And it is just as likely to go down to the lower end of the confidence interval. The meaning of the spread (properly called a confidence interval) is standard error. If we could give the same test to the same person many times (and have them not memorize it!), without re-test effects, there would be natural variation in their scores. The ci on most IQ tests is 95%, which could be explained as, if we were able to do the above repeated test experiment 100 times, we would predict that the score would fall within this range 95 times.
Also, unless the PG label gives you access to something of high value to you, there is negligible qualitative difference between a 141 and a 145.
Of course, if you want richer information about cognitive profiles, the WISC and SB are more in-depth instruments. And I've mentioned elsewhere that I'm not too fond of the RIAS, other than as a screening instrument.
