The NY Times has published a piece about a new NBER Working Paper that finds that the average judge appointed by a Republican issues more stringent sentences against African-Americans than the average judge appointed by a Democrat. Since such judges are randomly assigned to cases, this result is interpreted as a "treatment effect" (rather than as a "selection effect").
Such research cannot answer "why" this fact is a fact.
Given this point, let me turn to a more subtle issue.
Will this research "matter"? If we as a society do not want there to be "differential sentencing" for the same offense, then this research could change our society in two different ways.
Case #1: The Republican judges were unaware of their "hidden bias". Now that the NBER detectives have uncovered this result and taught the world about this bias, the Republican judges will converge back to the average sentencing norm.
Case #2 The Republican judges were aware that they were engaging in this behavior but the rest of us were unaware of this. This is the asymmetric information case. Now that the Republican judges know that their "secret" is out, the New York Times and other progressive outlets will hold them accountable. As they are now watched the muckraker press, there will be two effects. First, there will be a selection effect as judicial appointments will initially be more heavily scrutinized. Second, judges will have lost some of their discretion because the media will be more carefully watching their ruling. In this age of "Big Data", it is easier for young Assistant Profs of Law to write these types of papers to test in real time for differential sentencing.
I am very interested in economics of the principal/agent problem in the context of government and our institutions doing their job. Who is held accountable and what institutions create such "rules of the game"? An optimistic paper studying this is the Besley and Burgess QJE paper.
Besley T, Burgess R. The political economy of government responsiveness: Theory and evidence from India. The Quarterly Journal of Economics. 2002 Nov 1;117(4):1415-51.
Based on data from India, they argue that when people have more education --- they demand more sophisticated newspapers that have incentives to invest in Woodward and Bernstein Watergate style muckraking reporting. This creates accountability as political leader "private information" vanishes.