I have a problem working with adjacency and affiliation networks. I have a
dataset on actors (corporations in Japan) and ties between them. I also
have a dataset on ties between these actors and another well defined set
(banks), but no data on the ties between the other actors (banks). I'm
wondering how I can best utilize the corporate-bank data in conjunction
with the corporate-corporate data - or how can I simultaneously analyze an
adjacency matrix and an affiliation matrix?
The affiliation matrix can be converted to an adjacency matrix (using XX'),
but then I really have a matrix of reach 2 Although such a conversion is
in the spirit of connectivity and information flows it is no longer
comparable to the adjacency matrix. Running constraint measures on the
reach matrix might be defensible, but analyzing equivalency on it misses a
key intervening step - the bank.
I'd like to compute equivalency on the two matrices simultaneously without
doing the conversion. Using UCINET I can "join" the matrices, but CONCOR
and other routines refuse to run using matrices of different sizes and
non-square matrices. Should I simply expand the adjacency matrix to
include the banks and then have a large blank area in the matrix? Are
there any implementations of equivalency that can handle this type of data?
Thanks in advance,
Eliot Mason
|