A good deal of scholarship has been published in the legal and educational communities since the Supreme Court decided the Parents Involved case. In that case, the Court struck down the student assignment systems of Seattle, Washington and Louisville, Kentucky on Equal Protection grounds. Justice Kennedy (nearly universally regarded as the "swing" Justice, now that Justice O'Connor is gone) wrote the decision.
Most of the scholarship has either described the case, fit the case into existing jurisprudence, or made normative claims about its correctness or the likely consequences. Recently, though, I came across an interesting piece in the Teachers College Record addressing the case from the perspective of a social science researcher. The author, Professor David Armor of George Mason University, evaluates the portions of Justice Kennedy's opinion in which he accepted the conclusion of the "liberal wing" of the Court that desegregation was a compelling governmental interest, but rejected the justification for this conclusion, which the Justices in the "liberal wing" based on social science evidence indicating positive average effect sizes in test scores resulting from desegregation. Professor Armor makes some important points that practitioners and researchers of constitutional law would do well to understand. The article is here. In particular, I think Professor Armor gets at a thorny concept--statistical effect versus the practical meaning of such an effect--that confuses many legal decision makers.
I think that Professor Armor's points are particularly salient today because legal scholarship, and increasingly legal adjudication, are becoming permeated with social science methodologies and evidence. Few lawyers and judges have the training necessary to understand such evidence and its limitations, and much legal decision making involving such information illustrates this lack of training. I hope that more work like Professor Armor's will help legal decision makers to understand what social science evidence can and cannot show.