We compared the distribution of possible confounding variables between subjects with emmetropia, mild myopia, moderate myopia, severe myopia, and hyperopia using design-adjusted Rao-Scott Pearson-type χ2 and Wald tests for categorical and continuous variables, respectively. Multivariate logistic regression models were created to examine the independent association between refractive status and the binary outcome variables self-reported glaucoma, any visual field defects, and optic nerve vertical cup-to-disc ratio greater than or equal to 0.7, while adjusting for potential confounders. Odds ratios (OR) and 95% confidence intervals (CI) were calculated. A multivariate linear regression model was created to examine the independent association between refractive error and vertical cup-to-disc ratio as continuous numerical variables, while adjusting for potential confounders; subjects with hyperopia were excluded. We also performed a design-adjusted Rao-Scott Pearson-type χ2 test to determine if subjects with normal visual fields, early visual field defects, moderate visual field defects, and severe visual defects were distributed differently across subjects with emmetropia, mild myopia, moderate myopia, severe myopia, and hyperopia. To most accurately calculate CIs around estimates for the US national population, we performed all data analyses (Stata 12.0; Stata Statistical Software, College Station, TX) using weighted data, and SEs of population estimates were calculated by Taylor linearization methods.