Since 2012 I have been assembling a partnership database for a denomination of churches and studying each year's results using Gephi and NodeXL. My initial motivation was to look for correlations between partnership and missional effectiveness, but I've discovered other important insights along the way.
Here are some of the analytical tools I have used to engage the data:
- Relational patterns of contribution (OUT) vs prestige (IN)
- Innovation strategies as a function of reach (BETWEENNESS) and community strength (CLUSTERING)
- Consolidation (CLUSTERING and MUTUALITY) in times of crisis
As I near the 10 year mark, I am looking for ways to make comparisons between the years and test my hypotheses. As long as I compare degree distributions, there's no problem. Clustering likewise seems relatively safe since it is a percentage of total possible triangles. But what about comparing betweenness from one year to the next? As the graph changes from year to year (nodes and edges), so also the total number of possible paths, right? Is there a basis for comparing betweenness from one year to the next? Will normalizing be enough to solve the problem?
I don't have any formal training in mathematics or network theory so I would b very grateful for any wisdom from experts in the field or anyone with more experience.