Purchase this article with an account.
David A. Leske, Sarah R. Hatt, Laura Liebermann, Jonathan M. Holmes; Evaluation of the Adult Strabismus-20 (AS-20) Questionnaire using Rasch Analysis. Invest. Ophthalmol. Vis. Sci. 2011;52(14):5562.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Adult Strabismus-20 (AS-20) questionnaire is a 20-item patient-derived strabismus-specific instrument designed to evaluate health-related quality of life in adults with strabismus. The present study evaluates the AS-20 using Rasch analysis in a cohort of adult strabismus patients.
348 adult strabismus patients completed the AS-20 at the time of their clinic examination. Rasch analysis was performed using Winsteps software (ver 3.70), evaluating response ordering, local dependence, fit, differential item functioning (DIF; assessed for gender and age), targeting, and dimensionality. Since the AS-20 was designed to have 2 subscales (psychosocial and function), these two dimensions were analyzed separately.
Each of the 2 subscales met requirements for unidimensionality. For the psychosocial subscale, response categories were properly oriented and distributed. There was no local dependency of items. Five of ten items had suboptimal infit (mean squares >1.5) and were removed sequentially. DIF indicated no differential functioning by age but marginal differential functioning by gender (0.63 logits) for 1 item (not subsequently removed). The resulting 5-item psychosocial scale had appropriate targeting and demonstrated unidimensionalilty (77.2% of variance explained by measures). The 5-item psychosocial scale had a person separation index of 3.12 and reliability of 0.91. For the function subscale, response categories were ordered but the "rarely" response category was underutilized and therefore was combined with the "never" response. No local dependence was seen, but one item showed misfit and was removed. DIF indicated no differential functioning by gender or age. The resulting 9-item function scale demonstrated appropriate targeting and unidimensionality (58.2% of variance explained by measures). The 9-item function scale had a person separation index of 2.46 and reliability of 0.86.
Rasch analysis of the original 20-item AS-20 indicated that the questionnaire would benefit from item reduction in both subscales and condensing of the response categories in the function subscale. These changes to the AS-20 should result in an even more valid instrument.
This PDF is available to Subscribers Only