114 
CHAPTER 3. EXPERIMENTS 
Table 3.36: 
ANOVA for Kendall's ( (Exp 34) 
Source 
85 
MS 
E 
4 
Between subjects: 
0.255 
Block Design (BD) 
0.128 
2 
0.99 
3.479 
Error 
27 
0.129 
Within subjects: 
0.088 
Session 
0.044 
2 
2.11 
0.028 
BD by Session 
0.007 
0.25 
4 
1.542 
Session by Error 
0.029 
54 
Discrimination Cycles 
In Table 3.35 the average (-values with standard deviations for each group and session are 
listed. All groups had comparable average (-values in the first and more or less in the 
following sessions. In the first session subjects in Group S made choices under a resolution 
block design reaching an average (-value of .73, whereas subjects in Group P had an average 
value of .66 under the repetition block design. The subjects of Group N made choices under 
the random block design and reached a higher average (-value of .77. 
Table 3.37: Stepwise Discriminant Analysis on Dicycles (Exp 34) 
Removed 
Step Entered 
 
Number Partial R° Partial F pF 
Session 1 
No variables entered 
Session 2 
1 
0.143 
1 3-dicycles 
0.125 
2.250 
Session 3 
No variables entered 
The same type of analysis was applied to the number of intransitive triples using Kendall's 
(as a standardized measure (see Table 3.36).17 There was no significant effect of block design 
(F[2,27 « 1, ns) and no effect of session (Fs2,54 = 2.11, p = 0.131). 
For each session stepwise discriminant analyses were performed on the number of dicycles, 
i.e. the coefficients z of the polynomial %, and the number of ear dicycles from the ear 
decompositions by sequence. The numbers of k-dicycles were entered stepwise as dependent 
variables starting with dicycles which discriminated best between groups. The results of the 
analysis are given in Table 3.37. Variables were entered and excluded at a significance level 
of p « 0.15. 
Only in Session 2 the three block designs could be weakly discriminated by the number of 
3-dicycles whereas in Session 1 and 3 no variables were entered. The stepwise discriminant 
analysis on the number of ear dicycles revealed some differences between groups in all three 
sessions. The results are summarized in Table 3.38. 
17A sphericity test revealed no significant violation of homogeneity for the repeated measurements over 
sessions (approximate x22 = 5.25,p = 0.07).