Empirical Legal Studies

ELS Outside of the Ivory Tower

ELS--at least in its applied form--increasingly attracts interest outside of the academy. [Now former] NYC Mayor Bloomberg developed and recently released a web-based application that allows users to explore recidivism rates for specific sub-groups of New York City criminal defendants and help predict defendants’ likelihood of re-arrest.

Data Analytic Recidivism Tool ("DART") "enables users, including criminal justice professionals, program planners and researchers to select a group of defendants based on factors like age, prior criminal history and details about their criminal cases. DART will then produce a graph displaying three different one-year re-arrest rates for the selected group, including the percentage re-arrested for any crime within a year; the percentage re-arrested for a felony within a year; the percentage re-arrested for a violent felony within a year; and a comparison to the citywide average."

The Empirics of Attorneys' Fees and Securities Class Actions

While prior securities fraud class actions studies under the PSLRA have found that class counsel’s fee requests and awards are lower in cases in which the lead plaintiff is a public institutional investor, prior work has not explained the mechanism that underlies this reduction in agency costs. Specifically, do public funds negotiate better terms with their chosen counsel ex ante than do other lead plaintiffs? Or are judges responsible for the reductions in agency costs, suggesting that the PSLRA may not be working as Congress intended. Lynn A. Baker (Texas), Michael A. Perino (St. Johns), and Charles Silver (Texas) take up these questions in Setting Attorneys' Fees in Securities Class Actions: An Empirical Assessment. A summary of their findings follows.

"To learn how the mechanism created by the PSLRA is working on the ground, we studied securities class actions that settled between 2007 and 2011 in the three federal district courts that processed the largest numbers of these cases: the Central District of California, the Northern District of California, and the Southern District of New York. Briefly stated, we found little evidence that ex ante fee agreements play a role in the process for selecting lead plaintiffs. At the settlement-approval and fee-award stages, however, we found that lawyers more frequently invoked such agreements to support their requested fee and that courts deferred to attorneys’ fee requests more often in cases with evidence of an ex ante fee agreement. We further found evidence of an ex ante fee agreement or of a proxy for such an agreement (specifically, the presence of a public pension fund as the lead plaintiff) to be correlated with statistically and economically significant reductions in fee requests and awards, as well as with greater judicial deference to the requested fee. Overall, the court awarded a lower fee than the class counsel requested in about 18% of the cases we reviewed, a somewhat higher percentage than we had anticipated."

Ratio of IVs to N

I came across the following exchange on a listserv recently that engages with a question that I get a lot. Specifically, "[H]ow many independent variables [can] one regression can have?" A slightly edited (and redacted) version of the exchange follows.

Query:
"I came across a statement in a book I am using to teach a class on evaluation that says "a common rule of thumb is that 1 independent variable can be added for every 10 observations." (It goes on to say that this depends on multicollinearity and desired level of precision). The book does not provide a reference for this statement. Does someone know of a reference for this ratio, or perhaps a different ratio?"

Reply (1):

"This is often heard/said and is discussed in a number of places (e.g., Harrell's book on "Regression Modeling Strategies" on p. 61).

However, one implication of this simple statement is that it is acceptable to estimate a one-predictor model with and N of 10; I don't agree and would suggest a different rule (don't remember where I have seen this) such as 50 plus 10/predictor. Note that in any version of the rule, predictor should be read as "candidate" predictors at the start of the modeling process, not the final number of predictors."

Reply (2):

“Stata has excellent power tools that would be more useful than any single rule of thumb.”

ELS in the UK

A consortium of 15 universities across the UK is committing to ELS in a serious manner. Funded in part by the Nuffield Foundation, the "Q-Step" program seeks to fill "53 new posts to design and deliver specialist undergraduate programmes" in quantitative social science training. More information is found here.

Trends in Federal Civil Rights Litigation

Cornell colleague Ted Eisenberg finds that civil rights plaintiffs are making less -- and less successful -- use of federal courts over time in a recent paper, Four Decades of Federal Civil Rights Litigation. While much of the scholarly literature of late emphasizes recently-heightened pleading standards, Eisenberg notes the influence of increased settlements. The abstract follows.

"Civil rights cases constitute a substantial fraction of the federal civil docket but that fraction has substantially declined from historic peaks. Trial outcomes, as in other areas of law, constitute a small fraction of case terminations and have changed over time. The number of employment discrimination trials before judges has been in decline for about 30 years, a trend also evident in contract and tort cases. The number of employment trials before juries increased substantially after the enactment of the Civil Rights Act of 1991 but has been in decline since 1997. In constitutional tort cases, the number of judge trials has been declining for about 30 years; the number of jury trials has been reasonably constant over that time period. Civil rights plaintiff win rates at trial have been steady in both judge trials and jury trials for at least a decade. The success of civil rights litigation, as measured by trial win rates and settlement rates, has been quite low compared to contract and tort cases. Median awards in civil rights trials have increased more than the rate of inflation but median trial awards in both constitutional tort cases and employment cases are below the awards in contract cases and tort cases."