A couple of years ago I talked to the Dean of Liberal Arts (I think it was) at a college out on Long Island. He was a math guy (I'm thinking a mathematician, but he may have been a scientist).
When I asked him about students' knowledge of math he told me, "We can't assume students know anything we would want them to know."
This included being able to solve a linear equation with one variable.
(The phrase "decline at the top" isn't mine, but I don't remember who coined it. Waiting for Utopia discusses decline at the top.)
update: Diane Ravitch used the term in 1997.
Over the years researchers have debated the meaning of the decline in SAT scores. Some have concluded that it is solely a reflection of the democratization of American higher education meaning a growing number of minority, low-income, and low-ability students in the test-taking pool. Certainly, changing demographics contributed to the decline, yet something more was happening. Declines occurred at the top of the ability distribution, especially on the verbal part of the test. For example, in 1972 (the first year for which comparable data were available), 116,585 students - 11.4 percent of test takers -scored higher than 600 on the verbal test. By 1983 that number had fallen to only 66,292, or 6.9 percent of the total. Since then the proportion of high-scoring students has remained around 7 percent. By contrast, in mathematics the decline at the top was only temporary. In 1972, 17.9 percent of test takers scored over 600. That proportion dipped as low as 14.4 percent in 1981, but by 1995 it reached 21 percent - the highest proportion of students ever to exceed 600 on the math test.