The phrase "does not necessarily correlate" needs some unpacking. I think anyone would agree that for N=2 that's true, but for higher numbers it would indicate sample error.
From your own link:
"Since the statistical results are not distinguishable from zero, this suggests IQ test scores and net worth are not connected." (em. added)
That's a common misinterpretation of the results of statistical tests. You can never conclude from that type of test that two things are not correlated, you can only conclude that the data you have does not suggest a correlation ("fail to reject the null hypothesis").
http://www.quora.com/What-is-the-covariance-between-cognitiv...
The phrase "does not necessarily correlate" needs some unpacking. I think anyone would agree that for N=2 that's true, but for higher numbers it would indicate sample error.