Published:
Cambridge, Mass: National Bureau of Economic Research, 2020
Published in:NBER working paper series ; no. w28293
Extent:
1 Online-Ressource; illustrations (black and white)
Language:
English
DOI:
10.3386/w28293
Identifier:
Reproduction note:
Hardcopy version available to institutional subscribers
Origination:
Footnote:
System requirements: Adobe [Acrobat] Reader required for PDF files
Mode of access: World Wide Web
Description:
We compare predictions from a conventional protocol-based approach to risk assessment with those based on a machine-learning approach. We first show that the conventional predictions are less accurate than, and have similar rates of negative prediction error as, a simple Bayes classifier that makes use only of the base failure rate. Machine learning algorithms based on the underlying risk assessment questionnaire do better under the assumption that negative prediction errors are more costly than positive prediction errors. Machine learning models based on two-year criminal histories do even better. Indeed, adding the protocol-based features to the criminal histories adds little to the predictive adequacy of the model. We suggest using the predictions based on criminal histories to prioritize incoming calls for service, and devising a more sensitive instrument to distinguish true from false positives that result from this initial screening