***** To join INSNA, visit http://www.insna.org *****
Hey,

Thanks Christian for the reference on EpiModel (which I find really great!), although we are not the authors of it :) (you can thank Samuel Jenness, Steven M. Goodreau and Martina Morris for it!). That said, which way to go depends on what are you trying to model. netdiffuseR, which I maintain, has a bunch of alternatives, whereas you want to run a lagged regression model (for which you can use our exposure function), do permutation tests (which are implemented in the struct_test function), or simulate diffusion processes (which is provided by the rdiffnet and rdiffnet_multiple functions), you can do it with netdiffuseR. The last feature, simulating diffusion processes, can also be done in EpiModel, the main differences, from what I understand, are two: (1) EpiModel is significantly more flexible and complex, it uses ERGMs and tERGMs to simulate changes in the networks (which I find very cool) and is written in a modular way, so the user can define almost all dynamics in the diffusion/contagion process, whereas netdiffuseR is less flexible and has some pre-defined dynamics, and (2) Given the previous point, more flexibility comes with a cost, simulations are slower in EpiModel in comparison with netdiffuseR; in netdiffuseR simulating thousands of diffusion networks with thousands of vertices is relatively fast; something that, again, as far as I understand, is not the case with EpiModel.

HIH


George G. Vega Yon
+1 (626) 381 8171
http://ggvy.cl

On Thu, Nov 30, 2017 at 12:34 AM, Christian Steglich <[log in to unmask]> wrote:
***** To join INSNA, visit http://www.insna.org *****

Hi Fabrizio,

first and foremost, the triple links between data, modelling, and theory are crucial to get right before proceeding to any analysis. To illustrate: If your data consists of time-stamped e-mail traffic, you should probably not apply methods that require aggregation of data over hard-to-define time windows, because it implies a tremendous loss of data and relevant detail about timing (what came first, what came later in reaction to it). Or, if your theory doesn't address network dynamics but your data includes dynamically shifting network links, something is missing in the theory framework. Or, if ideas can be adopted and un-adopted again, a model that only considers adoption might not be the best choice.

In this triple linking, everything that misses or mismatches needs to be filled in by *assumptions that are typically hard to sell, and that people therefore tend not to write about - omissions that undermine trustworthiness and make a lot of contagion research hard to follow, not completely 'non sequitur' but getting close, while leaving fast readers with less doubts than they should have. It is better to be as explicit as possible about assumptions, and think a lot before proceeding to data parsing.

That being said... next to the excellent reference Tom Valente gave ("epimodel" software http://www.epimodel.org/, and George Vega Yon's work) there are agent-based modelling techniques (e.g., Netlogo-applications, which can be more or less easily calibrated to empirical data), *assuming dynamic or static underlying networks of empirical or theoretically *assumed topologies...

The approach that I myself am a bit familiar with is stochastic actor-based (or actor-oriented) modelling, for which there are software packages for discrete-time data ("RSiena" by Tom Snijders; see a list of papers applying the method here: https://www.stats.ox.ac.uk/~snijders/siena/siena_applications.htm) as well as continuous-time data sets ("Goldfish" by Christoph Stadtfeld & colleagues, see here: http://www.social-networks.ethz.ch/research/goldfish.html). In this approach, the main *assumption is "agency", i.e., that social actors propel network change by taking decisions to rewire their own ties, and they are the ones deciding about whether or not to adopt an idea, given the information at their disposal. Often, this actor-orientedness helps constructing a walkable bridge between analysis method and social science theory.

All the best, Christian


On 29-11-2017 22:16, Thomas William Valente wrote:
***** To join INSNA, visit http://www.insna.org *****

Fabrizio

You can use our R library netdiffuseR to model diffusion and contagion through a wide variety of theoretical processes.

There are graphing, simulation, statistical test procedures, and you tube videos of tutorials.

See:

https://usccana.github.io/

https://github.com/USCCANA/netdiffuseR

 

-Tom

 

Thomas W. Valente, PhD

Professor and Interim Chair

Department of Preventive Medicine

Keck School of Medicine

University of Southern California

Soto Street Building, Suite 330

2001 N Soto Street, MC 9239

Los Angeles CA  90089-9239

Email: [log in to unmask]

 

 

FOR FEDEX DELIVERIES please use zip code 90032

 

Phone:  323 865 0803

Fax:  323 865 0854

 

For appointments and scheduling please contact:

 

Luz Moncayo

Email: [log in to unmask]

Phone: 323 865 0401

 

Description:
              cid:<a href=[log in to unmask]" height="46" width="210" border="0">

 

CONFIDENTIALITY NOTICE: This transmission (including attachments) is intended only for the person or entity to which it is addressed and may contain privileged and/or confidential material. If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copy of this message is strictly prohibited. If received in error, please notify the sender immediately and delete/destroy the message and any copies thereof. Thank you.

 

***** To join INSNA, visit http://www.insna.org *****

Hello, I'm totally new to Social Network Analysis.

I have a dataset of a corporate network consisting of 3000 actors and I have the evolution over time of the network.

My data indicates the diffusion of the political ideas of the actors.

I'm looking for references to know which would be the best approach and which software to use to model a contagion of ideas and to study network effects. 

Thanks in advance.

Fabrizio Marini

[log in to unmask]

_____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.

 

_____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.

--


Interuniversity Centre for Social Science Theory & Methodology
Department of Sociology
Grote Rozenstraat 31
NL-9712 TG GRONINGEN
steglich.gmw.rug.nl
_____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.

_____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.