-
Introduction
Page 1
-
Materials and Methods
Pages 2 - 3
-
-
Discussion
Pages 8 - 9
-
Conclusions
Page 10
-
References
Page 11
-
Figures 1- 8
Figures 1 - 2 - 3 - 4 - 5 - 6 - 7 - 8
-
Tables I-IV
Tables I - II - III - IV
(Continued from previous page)
Dyadic sequence analysis of acts and postures was conducted on the less inclusive data sets arranged as 30 by 30 tables as described above. The analytic procedure was that devised by Brown [1974] and recommended by Fagen and Mankovich [1980] for exploratory purposes such as those of this study. The BMDP computer package for two- way frequency tables involving empty cells and departures from independence (BMDP2F) was used to determine those sequences that were greater than would be expected by chance. A control analysis was run on the within-animal data in which all diagonal cells (i.e. behavioral sequences in which the same behavior was repeated or continued) were excluded. The conclusions proved to be robust; of the 43 statistically significant sequences, only one failed to achieve significance when the diagonals were excluded, and three others were newly included with weak rankings, all three being variants on other sequences that were significant.
The ratios of obtained frequencies to expected frequencies were obtained from 2 x 2 contingency tables, using the same method as that described in Lehman and Adams [1977]. They were carried out for all behavioral sequences that were determined to be nonrandom according to the computerized analysis. Separate tables were constructed and analyzed for three subsets of the data in terms of the dominance relations of the two monkeys. Six such analyses were performed: sequences within dominant monkeys when facing subordinates; sequences within dominants facing another dominant; sequences within subordinates facing a dominant; sequences from dominant to subordinate; from dominant to dominant; and from subordinate to dominant. Triadic sequences (three consecutive acts or postures) were also analyzed for all triads that could be constructed from significantly nonrandom dyads of acts and postures. Such triads were those consisting of a significant dyad and either the preceding or the following behavior. The dyad was treated as a single unit and two-by-two contingency tables were analyzed as described above.
Vocalizations and facial expressions were analyzed separately. The more inclusive data sets were used; as noted above, a new behavioral event was recorded each time any eye orientation, facial expression, vocalization, or act or posture changed in one or the other monkey. Three types of analyses were conducted for vocalizations and facial expressions. First, both vocalizations and facial expressions were tested for the degree of association with particular acts and postures of the behaving monkey. A Chi square analysis was done on a 2 x 2 matrix consisting of a) the number of behavioral events containing both the particular act and the particular facial expression or vocalization, b) number of behavioral events containing the act but not the facial expression or vocalization, c) number of behavioral events containing the facial expression or vocalization but not the act, and d) number of behavioral events containing neither. Second, they were each tested for association with the preceding act or posture of the opponent. Third, vocalizations were tested for association with the preceding vocalization of the opponent. Chi square analysis was performed on these tests, and those that were significant at a probability level of 0.001 were inclined in tables for results and discussion.
 |
 |
 |