[Rsiena-help] undirected networks in RSiena

[Kevin Lewis]

greetings,

i wonder if anyone can provide an update on current functionality for 
undirected networks in RSiena? it appears that RSiena indeed recognizes 
my network as undirected, but this case is not as thoroughly documented 
in the most recent manual. in particular i am wondering about the following:

1) if i recall correctly, version 3 did not allow estimation by the 
score function method, and instead forced the "finite differences" 
method for all undirected models. however, when i run an undirected 
model in RSiena, it seems i am able to select the score function method 
without problem (and the output reports that standard errors are 
estimated with the likelihood ratio method, which i assume is the same 
thing). am i correct in interpreting this to mean that the score 
function method is now available for undirected networks in RSiena?

2) while RSiena will successfully run an undirected model, the "model 
options" screen still looks identical to the options screen for directed 
networks- and therefore i am unable to select which of the 6 undirected 
model types i would like to use (a possibility that was available in 
version 3). the output reports, however, that "Model Type 2: Forcing 
model" appears to be the default. is there a way to change this to 
another type, in particular the initiative-confirmation model? if not, 
is there reason to believe this will influence results anyway? (i recall 
reading somewhere that results did not vary much across different 
undirected model types.)

i am obviously also using the GUI, so perhaps this is something i can 
accomplish without the GUI.

thanks for any help with this!

-kevin

-- 
Kevin Lewis
Ph.D. Candidate in Sociology
Harvard University
33 Kirkland Street
Cambridge, MA 02138