Visual acuity: single-run confidence interval obtained via bootstrapping, implemented in the Freiburg Vision Test.

Abstract

Whenever measuring, it is good practice to report not only an outcome point value, but also a measure of dispersion. This is rarely done for visual acuity. We here demonstrate that a confidence interval (CI) can be obtained for visual acuity based on all staircase responses. The CI is obtained via bootstrapping to construct an estimate of the pragmatic distribution of outcomes. The method has been implemented in FrACT₁₀ (Freiburg Vision Test) and can calculate the 95% CI based on 18 trials, sampled 10,000 times. In two large datasets, the sampling-based confidence intervals were found to be very close to the test-retest limits of agreement (based on repeat runs). The bootstrapping approach might fruitfully be generalized to other psychophysical thresholds.