Jonah Rockoff

This dude is pretty sharp. Young scholar at Columbia Business School. Not Ed School. He wants you to know that. Quant guys can be touchy about affiliation. "Bam! Here are the numbers. No fuzzy emotion." That's the B-School vibe. Here's one of his recent papers.

Jonah opened at a New Schools Venture Fund conference in NYC this week. Below is some stuff he said. I think. I may be misrepresenting him. I was mostly listening...but also playing Scrabble online. Qintars.

1. 60% of all who get teaching degrees never end up...teaching.

We don't know much about these people. Did they sense during student teaching that they were bad, and therefore didn't pursue jobs? Did they try to get hired, but fail?

2. Experienced teachers, of all races, tend to move away from schools with poor, minority students over time. They take jobs instead in more affluent schools. (Charlotte, NC data).

3. It's not just teachers that seem to get better in their second year on the job. A teacher evaluator is better in his second year, too -- better able to predict the teacher's impact on learning, as measured by the pupils' value-added gains.

4. Evaluators are imperfect but useful. An evaluation in October of someone's rookie year of teaching is strongly correlated with that teacher's value-added gains of year 2.

But certain traits lead to higher evaluations (charisma) that don't correlate with kids learning more.

5. Are value-added scores fair?

Eye of beholder. If a rookie is bottom quartile during Year 1, then here's his chances for Year 2: a 50% chance to still be in bottom quartile; 35% chance to be in third quartile; 10% chance to be in second quartile; and a 5% chance to go from "worst to first" and be in the top quartile.

Let's say you hired 400 teachers for your school district. And at the end of the year, you fired the 100 rookie teachers in the bottom quartile.

Among the fired, you'd have 15 people who would have been above-average during Year 2, and 85 who would have been below average. Economists would say that is a "win" for kids in the sense that the "Replacement 100" is going to generate a better average teacher. But if you're one of those 15 teachers, it's not fair -- you were a false-negative.

Value-add is more reliable if you can look at 2 or even 3 years to make a judgment.

We need Jonah to run this data for bottom tenth of rookie teachers versus bottom quartile. More on why in a future post.

6. The best way to predict future teacher performance is to look at past teacher performance, combining both value-added gains and subjective evaluations.

7. If you really cared nothing about teachers, and ruthlessly sought to maximize student achievement, and had a fairly unlimited number of applicants...

...the optimal mathematical number of rookie teachers to weed out each year would be the lowest 80%.

This is kind of crazy at first glance. And second.

But it's sort of what Zeke is trying to do, build a charter school all-star team of teachers and pay 'em each $125,000 per year.

His hiring process wasn't to examine the value-added gains of his applicants, though. But someone will come along and do that, once these numbers start to become more transparent.

I think that will also help to humanize these high value-added teachers.

We once hired a teacher who, at her previous school, taught an 8th grade with what ended up to be the highest MCAS English proficiency rate in Massachusetts. I'm sure her value-added gains were ginormous. But this is a teacher who is fairly disinterested in test scores, and who visitors admire for her rich text-based discussions of plays and novels. I.e., instead of kids just blathering random opinions, she insists they cite the text as evidence. She's old school.

My guess is that if we could just watch video of many high value-added teachers, fear about "teaching to the test" would plummet.