A few years ago there was a great magazine advertisement that showed a picture of a napkin. On the napkin was written:
1. Build Computer.
2. Sell Computer.
Except someone had crossed out the “1″ and written in “2,” and then crossed out the “2″ and written in “1.” In other words, the business plan for Dell Computer that allowed Michael Dell to make a hundred jillion gazillion dollars by age 40 by having all his customers order and pay for computers over the Internet before Dell built and shipped them, saving vast sums of money on warehousing, inventory, etc.
I thought of this ad during the last few weeks while I watched Eduboss Andy Rotherham cajole the National Board for Professional Teaching Standards into finally coughing up a largely critical evaluation from value-added guru Bill Sanders. Sanders’ main finding was:
The amount of variability among teachers with the same NBPTS Certification Status is considerably larger than the differences between teachers of different Status.
At first, the implications here seem limited to NBPTS. But they’re actually much more significant and far-reaching, touching on the foundations of contemporary public policy as it relates to teachers.
The Sanders value-added system estimates teacher effectiveness by comparing the actual year-to-year growth in learning (as measured by standardized tests) of each teacher’s students to amount of learning growth those student would be expected to gain given their previous academic history. The more actual growth exceeds expected growth, the higher the rating.
In essence, the Sanders study found that individual NBPTS-certified teachers differ more from one another than they collectively differ from non-NBPTS teachers. On the surface, the imlications of this finding seem straightforward: the NBPTS folks have some work to do to improve their process and states that have tied many millions of dollars in salary bonuses to NBPTS status should give those policies a second look.
But both of these ideas flow from the same premise: that this study is just another step forward in the ongoing search for the characteristics of the effective teacher. A definitive list of such characteristics is the holy grail of teacher policy. If we only had that list, so the thinking goes, we could do all kinds of important and useful things. We could reshape education schools to impart those characteristics. We could set up certification systems to filter out teachers who don’t have those characteristics. We could design compensation systems that pay teachers with those characteristics more money.
In other words:
1. Identify effective teacher.
2. Hire effective teacher.
The problem is that this entire approach is flatly contradicted by the evidence. The essential findings of the Sanders study–more variation among teachers with or without NBPTS status than between teachers with our without NBPTS status–is also basically true for every other teacher characteristic that’s ever been studied.
Experience, education level, certification status, training, content knowledge, selectivity of the undergraduate institution, verbal ability — all of these things have been shown by some studies to influence teacher effectiveness (although some more than others–having a Master’s generally seems to have no effect, which is , given the resources devoted to putting millions of teachers through graduate school, an enormous problem).
But even in combination, these factors explain only a minority–often a small minority–of the overall variation among indvidual teachers in helping students learn. There are great inexperienced teachers and terrible inexperienced teachers, effective certified teachers and ineffective certified teachers, etc. etc. etc. It’s not that these things don’t matter. But other, unknown things clearly matter more.
Yet a lot of teacher policy conversations are based on the idea that if we just keep looking–or created elaborate processes like the NBPTS–we will, someday, finally nail down the prototype of the elusive effective teacher, and then proceed to stamp out millions of copies of him or her.
Underlying this idea is a kind of commodified view of teachers — that they all have pretty much the same job requiring pretty much the same characteristics to succeed. Obviously this varies somewhat by grade, subject, and type of student, but the essentials are supposed to be the same. This view is reinforced by most collective bargaining agreements, which enforce further uniformity on how teachers are hired, managed, and paid.
My strong suspicion is that this whole way of thinking will ultimately turn out to be profoundly wrong. That teaching is actually a much more complicated, difficult, and idiosyncratic process than our mainstream, characteristics-based teacher policy suggests. That knowing teacher characteristics like experience, training, etc., is useful, but only marginally so. That we could double, triple, or magnify tenfold our efforts to refine and expand things like the NBPTS and still never get close to identifying the effective teacher, for the simple reason that she doesn’t exist.
In the long run, I think we’ll eventually conclude that the best and only way to consistently and usefully identify teachers who are good at helping students learn is to assess how much teachers’ students are learning. We can and should wonder why they’re so successful, but we shouldn’t let the inherent limitations of our ability to do so limit the plain logic of shaping public policy around the fact that they are successful, or are not, or are somewhere in between.
In other words, I think we’ll utimately come to grips with the fact that the combination of knowledge, skill, motivation, work ethic, talent, focus, and myriad other factors that make for an effective teacher are 1) very different for different teachers who nonetheless produce similar results, and 2) far beyond our ability to identify and categorize for the purposes of crafting good public policy.
1. Hire effective teacher.
2. Identify effective teacher.
It worked for Michael Dell.