How Bad is Too Bad? 5% as the Emerging Consensus
This past week, Washington released its annual list of the lowest-performing 5% of schools, which are eligible for state School Improvement Grants. I'm not sure when or why 5% was chosen as the magic number of schools that are unacceptably low-performing, but that figure seems to reflect an emerging consensus on performance in public education.
This NY Times/Bay Citizen article on "grade inflation" in teacher evaluations reports that only 2.7% of San Francisco teachers received below-satisfactory ratings in the past five years, and that not a single tenured teacher has been fired in the past three years. Such a low number seems to strain credulity, especially when 17% of new teachers were not retained last year (perhaps due to budget limitations). If 17% seems like too many teachers to dismiss and 2.7% seems too few to rate unsatisfactory, 5% seems to have more resonance with the public.
Erik Hanushek has been widely quoted as saying that the US would jump from 30th to 1st in international comparisons of math and science scores if we replaced our worst 5-10% of teachers each year with "average" teachers. I'll lump principals in with teachers and pose the question: What if we did just what Hanushek proposes and fired the worst 5% of principals and teachers every year? Would we get the results we want, and would 5% turn out to be the right number?
Perhaps 5% has an intuitive appeal because it's a familiar number in several ways. We like to be 95% sure about things; in the sciences, 95% confidence is essentially equivalent to certainty that a finding is accurate. There may be a 5% chance that we're wrong, but that's a small enough probability that we'll still feel comfortable making decisions based on the available information. Of course, identifying the worst 5% and identifying them with 95% confidence are two distinct tasks, and I don't mean to imply that they are the same, just that the number is familiar. Value-added analysis has thus far proven incapable of reliably determining who the best and worst teachers are, so more work on teacher evaluation will be needed if a 5% policy is to become viable.
Critics often claim that the teachers who would replace the fired teachers would be no better, but this seems highly unlikely in the aggregate: if we are targeting the worst 5% for replacement, even hiring at random should raise performance dramatically. Given the number of talented and energetic new teachers who have been unable to find jobs due to the economy, I have no doubt that we could easily replace the worst 5% with above-average teachers in most subject areas.
What do you think? Could we live with the top 95%? Would replacing the worst 5% of educators every year bring about dramatic improvements in education? In what details might we find the devil when implementing such a policy?