Percentile is the proportion of a population above which a given score ranks, percentage is the proportion correct on a given assessment. While the above quote is nonsense, saying that 1 in 1000 test takers score at 95 percent or higher on a given assessment would be totally fine.
Not that I really think that this is what the author means...
Suffice it to say, slippery math like this shows up in an alarming proportion of social science and education research. It always makes my eyes cross a little, but often the author believes that the statistic is expressed correctly... even when it flies in the face of mathematical good sense.
In this case, I think the author was clueless. I searched for "percentile" in the book and didn't find anything about a particular test for giftedness having a score of 95% being at the 99.9th percentile. Also, here's a quote about bell curves from page 5 of that book (available on
Google Books):
Three standard deviations to either side would represent 99.5 percent [of the population]. Four standard deviations ...would show a representation of the middle 99.7 percent.
Err...think that's 99.99something, not 99.7.
