60 MEASURES OF LOCATION AND VARIABILITY
Figure 5.1 Box plot of hemoglobin levels of high-altitude mine workers.
Note that continuous variables tend to be interval or ratio data. Categorical vari-
ables can be nominal data or ordinal data since these two types of variables often have
discrete outcomes.
In the SAS/JMP statistical package the user can specify what type of variables
using the Stevens system they have. Here they are called modeling types and the three
types that can be specified are nominal, ordinal, and continuous. This information is
entered into a row that is above the row used for variable names. When this is done,
the statistics that are computed are appropriate for the type of variable. SPSS also
allows identification by the Stevens system.
5.5 A COMMON GRAPHICAL METHOD FOR DISPLAYING
STATISTICSStem and leaf charts and histograms for displaying the observations were discussed
in Chapter 4. Here, we present displays of one other commonly used graph, the
box plot (originally called a box and whisker plot). Box plots were designed to
display medians, fourths, and extreme values (see Tukey [1977]), but the various
statistical programs do not all display the same quantities, so users should check what
is displayed in their programs. In the box plot displayed in Figure 5.1, the horizontal
line inside the box displays the median, the upper edge of the box is Q3, and the
lower edge of the box is Q1. The length of the box is the IQR. The vertical lines
above and below the box with the horizontal line at their end are the “whiskers.”
The horizontal lines are calledfences. The upper fence is at (Q3 + 1.5(IQR)) or the