确定在每个学校的七个社交网络之间的相似性，我们估计一系列的二次分配程序（QAP）相关网络（博尔加蒂，埃弗雷特，& Freeman，2002；Hanneman和谜语，2005；Krackhardt，1987）。QAP是一个程序计算之间的社会网络关系。进行社会网络研究时，独立的统计假设被侵犯，因为个人之间的关系是嵌套的，并嵌入在同一网络。社会网络数据往往相互依存，从而限制了使用“常规”的统计技术，如皮尔森相关。QAP是设计上与社会网络数据的使用传统的相关性分析的变化。QAP遵循特定的过程。首先，皮尔森相关系数计算两个相应的细胞的两个矩阵，其中包含网络数据。然后，它随机置换的行和一次数百列矩阵（每次计算的一种新的相关系数），并比较了倍的比例，这些随机的相关性大于或等于原始观测相关。一个低比例（P。05）表明一个强大的矩阵之间的关系是不可能发生的偶然（贝克和休伯特，1981）。我们计算的相关性为七QAP网络内的每一所学校，然后汇总这些相关使用矩阵代数表示整体QAP相关七网络之间。这些聚集的QAP相关措施，代表七个网络之间的相似性在所有样本学校。
新西兰坎特伯雷论文代写:七个社交网络
To determine the similarity between the seven social networks within each school, we estimated a series of Quadratic Assignment Procedure (QAP) correlations in Ucinet (Borgatti, Everett, & Freeman, 2002; Hanneman & Riddle, 2005; Krackhardt, 1987). The QAP is a procedure to calculate correlations between social networks. When conducting social network research, statistical assumptions of independence are violated because relations between individuals are nested and embedded within the same network. Social network data are often interdependent, thus limiting the use of ‘conventional' statistical techniques such as Pearson correlations. The QAP was designed as a variation on conventional correlational analyses for the use with social network data.The QAP follows a specific process. First, a Pearson correlation coefficient is calculated for two corresponding cells of two matrices that contain network data. Then, it randomly permutes the rows and columns of one of the matrices hundreds of times (each time computing a new correlation coefficient), and compares the proportion of times that these random correlations are larger than or equal to the original observed correlation. A low proportion (p<.05) suggests a strong relationship between the matrices that is unlikely to have occurred by chance (Baker & Hubert, 1981). We calculated QAP correlations for the seven networks within each school, and then aggregated these correlations using matrix algebra to signify overall QAP correlations among the seven networks. These aggregated QAP correlations are measures that represent the similarity between the seven networks over all sample schools.
