Purchase this article with an account.
Marc M. Schulze, Trefford L. Simpson, Paul Stolee; Rasch Analysis Of Bulbar Redness Scaling And The Relationship Between Physical Redness (u’) And "Rasch Redness". Invest. Ophthalmol. Vis. Sci. 2011;52(14):1966.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To use Rasch Analysis to specifically examine the dimensionality of grading of the redness of images of the bulbar conjunctiva and examine the relationship between the Rasch scores of redness and the chromaticity of the images.
25 images of the bulbar conjunctiva were arranged on a table in order of increasing redness by 9 subjects naive about the purpose of the experiment. The position settings (redness ratings) of each subject were divided into 10 bins and Rasch Analysis was conducted using Winsteps software. Infit and outfit statistics (characterizing the discrepancy between the Rasch model and the ratings) were used to determine the dimensionality of the data and then the linear correlation between the Rasch transformed ratings and image chromaticity (CIE u’, measured using a Photo Research PR-650 spectroradiometer) was estimated using Pearson correlation coefficient.
Bulbar redness ratings were unidimensional. For the 10 binned scale mean (+SD) infit was 0.98+0.54 and mean (+SD) outfit was 0.97+0.55. The correlation between the Rasch scores of redness and the chromaticity was 0.93, between untransformed redness and Rasch redness was 0.98, and between untransformed redness and chromaticity was 0.92 (all p’s < 0.05).
Although Rasch analysis was useful in demonstrating the unidimensionality of the ratings (that is the redness ratings were in accord with Rasch theory), in this experiment, Rasch transformation of the data did not appear to provide much benefit perhaps because the untransformed ratings themselves were very well correlated with the physical redness.
This PDF is available to Subscribers Only