Short Forms and Computerized Adaptive Tests With Monosyllabic Words Can Efficiently Measure Speech Recognition.

Area of Science:

• Audiology
• Psychometrics
• Speech-Language Pathology

Background:

• The construct validity of Northwestern University Auditory Test No. 6 (NU-6) word lists and their shortened versions has not been evaluated using Rasch measurement theory.
• Assessing speech recognition accurately is crucial for diagnosing and managing hearing loss.

Purpose of the Study:

• To evaluate the fit of Rasch measurement models to NU-6 monosyllabic word lists.
• To determine if shortened NU-6 word lists and computerized adaptive testing (CAT) can effectively measure speech recognition.

Main Methods:

• A cross-sectional study involved 50 participants with mild-to-moderate hearing loss.
• NU-6 Word Lists 1 and 2 were tested for Rasch model fit.
• Pearson correlations and standard error measures assessed the accuracy and precision of short forms and CAT simulations against full lists.

Main Results:

• NU-6 word lists demonstrated unidimensionality with minimal misfit and high person reliability.
• 19-word short forms showed high linear association (r > .91) with full lists, while 11-word forms had moderate association (r > .81).
• CAT simulations achieved a correlation threshold (r > .90) after 15-20 words, with precision-based stopping rules using 18-20 words.

Conclusions:

• 19-word short forms of the NU-6 and CAT are viable methods for accurately and precisely measuring speech recognition.
• These findings support the use of efficient measurement tools for speech recognition assessment in audiology.