skip navigation

PUBLICATIONS

Register for Latest Research

Stay Informed
Register with NCJRS to receive NCJRS's biweekly e-newsletter JUSTINFO and additional periodic emails from NCJRS and the NCJRS federal sponsors that highlight the latest research published or sponsored by the Office of Justice Programs.

NCJRS Abstract

The document referenced below is part of the NCJRS Virtual Library collection. To conduct further searches of the collection, visit the Virtual Library. See the Obtain Documents page for direction on how to access resources online, via mail, through interlibrary loans, or in a local library.

 

NCJ Number: 72747 Add to Shopping cart Find in a Library
Title: Robust Estimation in Latent Trait Analysis - Final Report
Corporate Author: Bureau of Social Science Research, Inc
United States of America
Project Director: H Wainer
Date Published: 1980
Page Count: 158
Sponsoring Agency: Bureau of Social Science Research, Inc
Washington, DC 20036
US Dept of Justice
Washington, DC 20531
Grant Number: 78-NI-AX-0047
Format: Document
Language: English
Country: United States of America
Annotation: A new technique for parameter estimation in latent trait analysis is developed and applied to the prediction of recidivism among parolees.
Abstract: The theoretical aims of the project were to explore latent trait theory and robust methodologies in an attempt to derive estimation schemes that would yield correct and useful results in the face of contamination. In both biological and psychophysical applications of measurement models, the assumption of a single underlying dimension (a latent trait) has been coupled with some response function. This implies that as the latent trait escalates, the probability of a positive response on a specified dependent variable increases. In this study, the technique developed for parameter estimation in latent trait analysis combines traditional Jackknifing with the Sine M-Estimate to yield improved accuracy and efficiency of estimation. Asymptotically optimal methods of parameter estimation were found not to be necessarily optimal for small samples, nor were they found to be necessarily robust against departures from their assumptions in large samples. The new method, the AMT-Jackknife, was found to be superior to all other tested procedures (including maximum likelihood) in extensive computer simulations. In applying the new technique to the prediction of recidivism among a sample of Federal parolees, accuracy was improved significantly. Tabular data and four references are provided. (Author abstract modified)
Index Term(s): Mathematical modeling; Parole outcome prediction
To cite this abstract, use the following link:
http://www.ncjrs.gov/App/publications/abstract.aspx?ID=72747

*A link to the full-text document is provided whenever possible. For documents not available online, a link to the publisher's website is provided. Tell us how you use the NCJRS Library and Abstracts Database - send us your feedback.