Information Technology Reference
In-Depth Information
Table 11.1. Summary table for within-subjects and between-subjects effects for the two-way
numerical example
Between-subjects effects
Source
SS
df
MS
Subject differences ( S )
30.222
5
6.044
Within-subjects effects
η
2
Complete
Source
SS
df
MS
F
Sig.
Partial
Vehicle ( A )
2.778
1
2.778
1.761
0.242
Error ( A
×
S )
7.889
5
1.578
Drinks ( B )
157.389
2
78.694
52.658
0.000
.913
.774
Error ( B
×
S )
14.944
10
1.494
A
×
B
17.056
2
8.528
26.017
0.000
.839
.084
Error ( A
×
B
×
S )
3.278
10
0.328
Total within subjects 203.334
Complete based on the total within-subjects sum of squares.
11.3.2 WITHIN-SUBJECTS EFFECTS
The lower portion of Table 11.1 shows the within-subjects effects and sum-
marizes the within-subjects partitioning of the variance. Several aspects
of the table are worthy of note; we discuss these next in the context of the
effectsinwhichweareinterested.
11.4 EFFECTS OF INTEREST IN THIS DESIGN
Whether the study represents a between-subjects or a within-subjects
design, with two independent variables in the design there are three effects
of interest: the main effect of one independent variable, the main effect
of the other independent variable, and the interaction of the two inde-
pendent variables. In a within-subjects design, all three of these effects are
subsumed in the within-subjects variance.
11.4.1 MAIN EFFECTS
As can be seen from Table 11.1, the main effect of Vehicle is not statistically
significant. We therefore conclude that driving either a sports sedan or an
SUV does not result in a different number of driving errors overall.
The main effect of Drinks is statistically significant. This indicates that
the number of driving errors is related to the amount of alcohol students
drank. With three levels of this variable, however, we need to perform
pairwise comparisons of the means to determine which are significantly
different from each other. Because the strength of this effect as indexed
by eta squared is so substantial (as can be seen in Table 11.1), we would
probably want to follow up on this main effect even though we obtained
a significant interaction.
 
Search WWH ::




Custom Search