Any idea how reliable a single IQ score is? If you repeat the test a week or a month later, too soon for changes to reflect a significant change in cognitive abilities, but far enough away to get rid of effects like a poor night's sleep, stress from school or relationships, &c., by how much is the score expected to vary?
A proper IQ test done under the same conditions (well rested, etc) will give the same results. It's vulnerable to poor conditions - if you're not well rested, you're not going to do as well as you could otherwise.
In my childhood I did enough IQ tests [1] to be able to say "Oh, it's one of this class of problems again". I assume that people taking repeated IQ tests get better at them, because there's a limited selection of problem types.
[1] Only three or four. I'm sure there are lots of kids out there getting far more over-tested than I.
