C W McMonnies1, A Ho. 1. School of Optometry, University of New South Wales, Australia.
Abstract
INTRODUCTION: We tested the appropriateness of the assumption that charts, which only vary in having different sequences of five letters per line, chosen from the same 10 letter set, can be regarded as equivalent for the purpose of making valid comparisons of visual acuity. METHOD: Visual acuity findings from samples of 400 patients, for each of two nominally equivalent Bailey-Lovie charts, have been used to determine the relative legibility of individual letters and lines, using the percentage incorrect method of analysis. RESULTS: The chance of error for the hardest letter is approximately 13 times greater than for the easiest letter on each chart. Significant between-chart differences in error rates for the hardest letters ('F' and 'H') were found. Some letters that have adjacent ranks on the legibility scale were found to have significantly different legibility. Significant differences in difficulty can occur for the same nominal line on apparently equivalent charts because of chance combinations of easier or harder letters in that line. Uneven line-to-line scale intervals have been confirmed for the charts examined, by showing corresponding differences in the distributions of lines of threshold acuity. CONCLUSIONS: The use of varying examination distances may be inappropriate for Bailey-Lovie (or similar) charts that depart from their nominal interval scaling. When different versions of these charts are assumed to be equivalent, the discrepancy between repeated measurements may be significantly increased. There is the possibility of increasing measurement precision using charts having different sequences of the same 10 letters in each line to achieve equal scaling of line intervals. Equivalent charts can be validly constructed using different sequences of the same 10 letters in each line.
INTRODUCTION: We tested the appropriateness of the assumption that charts, which only vary in having different sequences of five letters per line, chosen from the same 10 letter set, can be regarded as equivalent for the purpose of making valid comparisons of visual acuity. METHOD: Visual acuity findings from samples of 400 patients, for each of two nominally equivalent Bailey-Lovie charts, have been used to determine the relative legibility of individual letters and lines, using the percentage incorrect method of analysis. RESULTS: The chance of error for the hardest letter is approximately 13 times greater than for the easiest letter on each chart. Significant between-chart differences in error rates for the hardest letters ('F' and 'H') were found. Some letters that have adjacent ranks on the legibility scale were found to have significantly different legibility. Significant differences in difficulty can occur for the same nominal line on apparently equivalent charts because of chance combinations of easier or harder letters in that line. Uneven line-to-line scale intervals have been confirmed for the charts examined, by showing corresponding differences in the distributions of lines of threshold acuity. CONCLUSIONS: The use of varying examination distances may be inappropriate for Bailey-Lovie (or similar) charts that depart from their nominal interval scaling. When different versions of these charts are assumed to be equivalent, the discrepancy between repeated measurements may be significantly increased. There is the possibility of increasing measurement precision using charts having different sequences of the same 10 letters in each line to achieve equal scaling of line intervals. Equivalent charts can be validly constructed using different sequences of the same 10 letters in each line.