Analysis of data from the International Outcome Inventory for Hearing Aids (IOI-HA) using Bayesian Item Response...

Area of Science:

• Audiology
• Psychometrics
• Statistical analysis

Background:

• Hearing aid outcome measures are often analyzed assuming interval-scale properties for ordinal responses.
• The Hearing Aid Outcome Instrument (IOI-HA) is a commonly used questionnaire for assessing hearing aid outcomes.

Purpose of the Study:

• To critically evaluate the assumption of interval-scale properties for IOI-HA ordinal responses.
• To compare conventional analysis methods with Item Response Theory (IRT) for IOI-HA data.

Main Methods:

• A Bayesian IRT analysis model was implemented.
• The model was applied to anonymized IOI-HA response data from 13,273 adult hearing aid users across 11 datasets.
• Data included Australian English, Dutch, German, and Swedish versions of the IOI-HA.

Main Results:

• Raw IOI-HA ordinal responses do not possess interval-scale properties.
• Conventional sum scores introduce a scale error of approximately 10-15% of the population's true standard deviation.
• Statistically credible differences were observed among the analyzed datasets.

Conclusions:

• Applying conventional statistical measures (mean, variance, t-tests) to raw IOI-HA ratings is questionable.
• Nonparametric statistical methods are recommended for group comparisons of IOI-HA results.
• The IRT approach is recommended for accurate analysis of individual IOI-HA results to avoid scale errors and potential incorrect conclusions.