Abstract
PURPOSE: To test the assumptions underlying Likert scoring of visual function questionnaires. METHODS Questionnaires were administered to 284 low-vision subjects by telephone. Each subject was administered two of four questionnaires: ADVS, NEI VFQ-25 plus supplement, expanded VAQ, and VF-14. RESULTS: Z-scores for cumulative frequency of using each rating category across subjects are not linear with rating category rank and items are not the same difficulty for any of the questionnaires. Guttmann coefficients of reproducibility ranged from 57% for the ADVS to 51% for the NEI VFQ-25. Cronbach alphas ranged from 0.92 for the VF-14 to 0.96 for the NEI VFQ; however, inter-item consistency coefficients ranged from 0.24 for the VAQ to 0.45 for the NEI VFQ. Likert scores were significantly correlated between instruments, ranging from 0.66 for NEI VFQ vs ADVS to 0.90 for the VF-14 vs. ADVS. CONCLUSIONS: The rating scales of all four questionnaires fail to satisfy Likert's assumptions. Also, ratings are probabilistic, rather than deterministic, which means that the Likert model is not valid for these questionnaires. However, Likert scores for all four instruments are intercorrelated, suggesting that they are monotonic with the latent subject trait distributed in the low vision sample.
Original language | English (US) |
---|---|
Pages (from-to) | 381-399 |
Number of pages | 19 |
Journal | Ophthalmic Epidemiology |
Volume | 11 |
Issue number | 5 |
DOIs | |
State | Published - Dec 2004 |
Keywords
- ADVS
- Likert scaling
- NEI VFQ
- VAQ
- VF-14
- Vision impairment
- Visual function questionnaires
- Visual function rating scales
ASJC Scopus subject areas
- Epidemiology
- Ophthalmology