Psychometricians will hate this answer (they think IQ measure general intelligence), but I contend that IQ tests measure academic skills and knowledge. When you look at the items on the Stanford-Binet or the WAIS, you see items that require knowledge typically gained in school: math, vocabulary, logical reasoning, even geography. In short, if you want to know who will succeed in academic settings, give the person an IQ test.
[WS:] Three points.
First, an attempt to prove this would typically suffer from selection bias. Kids who score high in IQ-like tests are the ones who get admitted to academic programs. A valid test of the above proposition would require testing the academic performance of kids who score low on IQ-type (i.e. academic "aptitude") tests, which is rather difficult, because these kids tend not to be admitted to academic programs. It is very likely that if all kids had an equal chance of being admitted to academic programs regardless of how they score on "aptitude" tests, there would be no statistically significant correlation between academic performance and scores on such tests.
There is a quite good substantive reason for that other than mere selection probability. The actual academic work rarely involves tasks that are tested on 'aptitude' tests. AFAIK, typical academic work involves reading assignments and writing term papers in your own environment (i.e. home, library, or dorm) and, somewhat less frequently, participating in collective activities (seminars, group research projects, etc.) involving information sharing - rather than picking up a right choice from the four or five alternatives spelled out by the testing authority and doing so in a strictly controlled and timed environment. These two are very different cognitive tasks, and only blind faith that both types of task involve some form of 'latent general intelligence' can establish equivalence among them.
The second point is that nowadays life tends to imitate art, or rather kitsch. This is true of the academic environment too. More and more, academic work involves standardized testing rather than teaching how to think independently, less alone critically. From that point of view, your claim seems to be true - not because IQ-like testing discovered something useful about human cognition (its contribution in that area is less than zero, negative, as it obscures more than it reveals, but because its practices are aped by academic institutions. And the only reason why the latter is happening is that testing is big business that is in a position of imposing demand for its product by modifying institutional environment of the nation. This is no different from other business practice in the US - the auto industry creating the demand for car by forcing government to undercut alternative modes of transit, the insurance industry forcing government to undercut public health care system, real estate and developer business constraining the supply of affordable housing - you get the drift.
I would go as far as saying that it is not true that IQ and kindred "aptitude" tests capture one of several important cognitive skills. They capture only a marginal one, while leaving all the important ones out. In short, they obscure more than they can possibly explain.
Finally, a thought that I believe Ravi would like. Psychometrics, like neo-classical economics and kindred brands of positivism/scientism more generally - is more religion than science. It is "science" in a superficial way only - by ritualistic adherence to certain procedures that are used by science. But while real science uses these procedures as the means to an end, scientists/positivists use them as an end in itself.
That is to say, real science would use any method that can discover the so-called "facts" or "truth" (i.e. "objectively" existing causal relations - in a Kantian sense, of course). Scientism/positivism, otoh, would limit its pursuit to the application of narrowly defined set of procedures that are considered "scientific," regardless of whether the results they produce have any predictive value or not.
In that way, they are a form of religion, where ritual is practiced regardless of whether it yields any practical results. This is a one way street indeed, where only confirmatory evidenced is allowed, and any contradicting evidence is explained away ("you did not pray hard enough"), compartmentalized away (that is "humanities" not "hard science") or altogether banned.
Wojtek