This doesn't seem to be reported here in a very illuminating way, which is reflected in the comments seeming to be everyone jumping to blame whatever pet peeve social issue they already have an axe to grind about.
The scoring system on this is set up such that 500 is the distribution median and 100 points is roughly 1 standard deviation. The US score is 517 in math, down from 535 the last time it was tested four years ago. Assuming IQ scores have to be whole integers, this is like having your IQ go from 105 to 103. Is that even a real difference or within normal test to test variation? This isn't like measuring height, where you expect almost exactly the same result every time.
You’ve described an individual instance (“your IQ”) while considering a population variation. You’re right in the first instance, individual 2 point variation between tests, say, no big deal. But the second, yes, you would be concerned if the population’s mean IQ dropped 2 points over some period.
The scoring system on this is set up such that 500 is the distribution median and 100 points is roughly 1 standard deviation. The US score is 517 in math, down from 535 the last time it was tested four years ago. Assuming IQ scores have to be whole integers, this is like having your IQ go from 105 to 103. Is that even a real difference or within normal test to test variation? This isn't like measuring height, where you expect almost exactly the same result every time.