It’s pretty tricky, and Matt Cutts has an excellent post that could be summarized thus: “lies, damn lies, and statistics”.
I’m not sure if this is a subtle reminder about the vagaries of toolbar PageRank, but he raises some good points about how to measure importance:
“To be completely sure in a metric, you need to know the raw incoming data and how a metric operates on that data to produce a number. And if you want to be more confident, look for similar metrics that should roughly agree. If different metrics agree, thatâ€™s a good sign. If they disagree, you should probably be cautious.“