On May 4, 2:17 pm, zim_...@hotmail.com wrote: > Hi, > I'm looking to test for a difference in center of mass between two groups of points. Each point has an x,y,z coordinate and I need to know if the means of the two clouds are significantly different, under the assumption that they are approximately Normally distributed. Can anyone point me to a test I can use? > thanks! > > ~z
When there are only two groups, there are three algebraically equivalent ways to test the hypothesis that the group centroids are the same. The easiest is probably multiple linear regression, with group as the dependent variable and the coordinate values as predictors. The F-test of the multiple R^2 is exactly equal to the F that you would get from a discriminant analysis or a one-way manova.