This test was a bummer for many. I don’t think complaining about it is a great idea at this point. All one can do is keep prepping for a few more weeks and take it again in August. There may be a lot of kids taking that test now, so I would get your seat while they are still available. I certainly would not say anything to colleges about your test date and your issues with the curve.
The reason that the math curve was so hard is because getting two easy questions wrong means that the student probably would have gotten 4-5 harder questions wrong on a harder test. That’s what “equating” does. This test is assuming that kids who got three “easier” questions wrong would have gotten maybe 9 questions wrong on a harder test because this test is missing those harder questions. For some kids that may be true, but that’s not true for all kids who could have gotten some of those harder questions correct and maybe just made a mistake or two on this test.
Can anyone statistically savvy explain the consequences of removing 4 questions from the R/W portion of the test?
I’m just curious. My daughter actually did OK on the test as she improved by 50 points (although that took an additional 22 correct questions compared to the May test to accomplish). She’s in the 90ish percentile and I’m kinda of thinking the test was an accurate reflection of her abilities. It was her second test so she was more familiar with the test and testing experience and she’d been prepping for a few weeks so some improvement was expected.
But I’m wondering how those 4 questions come into play.
With so much news about this keep in mind the AOs will be aware of this. They know what range the top scorers are in regardless. They know the difference between 750-800 is only two questions.
@123Mom456 I have heard that AOs do not see test dates…or sometimes assistants in the office fill out a “top form” that goes to the AOs and, on that form, they write superscored test scores. There wouldn’t be a date if that’s the case.
@suzyQ7 Even if they see the date, these AOs spend seven minutes on the whole app. I don’t think they are going to give special consideration to the test date…at least I would not bank on that.
And I get that people don’t like standardized tests and I’m not a huge fan either but it is what it is. Schools use these scores. Complaining to CB is not going to change those June scores. If our kids took this test, I would be upset for a few days but then we’d move on and take it again if needed. Life isn’t always fair.
Does anyone know whether 4 unscored questions is a lot compared to prior tests? Perhaps they were more difficult questions (though my understanding is that the unscored questions were reading and writing, not math).
The anecdotes where the student prepped more and scored lower - not terribly unusual for June - might call into question Reliability.
From College Board’s twitter statement on June:
vs the above-linked technical manual:
The anomalous nature of the curve, compared to prior New administrations (examples in Princeton Review blog) might indicate that the test forms were not already as parallel as possible.
I suppose that College Board feeling the need to make a statement at least means they have heard the feedback.
jujunette01: The June scores were reported five weeks after the test, not two weeks. As the Princeton Review piece noted, that could suggest post-test curving, not just pre-test “equating.”
For those who don’t understand why this particular test is a big problem, take the easy test issue to an extreme by imagining that it had been a basic, 100-question, single-digit, timed math facts test. If there were 100 problems and a student missed only one, can you confidently say that those who got a 100 are significantly stronger in math than the students who scored a 99? Maybe they are, and surely some of them are, but it’s impossible to know. On a harder test, most likely some of them would miss a few or even quite a lot of questions, whereas some of the 99ers would still only miss one “oopsie” question. On the easy test I might score a 100 and Einstein Jr. might score a 99. Am I the better mathematician? Maybe, maybe not. The point is, you just can’t know with a test that’s too easy, and an AO wouldn’t have a reliable means of differentiating us.
Another example shows why a curve isn’t effective if a test is too easy. In college I took a Calc 2 or 3 class. The first test was very easy. There were a lot of problems on it, but they were all pretty simple. Nevertheless, I ended up missing one (worked the problem perfectly, showed my work, but on the last step I accidentally left the negative sign off, even after writing it in the preceding steps). There is no doubt that I deserved a ding to my grade, but that one missed point translated to a C- on the curve. I absolutely understood why that happened, and I didn’t complain about the use of a curve, but I objected to the fact that the test was too easy to make it accurate. Professor agreed. Next test was much harder, and lo and behold, many—but not all—of the perfect scorers on the first test dropped down on the curve, and while some of the “C-” students remained at the C- level (or lower), many of them moved way up on the curve. Which test was more reliable in differentiating students? Obviously the harder test. Now, what if some students only took the first test and some only took the second, but their grades were compared as if the two tests had been the same? Clearly the tests were not properly equated. That is what appears to have happened with the June test (at least the Math portion).
You don’t have to have a horse in the race to understand why it makes sense to voice an objection. Failing to highlight the problem allows sloppy test creation to go unchecked and persist.
P.S. Students who did well on prior SAT sittings but didn’t take the June test have a good reason to complain, too. Again, taking an example to the extreme, if you got a perfect score on a linear algebra test, is it fair to assume a student who got a perfect score on a basic addition test is equally strong in math?
“P.S. Students who did well on prior SAT sittings but didn’t take the June test have a good reason to complain, too. Again, taking an example to the extreme, if you got a perfect score on a linear algebra test, is it fair to assume a student who got a perfect score on a basic addition test is equally strong in math?”
SAT Math is far more consistent across the numerous administrations, so the Lin. Alg. vs. basic addition example isn't relevant. Equating those two tests is meaningless. The Math sections were apparently relatively easy on the June test but overall the section still mimicked the other tests in terms of content and relative overall difficulty - especially compared to an extreme example such as comparing two completely different math subjects. Also, someone who happened to score 800 on a prior (and more difficult) test would naturally expect that score to impossible to achieve by missing the same overall number on an easier test. That's what "equating" is all about.
And your point is not relevant as the Adcom does not care if you miss one or score an 800.
And you have statistical proof of this?
That is a LOT; historically, only a rare test has an unscored question. To me, that is the big screw up by CB. (The fact that they had to throw questions out means that they were not properly pretested.)
This would be all well and good if the sole purpose of SAT scores were to evaluate students. Unfortunately, they are even more valued as a marketing tool. The colleges need those high scores to game the rankings.
I believe they always throw away questions. I know it happened on my kids’ tests.
I was under the impression that the new SAT was supposed to make the math more difficult so that they would not have to continue to implement the harsh curves with which the pre 2017 kids were quite familiar. If this one was significantly easier, essentially you’re dealing with the same variations in scores experienced by the kids who took the older version of the SAT. I know I’ve also heard this sort of complaint on different administrations of the ACT when the format, type of question, or difficulty varies from what’s expected. @JBStillFlying reports that it showed up with the fall 2016 test-new SAT. It happens.
I always felt that students would be better served by more difficult tests that aimed to separate the top scorers more reliably. But that’s not the goal of the SAT or the ACT. It seems that both tests are aimed at allowing an ever growing pool of students to score well. That’s great of course but it also creates a lot more anxiety over the whole process as more and more kids look alike between testing and high school GPA. If I were still dealing with this (thank goodness, I’m not), I’d put more energy into the intangibles that allow kids to learn, pursue their interests, and define themselves as individuals. If I really wanted to pursue more testing, I’d ask my kid to consider the Math 2 subject test where the curve is generous and it’s easier to show mastery.
^^Well sure, but do you really believe that USNews has a threshold between 750 and 760? Or that AdComs say, ‘ooh, this applicant only scored a 740, so reject’?
I Just think that there was something different about this specific test for example maybe the fact that it was really easy because a lot of my friends and even more thought it was very easier, so collegeboard might’ve messed something up and this resulted in a terrible curve.
Anyway, I thought the point of the SAT is to measure an individuals ability, understanding, and performance in Reading, Writing, and Math…so in reality, there should be a curve in the first place.
^ Results will always show up as some sort of distribution. They may not need to worry about fitting anything to a NORMAL curve if they were just testing mastery - results would probably look more like some of the AP test distributions. If they offered it once a year they wouldn’t even need to do “equating”.