


Measure  Marriage  Business 
Density  0.167  0.125 
inTransitivity  0.809  0.583 
Compactness  0.438  0.252 
Connectedness  0.875  0.458 
Wholeness  0.933  0.667 
Note that the marriage network is the more cohesive according to all measures.
Here's how to do it.
inTransitivity. Run NetworksCohesionTransitivity on Padgett dataset. The transitivity of each network will be displayed as "Transitivity: % of ordered triples in which i>j and j>k that are transitive". To calculate intransitivity subtract transitivity from 1.
For the remaining statistics, you will need to run DataUnpack on Padgett to create separate files for PADGM and PADGB. Then you will run the analytical procedures separately for each dataset.
Compactness. Run NetworksCohesionDistance on PADGM. Copy the compactness value. Repeat for PADGB.
Connectedness. Run NetworksCohesionKrackhardt's GTD on PADGM. Copy the connectedness value. Repeat for PADGB.
Wholeness. Run NetworksRegionsComponentsSimple on PADGM. Note number of components. Let's call that quantity C. Now calculate 1  (C1)/(N1) where N is the number of nodes in the network. Repeat for PADGB.
Since both networks in the Padgett dataset are disconnected (have multiple components), some measures of cohesion could not be calculated. In this exercise we will discard the isolates in each network and compute measures of cohesion on the remaining main components. The results should look like this:
Measure  Marriage  Business 
Density  0.191  0.273 
inTransitivity  0.809  0.583 
Compactness  0.500  0.550 
Connectedness  1.000  1.000 
Wholeness  1.000  1.000 
Average Distance*  2.486  2.382 
Diameter*  5.000  5.000 
Average Flow  1.695  1.618 
*inverse measure of cohesion
Note that the picture is now a bit more ambiguous, but in general the business network is more cohesive.
The first thing you need to do is extract the main components from each network. To do this, run DataExtract Main Component on PADGM and PADGB, creating new datasets PADGMMAIN and PADGBMAIN, respectively.
inTransitivity. Run NetworksCohesionTransitivity on PADGMMAIN. The transitivity will be labeled "Transitivity: % of ordered triples in which i>j and j>k that are transitive". To calculate intransitivity, subtract transitivity from 1. Repeat for PADGBMAIN.
Compactness, Average Distance & Diameter. Run NetworksCohesionDistance on PADGMMAIN. Copy the compactness value and the average distance value. Note the largest distance  this is the diameter of the network. Repeat for PADGBMAIN.
Connectedness. Run NetworksCohesionKrackhardt's GTD on PADGMMAIN. Copy the connectedness value. Repeat for PADGBMAIN.
Wholeness. Run NetworksRegionsComponentsSimple on PADGMMAIN. Note number of components. Let's call that quantity C. Now calculate 1  (C1)/(N1) where N is the number of nodes in the network. Repeat for PADGBMAIN.
Average Flow. Run NetworksCohesionMaximum flow on PADGMMAIN. This gets you a matrix of maximum flows between all pairs of nodes and saves a file called MAXIMUMFLOW. Then run ToolsUnivariate Statistics on the MAXIMUMFLOW matrix you just created. Select "matrix" as the dimension to compute, and make sure that "use diagonal" is set to "no". Copy the value labeled "mean". Repeat both steps for PADGBMAIN.
