Comparative Study of Item Selection Methods Utilizing Latent Trait Theretic Models and Concepts.

摘要

Latent trait theory provides the practitioner interested in developing tests with a far more powerful method of item selection than that provided by standard testing technology. This method involves the use of information curves as item selection criteria and allows the test developer to determine the contribution of each item to the test without knowledge of other items in the test. One problem that currently exists is that no well developed methodology has been established for the use of information functions in the test development process. The purposes of this investigation were as follows: 1) Provide some background on information curves for items and tests; 2) Develop several item selection methods and using a typical item pool (where items are described by parameters in the three-parameter logistic model), compare the score information curves resulting from application of these methodologies; and 3) Develop and compare several item selection methods for producing a scholarship exam and a test to optimally separate examinees into three ability categories. Keywords: Latent trait theory, Psychological tests, Mathematical models, Aptitude tests. (SDW)