Sunday, February 15, 2009

Sunday Commentary: What to Incentivize?

Teachers should be encouraged to do what is best for students, and rewarded when they do -- which is why performance incentives make perfect sense on paper. But a million little details need to be solved before they can meet their promise. Perhaps most notably we need to be sure we incentivize the right behaviors. And teaching is far from the only field that struggles with this.

Today's NY Times Magazine has a fascinating article basketball statistics. More specifically, it explores the relationship between statistics and good basketball. The story centers around Shane Battier -- a player with below average stats but whose teams win when he is on the court. The Houston Rockets have apparently developed some ways to measure things that contribute to the team in place of the typical stats (points, rebounds, assists, etc.).

The most interesting (and relevant) part of the article is this: "It turns out there is no statistic that a basketball player accumulates that cannot be amassed selfishly. 'We think about this deeply whenever we’re talking about contractual incentives,' [Rockets GM Daryl Morey] says. “We don’t want to incent a guy to do things that hurt the team.”

In other words, basketball teams face the same problem (though in a different way) as do schools: creating incentives that encourage people to do what's best for the team/school. A basketball team does not want to promise Player A a million dollars more if he averages 20 points per game and then watch the team lose because Player A takes a ton of wild shots instead of passing to his teammates.

I don't think the exact same problem exists in school -- Teacher A probably isn't going to hurt the school by trying to raise test scores in their classroom. But a similar problem exists: we're not sure exactly what actions we want people to take, or we're at least not sure how we would measure such actions.

Can you imagine trying to rate a basketball player on whether he hustles, is in the correct position, passes to others when they're open, takes shots when he is open, communicates well with his teammates, and so forth? Similarly, how do we rate teachers on how well they motivate their students, keep students focused, prepare lesson plans, offer useful comments when grading, and so forth?

In basketball, one could give up on trying to measure what a player contributes and simply reward players based how many games the team wins. It is unclear whether players would want to win more if they had money riding on it, but it seems logical that such a scheme would at least discourage selfish behavior.

What about teaching? One of my colleagues once said that teaching is amenable, but not reducible, to measurement -- and I agree. The only possibility I see of truly measuring how well somebody teaches would take far too much time and effort to ever be feasible. But if we are committed to rewarding teachers based on what we measure, we have to find some sort of proxy. Unfortunately, we do not have a won-loss statistic in education, so there's no similarly simple solution. We do have test scores, but most teachers don't teach tested subjects. As such, we need to recognize that we cannot judge or reward most teachers based on their students' scores.

So what do we want to incentivize? Basketball GM's are starting to figure out that incentivizing ends may not justify the means. Rewarding teachers with high test scores may lead them to do all sorts of things that they should not. But rewarding means is almost impossible? Do we really want to sit around measuring all the little things that make a teacher special?

Whatever the measure is, I think we can learn one thing from basketball. If everyone in the school is rewarded for it then at least there is a possibility that it would foster cooperation and encourage people to raise their expectations of their colleagues.

Rewarding a player for a team's wins may not mean that he scores more points, but it might mean that he corrects the hitch in his teammate's jump shot or sets a valuable pick. Rewarding a teacher for a school's performance may not lead them to transform their teaching style, but it may mean that conversations in the teachers' lounge are a bit more productive, that staying after school to tutor a student (without pay) is less crazy, that the dynamite lesson plan on photosynthesis is shared with all of the 3rd grade teachers, and that the kid from Mrs. Smith's class who is wandering the hall becomes everybody's responspility. If we can just figure out those other 999,999 details, performance pay should be good to go.

______________________________________________________
Corey Bunje Bower is a Ph.D. student in education policy at Peabody College, Vanderbilt University. Before beginning his studies he taught sixth grade at a low-performing middle school in the Bronx that has since been shuttered. His research focuses on issues surrounding high-poverty urban schools -- including teacher retention, discipline, and school climate.

Sunday Commentary is a running feature on Thoughts on Education Policy. Submissions are open to all who are knowledgeable about education and willing to write a concise, thoughtful piece. Submissions may be sent to corey[at]edpolicythoughts.com.

6 comments:

Rachel said...

Rewarding a player for a team's wins may not mean that he scores more points, but it might mean... Rewarding a teacher for a school's performance may not lead them to transform their teaching style, but it may mean...

The catch to this approach is the same for teaching and basketball. It not only motivates players/teachers to act for the good of the team/school, it motivate them to look to move from "losing" teams to "winning" teams where they will be more highly rewarded for the same skills.

The advantage schools have is that achievement isn't a zero sum game -- but a reward structure needs to be carefully constructed not to be one.

turducken said...

Check this out post at another blog.

Nancy Flanagan said...

Corey: "The only possibility I see of truly measuring how well somebody teaches would take far too much time and effort to ever be feasible. But if we are committed to rewarding teachers based on what we measure, we have to find some sort of proxy."

Nancy: There's National Board Certification--as good a proxy as we have right now. Most people who readily dismiss NB Certification have no clear and detailed idea what the assessment process entails or measures (including some Famous Researchers).

One of the portfolio entries in NB Certification measures "groupiness"--requiring candidates to provide convincing evidence of how their work with colleagues and parents impacts student learning. This is a challenge for every candidate. We're not used to reflecting on anything but our own work results. This piece of the assessment has historically earned the lowest score, across all disciplinary groups and grade levels. But it exists--and the idea that tens of thousands of teachers have wrestled with this concept is encouraging.

There are many other misconceptions. Some believe NBC measures only pedagogy, not student learning outcomes, for example--when student learning is at the heart of the assessment process. And 40% of the assessment is on-demand testing of content knowledge.

So--it's out there.

Corey Bunje Bower said...

NF: Right, but that's a measure of how good a teacher is -- not how well they performed this year.

Nancy Flanagan said...

Are you trying to make a distinction between a teacher's general ability to increase student learning, and year-to-year test data, which will fluctuate? Which of these is more important to incentivize?

Corey Bunje Bower said...

I'm drawing a distinction between general ability and year-to-year performance. While they should be highly correlated, I'd probably lean toward rewarding the latter (if possible). But that's a good question that I need to mull over.