Comment

When I was a senior in high school, I was placed in a remedial English class. My guidance counselor was very apologetic. It could not be helped, she said. I was taking three classes that each had but a single section—honors French, honors biology, and an honors course in modern algebra for the students the math department felt were not strong enough for Advanced Placement calculus. Plus, I was the features editor for the school newspaper. And when all of these constraints were taken into account, the only available English class was a remedial class held during the last period of the school day.

I recognized how incongruous this placement was. I was thrust into a class of students whom I had barely seen in gym class, let alone other core academic classes. I spent much of the year cracking jokes that went over the heads of most of my classmates, but the teacher seemed to get a kick out of them.

I do not recall mentioning the situation to my parents. Even if I had, they likely would not have complained or taken any other action. On academic matters, they were content to defer to the authority of the school. From Maureen T. Hallinan's paper, however, I have come to fully appreciate the damage that was done to me. I enjoyed the relaxed pace of the class and basked in the presence of amiable, drug-addled classmates who were fans of Ozzy Osbourne when he was a founding member of the band Black Sabbath, three decades before mainstream America discovered him. I now know that I was shortchanged and that my academic achievement would likely have been higher had I been placed in a class that was more challenging academically.

But I am not bitter. To the contrary, my misassignment, and the group assignments of many other students, may not be easily manipulated via building-, district-, or state-level education policies.

I want to say something about the technical features of Hallinan's analysis, because she is a wise researcher and her approach deserves to be highlighted. I appreciate her care in describing some of the limits of her estimates of group effects. For example, she says that the predicted achievement of students hypothetically placed in a particular group is likely to be accurate when the characteristics of those students are similar to the actual characteristics of students in that ability group. When the student characteristics are not similar to students in that ability group, the predictions are less accurate. As a result, the further one moves from a student's actual group assignment, the less reliable the predicted mean achievement score will be.

What this suggests is that the overlap across groups in students' characteristics is critical to the ability to estimate group effects. It is fortunate that there is so much overlap in Hallinan's sample. If the groups were homogeneous with respect to prior achievement and other student characteristics, a persuasive argument could not be made that a student in one group would have performed better or worse if he or she had been in a different group.

The gold standard here, not surprisingly, is random assignment of students to groups, which would ensure, subject to sampling variation, that the measured and unmeasured characteristics of students are distributed identically across groups. But students have never been randomly assigned to basic, regular, honors, and advanced groups, and they probably never will be. Schoolteachers and counselors believe that the most capable students belong in the honors and advanced groups, and savvy middle-class parents want to position their children in the higher groups, thinking that this will enhance their children's chances for attending a prestigious selective college. Even if the process by which students are assigned to groups cannot be fully specified, the process clearly is not at all random.

To take account of the nonrandom selection of students into groups, Hallinan uses multiple regression methods, a time...