Key points are not available for this paper at this time.
results, one is immediately impressed with the power of graphical representation as a means of conveying relatively complex ideas and relationships. Unfortunately, as soon as the analysis involves more than two variables this valuable tool is no longer available except for showing isolated partial projections, etc. The discussion which follows does not attempt to draw n-dimensional diagrams but it does follow the line of seeking a graphical representation for some complex ideas involving the interrelations among several variables. To be more specific, in multiple regression analysis non-orthogonality among regressors makes comprehension of the meaning of statistical tests and understanding of the consequences of changing the set of regressors very difficult. These topics are so difficult that many students carry away from statistics courses the notion that nonorthogonality is a nasty aberration instead of the usual and only case worth serious consideration. Similar difficulties are often encountered in understanding the intricate relations among simple, partial and multiple correlations. The device proposed below has been found useful in helping students gain insight into the inter-relations among measures of inter-relatedness as well as in providing a concise and easily assimilated summary of statistical information. By way of illustrating its usefulness, an application of the tool is provided at the end.
Harold W. Watts (Fri,) studied this question.