Mr1—Multiple Regression Analysis on the IBM 7070

MRl will compute the appropriate statistics to a multiple regression analysis for up to 129 independent variables. It operates in a stepwise fashion, first computing these statistics for the one “bestJJ i.v. It then selects the “best” of the remaining variables from the pool of independent variables (from which have been partialed out the first variable that had been selected). I n this fashion, i.v,’s may be designated as “in7’ or Grout" of regression, and the program attempts to approximate the optimum set of i.v.’s by “entering into regression,” one variable per cycle, all i.v.’s satisfying a specified criterion of significance. During each cycle, the program may also “delete from regression” any variable previously entered which, as a result of other variables being entered, is no longer significant.