Generalization of the Method As many business problems will yield to similar methods of analysis, the particular approach found useful here may be generalized as a sequence of steps as follows: (1) Following a study to determine the economics of the problem, a measure of effectiveness was selected. (2) A mathematical model of the problem was built around this measure of effectiveness and included, those variables which most appeared to influence the measure of effectiveness. (3) The coefficients in the model were chosen by mathematical manipulation (multiple regression) to make it as accurate a symbolic description as possible. (4) The model was "tested" by statistical means (correlation coefficient). (5) Again by a mathematical manipulation (differentiation), the model was minimized with respect to the factor to be used in the decision rule. That is, an expression for area was determined, which should give the minimum warehousing cost per dollar's worth of goods distributed in each particular area. (6) In the special case of the plant wares house area, an equation of marginal analysis was employed to establish the optimum radius to be served directly from the plant. Variations Within the Method Some variations within the same general framework of analysis might be considered. (a) The measure of effectiveness might have been cast per pound of goods distributed or some other similar criterion (b) The model might have included more variables such as warehouse design, etc. (c) The coefficients in the model might have been determined from an engineering or cost accounting type, of approach. From a chart of accounts and past records and budgets, a, b, c, etc., might have been determined. This type of approach is more common in business today. However, by definition (the statistician's), these values could have been no better and might have been poorer. (d) A tabular or graphical comparison of costs "predicted" by the model to actual warehouse costs might have been used rather than the correlation coefficient. (e) A tabular, graphical, or trial and error method might have been used to determine the A (area) which minimized the cost expression. However, this would vary with K--the sales density--and, therefore, it would have been necessary to repeat this procedure for selected values of K. (f) Rather than set up an equation for marginal analysis in the special case of the plant warehouse, it would again have been possible to tabulate, graph, or try many radius distances with their associated costs to determine the best one. Implications of the Method The length of time necessary and the accuracy of results for these different methods of analysis would probably vary with the person using them. It is suspected that the more conventional approach would have been substantially more time consuming and probably less precise. Possibly the best advice is: if the more elegant shoe fits, wear it. [ABSTRACT FROM AUTHOR]