Statistical Methods for Vectorcardiographic Directions

Current statistical methods for treating vectorcardiographic directions are invalid because they treat the directions as points on a line or plane. It is more appropriate to consider the directions as points on a circle or sphere; the direction toward the center of gravity of such points is then a measure of the prevalent direction, and the distance from the center of the circle or sphere to the center of gravity is a measure of precision. The precision is a measure of how closely the directions cluster about the prevalent direction. The Fisher distribution is suggested as a probability model for spatial directions, and its usefulness is illustrated by 1) a test statistic for determining whether two samples of directions come from the same population, 2) the construction of a ``confidence circle' for the spatial prevalent direction, and 3) a method for obtaining planar prevalent directions and precisions from the spatial ones.