Information Technology Reference
In-Depth Information
Extreme Values
a
therapy
1
Case Number
Value
GAFscore
Highest
1
2
1
2
1
2
1
2
1
2
1
2
4
72
3
68
Lowest
1
50
2
55
2
Highest
9
55
7
50
Lowest
6
35
8
39
3
Highest
11
72
15
71
Lowest
12
69
70
b
14
a.
The requested number of extreme values exceeds the number
of data points. A smaller number of extremes is displayed.
b.
Only a partial list of cases with the value 70 are shown in the
table of lower extremes.
Figure 5.14
Extreme values output from
Explore
.
output as it pertains to cases that have
Extreme Values
on the depen-
dent variable partitioned by levels of the independent variable is shown
in Figure 5.14, which lists the two largest and the two smallest cases with
Extreme Values
on our continuous dependent variable (
GAFscore
)bro-
ken by levels of the independent variable (
therapy
). Note that the SPSS
default is normally the highest and lowest
five
cases, which was truncated
as a result of the small sample size. The output in Figure 5.14 provides
both the
ExtremeValues
and the
CaseNumber
to facilitate evaluation and
identification of potential outliers. In the present example, we deem these
values to be within published ranges, and hence we can ignore them. Con-
versely, had we discovered any unusual outliers, we could have considered
them possible candidates for deletion.
5.5.4 SPSS NORMALITY ANALYSIS
We begin our assessment of normality by examining the skewness and
kurtosis values of our continuous dependent variable (
GAFscore
) parti-
tioned by the three levels of the independent variable (
therapy
)asshown
in Figure 5.15. An examination of the
Descriptives
output indicates that
the skewness values are within the normal
1.0 range. However,
negative kurtosis is associated with the
brief
and
psychodynamic
conditions
(
+
1.0 to
−
1.557
, respectively). We suspect that this kurtosis value is
more a function of the small sample size (
n
−
1.869
and
−
5) than the manipulation
itself, and so we do not attempt to transform the dependent variable in
the present example. For discussion of how to make these transformations
with SPSS, see Meyers et al. (2006).
=
