Graphics Reference
In-Depth Information
Figure
.
.
Bar plot for the hospital data
Figure
.
.
-D-bar chart for the hospital data
provides more detailed information on this topic. Consider our example of the hos-
pital data from above. Step
consists of splitting a square according to the marginals
of one of the variables. To be consistent with the textual representation, we choose
LOS with vertical splits (see Fig.
.
). he result is similar to a bar plot, but in this
case the width rather than the height is adapted to visualize the counts foreach level.
Such a plot is also called a spine plot (Hummel,
). From this plot, we see that
the number of patients decreases with the length of stay. Step
involves performing
further splitting in the other direction (resulting in horizontal splits) for the second
variable. his means that each vertical bar is split according to the marginals of the
secondvariable, giventhefirstvariable (seeFig.
.
).heresultingplotvisualizes the
contingency table suchthat each cellhas a size proportional tothe corresponding ta-
