Page - (000164) - in Knowledge and Networks

157 central, which roles are peripheral, and which trajectory types play a brokerage role. The results of this analysis are shown in Fig. 8.2a–h. I have left the original cluster numbering visible on the sociograms. White nodes denote below-average-density cells, which can be interpreted as groups with low cohesiveness. I also verified how similar each period was to any other period of time. Webster, Freeman, and Aufdemberg (2001) argue that actors in similar social contexts will establish similar network patterns. Following this analytical approach, I used Ucinet’s QAP Matrix tool to obtain a matrix of Pearson correlations among all peri- ods (Table 8.5). The diagonal under the main diagonal (with 1 s) represents the correlations between subsequent periods. Given the high incidence of right and left censored trajectories in the periods 1930–1934 and 1965–1969, I warn against pos- sible misinterpretations of the very low Pearson correlation at both extremes of this diagonal. Hence, I excluded Clusters 7 and 8 (left and right-censored trajectories) from this analysis in order to mitigate their phasing out and phasing in effects. Table 8.5 QAP matrix Pearson Correlation between periods Period Period 1930– 1934 1935– 1939 1940– 1944 1945– 1949 1950– 1954 1955– 1959 1960– 1964 1965– 1969 1930– 1934 1 1935– 1939 0.186 1 1940– 1944 0.295 0.434* 1 1945– 1949 0.405* 0.691*** 0.386* 1 1950– 1954 0.473** 0.46** 0.371 0.754*** 1 1955– 1959 –0.027 0.354** 0.306* 0.092 0.406** 1 1960– 1964 0.036 –0.163 0.364*** 0.117 0.405*** 0.189 1 1965– 1969 –0.289** –0.132 –0.163 –0.058 0.004 –0.089 0.333** 1 1930– 1934 1 Note * p < .05; ** p < .01; *** p < .001 Two-tailed tests. 8 Trajectory Types Across Network Positions: Jazz Evolution from 1930 to 1969