Using data to predict students headed for trouble

95457 full
95457 full

Principal Kelley Birch’s office at Vacaville, Calif.'s Willis Jepson Middle School has the usual stuff: elaborate scheduling calendars, photos, and a neat stack of papers. 

What you won’t see, unless you walk around to Birch’s desk, is a whiteboard with handwritten names of the 56 students at Willis Jepson who have been struggling, the 7th and 8th graders who might not graduate high school a few years down the road.

Next to each name on the list are their trouble spots, things like poor grades, poor attendance, serious behavior issues. On the same list are notes about the ways the school has tried to help. The more marks next to a kid’s name, the more interventions the teachers and counselors attempted. 

“When I see that board,” says Birch, “I have an urgency that these kids need something now.” 

In the past, Birch says, getting a full view of which kids were in trouble took time. “We would wait for the teacher. And the teacher would go to the counselor and say, 'I have this student and they aren’t doing well," Birch says. "And the counselor would go look and say, 'Yeah, they aren’t doing well.' But by then, it’s a quarter into the school year, a semester into the school year.” 

Now, Birch uses what’s known as an Early Warning System. Her team gathers and processes a steady stream of student data, like GPA, attendance, demerits, and test scores, to peer into the future and spot the 7th and 8th graders most at risk of dropping out of high school in the future. 

An example of Early Warning System

An example of an Early Warning System. (Maine Department of Education)

“It’s about using data that are available to predict which students are at risk, identify them, and then provide supports and interventions so they can get back on track,” says Susan Bowles Therriault, Principal Researcher for the education program at American Institutes for Research. 

The national graduation rate has been climbing steadily. Today, about 80 percent of public high school seniors will graduate. A decade ago, that number was closer to 70 percent, but educators, parents, and politicians all want to see that number increasing. 

Early Warning Systems are one way schools are trying to make that happen. 

Therriault says the proliferation of individual-level student data has made these systems possible, even common. “There are probably schools  and districts in every state across the nation that are using Early Warning Systems in some format,” she says. 

Research shows that the two most important factors when trying to predict whether a kid will graduate from high school on-time are academic performance and attendance. But, different schools, districts, and states have their own models. They might include their own variables, or they start looking for signs a kid is at risk at different points in their education. And they flag them in different ways.

Early warning system graphic

In Wisconsin, every 6th grader in the state is given a score between 0 and 100 that represents the child’s expected chances of finishing high school on time. Students under 78.5 are flagged “high risk,” and highlighted in red in the state’s student database.

“It’s early, and that’s the real advantage of it,” says Jared Knowles, a Research Analyst at the Wisconsin Department of Public Instruction, which calculates the scores.

Knowles says predicting the path of a 6th grader gives teachers a long lead time to change that path. Plus, he says, it can be cheaper to intervene early, before problems multiply.  

But Knowles acknowledges, using data this way, to mark kids as potential dropouts, has risks. “We do a lot of work to communicate about the limits of the prediction,” Knowles says, “that it’s not destiny.”

Not destiny.

You don’t want a teacher to see a kid with score of 20, and give up, in order to help the kids with 80s. You don’t want a kid to know she’s got a red flag next to her name, and give up on herself.

The data, says Knowles, is just a snapshot of how a child is doing. It’s a symptom of trouble, but it’s not a diagnosis, and it’s not a cure. “The school has to do that hard work to re-engage them back into the education system,” says Knowles.

It’s that work that’ll actually change an outcome for a struggling student, not data or data systems.

These school interventions take a lot of forms, everything from special-ed evaluations, to behavioral counseling, to mentoring, to intervention classes in a subject area back at Principal Birch’s middle school in Vacaville.

In the English intervention class, about a dozen students are going over the basic of reading comprehension. In the Math intervention class, students are struggling to calculate discounts and tips.

These classes take resources, and Willis Jepson Middle School didn’t have extra money, so Birch came up with an elaborate bell schedule to squeeze them into the day. She also made some classes a little bigger, to free up teachers to run these interventions.

Birch says all the extra work like data crunching and schedule crunching is worth it to get kids back on track.

And, hopefully, erase their names from that list in her office.

Help KPCC improve our comments section! Take a 5 minute survey

blog comments powered by Disqus