Making Fixed-Precision Between-Item Multidimensional Computerized Adaptive Tests Even Shorter by Reducing the...

Area of Science:

• Psychometrics
• Educational Measurement
• Computerized Adaptive Testing

Background:

• Fixed-precision multidimensional computerized adaptive tests (MCATs) are increasingly utilized.
• Current item selection rules optimize multivariate precision, while stopping rules target fixed marginal precision per dimension.
• This creates an asymmetry between selection and stopping criteria in MCATs, unlike in unidimensional tests.

Purpose of the Study:

• To address the selection-stopping asymmetry in fixed-precision between-item MCATs.
• To introduce and evaluate novel item-selection approaches.
• To develop effective MCATs with fixed marginal precision that are shorter in test length.

Main Methods:

• Discussion of the selection-stopping asymmetry and its implications.
• Introduction of three alternative, computationally inexpensive item-selection methods.
• Evaluation of the proposed methods for fixed-marginal-precision MCATs.

Main Results:

• The proposed alternative item-selection approaches are computationally inexpensive and easy to implement.
• These methods result in effective fixed-marginal-precision MCATs.
• The new approaches yield shorter test lengths compared to current methods.

Conclusions:

• The identified selection-stopping asymmetry in MCATs can be effectively addressed.
• Alternative item-selection strategies offer practical advantages for developing efficient MCATs.
• The proposed methods improve the precision and reduce the length of fixed-precision MCATs.