Awkward Utopia

[Note: I started typing this post in January but lost interest in it; so I am pushing it out now.]

From several potentially very interesting sessions on the early morning of the first day of AEA 2008, I picked the “Understanding School Performance” discussion because it related to the UES summer mission of working on a private school. Although the project has fallen by the wayside, despite keen interest on the part of the old UES vanguard, I figured this could still help inform our project should it ever move forward again. Additionally, the first speaker, Dr. Hanushek, is the author of several papers that have been indispensable to our project thus far. Dr. Hanushek presented a paper that he wrote with Steven Rivkin about the “School Quality and Black-White Achievement Gap.” Please pardon the ambiguities of this post as they are translated from very messy, unorganized written notes.

Introduction to the Achievement Gap

In the introduction, Hanushek mentioned that the achievement gap in schools robustly explains later earnings gaps between race groups, so it is a very important topic to examine. Of particular interest: does it expand or contract with schooling? There are serious selection problems in studying these issues because, amongst other reasons, there are differences in who takes tests and retention in grade levels differs among race as well. (Indeed, Dr. Figlio spoke to UES in 2002 on this topic, suggesting several ways that school administrators successfully eliminate certain test-takers to alter the school’s performance on tests.)

The gap itself is huge. One way of looking at the gap is by standard deviations. I didn’t write all the parameters of the graph down, but I believe Hanushek displayed a graph illustrating average black achievement decreasing from 1.25 to 1 or 0.75 standard deviations from average white achievement by 1985-90, although it promptly began going back up and continues to this day. The gap grows significantly across grade levels, although most of the growth is between schools — meaning, I think, that schools tend to have much larger gaps than others.

According to Hanushek, most if not all of the gap is explained by:

1. Student “churning” in schools (turnover), which creates an externality
2. Inexperienced teachers (schools serving more disadvantaged kids)
3. Racial concentration: the larger a concentration of black students, the greater the gap

The Data and Teachers

The data used come from Texas, as usual, and the authors found that the changes in the gap are most pronounced for the top black achievers. If anything, the data may underestimated the effect because blacks are more likely retained in grades. The data seemed to show that blacks perform better when there is more racial diversity, and that gap gets worse as the students get older, particularly 5th through 8th grade. Teacher experience is a large factor. Dr. Hanushek mentioned a few times the fact that blacks will tend to have teachers with less experience and that teacher turnover is more frequent in their schools. Apparently, teacher turnover rates matter a lot. (This seems like it may have interesting consequences for the Teach for America program, whose teachers often leave after 1-2 yrs.) This teacher effect is most pronounced in elementary school and is larger with lower achieving students.

The discussant, whose name was not on the program I don’t think, highly praised the data. He discussed “attrition” issues, saying that students who end up leaving are not the best and the brightest. In hindsight, and late in the day, I am unsure what this means. The discussant wanted more information on demographics and spoke a lot about wanting more details on the student “churn.”