Search

In a recidivism prediction context, there is no consensus on which modeling strategy should be followed for obtaining an optimal prediction model. In previous papers, a range of statistical and machine learning techniques were benchmarked on recidivism data with a binary outcome. However, two important tree ensemble methods, namely gradient boosting and random forests were not extensively evaluated. In this paper, we further explore the modeling potential of these techniques in the binary outcome criminal prediction context. Additionally, we explore the predictive potential of classical statistical and machine learning methods for censored time-to-event data. A range of statistical manually specified statistical and (semi-)automatic machine learning models is fitted on Dutch recidivism data, both for the binary outcome case and censored outcome case. To enhance generalizability of results, the same models are applied to two historical American data sets, the North Carolina prison data. For all datasets, (semi-) automatic modeling in the binary case seems to provide no improvement over an appropriately manually specified traditional statistical model. There is however evidence of slightly improved performance of gradient boosting in survival data. Results on the reconviction data from two sources suggest that both statistical and machine learning should be tried out for obtaining an optimal model. Even if a flexible black-box model does not improve upon the predictions of a manually specified model, it can serve as a test whether important interactions are missing or other misspecification of the model are present and can thus provide more security in the modeling process. [ABSTRACT FROM AUTHOR]

Are public officials more responsive to requests from affluent or poor constituents? A growing body of evidence suggests that lawmakers are more responsive to the rich when they craft policy. However, some scholars theorize that officials also exhibit a corresponding bias in favor of the poor when they handle casework, essentially giving policy to the rich and services to the poor. In this paper, we test this casework prediction using four experiments in which confederates sent simple requests to state or local officials. In each, our confederates’ reported social classes were randomly assigned and signaled with a brief introductory statement mentioning the sender’s occupation or economic situation. Across our samples, we find precisely-estimated null effects of social class biases: the officials we studied were equally likely to respond regardless of the constituent’s class. These findings raise doubts about whether casework is really a class-biased process. [ABSTRACT FROM AUTHOR]