Crunching the numbers on a presumption of reasonablenss

I was very pleased to learn of this new article on SSRN, which examines empirically the impact of the presumption of reasonableness for within-guideline sentences.  The article by Alex Robbins and Lynda Lao is entitled "The Effect of Presumptions: An Empirical Examination of Inter-Circuit Sentencing Disparities After United States v. Booker," and here is the full abstract:

In the two years since United States v. Booker, the circuits have divided over how to use the Federal Criminal Sentencing Guidelines when reviewing sentences imposed by district courts. Seven circuits have held that a sentence within the Guidelines range is entitled to a presumption of reasonableness on appeal; five have held that it is not.  Although the Supreme Court's recent holding in Rita v. United States allows the courts of appeals to adopt a presumption of reasonableness for within-Guidelines sentences, it does not require them to, and so the circuit split remains.

Using this circuit split as a natural experiment, we undertake what we believe to be the first statistically robust analysis of the effect of a presumption of reasonableness on sentences imposed at the federal district court level.  Specifically, using 145,047 individual-level observations recorded by the United States Sentencing Commission (comprising all recorded federal sentences in all twelve circuits for a one-year period beginning in November 2004 and ending in October 2006), we perform a multivariate regression analysis to determine how a circuit's adoption or rejection of a presumption of reasonableness for within-Guidelines sentences affects the frequency with which district courts impose below-Guidelines sentences.  We find that a circuit's adoption of a presumption of reasonableness decreases the frequency of below-Guidelines sentences by less than one percent, although this result is statistically significant.

Our results do not, however, robustly support the inverse hypothesis that a circuit's rejection of a presumption of reasonableness increases the frequency of below-Guidelines sentences. The effect of such a rejection is insignificant when we control for circuit-specific fixed effects, and appears to be driven almost entirely by the particular behavior of the Second Circuit.  Finally, we are similarly unable to find robust empirical support for the hypothesis that intercircuit differences in sentencing after Booker can be explained simply by a circuit's underlying characteristics.