This content is PDF only. Please click on the PDF icon to access.
Abstract
An apparently common error in statistical analysis of ophthalmic data is to perform statistical tests that do not account for the correlation generally present between observations made for the right and left eyes of a subject. This error has as a consequence an overstatement of the precision of the study, resulting in incorrect P values which indicate a greater measure of statistical significance than the data warrant. As measures to reduce the occurrence of this serious error in statistical analyses, the authors recommend increased emphasis on educational programs for investigators, stimulation of nontechnical articles reviewing statistical methods, and a sharper focus upon statistical analysis in the peer review process.