Over the last decade, Los Angeles has made the CompStat crime mapping system the foundation of the Police Department, with crime statistics informing decisions on where to deploy officers and how to make the city safer. The LAPD's data-driven policing has been credited with reducing crime for 11 straight years, and has been held up as a model for managing other city agencies.

But as with any statistical analysis, CompStat is only as good as the data it's based on. That's why city leaders should be very troubled by a Times investigation that found the LAPD misclassified nearly 1,200 violent crimes as minor offenses during a recent one-year period. The analysis determined that hundreds of stabbings, beatings and robberies were reported as simple assaults, meaning they were not included in the tally of serious crimes — and that L.A. was less safe than official statistics suggested.

Department brass characterized the misclassifications as inevitable in a complex data collection system. Chief Charlie Beck said the process was "subject to human error." However, the analysis found that offenses were almost always miscoded to turn a serious crime into a minor one, not the other way around, indicating that there may be a deeper problem than human error. Some department insiders pointed to the tremendous pressure on captains and lieutenants to meet the crime reduction goals set by higher-ups, and said that deliberate miscoding is a common practice. If so, it undermines the LAPD's data-driven strategy. Command staff can't identify violent crime trends if incidents are misclassified.

It's also worrisome that audits flagged this problem three years ago, but that the department took steps only this year to improve the accuracy of its crime statistics through training and increased scrutiny — after inquiries by The Times. Several months later, the department reported that aggravated assaults were up 12% for the first six months of the year — a steep increase that Mayor Eric Garcetti downplayed as the result of "more aggressive reporting."

Now, Garcetti must demonstrate that he will hold Beck and the department accountable for inaccurate statistics. The mayor wants to introduce the CompStat model to all city departments as a way to improve services and to measure performance, but the LAPD example should raise red flags about how metrics can be misused. It's unfortunate, but perhaps inevitable, that in a high-pressure, results-driven system, some individuals will cheat.

The Police Commission, which is considering Tuesday whether to reappoint Beck, should press the chief on how he would prevent deliberate misclassification in the future. The concern is that Beck and his command staff may have created an environment in which personnel believe they have to cook the books in order to succeed. If that's the case, the LAPD is on a shaky foundation.

Follow the Opinion section on Twitter @latimesopinion