LAPD’s big data covered up big mistakes

At the leading edge of the “big data” trend, the Los Angeles Police Department has found itself in hot water. From underreported murders to misclassified assaults, a fresh spate of scandals has started to brew, just as Police Chief Charlie Beck was reappointed to a second term Tuesday.

Los Angeles’ police department isn’t alone in leaning hard on the use of data. Over the past several years, law enforcement officers around the country have turned to numbers-crunching to help solve — and even prevent — crime. From Washington on down to the municipal level, the trend has been clear.

The idea is simple: Collecting large amounts of information can aid the Department of Homeland Security and a small-town police department alike. Powerful conclusions can be drawn from data compiled from crime reports, emergency calls and the like.

But as L.A.’s experience shows, there are perils. First, human error can slip by undetected, causing cities and citizens to put misplaced confidence in numbers and trends.

Second, more deliberate “mistakes” can be made, fostering a kind of false confidence with potentially damaging consequences.

Third, cities and police forces themselves can fall into patterns of misconduct that good-looking data is used tacitly to justify.

All three of these problems have worked their way into the LAPD’s use of big data.

Misclassified crimes

In a bombshell story, The Los Angeles Times revealed the results of an exhaustive report into misuse of data on the force. From Sept. 2012 to Sept. 2013, L.A. cops “misclassified nearly 1,200 violent crimes” that included “stabbings, beatings and robberies.”

In fact, almost “all the misclassified crimes were actually aggravated assaults” that wound up “recorded as minor offenses.” The result? The crimes “did not appear in the LAPD’s published statistics on serious crime,” which “officials and the public use to judge the department’s performance.”

In interviews, current and retired police officers gave the Times two different explanations for the systematic discrepancies. Some said they were merely “inadvertent.” For others, however, “the problem stemmed from relentless, top-down pressure to meet crime reduction goals.”

As the Times explained, top cops “set statistical goals” for crime reduction at the beginning of every year. That overarching plan leads to the creation of smaller, but just as data-driven, objectives. “As part of that process, the department’s 21 divisions are given numerical targets for serious crimes each month.”

Instead of taking the emotion and uncertainty out of reducing crime, however, at least one source suggested the opposite began to happen. Since the numbers became paramount, opportunities arose to use terminology to change what it was the numbers indicated. Crime could appear to decrease, not by pushing the numbers down, but by altering the classifications that made sense of the numbers.

Official reaction

Initially, the LAPD balked at the Times’ report. Officers had already been pulled into a controversy by the Daily News, which recently questioned city murder statistics. The Daily News wanted to know why the LAPD had reported such low rates of crime-solving. (Chief Beck had gone on record saying cops had solved substantially more murders than reported to federal and state authorities.)

After first promising to “open up the books,” Deputy Chief Kirk Albanese backtracked, claiming the LAPD’s inspector general, Alex Bustamante, would dig into the matter and publicly present his findings to the Police Commission.

Bustamante confirmed he would be reviewing that matter. But he also announced he would lead an investigation into several years of statistics to shed light on the misclassifications of crime that Times report had uncovered.

That includes a full investigation into COMPSTAT, the centerpiece of the LAPD’s big data program. Bustamante told the Daily News that he’d far exceed the scope and detail of the Times investigation, in a “much more expansive” look at “thousands and thousands” of cases.

In the event that wrongdoing is found, the LAPD has indicated those responsible will be disciplined. A larger question remains, however. If big data-driven policing is here to stay, what steps must be taken by officials to restore the trust of Angelenos and their elected representatives?