6.2 Network analysis of cases

At present one of the hidden worksheets in EvalC3 is a cases x attributes matrix. It is used to calculate the similarity measures in the View Cases worksheet. The same matrix, if imported into a Social Network Analysis (SNA) software package (like Ucinet&Netdraw) can be treated as a “two-mode” network. It is possible to construct two types of network visualisations with this data, showing:

  • How cases are connected to cases, where link strength between any two cases reflects the number of the same attributes they share
  • How attributes are connected to attributes, where link strength between any two attributes reflects the number of the same cases they connect.

After calculating the average link strength, for either type of network, it is then possible to filter out the below average links, thus highlighting the clusters of cases (or attributes) that are co-occurring at above the average levels.

In the cases x cases network example below there are two distinct clusters. One is the densely connected group on the left, the other is the much less densely connected group on the right. All the members on the right were evaluated as “less successful” than those in the left, on a measure that included self-assessments by the NGOs themselves. It appears that the “less successful” group was more diverse in its attributes than the more successful group.

[Network diagram to come]