Gifted Issues Discussion homepage Forums Identification, Testing & Assessment Regression to the Mean--Why test scores change

This was an informative read -- thanks for sharing!
To be honest, the conclusions drawn in the paper coincide pretty strongly with what I see in my work. I have many, many children who come in with previous IQs at 130+, and are consequently in gifted education, but end up in the 100s to low 120s range. In fact, I had a case with this presentation just this week. The parents tend to blame anxiety, or something along those lines, for the drop, but given that I give multiple assessments of a similar kind every day, I have a pretty decent grasp on when a child's testing is an accurate representation of their abilities.
I tend to chalk it up to two factors: 1) the school districts and even some of the psychologists in our state are generally pretty awful at administering highly standardized assessments...I recognize that there is an assumption that it's impossible to "fake good," but the tester can absolutely make that happen; 2) regression to the mean is, indeed, a reality. On the clinical side of things, unresolved language issues (expressive or receptive, but especially expressive) tend to coincide with a decrease in IQ scores over time.
Of course, none of this makes telling the parents, or even the children depending on their age, any easier.
As aeh said, the same can be true for the other end of curve, as well, especially because IQ is incredibly unstable at any early age, and there's several compounding factors involved (even a typical child with few behavioral issues is bound to get bored with the WPPSI at three or four). Obviously, in these situations, an increase can be a positive change to see.