## SOCNET@LISTS.UFL.EDU

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font

Subject:

Regular blockmodels with valued relations

From:

Date:

Thu, 9 Dec 2004 22:30:51 +0100

Content-Type:

text/plain

Parts/Attachments:

 text/plain (51 lines)
 ```***** To join INSNA, visit http://www.insna.org ***** Hi all, I have checked my standard textbook (Wasserman/Faust 1994), looked in Borgatti/Everett 1992 paper (Regular blockmodels of multiway, multimode matrices) and elsewhere, including Googling on the web, but I haven't found any answer so I resort to this list: How to interpret regular blockmodels based on valued (non-categorical) directional relations? I haven't seen it being done anywhere. References anyone? I understand how to interpret regular blockmodels (i.e blockmodels where the different positions are established through the notion of regular equivalence and, for instance, REGE) for dichotomous (binary) values: as regular equivalence is concerned with equivalent actors having relations to actors which in turn are equivalent (instead of the strict requirements of structural equivalence), the submatrices making up inter-positional relations are interpreted differently in order to create block images. However, how to do this for valued (non-categorical), directional graph? Is something like this feasible: 1. Establish discrete role-equivalent positions using regular equivalence (as implemented in the REGE algorithm) 2. Sort the rows/cols in the original matrix according to positional membership (i.e create a standard "blockmodel matrix") 3. Dichotomize this matrix, using the mean density of the network as the cut-off value! 4. Interpret the blockmodel as it is done for regular blockmodels. Would something like this be feasible? Number 3 here is of course what might be problematic. Perhaps using some other cut-off-value? Density of relations within positional submatrices in some way? Yours, Carl --- Carl Nordlund, BA, PhD student carl.nordlund(at)humecol.lu.se Human Ecology Division, Lund university www.humecol.lu.se _____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.```