outliers. However, boxplots are normally used as informal heuristic devices, and subtle dif-
ferences in definition are rarely, if ever, a problem. I mention the potential discrepancies
here simply to explain why analyses that you do on the data in this book may come up with
slightly different results if you use different computer programs.
The real usefulness of boxplots comes when we want to compare several groups. We
will use the example with which we started this chapter, where we have recorded the reac-
tion times of response to the question of whether a specific digit was presented in a previ-
ous slide, as a function of the number of stimuli on that slide. The boxplot in Figure 2.15,
produced by SPSS, shows the reaction times for those cases in which the stimulus was ac-
tually present, broken down by the number of stimuli in the original. The outliers are indi-
cated by their identification number, which here is the same as the number of the trial on
which the stimulus was presented. The most obvious conclusion from this figure is that as
the number of stimuli in the original increases, reaction times also increase, as does the dis-
persion. We can also see that the distributions are reasonably symmetric (the boxes are
roughly centered on the medians, and there are a few outliers, all of which are long reac-
tion times).
2.10 Obtaining Measures of Central Tendency and Dispersion Using SPSS
We can also use SPSS to calculate measures of central tendency and dispersion, as shown in
Exhibit 2.1, which is based on our data from the reaction time experiment. I used the
Analyze/Compare Means/Meansmenu command because I wanted to obtain the descriptive
statistics separately for each level of NStim (the number of stimuli presented). Notice that you
also have these statistics across the three groups. The command Graphs/Interactive/Boxplot
produced the boxplot shown below. Because you have already seen the boxplot broken down by
NStim in Figure 2.14, I only presented the combined data here. Note how well the extreme
values stand out.
Section 2.10 Obtaining Measures of Central Tendency and Dispersion Using SPSS 51
1
30.0
40.0
RxTime
50.0
60.0
70.0
80.0
90.0
100.0
3
NumStim
5
O12
O43
*^35
*^46
O140
O110
O102
O212
O239
Figure 2.15 Boxplot of reaction times as a function of number
of stimuli in the original set of stimuli
