Truth (but alas, no mod points.) Though the close correspondence between the two in the research would seem to initially imply that there are a minimal number of confounding factors. If you tell me there is a statistically significant link between scores on test A and trait B, the simplest conclusion is that test A tests for trait B. Doesn't make it true; just the simplest answer. Any longer train of logic requires more assumptions and/or discrepancies to explain away. Case in point; if we assume IQ tests measure pattern matching and noise filtering and X, then we have to explain why we don't (if we don't) have a bunch of people who are lousy at pattern matching, but score high on IQ tests anyways (because they're good at X). There are lots of perfectly sensible answers for that, but we'd better go find one.
Of course, if we _do_ have a bunch of people like that that need explaining, then its good evidence for the existence of one or more X's...