Ben Mazzotta
2009-Aug-17 21:04 UTC
[R] multiple values for a factor using nodematch in ergm()
Greetings, and many thanks for any insight you can offer. I have a network object on which I'd like to simulate an ergm() using {statnet}. Since the network object is large, I have not copied it here. Each vertex has an attribute "language" which can take on integer values greater than or equal to 1. Is there a way to use the nodematch argument of ergm(), such that it searches for any matches in the vertex attributes, rather than a 1-to-1 match? For example, given mat <- matrix(1, ncol=4, nrow=4) net <- network(mat) lang <- list("A", c("A","B"), "B", "C") net %v% "languages" <- lang The truth table for the matches should read H T V (for head, tail, value) 1 2 1 1 3 1 1 4 0 2 3 1 2 4 0 3 4 0 This is the standard implementation of ergm(... , nodematch()), I believe. model2 <- ergm(net ~ nodematch(languages)) In the course of writing this post, I discovered that the net %v% "languages" command above coerces the list "lang" to a vector. I also have a dataset that specifies exhaustively whether dyads share a language or not; but I am not sure how to enter dyad information directly into the ergm() function. -- Ben Mazzotta PhD Candidate Fletcher School, Tufts University 160 Packard Ave, Medford MA 02155 benjamin.mazzotta at tufts.edu +1.617.462.4486