A General Permutation-Based QAP Analysis Approach for Dyadic Data from Multiple Groups1

The QAP approach has been extended from the analysis of bivariate correlations to multiple regressions, and is assuming the “workhorse” role for social network analysis analogous to that played by the OLS linear regression for non-network analysis. But there are severe limitations to current implementations, namely the restriction to one group at a time, and to linear regressions. Contemporary computing power makes feasible an implementation that has neither of these restrictions. This paper describes the various alternatives to such an implementation, makes public a user-friendly program for linear and non-linear QAP regressions for data that may come from more than one group, and illustrates its use with an example that sheds some light on the dynamics of influence processes in naturally occurring groups.