SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to
Hope to see you soon in Montreal.
Keywords: multiple regression, quadratic assignment procedure (QAP), permutation tests, network data, exogenous groupings, time-series and geographic network data.
This workshop focuses on 3 topics regarding the multiple regression quadratic assignment
procedure, which encompasses permutation-based tests for network data in the regression
analyses framework. First, the workshop provides an introduction into permutation-based testing
by explaining the necessity to deal with structural dependencies when analyzing network-like data and demonstrating the mechanics of permutations of data organized in square matrices. Second,
in the workshop we will be elaborating on the use of different permutation scheme's for different
tests. F-test for model significance and t-tests for coefficient significance require different
permutation approaches. Specifically, the different uses of Y-permutation and DSP-permutation
will be clarified. Third, an extension of linear models dealing with exogenous grouping will be
presented to the participants. The usefulness of introducing further restrictions on permutations
to deal with exogenous groupings of ties demonstrates the versatility of the approach. An
application with time-series and geographic network data is presented. Participants may expect
to feel more proficient in using permutation techniques and encouraged to explore the wide
range of possible applications for permutation testing that remain unutilized to this day. Technical
requirements involve a basic understanding of correlation and regression analyses, and, access to
UCInet and R's statnet.