The Scholastic Aptitude Test (SAT) for several decades was a precise measure of students applying to college. While we were never sure what was being measured, students with high SAT scores were smarter than students with low scores. One aspect of this different was that they got better grades when they got to college. (I refer to the SAT in the past tense because the current SAT was compromised at least twice, by legal requirement that they reveal the questions and by social pressure to dumb-down the mathematics section. There may be greater compromises in the last ten years.)

     Whatever concept of "smart" a college admissions department was looking for, they could be fairly confident that a 1200 SAT student had a lot more of it than an 800 student. Even a 100 point difference had statistical significance, so 1400 is somewhat smarter than 1300 and 1600 is somewhat smarter than 1500.

     The folks at the Educational Testing Service (ETS) scored the math and verbal components from zero to 1000 and then truncated them from 200 to 800 because they felt that the differences at these outer ranges were not illuminating. What they were saying was that they felt confident that 1600 is likely to be smarter than 1500 and 1700 is not so likely to be smarter than 1600.