Print

Print


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

Friedkin defined the unique in and out degree nodes as distinct points.
-Tom

Carl Nordlund wrote:

> *****  To join INSNA, visit http://www.insna.org  *****
>
> Hi SocNetters,
>
> I have a question on a centrality index - don't know if I just 
> invented a new one, or whether I have reinvented an old wheel of 
> unknown quality...
>
> In my PhD thesis, I play around with intl trade flow matrices, i.e. 
> directional valued matrices with quite large value-spans between 
> different actors. While primarily looking at the role-structures 
> (conceptualized through regular equivalence), I am also interested in 
> looking at the relationship between different indices of "centrality" 
> and more mainstream (actor-based) attributes on economic well-being.
>
> As I am especially interested in the number of trading partners each 
> country has, both with regards to exports and imports, I have 
> experimented with a simple heuristic for counting the number of 
> significant trading partners for each country. However, I don't know 
> if this same procedure has been done before - and if so, what this 
> type of degree-style centrality index is called.
>
> How I calculate the number of partners an actor has:
> First, I look at the "outflow profile", i.e. the row-vector for each 
> actor. Each outflow that is larger than 0.5 percent of the total 
> outflow is included in my count. I.e., if an actor has a total 
> outdegree of 1000, I count the number of outflows which are higher 
> than 5. Secondly, I do the same thing for the "inflow profile", i.e. 
> by looking at the col-vector for each actor. Similarly, I count all 
> inflows that are above 0.5 percent of the total inflows to the actor. 
> Then I calculate total number of partners by adding the number of 
> inbound and outbound partners. (The 0.5 percent "significance 
> threshold" is of course arbitrary). (As the same partner actor could 
> be counted twice - inbound as well as outbound - one could choose to 
> only count each trading partner once).
>
> What I thus try to do is to only count the number of "significant" 
> partners, where significance is defined on a per-actor basis. The 
> resulting index is very interesting for my data-set: I find a 
> statistical association between total number of trading partners 
> (weighted to total trade flows) and economic well-being.
>
> What is this centrality index called (as I assume the above has been 
> done before)?
>
> Yours,
> Carl
>

-- 
To learn more about my evaluation book go to:
http://www.oup-usa.org/isbn/0195141768.html
My personal webpage:
http://www-hsc.usc.edu/~tvalente/
---
Thomas W. Valente, PhD
Director, Master of Public Health Program
http://www.usc.edu/medicine/mph/
Department of  Preventive Medicine
School of Medicine
University of Southern California
1000 S. Fremont Ave.
Building A Room 5133
Alhambra CA 91803
phone: (626) 457-6678
fax: (626) 457-6699
email: [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.