```*****  To join INSNA, visit http://www.insna.org  *****

Dear Philip,

see Wassermann/Faust (1994), chapter 14.3, and the literature cited there.

In our book (but German!), we give an example using the Newcomb Fraternity,
2nd year, 3 most preferred choices, week 14:
Mark Trappmann, Hans J.Hummell, Wolfgang Sodeur (2011): Strukturanalyse
sozialer Netzwerke. Konzepte, Modelle, Methoden. 2nd edition, first 2005.
Wiesbaden: Springer Fachmedien (1st edition VS-Verlag).

If your network is small (about 200 vertices) and has a relatively weak
density, I can offer a quick-and-dirty method by means of random network
generators, which were designed for more complicated conditions than MAN
originally.

We try to bring our old network generators to R at the moment. Programs
(Fortran) are about 30 years old. The intension was to test the triad
census (and other structural properties of networks) by comparing them with
a sufficient number of biased random networks,. At the moment we are still
on the level of PC-DOS and the mainframes of this former time. The export
to Windows and/or UNIX did not start until now.

As conditions for generating the biased random networks we used (a) the
>distribution< of outdegrees, (b) both >distributions< of indegrees and
outdegrees simultaneously, (c) the same >distributions< as (b) and among
these the distribution of reciprocated degrees (symmetric).

Before we will publish these old random generators we try to test them by
comparisons with analytic results (see above, Wasserman/Faust). This is the
point, where we try to use the old programs to generate networks under less
restricted conditions: With the old programs (here (c)) we try to generate
less restricted random networks like MAN.

Tell me the size of your network and the number of mutual and asymmetric
ties. If your empirical network does not have too many vertices and is not
too dense, I can give vou the expected means and standard deviations of the
often.

Best wishes
Wolfgang Sodeur

------------------------------------