Good point about the Raven's Progressive Matrices, but I still think the parent's hypothesis that we have improved at test taking is relevant. Outside of verbal reasoning there are a lot of soft factors that influence test taking ability. People who haven't taken a timed test before would probably perform worse than expected simply because it is an unfamiliar context, and unfamiliarity tends to create higher stress/anxiety which is known to be bad for cognitive performance in novices. A group that has taken hundreds of tests will probably score better on a test in an unfamiliar subject than another unprepared group that isn't used to taking tests simply because they have some meta-knowledge about test taking.