Mitchell Centre Seminar Series: Tom Snijders
|Starts:||16:00 1 Oct 2014|
|Ends:||17:30 1 Oct 2014|
|What is it:||Seminar|
|Organiser:||School of Social Sciences|
|Who is it for:||University staff, Current University students|
|Speaker:||Tom A.B. Snijders|
The Wonders of Specification Possibilities
of Stochastic Actor-Oriented Models for Network Dynamics
The Stochastic Actor-Oriented Model for Network Dynamics (SAOM), as implemented in the RSiena package, has a wide array of so-called effects, which is the name for the functions of the network and the covariates on which the network dynamics is supposed to depend. There are many possibilities which are rarely used. This presentation is about two of these: the possibility to differentiate between effects determining creation of new ties, and maintenance of existing ties (the “creation” and “endowment” effects); and effects depending on what happens at distance two from the focal actor.
The basic SAOM specification uses an “evaluation function” to determine the probability of tie changes. With this specification, terminating a tie is just the opposite of creating it. When creation and/or endowment effects are used this is not the case, and effects may differ, or the strength (coefficients) of effects may differ depending on whether a tie is created or terminated.
Homophily, i.e., the tendency that similar actors have a greater probability of being tied, is a well-known phenomenon and is studied a lot in social network analysis. Theoretically, this extends naturally to the expectation of similarity between actors who are tied indirectly, e.g., those who have a sociometric distance of two. Is there evidence for such indirect homophily, over what may be expected as a direct consequence of regular homophily?
The presentation discusses how these two possibilities may be investigated in medium-size networks using RSiena. This is exemplified by two well-known and high-quality data sets: the MBA students (Torlò – Lomi) and the Glasgow school (Teenage Friends and Lifestyle Study, West – Michell – Pearson).
Tom A.B. Snijders
Role: Professor of Methodology and Statistics
Organisation: University of Groningen
Travel and Contact Information
Humanities Bridgeford Street