EFFECTPLOT Statement

The EFFECTPLOT statement produces a display (effect plot) of a complex fitted model and provides options for changing and enhancing the displays. One simple effect plot is the display for a linear regression of the response on a single predictor : the regression line is drawn with the predicted response on the Y axis and the covariate on the X axis. The regression line can be enhanced by displaying the observations and adding confidence and prediction limits. When your model is more complicated—with more continuous and categorical covariates, nestings and interactions, and link functions—the effect plots display the behavior of some covariates over their ranges while fixing other covariates at some fixed values; this can enable easier interpretation and explanation of the resulting model.

By default, a single plot is produced based on the type of response variable and the number of continuous and classification covariates in the model. You can also specify options to do the following:

• select the variables to display on the plots

• produce multiple plots based on the following: the levels of classification covariates; the minimum, maximum, mean or middle (midrange) value of continuous covariates; and specified values of the covariates

• specify different fixed values for continuous and classification covariates that are not displayed on the plot

• panel and unpanel plots

• select variables to slice or group by

• display (or remove from display) observations and confidence limits