Advice about analyzing a group on network level
- Hello everyone,
For our company we have done a SNA of several groups. We have asked
several SNA questions which measured some of the dimensions which
could have influence on the knowledge sharing of the group (e.g.,
Instead of analyzing the relations of individual nodes we want to
analyze the group (and thus network) as a whole and particularly how
to increase the performance of the group(s).
For example, we have measured who knows who. In this question there
was the possibility to answers on a 5-point scale (1- hardly know
each other 5= know each other very well). If we increase this value,
the network (relations between nodes) changes significantly.
Moreover, if two (central) nodes are removed from the network only 3
cliques will exist and not a network as a whole. Therefore, we want
to answer to following question: Which relation between two nodes
would be most effective for the network (e.g., decreases the average
shortest path). Which program do you use to calculate the average
The most central nodes in the network do not have to be the
best option. Because they could have the shortest path to a number of
nodes, but the nodes at the outside of the network do not have a
short path to these central nodes. Therefore a `new' relation between
some nodes at the outside of the network could be more efficient. But
how can this be calculated?
Furthermore, `know each other' does not have to be the only
independent variable of how to measure performance (knowledge
sharing). Also `trusting the other person', the `availability of the
other person' and `the value of information getting from the other
person' could well be dimensions of knowledge sharing (we argue this
based on scientific literature). At this moment we are also
developing a formula which takes the values of these several SNA
dimension into account. We would appreciate it very much that in case
you have any advice in this matter you would be able to share it with
At this time we use the programs UCINET, NETDRAW and SPSS.