That’s true. And, I’m not sure how much that matters: I’d suggest there’s a difference between ability, in the sense measured by MSCEIT, and competency or skill — and a further difference with actual behavior. The ability tests are trying to measure something like “capacity” — like the raw processing power. Layered on top is competency — the learned capabilities for using this capacity. What’s intriguing is that neither capacity nor competency necessarily dictates actual performance. EG, a child with high mathematical IQ is not, necessarily, good at solving algebra problems. A self report measures something in between, and is tied somehow to self-concept (because it’s a self-report) — and I suspect this is, across a good sample population, closely linked to how people perform in the real world.