Purchase this article with an account.
Gislin Dagnelie, Pamela E Jeter, Kemi Adeyemo, Collin Rozanski, Amelie-Francoise Nkodo, Robert W Massof, ; Psychometric properties of the PLoVR ultra-low vision (ULV) questionnaire. Invest. Ophthalmol. Vis. Sci. 2014;55(13):2150.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To calibrate the initial version of a survey for individuals with ultra-low (<20/500) vision assessing the difficulty of performing everyday visual tasks. This survey was developed as part of the Prosthetic Low Vision Rehabilitation (PLoVR) curriculum. Current standardized visual functioning questionnaires include few if any items that can be accomplished with rudimentary vision.
Based on transcripts from focus groups representative of the ULV population, a survey of 149 items was developed, collecting difficulty ratings for tasks classified as relying on high (N=63) or low (N=22) contrast, lighting conditions (N=27), or form and movement (N=37). The initial survey was administered on-line or by phone to 81 individuals with minimal or formerly minimal vision. A Rasch analysis was performed on all data combined, and on each of the 4 item classes, yielding person ability and item difficulty measures on a logit scale. An iterated principal axis factor analysis was applied to the person measures obtained for the item subclasses to examine whether the emerging factors differentially affect the item classes.
The overall Rasch analysis result showed high internal consistency in person (0.99) and item (0.96) reliability scores, which was confirmed by the infit statistics: Only 4 persons and 4 items had Underfit Z-scores >4. The item subclass factor analysis showed high internal correlations (0.917 - 0.990) between item classes, and between each class and the overall item bank. Varimax rotation confirmed that under no transformation the differential factor loadings were significant. Item measures in this initial survey did not encompass the full range of person measures, and the median person measure falls 1.6 logits below the median item measure; this indicates a need for item modification towards even less demanding tasks.
The first version of the PLoVR ULV questionnaire demonstrates excellent psychometric properties, with the exception of 4 underfit items, but is lacking in items with minimal visual demands. In developing the 2nd version the emphasis will be on shifting items towards less demanding tasks.
This PDF is available to Subscribers Only