When new leadership took over at Maryland's Department of Juvenile Services, the first priority was building real structures for data collection and review. Not dashboards for their own sake. Data that leadership actually looked at, and acted on when something did not add up.
Early on, the focus was overcrowding and violent incidents inside facilities. Those got fixed through the obvious levers: stronger facility leadership, better staff accountability, smarter hiring. Once that ground stabilized, the next question was length-of-stay. In juvenile justice, the goal is for a youth to stay in a treatment facility only as long as necessary. Stay too long and treatment gains erode. Some youth leave secure treatment worse off than a shorter, well-timed stay would have left them.
I served on the committee that set length-of-stay guidelines for DJS secure treatment facilities. We started, naturally, with the data. We pulled a year of releases and cross-referenced length-of-stay against each youth's validated risk level: low, medium, high.
We expected high-risk youth, who need the most treatment, to stay longest. The data said the opposite. High-risk youth had the shortest stays. Low-risk youth were staying weeks and months longer than they needed to.
That is not a rounding error. That is a system doing the opposite of what it was designed to do.
Following the Data to the Real Problem
The committee went back to facility behavior and incident reports. High-risk youth, as expected, had the most conduct violations. But their behavior points, the ones tied directly to release timing, were rarely being deducted, in violation of policy.
Interviews with facility staff explained why. A youth's release depended on accumulated behavior points, and the higher the points, the sooner they went home. Some resident advisors were not deducting points from the most difficult youth, because a higher score meant that youth left faster. Meanwhile, lower-risk youth, who were easier to manage and more pleasant to work with, lingered in secure treatment far longer than the guidelines called for.
The incentive was not malicious. It was human. But it had quietly inverted the entire purpose of the release system, and nobody would have caught it without pulling the data and asking why the numbers did not match expectations.
What Changed
Once the gap was identified, DJS leadership moved fast. Two reforms fixed it. Release was tied to documented treatment progress instead of behavior points. And release decision-making authority moved to certified behavioral health staff, not resident advisors.
Neither reform required new technology or new funding. It required looking at the data closely enough to see that the system was not doing what everyone assumed it was doing.
The Takeaway
Every agency has a version of this gap somewhere. A well-intentioned policy that has drifted from its purpose because nobody checked whether the incentives on the ground still matched the goal on paper. Data does not fix that by itself. But it is the only reliable way to find it before it costs a young person months they should not have lost.