Annica--
 
There are even some more fundamental issues about what density tells you about cohesion within groups of similar size, let alone across groups of varying sizes.  For example, two groups of size 30 that have identical densities can have very different network structures that may influence a number of outcome variables.  One group may have connections distributed throughout the network with little clustering (e.g., core-periphery structure) while the other may have concentrated connections among several subclusters in the network (e.g., clique structures).  Both may have identical densities, but the theoretical implications of the structures may be wildly different.
 
This is discussed in a recent paper along with a look at some alternative approaches.  
 
 

J.C. Johnson, J. Boster, and L. Palinkas (2003). "Social Roles and the Evolution of Networks in Isolated and Extreme Environments. The Journal of Mathematical Sociology  Volume 27/Numbers2-3: pp. 89-122. .

 
Hope this is helpful.
 
Jeff Johnson
 
-----Original Message-----
From: Annica Sandström [mailto:[log in to unmask]]
Sent: Wednesday, June 11, 2003 3:44 AM
To: [log in to unmask]
Subject: Comparing the level of cohesion between networks of different sizes

Dear list members,

 

I am about to perform a comparative analysis of five different networks regarding their level of cohesion. Three of them contain about 20 actors each, the other two 40. Density as well as distanced based cohesion will be used. From what I learned, comparing cohesion between networks of different sizes is somewhat problematic since a low density within a large network might indicate a higher level of cohesion than a higher density within a small network?

 

My question is how others have tackled this problem? Is there some general way to translate the density measure between networks of different sizes? If, for example, a network with 42 actors have a density of 0.16 and a network with 19 actors have a density of 0.25? How should one interpret the values?

 

Would be very grateful for any guidance or literature recommendations!

 

 

Annica Sandstrom

Phd student

Division of Social Sciences/ political science

Lulea University of Technology, Sweden

[log in to unmask]