Several months ago, I proposed to a few AmLaw 50 firms that we try to develop our own version of "Associate Moneyball," looking through firms' data about associates' backgrounds, who left and how soon, who made partner and who went inhouse, etc., to attempt to derive our own portfolio of characteristics that might predict success in BigLaw—assuming that hiring "the top 10% from the top 10 schools" is an exhausted strategy. What if, just to make up some hypothetical's, we were to discover that taking one or more years to work between college and law school correlated with lower attrition? What if coming from inherited wealth correlated with the reverse? Were JD/MBA's a better or worse bet than plain old JD's? What if moving across the country from one's home town roots [did/didn't] increase a propensity to stay? What if... You get the idea.
This may be one of those potential insights that is simultaneously self-evident once you think of it, immensely desirable of pursuit, and impossible to execute in reality. Among the obstacles we encountered were anti-discrimination issues (you might want to know the answers to questions you're not allowed to ask), and, perhaps more fundamentally, the difficulty of finding a real "counterfactual group" of law students at these top AmLaw firms who did not come from the top of their classes at the very best schools. Since the firms weren't hiring sub-median students from sub-median schools, the data as to how those people might stack up doesn't exist—and in all likelihood never will.
Project Associate Moneyball, however, remains deeply intriguing to me, and if anyone has suggestions for how to surmount the practical obstacles to doing serious analysis, please let me know. One possibility is that subtler patterns could be teased out even from the top students/top schools cohort to suggest ways to recruit more intelligently. One thing we can be confident of is that there's enough money at stake to make it worthwhile.