A Modified Doolittle Approach for Multiple and Partial Correlation and Regression

A method for obtaining multiple and partial correlation and regression coefficients on desk calculators is presented that is efficient and highly flexible. Insofar as possible, the computations parallel those used when fitting interrelated systems of equations by the limited information approach. This method is easier for beginners to understand than the usual Doolittle approach because no back solution is required. Efficient methods for (1) coding the data, (2) interchanging, adding or eliminating variables, and (3) obtaining standard errors of the function or of forecasts are included.