To the Editor. —We would like to draw attention to the statistical interpretation of data represented by Quigley et al¹in the October 1991 issue of theArchives. In Fig 1, the relationship between optic nerve fibers and optic disc is plotted with a linear regression correlation coefficient (r=.47,P<.05). On examination of the plot, these data seem to have a poor linear fit and, in fact, thervalue of.47 confirms this observation. Ifr=1, there is a perfect linear relationship and all the points lie on a straight line. Ifr=0, there is no linear relationship. Even though the correlation coefficient is greater than zero and is significant, this does not imply a good linear relationship. The real strength of a linear relationship is best indicated byR², which is a measure of the proportion of variation accounted for in the linear regression.
No takes yet. Share an insight, caveat, or question.
Francis Y. Falck (1992) studied this question.