Print

Print


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

Dear Nick,

for RSiena / stochastic actor-based etwork evolution models, the issue 
of relative importance of effects (taking into account prevalence of the 
effect statistic) was addressed - and a solution proposed - in this paper:
> Indlekofer, Natalie, and Ulrik Brandes. "Relative importance of 
> effects in stochastic actor-oriented models." /Network Science/ 1.03 
> (2013): 278-304.
I am not aware of similar work on ERGMs, but the ideas from this paper 
can to a good degree be transferred to the contribution of effects to 
tie swap probabilities in an ERGM.

In terms of 'coefficient of determination', I would think it is hard to 
get a meaningful one. Here is one proposal, relying on spectral analysis 
/ the vector of ordered eigenvalues:
> Shore, Jesse, and Benjamin Lubin. "Spectral goodness of fit for 
> network models." /Social Networks/ 43 (2015): 16-27.
The authors claim universality, but similar R-square-type indices can be 
constructed for any other set of network statistics (triad census, 
geodesic distribution, degree distributions, etc.)

Best,
Christian


On 19-Mar-16 06:42, Nicholas Harrigan wrote:
> ***** To join INSNA, visit http://www.insna.org *****
> Dear SOCNET,
>
> We are trying to approximate a coefficient of determination for an 
> ERGM (actually the relatively contribution to the coefficient of 
> determination for each of the individual parameters of the ergm).
>
> We would also be happy with something like an effect size, but which 
> took account of the relative prevalence of the graph statistic (since 
> rare statistics may have a big effect, but not a major contribution to 
> a model because of their rarity).
>
> Does any one have suggestions?
>
> We will also want to solve this problem for an RSiena model in the 
> near future, so suggestions for RSiena would be welcome too.
>
> Thanks
> Nick
>
>
> -- 
> Nicholas Harrigan
> Assistant Professor of Sociology
>
> Level 4, Room 4-039
> School of Social Sciences
> Singapore Management University
> 90 Stamford Road
> Singapore 178903
>
> Office: +(65) 6828 0842
> Mobile: +(65) 9029 9499
> Fax: +(65) 6828 0423
> www.smu.edu.sg <http://www.smu.edu.sg>
>
> The information contained in this email (including all attachments) is
> confidential and may also be privileged. It is intended for the use 
> only of
> the addressee(s) named. If you are not the intended recipient, kindly note
> that any dissemination, distribution, publication or copying of this email
> is strictly prohibited. If you have received this email in error, please
> accept our apologies and notify us immediately. Thereafter, kindly delete
> all copies from your system. Although reasonable precautions have been 
> taken
> to ensure the integrity of this email and that it is virus free, SMU 
> cannot
> be responsible for any change made to this email or the presence of any
> virus therein without its knowledge and/or consent. Unless it relates 
> to the
> official business of SMU, any opinions, views and other information
> expressed in this document are those of the individual sender.
> -- 
> --
>
> Teaching consultations: http://bit.ly/1RfKV8j
>
> Nicholas Harrigan
> Assistant Professor of Sociology
> School of Social Sciences
>
> Singapore Management University
> 90 Stamford Road, Level 4
> Singapore 178903
>
> Office: +(65) 6828 0842
> Mobile: +(65) 9029 9499
> Fax: +(65) 6828 0423
> www.smu.edu.sg <http://www.smu.edu.sg/>
>
> The information contained in this email (including all attachments) is
> confidential and may also be privileged. It is intended for the use 
> only of
> the addressee(s) named. If you are not the intended recipient, kindly note
> that any dissemination, distribution, publication or copying of this email
> is strictly prohibited. If you have received this email in error, please
> accept our apologies and notify us immediately. Thereafter, kindly delete
> all copies from your system. Although reasonable precautions have been 
> taken
> to ensure the integrity of this email and that it is virus free, SMU 
> cannot
> be responsible for any change made to this email or the presence of any
> virus therein without its knowledge and/or consent. Unless it relates 
> to the
> official business of SMU, any opinions, views and other information
> expressed in this document are those of the individual sender.
> _____________________________________________________________________ 
> 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. 


-- 
_ __ ___ ____ ___ __ _ __ ___ ____ ___ __ _

Christian Steglich, University of Groningen
Faculty of Behavioural and Social Sciences
Grote Rozenstraat 31, NL-9712 TG Groningen
http://www.gmw.rug.nl/~steglich/
_ __ ___ ____ ___ __ _ __ ___ ____ ___ __ _


_____________________________________________________________________
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.