The previous two stories show just how much activity there is these days on the teacher evaluation front, and as more evaluation systems emerge that measure student learning, it is likely we will see more stories like New York Times columnist Michael Winerip's recent profile of New York City teacher Stacey Isaacson. The question at hand is, what can be done when value-added scores are at odds with commonly held perceptions about a particular teacher?

Isaacson, a popular, hard working young teacher in a high-performing magnet middle school also happens to have terrible value-added scores. Based on these data, Isaacson does not meet New York City's criteria for receiving tenure, and could face dismissal for poor performance. However, her principal sees things very differently.

The principal points to the fact that in her first year of teaching, all but one of Isaacson's students scored at the proficient level on the state test. She received positive reviews during her two and a half years at the school, and even volunteered in the classroom while on maternity leave.

So do the value-added data, as the Times suggests, lie?

First, it is possible that Isaacson just didn't accomplish much with her students. They may have entered her classroom as proficient students and left the same, without Isaacson having made any real impact.

Second, no one has ever suggested that value-added measures are perfect (or that any single measure is). There will be teachers who are better than their scores suggest, but in assessing the likelihood of locating them, it's important to note Winerip's own history on accountability issues (see here and here, for example). He no doubt worked hard to find Stacey Isaacson. If there were a long list of other teachers aggrieved by the system, we would have almost certainly heard all about them.

Nevertheless, the red flag is raised. When objective and subjective evidence diverge so radically, how do you keep teachers and the public from losing confidence in accountability? There is no easy answer, but it's a question that states and districts ignore at their peril.