November 5, 2006
Posted by Sam Jackson
Class of 2006 Senior SAT percentile scores number-crunching!
That's right folks, watch and be amazed as I process and condense charts and figures for your consumption! Compiled score report data for 2006 seniors was released at the end of August but I'm only posting about it now. You might recall the hub-bub back then about the score decline--that information was included in the general purpose report. It was explained as being a consequence of the new test. College Board press release on the subject of the report is available here: College Board Announces 2006 senior data. If you're a student this might make you feel worse or better about your test performance; I really couldn't say.
Though we would all like to think the College Board reports percentiles because it likes to induce inferiority complexes, it's more because they want to make their test 'useful' for sorting students. No one knows the difference between a 2400 and a 1600 unless they know the first is above the 99th percentile and the second is the 61st. Here's how College Board describes their statistical nonsense:
SAT Program data provide important information about test characteristics, prediction, validity, and reliability. Additionally, college-bound senior data provides insightful information about the test takers, i.e., subgroup performance, college plans and goals, and high school background.
Sadly, the "Test Characteristics (Reliability, Difficulty Levels, Completion Rates)" mentioned are still "coming soon." This is the section of the report which is meant to demonstrate "...that the test is reliable and appropriately difficult and that a sufficient amount of time is allocated for each test section." Lacking this table I suppose we'll just have to assume that these claims to fairness are untrue. Moving on...
Scores are broken down by gender, ethnicity, and ESL. Each subscore has its own set of sub-data, too. There was even a bogus section claiming that the test was barely coachable based on a somewhat flawed survey taken ten years ago. Very up-to-date, College Board. In any case, the real meat was in the scoring breakdowns; keep reading.
First, the observation which most closely ties to stereotypes! : Asian, Asian Americans, and Pacific Islanders (one subgroup) had their percentiles pushed 'up.' This was most striking for math, where the top 98th percentile was reported at 800. In the group "males" and "other (reported ethnicity)" this was 99th, but elsewhere it was only 99+. This perhaps supports the hypothesis that many Asian-Americans report their ethnicity as "other" to shield themselves from being considered as "over-represented" minorities. I won't make any real conclusions, but it is interesting. However, critical reading and writing subscores were also 99th percentile at 800 points, which outpaced almost all other groups in one or both places. See the ethnicity chart breakdown for yourself.
785,019 (54 percent) of SAT takers were female and 680,725 (46 percent) were male. The female / male score gap, which shrank with the new test to only 26 points from 42, (females scored 11 points higher on the writing, which helped a good deal) has been reported on heavily in the mainstream press so I won't particularly bother with it. Another note on the gender issue: "Thirty-five percent of SAT takers were first-generation college students. Fifty-eight percent of first-generation college students were female. This female dominance is true for all the racial/ethnic groups."
Females lagged behind males overall, but at the top of the spectrum boys were outperforming girls percentage-wise--i.e., boys represented more than 46% of the top-scorers, apparently. The top score brackets had more girls (because there were 8% more female test-takers than boys) but the girls had lower percentiles beneath 97/98 or so, across Critical Reading, Math, and Writing.
For example: Of 8,862 people who scored 800 on the critical reading, 4,760 were female--normal given the demographic breakdown (rounds to 54%), but if we follow along the right-hand column we see that the female percentile for score 720 on CR is 97th, whereas for boys it is 96th by that time. Furthermore if for CR we add all the scorers up for the top 99th percentile girls represent only 53% of that subsection although they represented total test takers 54%. This goes with what is I think understood to be a distribution for males which tends more towards the extremes on standardized testing and intelligence-type things--boys over-represent the very bottom percentiles, too.
But wait! At the same time, males totally dominated the composite score breakdowns--there were far more high-scoring boys than girls (count them up!) However it might be true that given the 26 point score differential, when weighted for the number of test takers, the girls were still doing very well at the top; I don't have enough of a statistics background to say. The curiosity is that if there are more females for each subscore at the top (800) level, why are there fewer 2400 scoring girls? It must simply be true that there were more boys who had high scores across all subsections. Again, interesting.
I could go on and on but there isn't all that much else to say that's terribly exciting. Of additional interest, however, are the state-by-state breakdowns. Check out your own state and see how you (or your state) compare. Massachusetts was looking pretty good. There is SAT2 data around there if you want to look after it.
Also, don't miss the "other" tables section. Some good aggregate data is available there, pre-processed for you--mean GPAs by ethnicity (reported) over the last 10 years, for example.
That's all for now, too many numbers for me.
I'm currently a rising senior at Yale University and I've been blogging about college admissions and higher education marketing trends since I began my college application process in 2005. I now also write about my experience here at Yale.
2 Trackbacks
6 Comments
November 5, 2006
Hey Sam. I'm a long-time reader, but a first-time commenter.
I saw these numbers when they came out in August, but I didn't realize they had state data up. As an Iowan who only took the SAT test to fulfill the National Merit requirement, I was quite curious to read about the performance of Iowa SAT-takers. Since the ACT test is compiled in Iowa, it always proves far more popular (there were only four students taking the SAT Subject Tests at my center yesterday).
1,477 Iowa students took the test in 2005-06 (4% of high school graduates), so you would expect those who -do- take the test to score fairly well.
There were some unexpected results, though: males outpaced females by 16 points in reading, 50 points in math, and 8 points in writing. Public schools outscored private or independent schools in all three categories. The highest-scoring reading ethnicity was Mexican or Mexican-American.
I thought the most interesting statistic, however, was wedged into the College Plans section. Of the 1,477 test-takers, 217 sent their scores to NMSC (the third-most-frequent recipient).
November 5, 2006
I love how .45 a percentile can make a big enough difference that College Board was compelled to report it.
It makes sense that females dominate in all groups because females are born more often than males (I think it's something like two-thirds of the world's population is female).
I'm not sure if you had the time, but did you see ABC News' two-hour special on money and prestige on Friday? A segment was done on college admissions. It wasn't anything new, but they were basically blasting developmental admits, claiming that other people deserve their spots more.
November 5, 2006
Kevin: Thanks for your readership and especially now your very interesting comment--that is quite fascinating. A very self-selecting group there. I wonder how those numbers go across the rest of the ACT-dominant states / counties? On the ACT generally: I've seen some practice tests of it lying around and it looks like a lot more fun than the SAT. That might not be true under test conditions, I don't know. I won't have a chance to take it because it isn't worth getting up at 4 am to take a bus to maine from school to do so. That only happens twice a year, anyways. Too bad...
Stephen: There are more females who have these higher scores because there are more girls taking the test, true--but they still aren't representative percentage-wise. I suppose that percentile difference is still report-worthy because it is a longstanding fact; also because it changed so much. And I did not catch that ABC segment... I don't get to watch very much TV here at Exeter because it usually isn't worth fighting for channel rights with other people in the dorm. Sounds like the mainstream media was making a good point though, so that's good.
November 5, 2006
Based on the length of the test alone (under three hours), the ACT is far more pleasant. The essay is the last portion, which I liked (although the topic was less fun than the SAT's). The reading section can be a bit grueling because all four passages are fairly lengthy, and you're only given 35 minutes for 40 questions.
About 1.2 million people took the ACT last year, so it's still a bit less popular than the SAT. My high school's center typically sees about 150 test-takers from the surrounding area on each date; there are probably about 2,000 high school students in the county. The December test is offered in 60 locations across Iowa and only 29 across Massachusetts.
I scored relatively higher on the ACT than the SAT, and I didn't bother to send my SAT score to most of the Midwestern schools to which I'm applying. The ACT seems to be far more respected here, even at the most selective universities (I got that impression at the U of Chicago, at least).
November 6, 2006
I didn't realize that it was actually difficult to take the ACT out on the east coast. Most people I know take them both; I definitely preferred the format and length of the ACT and also did significantly better on it, which is actually rather irksome. I like that they include science on the ACT, and keep each subject in one section, rather than splitting each into several micro-sections as they do in the SAT. I still find the geographical variances in test preferences (among students and colleges) pretty interesting and unusual. Do you know which colleges only accept the SAT?Another odd bit is that several colleges will accept either the ACT alone OR the SAT with an additional required number of SAT subject tests.
November 6, 2006
I don't know anywhere that only accepts the SAT, just because colleges don't want to limit possible students (to reject, depending on how you look at it). Same thing with the subjects--they want to give more students an opportunity to apply. Probably doesn't hurt that it gives them more opportunities to give college board money, either. I keep feeling the schools and CB are all colluding together to make us take too many tests or something like that. I am still reading about the history of college admissions in The Chosen but I'm only at 1930 or so now, so when I get to the SAT and all that maybe the geographic preferences will be explained. It has always puzzled me.
I think that in more... urban areas, like back home near Boston, it would be rather easier to take the ACT. It's only that here in Exeter, there is nowhere at all nearby. NH has so little demand, I suppose--MA definitely I could understand having more people who would want to take it.
Leave a comment