An examination of the precision of measurement of computerized adaptive tests with limited item pools.
Item
-
Title
-
An examination of the precision of measurement of computerized adaptive tests with limited item pools.
-
Identifier
-
AAI9530878
-
identifier
-
9530878
-
Creator
-
Halkitis, Perry N.
-
Contributor
-
Adviser: David M. Rindskopf
-
Date
-
1995
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Psychometrics | Education, Tests and Measurements
-
Abstract
-
This paper describes a method for examining the precision of a computerized adaptive test. When item pools are ideal in composition, CAT has been proven to increase efficiency and accuracy in both live and simulated testing situations. The precision of measurement of a CAT, based on IRT methodologies, is a function of the information that is provided by the individual items and the overall item pool. When pools are limited in their composition the results are less ideal. The precision and efficiency of the CAT process may be limited by the test construction process and speaks to the limitations of CAT in environments where the generation of large and extensive item pools is not possible. To further examine the precision of a CAT, standard errors of measurement ascertained in the testing of simulees in a CAT with a limited item pool were compared to those results obtained in a live paper-and-pencil achievement testing of 4371 nursing students on four versions of an examination of calculations of drug administration. Item difficulties and examinee abilities were ascertained with maximum likelihood estimation procedures based on the Rasch model. Analyses examined the extent to which indices of precision were enhanced when this achievement testing was administered in a CAT format given the realities of an item pool. CAT measures of precision were considered when the simulated examinee pools were uniform and normal. Results suggest that regardless of the size of the item pool, CAT provides greater precision in measurement with a smaller number of items administered even when the choice of items is limited, but fails to achieve equiprecision along the entire ability continuum.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
