Hmm, interesting. I guess the question is this: if IQ is not changeable but people score dramatically differently at different times, which do we consider to be their "real" IQ? It seems unlikely that someone could accidentally score significantly higher than they ought to if the test is properly administered. It seems more likely that someone might test lower due to a variety of factors: maturity, distractions, not enough sleep, anxiety, etc. So I would probably lean towards the higher number being the more accurate.
However,these changing scores raise alot of questions about the test itself and what it is actually measuring.
