Fretting about good research

See all posts

Angela Duckworth caught our collective imagination a few years ago when she put forward the notion that grit is a greater contributor to student success than more traditional intelligence measures. Grit, now more commonly referred to as persistence (I think mostly to avoid the cliché), is having its day in the sun in education circles.

What's a whole lot less clear is how teachers can instill kids with grit or self-restraint or any of those other soft skills we know matter. My own theory, though it's not much in fashion these days, is we take a huge risk in divorcing such strategies from student learning. Want to know how to suck all the motivation out of a child? Don't teach him to read. I fear that many schools are approaching these skills with the same lack of imagination as was the case in character education: "okay class, let's share some memories of when we persevered." I'm envisioning school assemblies featuring children jamming to badly repurposed songs, like "let's all…boom! boom!... drive to the Executive Function junction!!"

I digress. What prompted me to write about this topic was a fascinating new study from the prolific Northwestern economist C. Kirabo Jackson, What Do Test Scores Miss? with a nice version minus the math in this quarter's Education Next. Jackson finds a way to actually measure if a teacher has a lasting impact on her students' behavior, an especially challenging task given that he tackles high school teachers who see their students one period a day.

With 9th grade classrooms in North Carolina as the experiment, Jackson devises a teacher's "behavior index," a combination of her students' GPAs, their attendance and suspension rates, as well as on-time grade promotion. He finds a way to hold a single teacher accountable for attendance or suspensions, nicely illustrated here in this GIF tweeted by Jackson.

The findings make for quite a headline, that a teacher's ability to change student behaviors is far more important than her ability to impact test scores, at least in terms of these three outcomes: taking the SAT, graduating from high school, and a stated intent of going to college. In fact, teachers who can change their students' behavior have ten times greater impact on those outcomes than teachers who are good at raising scores. (And it's important to know here that not as many teachers are good at both as you might have imagined.)

Great stuff, but here's what I am worried about. The headlines. Not the study.

By pitching a teacher's behavior index against her ability to raise test scores, the top line findings might suggest that schools are better off focusing on soft skills at the expense of building students' cognitive skills. Test scores are the only measure Jackson employs for teacher contribution to cognitive skills, though no one argues for their exclusive use particularly when evaluating teachers, so in that sense I think Jackson might have considered a more expansive definition. More importantly, few teachers would assert that the state tests accurately capture their cognitive contributions in their students, rather that GPA is a better reflection of their impact in that domain. Yet Jackson assigns GPA entirely to the behavior side, dooming cognitive contributions to irrelevance. Jackson, in an email to me, fully acknowledges that GPA is a measure of cognitive skill, that there is clear overlap. He writes: "The part that is able to explain high school graduation (that is not explained by test score) must be by definition the behavioral component of GPA. We don't need GPA and test score to be unrelated, we just need that there is some additional signal in GPA that is not captured by test scores." But it's also worth noting that once Jackson removed the GPA from the index, a teacher's behavior index was no longer statistically determinative of these outcomes.

Jackson indicates much more to come along the lines of this work, and I personally am rooting for a similar study examining early elementary grades. In any case, pray that misinformed, agenda-driven tweets are not how we decide what's important about this study and welcome this advancement in our understanding of what makes teachers great.