It's a great point. To compare benchmarks, we'd have to consider...6 months in behavioral changes since the Pew survey; the likelihood that long tail users also use the main social properties; what we actually classify as a social network; medium effects on answer bias and more.

I couldn't agree more with your points about the importance of survey design. Our hope with GCS is to provide a real time, affordable, iterative mechanism for conducting this type of research.

The challenges faced when benchmarking against Pew and others, is, I believe, an indication of the value of continually collecting and updating data.