multitude of factors, including but not limited to poverty rates, income inequality, demographics, recent history, employment rates and even factors as banal as the weather or statistical volatility. Perhaps de-policing did play some role in the murder spike. Perhaps growing discord between police and the community did. Perhaps both.

It’s the implications of the argument that Sessions and others make that are most unsettling: If we want to be safe, they say, we should focus less on holding bad cops accountable, allow police departments to be less transparent and stop drawing attention to police abuse. We should be comfortable with some incidents of racial profiling, police corruption and extrajudicial “street justice.” And if we refuse and if police stop doing their jobs as a result, and crime then soars, it’s the people demanding accountability who are to blame. Of course, the people making these arguments tend not to belong to the groups most likely to be on the receiving end of police misconduct.

In the end, the USA Today study injected some interesting and compelling data into this discussion, but it doesn’t lend itself to broad conclusions. On some level, it feels unsatisfying to say there’s nothing to be learned here. But that in itself is worth knowing, and it’s far better than learning the wrong lessons. In a free society, we should strive for police tactics that utilize the least amount of coercion possible. Any coercive tactics used should be shown to be effective in promoting public safety. Even here, those tactics need to be balanced with individual and constitutional rights. All we can say for sure about Baltimore is that the city’s police have long had a reputation for aggressive, proactive, law enforcement and a reputation of disregard for individual rights. Probably not coincidentally, city police also have long had a contentious and adversarial relationship with the city’s black and Latino residents. All the while, the city has remained one of the most dangerous in the country. Maybe it’s time for a different approach.