T1 Scaling: A Mathematical Programming Approach to Thurstonian Scaling (Classic Reprint) - Hardcover

James M. Lattin

Hardcover

ISBN 10: 0666408025 ISBN 13: 9780666408020

Publisher: Forgotten Books, 2018

Synopsis

Efficient, robust scaling for judgment data

Discover a resilient method for Thurstonian scaling that stays reliable even when data are imperfect. This book explains how T-scaling reduces sensitivity to inconsistent judgments and outliers.

The text introduces a robust approach that reframes scale estimation as a network flow problem, offering a practical alternative to traditional least-squares methods. It shows why minimizing absolute deviations can yield more trustworthy scale values, and compares the new method with existing techniques using real data examples. Readers will see how the approach handles incomplete data, includes diagnostic tools, and applies to classic scaling problems in psychology and behavioral research.

• Understand the limitations of standard least-squares scaling and why a robust alternative matters
• Learn how the T-scaling problem translates into a capacitated network flow model
• Explore diagnostic methods to detect limited inconsistency and intransitivity
• See applications to published scaling problems and what the results imply

Ideal for researchers and students in psychometrics, statistics, and social science fields who want a practical, computation-friendly scaling method that stays stable in the face of imperfect data.

Edition details

• Publisher: Forgotten Books
• Publication date: 2018
• Language: English
• ISBN 10: 0666408025
• ISBN 13: 9780666408020
• Binding: Hardcover
• Number of pages: 56