Impact of Test Design, Item Quality, and Item Bank Size on the Psychometric Properties of Computer-Based Credentialing Examinations

Computer-based testing by credentialing agencies has become common; however, selecting a test design is difficult because several good ones are available—parallel forms, computer adaptive (CAT), and multistage (MST). In this study, three computerbased test designs under some common examination condi...

Full description

Saved in:
Bibliographic Details
Published in:Educational and psychological measurement 2004-02, Vol.64 (1), p.5-21
Main Authors: Xing, Dehui, Hambleton, Ronald K.
Format: Article
Language:English
Subjects:
Adaptive Testing
Biological and medical sciences
Computer Assisted Testing
Computer based modeling
Computers
Credentialing
Design
Fundamental and applied biological sciences. Psychology
Item Banks
Psychology
Psychology. Psychoanalysis. Psychiatry
Psychology. Psychophysiology
Psychometrics
Psychometrics. Statistics. Methodology
Statistics. Mathematics
Test Construction
Test Items
Tests
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Computer-based testing by credentialing agencies has become common; however, selecting a test design is difficult because several good ones are available—parallel forms, computer adaptive (CAT), and multistage (MST). In this study, three computerbased test designs under some common examination conditions were investigated. Item bank size and item quality had a practically significant impact on decision consistency and accuracy. Even in nearly ideal situations, the choice of test design was not a factor in the results. Two conclusions follow from the findings: (a) More time and resources should be committed to expanding the size and quality of item banks, and (b) designs that individualize an exam administration such as MST and CAT may not be helpful when the primary purpose of the examination is to make pass-fail decisions and conditions are present for using parallel forms with a target information function that can be centered on the passing score.
ISSN:0013-1644
1552-3888
DOI:10.1177/0013164403258393