Computer-Adaptive Testing: CAT Fit Analysis

Fit is inferred from the consistency between examinee responses and item difficulties. Improbable responses signal inconsistent activity. Computer assisted testing (CAT) usually tailors item difficulties to match examinee ability, but when examinees are administered items selected to give them a 50% chance of success, both right and wrong answers become equally probable, precluding a basis for the usual fit analysis.

There are, however, two statistics which, in combination, can help detect misfit in CAT. The first is the observed percent correct on the test. If CAT targeting is set for 50% success, then instead of reviewing a person's response pattern item by item, we can compare the obtained percent correct with the intended 50%. A significant difference implies a fit problem.

The second statistic is the mean of the differences, (B-D), across all items administered to a person, between the person measure, B, estimated at each step, and the item difficulty, D, of that step. This mean (B-D) difference indicates how well the item bank has provided on-target items. When targeting has been set at 50%, its expected value is zero. When, however, the items provided have been too easy, the difference will be greater than zero, and vice-versa.

The mean (B-D) difference can be converted to an "expected" percent correct, e.g. when the mean difference is +.4 logits, the expected percent correct is 60%. If the expected percent correct deviates from 50%, the item bank is not providing correctly targeted items, and the examinee's expected success rate is no longer 50% but the expected percent correct. Failure of the item bank to provide correctly targeted items usually occurs for very high or very low ability examinees.

Person misfit is implied when the expected percent correct is near 50%, but the examinees observed success rate deviates from 50%. The item pool has provided correctly targeted items but the behavior of the examinee has been inconsistent.

In summary, comparison of observed percent correct with expected percent correct provides a useful check on item bank depth and permits the identification of some types of examinee inconsistency.

Computer-Adaptive Testing: CAT Fit Analysis, B Bergstrom … Rasch Measurement Transactions, 1990, 4:3 p. 112

Please help with Standard Dataset 4: Andrich Rating Scale Model

Rasch Publications
Rasch Measurement Transactions (free, online)	Rasch Measurement research papers (free, online)	Probabilistic Models for Some Intelligence and Attainment Tests, Georg Rasch	Applying the Rasch Model 3rd. Ed., Bond & Fox	Best Test Design, Wright & Stone
Rating Scale Analysis, Wright & Masters	Introduction to Rasch Measurement, E. Smith & R. Smith	Introduction to Many-Facet Rasch Measurement, Thomas Eckes	Invariant Measurement: Using Rasch Models in the Social, Behavioral, and Health Sciences, George Engelhard, Jr.	Statistical Analyses for Language Testers, Rita Green
Rasch Models: Foundations, Recent Developments, and Applications, Fischer & Molenaar	Journal of Applied Measurement	Rasch models for measurement, David Andrich	Constructing Measures, Mark Wilson	Rasch Analysis in the Human Sciences, Boone, Stave, Yale
in Spanish:	Análisis de Rasch para todos, Agustín Tristán		Mediciones, Posicionamientos y Diagnósticos Competitivos, Juan Ramón Oreja Rodríguez

Forum

Rasch Measurement Forum to discuss any Rasch-related topic

Go to Top of Page
Go to index of all Rasch Measurement Transactions
AERA members: Join the Rasch Measurement SIG and receive the printed version of RMT
Some back issues of RMT are available as bound volumes
Subscribe to Journal of Applied Measurement

Go to Institute for Objective Measurement Home Page. The Rasch Measurement SIG (AERA) thanks the Institute for Objective Measurement for inviting the publication of Rasch Measurement Transactions on the Institute's website, www.rasch.org.

The URL of this page is www.rasch.org/rmt/rmt43b.htm

Website: www.rasch.org/rmt/contents.htm