Scatterplots are a convenient tool to represent the relationship between two continuous variables. Often, it is necessary to classify this relationship according to values of a third categorical variable. There are several ways to do this (see, for example, Cox [2005]). Here we consider a variation of these graphs, sometimes referred to as bubbleplots, where an additional dimension of the data is represented in the size of the markers. In Stata, one can create such a graph by explicitly specifying a weight in the standard scatterplot syntax (see [G-2] graph twoway scatter). As an example, we will use auto.dta. Suppose we want to create a scatterplot of mileage and weight with markers weighted by repair record. Suppose further that we want to compare domestic (American) and foreign cars. A problem arises if we do not observe all values of the weighting variable in each of the groups defined by the categorical variable. Consider a cross-tabulation of repair record and car type.1