It's the most reliable source, but to a even a casual observer is deeply flawed.
I wish there was something better, but unfortunately the only hard data we have is so horribly skewed (and worse, the skew is unknown) so as to be useless for drawing conclusions that require even small assumptions regarding generality.