SSAT Math Scores Skewed?

<p>I have seen a lot of talk on this forum about SSAT Math scores being skewed. Can anyone explain this to me? I got in the 99th percentile last year on the math section, but I assure I am not as smart as some people say you have to be to get a 99.</p>

<p>What people are saying is that so many kids do well on the math section, that the percentile is not a real indicator of your skill ability anymore. For instance, (I’m just making up numbers here): say you got 5 wrong on the math section. That’s not too bad. You’re probably still pretty proficient at math. However, the percentile will be really low…like a 65%. and the percentile change between getting a 90% and a 70% can be the matter of just a few questions. People attribute this to the large number of people who are basically geniuses at math taking the exam (which is more common than a “LA Reading” genius per say. So yeah, it just means that the percentiles don’t reflect the ability or score anymore.</p>

<p>It means the math section is too easy, really. My D got a 770 and it was something like 87th percentile. Nuts, eh?</p>

<p>I think for 10th grade the PSAT should be required.</p>

<p>@putnamehere,</p>

<p>The scoring for the math is skewed b/c if you miss a few questions, your percentile goes down significantly.</p>

<p>many students from china and korea are applying for high schools in america, so when they take the SSAT’s they almost ALWAYS get 800 (or at LEAST above 780) in math section. thats my theory… since the math section really is, SUPER easy.</p>

<p>vivsters, The test is only easy for about 10% and SUPER easy for another 10% of the kids who take it. And for them, I think it really should be harder. Maybe that’s why after the 80th %tile, it really becomes not as important.</p>

<p>I wonder when the last time this test was normed? It may be time to re-center it. The score range is 500 - 800, so assuming there is a normal distribution, the maximum of the bell curve should be at 650. According to the the score reports, the average for:</p>

<p>9th grade girl </p>

<p>V668
M696 - OP, notice how high this is.
R662</p>

<p>8th grade boy (last year’s report)</p>

<p>V662
M689
R650 </p>

<p>I am NOT going to figure the standard deviation :), but I’m fairly confident the math averages are off enough to be statistically significant.</p>

<p>So…that is the Very Long answer to what people mean when they say that the math scores are skewed.</p>

<p>Schools care about the percentiles not the raw scores.</p>

<p>The percentiles are based on the scores. So when the median moves to the right, the percentile does as well.</p>

<p>Yeah sort of unfair but that’s life. I got a 743 and that was in the 60’s. I’m a 9th grade boy though.</p>

<p>@neato:
The scores are centered based on the median, not the average. Anybody who got ~50 percentile care to disclose their math score.</p>

<p>I was assuming a normal distribution (where mean, median and mode are equal). That’s really the only way that percentiles can be accurate, correct? </p>

<p>I was also assuming that the center of the scale should equal the median (and thus the mean IF it is indeed a bell.)</p>

<p>I am no statistician, so please forgive my errors (and point them out so I can learn).</p>

<p>@neato: it appears from the CC discussion that the distribution is bimodal, with one sharp peak at 790-800 and the other peak, much wider, at a much lower score. In this case, mean, median, and mode are going to be widely separated. Recentering won’t change the shape of distribution curve, while the other proposal - making the test harder - will just shift the second peak to the left without having much effect on the sharp peak at 790-800. These kids are good - they will score near perfect with a much harder test.</p>

<p>I wouldn’t be surprised if there isn’t a similar second peak on the Math SAT. SAT II Math II subject test is a bit tougher, but not much. Question is if you’ve identified the top 5% of students, is there any reason to construct a tougher test to identify the top 1%?</p>

<p>I could see how combining section scores could create a distorted combined percentile. However, I think the AO’s are a bit more thoughtful. I’m sure they are much more interested in balanced SSAT performance than the highest overall percentile.</p>

<p>One of the sad realities of corporate America is that math geeks are a dime a dozen. We have armies of them buried away in cubicles crunching numbers. Need a lot more than a 99.9% performance on the Math SSAT or SAT to succeed in this world.</p>

<p>My son just got his Jan. 8 scores. His math was 710, 44th percentile. He is in 9th grade and a solid but not great math student. However, of the 50 math questions, he got 14 wrong and omitted only one. He is only halfway through geometry and my guess is that there were at least 5 or 6 questions that covered material he hasn’t had yet. I expect if he had omitted some of the ones he had gotten wrong, his percentile would be significantly higher. He understood the strategy in the verbal—he omitted 10 answers there and still did reasonably well. But he went for broke in the math and it didn’t pay off. Nevertheless, I can’t imagine that a few questions are a big deal for most admissions officers. We’re looking mainly in the South, so it’s generally less competitive.</p>

<p>no, it’s not based on those percentiles. The only scores that the school looks at are the percentile based on other students in your grade/gender. So if you got a 720, but everyone else made lower than you, you would still have a 99%</p>