stepsetup([Model] [,silent:T] [,allin:T or in:logVec]), Model of form "y = x1+x2+...+xk" or "y = x1+x2+...+xk - 1", logVec a LOGICAL vector of length k |

stepsetup(Model), where Model is a CHARACTER scalar specifying a regression model, initializes a stepwise regression process. It creates invisible variable _STEPSTATUS and prints the F-to-enter statistics and P-values for all the variables in the model. It returns _STEPSTATUS as value. This can be assigned (stuff <- stepsetup("y=x1+x2+x3+x4")) but is not printed. See topic '_STEPSTATUS'. stepsetup(), with no model specified, does the same, except it uses STRMODEL, the model for the most recent GLM command, as model. stepsetup([Model,] silent:T) does the same except nothing is printed. stepsetup([Model,] allin:T [,silent:T]) does the same, except all the variables are entered in the model immediately so that backward stepwise regression can be done. stepsetup([Model,] in:In [,silent:T]) does the same, except the variables specified by In are entered in the model immediately. In must be a LOGICAL vector the same length as the number of independent variables in Model. Variable j is considered in the model if In[j] is True. When variables are initially entered in the model, stepsetup() prints an overall F statistic and its P-value, Mallow's Cp statistic, adjusted R^2 and R^2. The F-statistic tests the null hypothesis that the coefficients of the "in" variables are 0. Examples: Cmd> stepsetup("y=x1+x2+x3+x4+x5") starts stepwise regression process with no variables in the model. Cmd> stepsetup("y=x1+x2+x3+x4+x5+x6-1",in:vector(F,F,F,T,T,T)) starts stepwise regression process with variables x4, x5 and x6 in the model. Because of "-1" in the model, no intercept is included. See also stepstatus(), entervar(), removevar(), steplook().

Gary Oehlert 2003-01-15