The Thinking 

Some Basic Facts on the SAT

September 19, 2011


YOU MAY have heard that American college students are doing worse on the SATs, but a look at the official breakdown of scores according to race reveals that whites, who represent the majority of test-takers, are doing better in math than they were in the early 70’s, when scores overall were at their highest, and are doing only marginally worse in reading scores.

The College Board recently published a report that examines scores since the early 1970s. The significant drop in total mean scores can be explained by the increase in Hispanic and black test-takers, whose lower scores are not offset by the scores of Asians, who do significantly better than whites on the math portion of the test and whose numbers have also increased.

In 1972, the total mean score for all test-takers was 530 points in critical reading and 509 in math. The total mean score for whites as of 2011 had declined modestly in reading to 528 points and rose to 535 in math.

The mean scores for Asians this year were 517 in reading and 595 in math; for blacks, 428 in reading and 427 in math; and for Mexican-Americans, 451 in reading and 466 in math. Asians made up about 184,000 of the more than 1.6 million test-takers. Blacks, though a much larger percentage of the population, accounted for about 216,000 of those who took the test.

The College Board also breaks down scores by sex. For whites, boys scored a mean of 531 on reading, as compared to 526 for girls.  The mean for boys was 552 in math, as compared to 520 for girls. And for the writing portion of the test, white males scored a mean of 507, compared to 524 for girls. Many more girls take the SAT now than boys. The total number of female test-takers (876,000) exceeded the number of male test-takers (770,000) by more than ten percent.


                                   — Comments —

John writes:

Unfortunately, the information you provide about SAT scores of white test-takers remaining stable is not accurate, due to the fact that the College Board has changed the test and re-normed the scoring several times since the seventies. The numbers look the same now as they did back then because the College Board moved the bell curve of scores in order to achieve exactly that result, not because test takers are doing as well today as they did previously. The Wikipedia page on the SAT test has a great deal of information about various changes made to the tests. Here is a short synopsis of the most significant change:

The test scoring was initially scaled to make 500 the mean score on each section with a standard deviation of 100.[26] As the test grew more popular and more students from less rigorous schools began taking the test, the average dropped to about 428 Verbal and 478 Math. The SAT was “recentered” in 1995, and the average “new” score became again close to 500. Scores awarded after 1994 and before October 2001 are officially reported with an “R” (e.g. 1260R) to reflect this change. Old scores may be recentered to compare to 1995 to present scores by using official College Board tables,[27] which in the middle ranges add about 70 points to Verbal and 20 or 30 points to Math. In other words, current students have a 100 (70 plus 30) point advantage over their parents.

Laura writes:

Thank you for that information. I thought that the recent College Board figures from the 70s accounted for the renorming. The figures are here. This is how the College Board explains how the renorming is reflected in the figures just released:

For 1972–1986 a formula was applied to the original mean and standard deviation to convert the mean to the recentered scale. For 1987–1995 individual student scores were converted to the recentered scale and then the mean was recomputed. From 1996–1999, nearly all students received scores on the recentered scale. Any score on the original scale was converted to the recentered scale prior to computing the mean. From 2000–2011, all scores are reported on the recentered scale. Cohort data presented prior to 2007 include students testing through March of the senior year, while cohort data from 2007 to present include students testing through June. For further information see

The other factor I did not mention is that the test is considered by some to be easier today. There are conflicting views of this; a survey of math teachers found that, when the teachers were given sample questions from tests given from the 1980s to 2005, they tended to believe the questions had gotten harder.

Diana writes:

“The total number of female test-takers (876,000) exceeded the number of male test-takers (770,000) by more than ten percent.”

This may have something to do with the fact that the number of black and Hispanic test takers has risen dramatically, artificially inflating the number of female test-takers. I suspect that the breakdown among whites is 50/50. Just a suspicion, though, no hard data.

Laura writes:

Among whites, 410,811 boys took the test, as opposed to 454, 849 girls, so the discrepancy is fairly high. Among blacks, 95, 939 boys were tested as opposed to 119,877 girls, which represents a larger gap between the sexes.

Clark Coleman writes:

A key point is missing from the SAT score discussion. It has been proven that the declines in scores are NOT merely due to an increased number of test takers, or an increased number of minority test-takers, or any similar reason. The proof is that the absolute number of students scoring above certain levels declined during the 1970s, with the comparison being based on a test with a fixed norming throughout the period. So, if the absolute number of students scoring above 700 on the verbal portion (or the math portion) declines, this is not explainable by adding in some low scorers to the test taking pool, which is an explanation that only works for declines in average scores. 

See a discussion here with footnotes. Quoting from the article: 

During the 1960s and early 1970s, spending on education soared, but scores on the SAT plunged after peaking in 1963-64. In part, this was because the pool of test-takers increased, but in 1977, a panel organized by the College Board found that the decline was also caused by lowered expectations, reduced homework, and increasing numbers of non-academic courses.26 In fact, from 1972-1994, there was a 37 percent decline in the number of students who scored above 700 and a 50 percent decline in the number of students who scored above 750 on the SAT; since this is a decline of absolute numbers with high scores, not of the percentage with high scores, it cannot be explained by the larger pool of students taking the test.27 

26: Dianne Ravitch, “Defining Literacy Downward,” New York Times, Aug 28, 1996.

27: Gatto, Underground History of American Education, p. 55. 

When I examine reference 27, I find this quote: 

What should be made of a 50 per cent decline among the most rarefied group of test takers, those who score above 750? 2,817 American students reached this pinnacle in 1972; 1,438 did in 1994 — when kids took a much easier test. Can a 50 per cent decline occur in 22 years without signaling that some massive leveling in the public school mind is underway?

Laura writes:

Interesting. The latest College Board report does not include the absolute numbers of students in this high-scoring group.

On a related subject, here is a Wikipedia entry on the debate over why math scores have tended to improve while verbal scores have declined or remained stagnant. 

Diana writes:

Thank you for the correction. The sex ratio gap between white female and male is dismaying; the black/white sex gap is appalling.

Another worrying factor is that this downward trend has occurred even as the test-prep industry has boomed. In my day (I graduated from high school in 1972), we took a preliminary PSATs, then SATs. Of course, we all crammed and studied at home, but I don’t remember anyone’s parents paying for tutoring. And I went to a stereotypically academically pressured high school.

So, we have more girls than boys taking the test, in some racial groups girls outnumbering the boys massively. And on the whole doing worse, despite being intensively drilled beforehand. Oh, I nearly forgot; a dumbed-down test.

Laura writes:

As I mentioned above, math teachers contend the math section has gotten harder since the 80s. The SAT is still believed to be a reliable indicator of IQ, and that is its purpose. (See “Correlatons with IQ” in the Wikipedia entry.) It shouldn’t be considered a complete measure of the learning of high school students. 

One thing to bear in mind in connection with the differences in the relative numbers of boys and girls is that more boys go to vocational schools, directly into the trades and into the military.

The high incidence of test preparation is another factor complicating comparisons with scores in the 70s. Also, some students today receive special test conditions due to “learning disabilities,” and have more time allotted for the test.

Thomas F. Bertonneau writes:

Source A says one thing about the SAT’s and Source B says another. Commentators confuse the matter by offering additional, competitive interpretations. 

The confusion is itself the product of endless cynical tinkering with what should be a straightforward measurement of intellectual capacity and basic knowledge. I can tell you this: Many of my college students have trouble making simple inferences from a statement; many haven’t the slightest notion of an historical chronology, and few have a grasp of the fundamentals of English grammar. However, their scores on “standardized tests” like the SAT’s have qualified them for admittance to higher education.

Laura writes:

There is one thing to be said for Mr. Bertonneau’s students. They probably perform at a higher level than would students who did significantly worse on the SAT.

Laura adds:

See a reader’s comment disagreeing with my point above here.



Share:Email this to someoneShare on Facebook0Tweet about this on TwitterPin on Pinterest0Share on Google+0