[Rsiena-help] Longitudinal network data based on multiple networks with changing actor composition

[Tom Snijders]

Dear Mathias,

This type of question is more appropriate for the yahoo discussion list http://tech.groups.yahoo.com/group/stocnet/ than for the current list, which is meant for issues concerned with programming the RSiena package (as opposed to using the package).

But let me have a go at answering you, with the suggestion that if problems remain, you put the next post at the other discussion list.

In the first place, take care that your network ties are states, not events, so that the Markov assumption is reasonable, that probabilities of network change depend on the current state of the network.

For the changing composition, read Section 4.8 of the manual http://www.stats.ox.ac.uk/~snijders/siena/RSiena_Manual.pdf
(but  I think you already did this).
The countries must be identified by numbers 1 to n, where n is the total number of countries that are ever present in the data. The composition change (i.e., exogenous events) file must have n rows, one for each country. There is no possibility for using country names.

Whether there is a way to read in the data without splitting them into waves is not totally clear to me. I shall ask Ruth about the possibility of using the Siena format in sienaDataCreateFromSession for this purpose.

Note the example R scripts at the Siena website, e.g., for specifying models with multiple dependent networks. You mention that you have 7 networks. Usually, it is advisable to start working with one network to get an overall impression of the data set and how it can be modelled, before continuing with a multivariate longitudinal analysis.

Best regards,
Tom

================================================================
Tom A.B. Snijders
Professor of Statistics in the Social Sciences
Department of Politics and Department of Statistics
Nuffield College
University of Oxford
tel. +44-01865-278599


-----Original Message-----
From: rsiena-help-bounces at r-forge.wu-wien.ac.at [mailto:rsiena-help-bounces at r-forge.wu-wien.ac.at] On Behalf Of Mathias De Roeck
Sent: 28 February 2012 08:42
To: rsiena-help at r-forge.wu-wien.ac.at
Subject: [Rsiena-help] Longitudinal network data based on multiple networks with changing actor composition


Dear all

At the moment I'm learning to work with RSiena, but given the nature of my data I experience problems.

I have longitudinal network data (from 1965 to 2005, so 41 waves) about political, economic and military relations between countries (a total of 7 networks). Currently, the data is in edgelist format like for example:

Year  CountrySender   CountryReceiver  Relation
1965  AFG             USA               1

When there is no relation between two countries, there is no information in the datafiles (that means it is not coded as 0 or missing, but simply not present in the datasets). The world system is a network with changing actors, since countries are entering and leaving the system. Now I have some questions:

1. After checking the RSiena manual, it seems that I need to make an exogenous events file. Is it possible to name the actors in this file, so that I now that Afghanistan is Afghanistan and the US is the US?
2. Does the exogenous events file imply that I will incept the datasets within this broader dataframe?
3. Is there a way to directly read in the data, without splitting them up in different waves?

These are my questions. I would really appreciate any comments and suggestions.

Thanks in advance

Mathias De Roeck
Junior Researcher
Center for Social Theory
Ghent University

_______________________________________________
Rsiena-help mailing list
Rsiena-help at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help

Nuffield College is a Registered Charity No. 1137506. Registered Office: Nuffield College, New Road, Oxford, OX1 1NF