MDSpire - Summary
From the Journals

Accounting for Eye Correlation Improves Statistical Accuracy

Share

A simulation study in Optometry and Vision Science examines the impact of statistical approaches on ophthalmic research, particularly concerning correlations between a patient's two eyes. The study, conducted by researchers from Southern College of Optometry and the University of Memphis, evaluated statistical methods including single-eye analysis, averaging, mixed effects models, and generalized estimating equations (GEEs). Findings indicate that treating the eyes as independent can inflate false positives and mislead researchers. Mixed effects models and GEEs are recommended for eye-level predictors, while averaging is suitable for subject-level predictors, emphasizing the importance of proper model selection.

Original Source(s)

Related Content