Graphing Multiple Confidence Intervals

This study proposed a graphical method for displaying informative confidence intervals for any data set involving a categorical determinant, an outcome that could be either continuous or categorical, and an optional covariate. For continuous outcomes the graph is similar to the conventional plot of confidence intervals for individual means, but focuses on specified natural contrasts. The analogous graph for categorical outcomes involves graphing confidence intervals for specified natural odds ratios. In each case the method can be extended to include a covariate, and the extent of confounding can be illustrated on the graph. The confidence intervals are adjusted for multiplicity.