|
Abstract:
|
Case law surveys form a vital part of legal practice, allowing judges and attorneys to ascertain the law and to predict future court decisions. In some legal contexts, however, there are reasons to believe that the case law suffers publication bias. In this paper, we propose a method for detecting and correcting legal publication bias based on ideas from multiple system estimation (MSE), a technique traditionally used for estimating the size of hidden populations. On a simulated dataset involving biased observations of evidentiary admissibility rulings, our method is able to both detect the bias and recover the true admissibility rate. On a newly collected dataset of admissibility rulings on false confession expert testimony, the model estimates that an observed 16% admissibility rate may be in reality closer to 28%.
|
