Development and Simulation Testing of a Computerized Adaptive Version of the Philadelphia Naming Test

Area of Science:

• Neuroscience
• Psychometrics
• Speech-Language Pathology

Background:

• The Philadelphia Naming Test (PNT) is a widely used tool for assessing naming abilities.
• Assessing naming in individuals with aphasia can be lengthy and burdensome.
• Item-response-theory (IRT) models offer potential for optimizing test efficiency.

Purpose of the Study:

• To develop a computerized adaptive test (CAT) version of the PNT.
• To reduce the length of the PNT while maintaining or improving measurement precision.
• To extend previous work fitting the PNT to a 1-parameter logistic IRT model.

Main Methods:

• Archival data from participants with aphasia were used.
• Two PNT-CAT versions were simulated.
• Simulated PNT-CAT ability estimates were compared to full PNT scores and static short forms.
• A jackknife procedure ensured sample independence for item estimation and CAT simulation.

Main Results:

• The PNT-CAT demonstrated equal or superior accuracy in recovering full PNT scores compared to static short forms.
• Measurement precision was higher for the PNT-CAT than for static short forms.
• Minimal differences were observed between adaptive and static non-overlapping alternate forms.

Conclusions:

• Computerized adaptive testing (CAT) shows promise for efficient naming assessment in aphasia.
• PNT-CAT has the potential to reduce patient test burden.
• CAT maximizes the accuracy and precision of naming ability score estimates in aphasia.