Introduction to
Stochastic Actor-oriented Models
The intuitionClassification And Types Of Burns
András Vörös
Department of Social Statistics, University of Manchester
SOST71032 Social Network Analysis
* Thanks to Christoph Stadtfeld for many nice figures and examples.
First of all: some terminology
 SIENA
 Simulation Investigation for Empirical Network Analysis
 the software in which “the method” is implemented
 the debut of the method (and software) was at a conference in
Siena, Italy in the mid-90s – hence the name
 RSiena
 this is version 4.0 of the SIENA software, implemented in R
 SAOM
 Stochastic Actor-oriented Model
 this is “the method” that is applied on the data when you use
RSiena
 there are a few different types of models, which are referred to
as “SAOMs” or, sometimes, “SIENA models”
The problem: how do networks change?
 Which social processes may shape this friendship network of
students over time?
A tool to address the problem: the SAOM
Discrete data
Continuous-time
model
What is the SAOM?
 The SAOM is an agent-based (actor-oriented) model
 It models the chain of (unobserved) network states
between two observations at t1 and t2 as a stochastic
process…
 … through simulating sequences of the smallest
possible steps in network change (“ministep” or
“microstep”): an actor changing one outgoing tie
 It compares the results of the simulations to the
observed data at t2…
 … and optimizes model parameters (estimates) so that
at the end of a simulated chain we reach a network that
is as similar as possible to the observed network at t2
What is the SAOM?
 Each simulated chain of network states represents what
could have happened: how the network could have
evolved between t1 and t2
 The answer we get to “what could have happened?”
depends on what kind of processes we thought about
– the variables we include in the model
Please continue with the next topic.