ABSTRACT
Vector geometry was ‰rst used in 1915 (Fisher 1915) by the great statistician Sir Ronald Fisher to address the problem of deriving the statistical distribution of the Pearson product-moment correlation coef‰cient. Fisher used his great geometric imagination to demonstrate that the correlation between two variables is the cosine function of the angle between two vectors in an n-dimensional space. However, as most statisticians do not have the same great geometrical insights as Fisher did, and most statisticians consider an algebraic approach more mathematically rigorous, the geometric approach has not thus far received as much attention as it perhaps deserves (Herr 1980). Nevertheless, some teachers of statistics have found a geometric approach to be more intuitive for people without a mathematical background and a very useful tool to develop the understanding of linear regression analyses. A few statistical textbooks (Saville and Wood 1991, 1996; Wickens 1995; Carroll et al. 1997) are written using mainly vector geometry, and some textbooks of statistics or econometrics on linear statistical models (Wonnacott and Wonnacott 1979, 1981; Fox 1997; Draper and Smith 1998) have chapters on vector geometry. However, vector geometry seems to have been used rarely to illustrate or explore speci‰c methodological problems within biomedical research.
