Over the past few years, the Seattle school district has been developing a value-added model, not for the purposes of compensation reform, but reportedly to help district officials identify if slow, average, and advanced students are being assigned equally effective teachers. Unless Seattle is a lone exception out of 15,000 US school districts, they are undoubtedly learning what any researcher could have told them up front: that their most effective teachers gravitate to high-performing students, both because those are the students most teachers like to teach and because parents work the system to make sure that those great teachers are assigned to their children.

These two practices are referred to as "teacher sorting" (the former) and "teacher shopping" (the latter). And these two practices greatly hinder research attempts to learn more about teacher quality. Without more information than researchers typically have--information which would allow them to control for those two factors--any attempt to decide what teacher attributes matter is compromised to a certain degree.

In a fascinating and unique study of nearly 4,000 fifth grade teachers in North Carolina, "Teacher Sorting, Teacher Shopping, and the Assessment of Teacher Effectiveness," researchers Charles T. Clotfelter, Helen F. Ladd and Jacob L. Vigdor were able to control for these two factors by having access to such information as how many hours students watch TV, read, and do homework, as well as the education level of their parents--in effect, all those qualities that are telltale signs of high-performing students. They then set out to examine a number of teacher attributes to find out what really matters.

In some respects--but not all--their results mirror other studies' findings.

* Master's degrees add no value to student achievement, despite the fact that districts keep providing monetary incentives for them. In fact, teachers with advanced degrees produced lower math achievement gains than teachers without them.

* The scores teachers get on their licensing tests matter a great deal. This finding refutes researcher Eric Hanushek's long insistence that these tests don't equate to teacher effectiveness.

* There is a more linear relationship to the value of teacher experience than other research has found. Whereas other research hasn't distinguished much difference between a teacher in year 3 and one in year 15, this one does. However, as found in other studies, most of the value from experience is gained within the first year or two of teaching.

* National Board-certified teachers do not add value in terms of student achievement, at least at the 5th grade level. This finding is somewhat ironic, since the study that did find some value was also done in North Carolina (Goldhaber and Anthony, 2004). We'll try to get to the bottom of that one.

* The selectivity of the college that a teacher attended didn't contribute to student achievement gains, directly refuting the majority of studies looking at this factor.

Now, let's get back to the Seattle question, which raises a real quandary for schools. The teacher shopping study found that highly effective teachers consistently produced greater learning gains in high-performing students from well educated homes than they could in low-performing students, posing a sort of moral dilemma that schools won't want to touch with a ten-foot pole.

It's best illustrated by looking at two classrooms. For example, Mrs. Jones, stellar teacher that she is, consistently achieves a 10% learning gain from her classroom of mostly white students from well-educated homes. The school then decides to reassign her to the remedial class that is usually taught by mediocre Mr. Smith, in which he typically achieves a 2% gain. She does a lot better than he did, but still only gets a 5% gain. Mr. Smith, now reassigned to the class of high performers, lets them slack off and gets only a 3% gain. The remedial class has in fact benefited from Mrs. Jones' presence, but the school's overall numbers have dropped. The average gain of the two classrooms had been 6% but has dropped to 3.5%.

Ask yourself this: If you were the principal of a school that was in danger of not meeting AYP goals or even state accountability goals, what would you do? Drop us an email at tqbulletin@nctq.org, and we'll post the responses.