Diedre Bloom posted on the Harvard Social Science blog http://www.iq.harvard.edu/blog/sss/archives/2009/10/sources_of_rand.shtml
"One member of our group stated that he would not trust a particular source of data to provide useful estimates of population means, but he would trust it to estimate regression coefficients. This puzzled me..."
It's not such a puzzle, and a comment by Cyrus explained this clearly. I also chimed in.
But in addition to the two real comments there were 4 bogus ones, a sort of Randomness - Spamdomness in which a spam site tries to draw attention to itself and improve its mark on search engines by getting its site "referred to" on other sites. It's also cheaper than real advertising.
Here are the comments, leaving out the URLs they were trying to place:
I agree. Very well written. You know what you are talking about. I hope you plan to write more on this topic.
This topic is very good and is very interesting, I hope that you write more about it, thanks.
I agree this topic is really interesting to me too!
I am not sure I agree. I am still pondering the analysis.
They could almost be real. They are from four different spam sites, although possibly the same ultimate spammer.