it is possible to disallow individuals with certain attributes (e.g. being male) to make any mini-steps in specific behaviours (e.g. pregnancy) by setting a rate-effect (RateX) to a large negative value and fixing it at this value. This way you would only model the behavioural evolution of girls, as boys would never get the chance to update this variable. An example of how this rate fixing is done is given in one of the scripts on the Siena homepage towards the very end (https://www.stats.ox.ac.uk/~snijders/siena/RscriptSienaTwoModeAsOneMode.R, starting with the line “# A major thing that we have to do is to define...“).
I can recommend using the Siena mailing list for more detailed answers to these kind of questions (https://groups.yahoo.com/neo/groups/stocnet/info); answers there tend to be very helpful and more extensive.
I'm working with longitudinal adolescent network data and planning to use stochastic actor oriented models to study the coevolution of attributes and friendship behavior.
One of the attributes that we are interested in is pregnancy. This presents a challenge in that many of the actors are unable to take on this attribute. I can include a gender attribute of course and try to model an interaction effect, but I feel that the inability of some actors to take on this attribute might require a different approach.
Has anyone run across this issue before? Are there any existing approaches accepted in the literature?
Ian McCulloh, PhD
Johns Hopkins University
Sent from my phone-please excuse typos and brevity