How the method used to measure refractive error is expected to affect statistical results

Values of refractive error (RE) observed for subjects and groups depend on the degree to which the measurement method used relaxes accommodation. This affects statistics calculated on RE measurements in predictable ways. Comparing noncycloplegic (NC) with cycloplegic (C) methods, myopia is expected to be more frequent, differences between group means should be less, and correlations and regressions relating RE to other variables may be reduced slightly. Tests based on the categorization of RE will show weaker associations with putative influencing variables. The implications of this are discussed for comparing published studies and choosing measurement methods in epidemiological and genetic studies.