Dichotomous logistic regression with leave-one-out validation

In this paper, the concepts of dichotomous logistic regression (DLR) with leave-one-out (L-O-O) were discussed. To illustrate this, the L-O-O was run to determine the importance of the simulation conditions for robust test of spread procedures with good Type I error rates. The resultant model was then evaluated. The discussions included 1) assessment of the accuracy of the model, and 2) parameter estimates. These were presented and illustrated by modeling the relationship between the dichotomous dependent variable (Type I error rates) with a set of independent variables (the simulation conditions). The base SAS software containing PROC LOGISTIC and DATA step functions can be making used to do the DLR analysis.