kitchen table math, the sequel: get your recentered SAT scores right here

Monday, February 19, 2007

get your recentered SAT scores right here

Thanks to David I have a link that works.


2000 SAT I Mean Score Equivalents


Under the heading "how politics works," contributors and readers who are attempting to improve their schools should probably do screen grabs of this page for safekeeping. Very few parents know the SAT was recentered (I didn't), and the 70-point gap between the old Verbal scale and the new is a reality check.


Defining Literacy Downward

Diane Ravitch

Originally called the Scholastic Aptitude Test, the SAT was designed in the 1920s as an inexpensive, objective predictor of students' ability to do college-level work. It was developed as a multiple-choice alternative to written examinations in subjects like literature, history, science, mathematics, and foreign languages. The old "college boards" were subject-matter examinations, individually graded by teachers and professors. By contrast, the SAT was an intelligence test, graded by machine. When World War II broke out in 1941, members of the College Board decided to adopt the multiple-choice SAT in place of the time-consuming written examinations.



Scores on the SAT range from 200 to 800, with a mean of 500. In 1963-64, average scores reached an all-time high of 478 on the verbal portion and 502 in mathematics. After that year, however, the scores dropped steadily until 1980, when they leveled off with verbal scores at 424 and the mathematical average at 466.



In 1977, a blue-ribbon panel commmissioned by the College concluded that the steep erosion in SAT scores was due in part to the diversification of students taking the test, but also to lowered expectations, an increase in nonacademic elective courses, the assignment of less homework, grade inflation, and a diminution of thoughtful reading and writing. It concluded that the score decline was significant but not irreversible.



In recent years, states and colleges have raised standards in order to improve academic achievement. In response, students are now taking more advanced courses in mathematics and the sciences. Because of higher enrollments in courses like algebra and geometry, SAT mathematics scores have improved. Verbal scores, however, have not. As of 1995, the SAT mathematics score had increased to 482 (and it was up another two point in 1996)--not very far from its historic high in 1963-64-- but the verbal scores scarcely budged, never rising above 431. (This is not, as some might think, caused by the increased diversity of test-takers, the verbal scores of non-Hispanic white students have not risen above 450 in twenty years).



There are three clear lessons to be drawn from the very different trend lines on the SAT's mathematics and verbal tests. First, increased numbers of racial and ethnic minorities in the college-going pool were not the primary cause of the long score decline. In fact, by 1995, the proportion of students who scored over 600 (21%) in mathematics was higher than it had ever been (and the proportion of students who scored over 600 on the verbal test was consistently small at only 7 percent) . Second, students who take advanced courses in mathematics do better on the SAT than those who do not; or, effort in school is rewarded with improved test scores. Third, for the past generation, the United States has had a huge and unaddressed problem in the teaching of the English language.



Why did verbal scores remain in the cellar? One need only read the abortive "national English standards" to see the incoherence and jargon that has invaded the field of "language arts" in recent years, apparently driving out concern for careful use of the English language. If English teachers believe that any usage is acceptable, then it is not surprising that students would perform poorly on a test that requires knowledge of the English language's grammar, structure, and vocabulary.



Faced with sharply dissimilar trends on the verbal and mathematical portions of the test, the College Board chose to "recenter" the SAT. By psychometric and statistical fiat] The Board made the de facto averages in mathematics and verbal tests similar and higher. Beginning in 1995, it arbitrarily declared that the abysmal verbal average in the 420s was now to be treated as a 500, while the respectable 479 on the mathematics test was also converted to a 500. The reasons, the College Board said, were entirely technical.



What has been the practical result of the "recentering"? The SAT has "normalized" poor performance, both in comparison with the past and in comparison between the verbal and mathematical tests. Using the old mean, the mathematics scores are consistently higher than the verbal scores over the past thirty years; now the once-embarrassing verbal scores appear to be substantially higher on the trend lines than mathematical scores until the past few years, which is the exact opposite of the truth.



The latest report shows that the new national verbal score is 505 and the mathematics score is 508. Now we are to believe that verbal performance is slightly above average when it was more than 70 points below average only last year. The difference between excellent performance in mathematics and truly terrible performance on the verbal test has been eliminated by nothing more than legerdemain (even though the College Board made the verbal test easier last year by banishing questions about antonyms). On the recentered scale, the verbal "average" was recomputed by the College Board as a phenomenal 530 in 1972 at the very time that scores were falling. Thus, recentering not only falsifies the quality of performance in the present but in the past as well.

The old average was a standard that American education aspired to meet; the new averages validate mediocrity. We now know what the average of today's student performance really is, but the SAT no longer serves as a standard toward which today's students may aspire. (This is comparable to the difference between NAEP's proficiency levels--which describe what students know and can do--and its achievement levels--which compares students' performance to what they ought to know and be able to do.)



Does it matter? The value of standards is that they cause us to exert effort. Whatever our line of work, we need performance standards that show what can be done with effort; without realistic and high expectations, we can't expect improved performance.

The official word here in Irvington, with SAT scores of 578 V & 602 M, is, "Our scores are outstanding."

That's it.

The middle school principal actually said, at a Board meeting, that the kids who are "struggling" are the only ones he's concerned about.

The district has no apparent goal of raising achievement levels for children scoring 3 or 4 on the state test, though lip service is paid to raising 3s to 4s.

Precious time and energy are directed toward the goal of keeping kids out of Honors courses, accelerated math courses, Earth-Science-in-the-8th-grade courses, etc.

I have never heard a single administrator talk about improving instruction so that more students can benefit from advanced work.

Not one, not ever.


SAT test
sample SAT test
get your recentered SAT scores right here

7 comments:

Anonymous said...

The link above is for recentering mean scores (e.g. mean verbal score for Irvington School District). If you want to find out what your own individual recentered score is, try this other link to the SAT I Individual Score Equivalents table.

(first post on newktm; trying out the "other" option)

Catherine Johnson said...

Google Master!

I'd been wondering where you are!

Good to see you!

Catherine Johnson said...

oh!

I thought I posted that link!

Thanks!

Catherine Johnson said...

Those are the recentered scores.

For now I just want to know what the 88th percentile on the ITBS would be on the SAT.

I don't know, obviously, but I can see that the 88th percentile on SAT Math is a decent score.

Anonymous said...

For now I just want to know what the 88th percentile on the ITBS would be on the SAT.

First you need to determine the population being tested. Who takes the ITBS, all students or only college bound students?

Let's assume that you're talking about an 88th percentile within the population of Irvington schools. That makes it easier since I think maybe we can assume that almost everyone in Irvington goes to college.

So, we need to find a score distribution for college-bound seniors taking the SAT in New York.

Oh look, here's one.

Table 6: Score Distributions.

Click the Select toolbar button, drag-select and copy the contents of that table into Excel. Once in Excel, Data / Text to Table / Delimited by spaces to convert it to columns.

Add a column for the running total (bottom to top) and a column for the percentile (running total / grand total), and you've got percentiles.

Looks like 91st percentile is 600-649. So... I've gotta run, so can't stop to figure out which one, but 88th would be either 600-649 or 550-599. That's critical reading, which I think is what used to be called Verbal. Can do the same with Math.

Anonymous said...

I also have a question with regards to the effects that recentering the SAT had on percentile ranks, based the table at: http://www.eskimo.com/~miyaguch/sat.html

I sent for my archived 1972 SAT scores. I couldn't remember my scores but I recalled that the percentile was in the high 90's. I was surprised that my score was only a 640 in math. On the table from 1974, a 600 was 94% nationally, and 650 was 98%. However, when using the CollegeBoard conversion tables for the new recentered score, a 640 stays a 640, which is now an 83% nationally.

Yet CollegeBoard claims in a research paper The Effects of SAT Scale Recentering on Percentiles
that "Percentiles will remain virtually unchanged." at:
http://professionals.collegeboard.com/data-reports-research/cb/sat-recentering-percentiles

Any insight into this discrepancy?

Anonymous said...

"Any insight into this discrepancy?"

The math scores have been trending upward for a number of years. This has nothing to do with the recentering. What it means is that a score that would have been good enough to be in the top 4% in 1972 would only be good enough to be in the top 17% today. More students are doing better at math. This is a good thing (assuming that it isn't because the test is being successfully gamed).

This sort of thing is quite common in sports. In 1972, for example, a time of 51.47 seconds for the 100 meter freestyle long course swimming would have put you in the top 1 in the world. It was the world record. The new world record is 46.91 seconds and I suspect that a time of 51.47 seconds wouldn't even get you into the final round of many world competitions.

The 51.47 seconds in 1972 didn't get any slower ... the competition just got faster.

Similarly, your 640 in 1972 was good enough to be in the top 4% in 1972. Today, it isn't. Because today's students are scoring better in an absolute sense.

-Mark Roulo