My previous post on our prototype for mapping the PreferenceSphere showed preference data for social networking sites expressed as a 2×2 graph, which I said was derived from a directional graph. Here’s that underlying graph.
DISCLAIMER: There are many duplicates in the Internet data for our prototype. Don’t take these scores too seriously until we integrate this feature into our ConsumerBase product, by which time we will have corrected the problem. However, the duplicate problem works in both directions and tends to cancel itself out, so we don’t believe it’s skewing the results.
In this graph, the thickness of the lines corresponds to the number of expressions of preference—the more people who say they prefer one brand over another, the thicker the line. So the thick line originating at Facebook and ending with an arrowhead at MySpace shows that a large segment of people prefer Facebook over MySpace. The much-thinner line going from MySpace to Facebook shows that a much smaller segment has the opposite preference.
A very useful feature of the prototype is that it allows you to drill down into WHY people have these preferences, which you can do by clicking on the lines and viewing the actual sound bites from consumers that were used to draw the graph.
Deriving the 2×2 Graph from the Directional Graph
We use data from this directional graph to draw the 2×2 graph discussed in my earlier post. For example, to derive the data point on the Net Preference axis in the 2×2 graph, we subtract all the preferences for another brand (indegree) from the preferences for the brand in question (outdegree). (For more on graphing terms, see this Wikipedia entry on Directed Graphs http://en.wikipedia.org/wiki/Directed_graph.) In the case of our social networking sites example, we subtracted the values for all the arrows pointing away from Facebook from the values of all the arrows pointing towards it.
Comments? Want to see other features? Please let us know.