[mswen]

Agree they are using this context for eigenvector without going into detail. Another way I have used PCA and/or cluster analysis as an analogy with a group is to get them to put all the questions, product or service features, or customer benefits on large sticky notes. Each person generates as many as come to mind in a 5 to 10 minute period.

Then taking turns each person gets up and places a couple of the questions or features they have generated on a wall or whiteboard, briefly explaining what they mean while also placing it relatively near or far in conceptual space from the items that are already up there.

As you get further along, you invite people to start reorganizing the sticky notes talking about why they think some of these questions are more or less similar to others. As they explain their reasoning you start hearing the underlying latent dimensions, the organizing vectors.

When the process is all finished you have what looks like a cluster analysis. Then as a group we name those clusters. This qualitative dimension reduction exercise often has the result of getting a group of people talking about the question or feature space in a common set of reduced labels and gives them a common framework.

I found these techniques very enlightening to use with product groups as we prepared to use research techniques to capture a Voice of the Customer.