OK, let's talk about intelligence. First there is the concept of g, as being a general sort of cognitive ability, which can never be accurately tested and may not exist. IQ tests all have specific tasks and abilities that they test, and may get slightly different results from each other.
Then there is Gardner's multiple intelligences and his objection to the privileging of two that happen to easy to measure: verbal and logical/mathematical. Gardner and I are of the opinion that the most important of the intelligences is actually interpersonal intelligence. My work experience and the time I spent on a graduate admissions committee are parts of the reasons I believe this.
Also, there is the question of how behavior affects future scores on the SAT. And for that, I reference Walter Mischel's marshmallow test. In his famous 1960s experiment, Mischel asked kids three- or four-years-old to wait 15 minutes with a marshmallow in order to get a second marshmallow when the researcher returned.
Mischel sometimes found this delayed gratification in very young children, but he believed (or believes) that it is a learned behavior. At last report, he was investigating those people that could not wait at four, but had somehow dodged the substance abuse problems and other issues that dogged many of the future adult selves of children that could not wait for the second marshmallow.
Mischel followed up on the marshmallow kids later and found out most of the kids who could wait scored an average of 200 points higher on the SAT. So it turns out that the SAT may also be a measure of the kind of behavior that will allow students to do well in college.
