Objective: To test the validity and reliability of latent trait measures estimated from ratings by low-vision patients of the importance and difficulty of selected activity goals.
Design: Validation of a telephone-administered functional assessment instrument using Rasch analysis of self-assessment ratings.
Setting: Telephone interviews of respondents in their homes. Participants Consecutive series of 600 outpatients with low vision.
Interventions: Not applicable.
Main outcome measures: Ratings of the importance and difficulty of achieving 41 activity goals. Person and item traits were measured with the Andrich rating scale model. Measurement validity and reliability were tested statistically by comparing response patterns and distributions with measurement model expectations.
Results: Patients could distinguish only 3 categories of importance and 4 categories of difficulty. The distributions of person and item measure fit statistics were consistent with 2 unidimensional constructs: value of independence estimated from importance ratings and visual ability estimated from difficulty ratings. However, 8 of 41 activity goals were poor estimators of value of independence and 7 of 41 activity goals were poor estimators of visual ability. Person measure distributions could be divided into 3 statistically distinct strata for estimates from both importance ratings and difficulty ratings. Item measure distributions could be divided into 21 strata for estimates from importance ratings and 7 strata for estimates from difficulty ratings.
Conclusions: The 2 variables that define visual disability-value of independence and visual ability-are valid constructs that can be estimated accurately and reliably from patient ratings of the importance and difficulty of activity goals.