Mon December 3, 2012
Nate Silver Meets Recess: Crunching Data In Schools
SPOKANE, Wash. - The kind of finely tuned data crunching that fueled the 2012 election is spreading to another venue: the classroom. You might have heard that campaign analysts can predict who you're likely vote for based on the magazines you read and the car you drive. Now, researchers are finding ways to predict who's likely to drop out of high school based on, say, a third grade attendance record. Schools are hoping a computer program will help them reach kids before it's too late.
Meet Billy. Now, there's a lot we don't know about Billy -- like who his parents are or where he lives or even his real name. What we do know is that Billy is in trouble.
“You're starting to see Billy disengage by not coming to school.”
This is Steven Gering, the chief academic officer for Spokane Public Schools. He can tell a lot about Billy -- from the graphs on his computer screen.
“So Billy is in the lower portion of students in math and reading and and you can see four unexcused absences this school year.”
That means Billy has a 60 percent risk of not graduating from high school. Gering can tell that now … even though Billy is a fourth grader.
“Billy needs support. And if we want him to graduate eight, nine years later, we have to do something now so that he does have hope for his future and believes that he actually can stay in school.”
This year, Spokane Public Schools launched a computer program called the “Early Warning System.” It uses the data schools collect on attendance, test scores and disciplinary action to calculate students' risk for dropping out. And it makes these calculations daily. The hope is that schools will be able to catch problems in the early stages.
The program is based on a study of Spokane students in the 2008 and 2010 graduating classes. It found half the dropouts could have been predicted by elementary or middle school. And the warning signs were embedded in data the schools already had.
“Nowadays, school districts have massive amounts of information on each individual student,” says education researcher Mary Beth Celio, who wrote the study. The challenge, she says, is identifying which pieces of information matter most.
If you've seen the movie “Moneyball,” this might start to sound familiar. The film describes how baseball statisticians upended the conventional wisdom that a good batting average is the best measure of a player. By the same token, Celio argues economic status and race really aren’t the best predictors of dropouts.
“It's the same kind of thing, in the sense that things that anecdotally or stereotypically we think are the most important indicators once you get down into the information, aren't.”
Celio has also done studies for schools in Portland, Seattle and Kent, Wash. She says the exact risk factors differ by district. But there are some common threads: three or more unexcused absences in one year in elementary school, a single suspension in middle school, and an F in the first semester of high school are each big red flags.
It's lunchtime at Sacajawea Middle School in Spokane. Guidance counselor Linda Delaney has developed a pretty good sense of which kids are having problems. So, when she first saw the Early Warning System, she initially wasn't surprised by the kids at the top of the list.
“And then, every now and again, there’s one that pops up -- in fact I was looking at it this morning and I was really surprised to see a student there because I would not have expected that student to be so high on the list. So, that’s really good for us because it brings to our attention kids we might not notice otherwise.”
But then, there's this example of one girl earlier this year.
“In the first month of school she missed two weeks, which is 50 percent of the school year, so she popped way up to the top because of that. But, when we looked further, she had mono.”
So, Delaney says the human factor is still an important part of the equation.
On the Web:
Early Warning System overview (National High School Center)