I don't think that the author understands the statistics he is quoting. My understanding is that the families who chose to take the gifted exam are self selected--meaning that not everyone takes it. So right there you're essentially eliminating the left side of the curve. That together with the probability that the smarter a family is, the more likely they are to be motivated to do the testing. So you have fewer kids of average intelligence than you would if everyone with an IQ >100 took the test.

I wouldn't surprise me if self selection was the major contributor to the seeming over-representation of kids at the 90th percentile and above. Hot housing just completes the picture.