In his recent piece in the New Yorker ("Most Likely to Succeed," December 15), Malcolm Gladwell likens finding good teachers for America's schools to the "quarterback problem" that pro football teams have in predicting which star college player will make it in the NFL.
Yes, I know it seems a far-fetched premise, but here is what Gladwell has in mind: just like team officials don't really know what precise traits a good college quarterback must have to succeed in the NFL's much faster, tougher game, education managers and experts (school principals, district superintendents, academic researchers) can't identify a priori the qualities of a successful teacher. In both cases, the only way to find out who has the goods is to draft them, try them out, and maybe one in four will make the grade.
As Gladwell sees it, the two situations are alike in another respect. The stakes are high in both. Good quarterbacks make the difference between a mediocre and good football team; a good teacher makes the difference between mediocre and engaged, high performing students. According to Gladwell's source--a magical "back of the envelope" estimate by the Hoover Institution's Eric Hanushek--if we could only get rid of the "worst" 6 percent of American teachers and replace them with "average" ones, in a few short years American students would be reading and solving math problems as well as Canadians and Belgians.
Okay, it's a stretch. The fact that successful NFL quarterbacks are media idols and get salaries equal to the entire budget of a small school district might influence the number of people willing to invest years practicing to get a shot at the pros. The last time I looked there were no Pop Warner teacher leagues with twelve year-olds honing their math pedagogical skills.
But the piece is off the mark in other ways. First off, the assertion that we don't know what makes good teachers before trying them out for a few years while measuring their students' test gains is simply not true. For example, careful studies confirm what has long been suspected: how much math teachers know and how well they are trained to teach math affects their students' math gains. I'm certain Gladwell believes that raising students' test scores would have a major impact on their future job productivity. So wouldn't it make sense that teachers would be more productive if they scored high on subject matter tests?
There's more. In a huge study of North Carolina students, researchers from Duke's public policy center identified a number of teacher characteristics associated with greater student test score gains. One was how well teachers did on a licensure test. But years of experience and just having a regular teaching license made the biggest difference, followed by National Certification--a particularly demanding set of requirements for those teachers who want to be able to teach anywhere in the country. No one of these characteristics alone made a huge contribution, but together they added up to about one-fifth of the test score difference between pretty good students and mediocre ones.
So using the get-rid-of-the-worst-teachers dictum, in North Carolina you would want to get rid of the unlicensed, least experienced, lowest-scoring-on-the-licensure-test teachers. But don't forget, to increase student performance you would have to replace those with licensed, experienced, high scoring teachers. Where to get them is a question Gladwell and his informants haven't gotten around to yet.
Beyond that, as Gladwell himself points out, people who know about good teaching of reading or math can observe a teacher (or even a videotape of a teacher) and judge how good they are. In most cases, a skilled trainer can make a teacher much better. Of course, it takes practice and a good coach, but isn't that what most talented football players have had even before they get to college? And don't they get more of that once they get in the pros?
Gladwell also ignores that students in some U.S. states actually do as well as Canadians or Belgians on international tests. Minnesota made such large math gains on the Third International Mathematics and Science Survey (TIMSS) in 1995-2007 that its students now do as well as the higher scoring countries in the world. Did Minnesota fire the worst performing 6 percent of its teachers? Did the state hire four young teachers for each one it kept? Not at all. Instead, with the help of Michigan State's William Schmidt, it improved its math curriculum and helped its teachers learn to teach it effectively.
Minnesota is obviously not Washington, D.C. or LA Unified. Almost all those who become teachers in Minnesota went to school there, and the state already had a good education system in 1995. So its potential teacher pool was pretty good to start with. Yet even Minnesota doesn't have four teacher candidates for each one they keep. When you get to the other end of the teacher supply chain in many DC or LA schools, you are often just trying to find any teacher--even untrained--to put in front of each classroom.
Yes, the stakes of having good teachers are high for the students in DC and LA, but who is going to pay the pro quarterback salaries to get the best trained and the brightest to teach in inner city schools? An urban district can turn to Teach for America for smart, cheap, inexperienced college graduates to spend a couple of years in a low-income school. But by the time they pick up the skills, they leave for greener pastures. Meanwhile, the poorest inner city kids are likely to get the lowest draft picks or young teachers who have never even played the game.
No, American schools don't have a quarterback problem. They can identify good teachers when they see them. What they have is a preparation problem. Right now, not enough colleges have programs producing good teachers, and many teacher preparation programs and schools don't have enough good young teaching recruits and teacher trainers to produce top-flight education graduates.
Imagine if the NFL had 100 teams instead of 32. Unless you cranked up the quality of the college game at Divisions II, the fifty worst teams in the NFL would not be very good. This is exactly what happens in education. The "best" school districts get the pick of the crop, and yes, they make mistakes, but still end up with a lot of good teachers because they know how to pick them. The least desirable districts get what's left. Eliminating certification is not going to solve this problem. It's like asking the NFL to go to local parks and find its players by scouting pick-up touch football games.
We just have to bite the bullet, Malcolm. If we want good quarterbacks we've got to make people want to be quarterbacks--and build the infrastructure to produce great ones. Learning to become a good teacher has to be as exciting and rewarding as becoming a good athlete. Someone--and maybe it should be the large urban school districts themselves--needs to develop the college-level and in-school teacher training systems to grow enough skilled players to make every school a good one.