The anxiety teenagers feel at the prospect of taking the Scholastic Assessment Tests is nothing compared with the terror that overcomes their elders as they await the annual disclosure of national and district scores.
The College Board released its report on the Class of 1995 Aug. 24, to sighs of relief from educators who find the continued failure of their initiatives and innovations a profound embarrassment.
Scores on the verbal part of the test rose five points, College Board President Donald Stewart announced, and three points on the math portion.
Impressed? Don't be. They changed the test. 
"The Class of 1995 was the first to take the new SAT-I: Reasoning Test," Stewart said in a press release, "which was carefully designed to have the same level of difficulty as the earlier SAT, but to more closely reflect what and how students are learning today."
Stewart's remarks are carefully designed to obscure the point, which is that a test that more closely reflects a less demanding curriculum will be a less demanding test, although students may well find it as difficult.
The College Board's collapse of standards will be complete next year, when it changes the scores as well, in a process it misleadingly calls "recentering."
Before I delve into the dirty details of recentering, I want to give the SAT people credit for being right about one thing, and for putting it front and center in their report every year: If you want to go to college, you have to prepare for college.
Students who have taken 15 or fewer full-year courses in academic subjects during high school, less than four a year, average 354 verbal and 406 math scores on the SAT. Students who have taken 20 or more, at least five academic courses each year of high school, average 474 verbal and 531 math.
Yes, there are also differences by ethnicity and gender and a number of other variables the College Board keeps track of. But those are things students can't change. They can control whether they register for consumer math or precalculus, for family relationships or biology.
Virtually all the improvement in this year's scores came among top students; scores for those in the bottom three-fifths by high school rank stayed level or declined. The improvement for top students, moreover, is attributed to increases in the number of honors courses students take.
The College Board is right about that, but it cautiously stops short of pointing out that its recommendations implicitly contradict the dominant philosophy of the establishment that one education fits all. The old "college prep" curriculum is being resurrected under a slightly more democratic name.
Perhaps a return to tougher courses would in time reverse the decline in scores that has occasioned so much hand-wringing over the last couple of decades. It's hard to be optimistic, though, because the grading in those courses is getting steadily easier, with 35 percent of this year's test-takers reporting A-minus to A-plus averages.
But we won't ever really know, because starting next year the College Board will report scores according to a new formula that will effectively prevent colleges from using their historical data for comparison.
The change, which the College Board calls recentering, will move the average score up about 75 points on the verbal scale and 20 points on the math scale, so both come out around the 500 midpoint between the top and bottom scores of 800 and 200. In fact, students who took the test in April of this year received "recentered" scores, although their scores were converted back to compute the 1995 averages.
The reasons for the change are unconvincing, to say the least. The new scores will be more "intuitive," said Bradley Quin of the SAT program in a March 1995 press release.
"With both averages near 500, students will immediately know where they stand in the test-taking population."
Tell me please, what is "intuitive" about a 200-to-800 scale? It's all a matter of what you're used to, like Fahrenheit temperatures. No one has trouble understanding either weather thermometers or fever thermometers just because they show different averages. No one should have trouble understanding that a good math score and a good verbal score aren't exactly the same. What's essential is relative ranking.
"After recentering, students will still be in front of or behind whoever they were in front of or behind before recentering," Quin's statement said, which is about as illuminating as telling us that water is wet.
"The percentile will continue to pinpoint student achievement no matter how scores are reported."
The other reason for recentering, the College Board claims, is that it was last done in 1941, "when some 10,000 predominantly white men from Northeastern states took the SAT and sent their scores to highly selective colleges."
But so what, if percentiles are all that matter? The real problem is that average verbal scores dropped 43 points from 1967 to 1994, recentered or not. Math scores dropped too, but not so far, and they have partially recovered. Fiddling with the formula doesn't change those unpleasant facts. 
Recentering doesn't just change the average, it also performs cosmetic surgery on the shape of the curve. The bulge in the number of very low scores, 200-240, is removed, while the sagging number of high scores, 600 or above, gets an artificial lift.
Unfortunately, recentering just makes it easier for schools and students to put off taking the College Board's excellent advice about signing up for a lot of challenging courses.
In education, you get what you work for.