Robust partial least squares regression for the evaluation of justice court delay

Abstract Justice court delay is found to be consistently related to pending workload per judge, as well, output per judge (productivity) is found to be strongly related to demand pressure (total available workload) and size of court. The aim of this paper is to investigate among the causes of different performances in courts, highlighting possible anomalies (outliers) among the judges’ behaviour of a court due to the distribution of the workload among court sections and judges. The statistical methods considered are classical and robust variants of Partial Least Squares regression.