文档介绍：19
IIII I I I I I I III IIIIII _ II III I I IIIII ]J[_
ITEM RESPONSE MODELS FOR
THE ANALYSIS OF EDUCATIONAL
AND PSYCHOLOGICAL
TEST DATA
RONALD K. HAMBLETON, FREDERIC ROBIN, DEHUI XING
College of Education, University of Massachusetts at Amherst, Amherst, Massachusetts
I. INTRODUCTION
Psychometric theory and psychological and educational assessment prac-
tices have changed considerably since the seminal publications of Lord
(1952, 1953), Birnbaum (1957), Lord and Novick (1968), Rasch (1960), and
Fischer (1974) on item response theory (IRT). Several unidimensional
and multidimensional item response models for use with dichotomous and
polytomous response data have been advanced, IRT parameter estimation,
statistical software, and goodness-of-fit procedures have been developed,
and small- and large-scale applications of IRT models to every aspect of
testing and assessment have followed (van der Linden & Hambleton, 1997).
Applications of IRT models were slow to be implemented because of their
complexity and a shortage of suitable software, but now they are widely
used by testing agencies and researchers.
These changes in psychometric theory and practices over the last 30
years have occurred because IRT models permit more flexibility in the test
development and data analysis process, they have more useful properties
than classical test models (., item statistics that are less dependent on
Handbook of Applied Multivariate Statistics and Mathematical Modeling
Copyright 2000 by Academic Press. All rights of reproduction in any form reserved. 553
554 RONALDK. HAMBLETON ET AL
examinee samples), and they allow psychometricians to more effectively
model the test data they work with.
Today, the development and refinement of the largest and most impor-
tant assessment instruments in the United States is guided by item response
models, including the National Assessment of Educational Progress (a na-
ti