<p>Haha the walkway…aime 2007 #2.</p>
<p>I have to agree with the OP. I find that SAT scores, beyond a certain point, don’t really mean anything because scoring 2400 vs. 2300 has more to do with chance than anything else. Furthermore, the test itself is ridiculously flawed. It tests more speed reading skills more than anything else. For example, in the math section, you have a bunch of elementary school level problems, instead of having a few more difficult math problems that requires some thinking and time. Similarly, the SAT subject tests basically tests how fast you can read and compute quick answers to very shallow test questions. (Seriously, when will you ever have to face on a real physics exam dozens of worthless questions in 60 minutes? The upper level physics exams at MIT have 4-6 questions to solve in 3 hours, and actually require some thinking)</p>
<p>For example the problem presented up here could be solved by most 4-6th graders who know basic algebra. It’s just they might not compute it fast enough. Most of the time, a slow but careful and insightful thinker will always be able to crack a problem, while a quick thinker might not. And the SAT favors the quick thinker. I know plenty of 2300’s+ students who are completely helpless when they are faced with a slightly harder problem than SAT-type because all they can do is think and read quickly on shallow problems.</p>
<p>I applaud MIT for not taking the SAT too seriously. People should not try to decry MIT etc… for rejecting 2400’s, but rather should ask ETS to make a test that actually is able to seperate the highest scorers (the AIME does a good job of doing so, the AP test can too, but the scaling of 1-5 leaves very little resolution and precision). Note I don’t necessarily propose to replace the math section of the SAT to be replaced by the AIME, however, because: AIME relies too much on tricks and cleverness, which is only one part of doing mathematics. I think a balance should be found between USAMO/AIME and AP Calculus/SAT, where both proof-writing skills, cleverness, math knowledge and careful thinking is tested (For example, present a problem where the solution requires you to consider multiple possibilities, assess each of them, and re-evaluate your answers for satisfying conditions you set) Such skills are very valuable for college mathematics.</p>
<p>PS: I was a fairly successful test-taker too in HS. I’m not here complaining about my scores, but more reflecting back on the flawed test-taking system in the US.</p>
<p>No one here is saying 2400 should have a significant advantage over a 2300-scorer. I’d say no college distinguishes between a 750 and a 800 in a specific subsection of the SAT. </p>
<p>The question is where you draw the line. Once you start getting to the lower 700’s and the 600’s on the math section, IMO it’s probably because of some deficiency. The caveats are, of course, if you are not beyond algebra II and geometry when you take the test, and if english is not your first language. For the Math SATII, the curve is (or was) so generous you could make 6 mistakes and still get 800. For that test, a score less than 800 would raise a red flag to me. </p>
<p>I think we all agree the math section is a remedial exam.</p>
<p>
</p>
<p>I agree. A crucial part of the exercise of word problems is to get to the equation. Some people will see “3 girls for every 5 boys” and think 3/5 boys instead of the correct 3/8 boys. You have to understand the difference between a ratio and a fraction. That’s like 6th grade math.</p>
<p>What I’m proposing is that we get rid of the SAT-type tests, and completely replace it with AP tests (with a different scale) or some other tests that has a healthy combination of knowledge, cleverness, depth of thinking.</p>
<p>The reasons are:</p>
<ol>
<li>prepping for such a test will not be a waste of time. You’ll actually use skills that will be useful for college</li>
<li>You’ll learn more when you prep for those tests</li>
<li>those tests will be able to separate the top scorers that the SAT can’t</li>
<li>I’d much rather learn electrodynamics from griffiths rather than Princeton Review… just saying (and SAT-subject tests encourage the latter rather than the former).</li>
<li>Those challenging type of tests are much more fun and exciting, both for the test-maker and the test-taker.</li>
</ol>
<p>Just a contrast of 2 types of questions.</p>
<p>8.022 exam question (I think I saw it somewhere, don’t remember though):</p>
<p>Suppose the field for a point charge is no longer isotropic (it has an angle dependance) but is still scaled by 1/r^2. Will Gauss law hold? Will the circulation of the E field =0? etc…</p>
<p>SAT-type physics question (made it up):</p>
<p>You have 2 charges of equal magnitude and sign. You reduce the charges by half. How close do you have to move them to have the same force?</p>
<p>Honestly, I’d rather be tested on the 1st question. What do you guys thinK?</p>
<p>Hey, I’m all for it. But people who can’t do math would all do abysmally. </p>
<p>And people would say the test discriminates against poor people who can’t take AP classes. People will always say that, no matter what the test is.</p>
<p>Maybe they should replace it with a human-sized maze in the middle of the Sahara desert, and time people to see how fast they can get out.</p>
<p>
</p>
<p>I think about the heads of about 95% of the country would explode after reading the first question.</p>
<p>Actually, I would leave WAY more time for the first question. It will ask for a complete proof argument. Like 5- 10 minutes at least. SAT physics would leave you 1 min for the 2nd question.</p>
<p>It’s not really that the magnitude of the difficulty is exponentially larger, but more the type of difficulty. Most intellectually oriented person would rather be tested on things that requires some thinking instead of pure time-crunch speed computation.</p>
<p>On the AP protest, I would say that most serious students at MIT would say they learnt the most outside of school than they did in school. Secondly, the change in the testing system will be reflected in a change in curriculum across the country. People, instead of mandating Halliday-Resnick-Walker for physics, will use sth more like Kleppner/Kolenkow-Griffiths. Or Stewart Calculus will be replaced by Apostol&Spivak. It will just be a change of focus. Of course the number of perfect scores will fall down considerably, but as we have seen having dozens of perfect scorers competing for the same spot isn’t really helpful to admission officers.</p>
<p>Um, I don’t think that’s really feasible. When it’s not hard to find a Calculus teacher that probably couldn’t get a 5 on the AP exam, I don’t think you’ll be able to make Apostol/Spivak standard for calculus.</p>
<p>My Apostol/Spivak Analogy was just for calculus. Many other things could be taught more rigorously, without involving too much high level stuff. Although I have to say, I like Apostol’s style better.</p>
<p>
</p>
<p>There are several moving parts, any one of which is at the 4th to 8th grade (US) level, but combining them effectively requires some more systematic understanding that is not an automatic product of having passed those grades. For example, going from “3 boys per 5 girls” to 3/5 or 3/8 isn’t completely routine, it requires figuring out whether one needs to multiply or divide by factors of 3 and 5 in setting up the equations, or does one divide 3 by 5, or something else. You also need to have an overall direction to the work, where further computations focus the given information (there are four equations in four unknowns) toward a solution. This requires some degree of planning based on intuitive understanding of what might lead to an answer. </p>
<p>It’s easy to dismiss the math SAT as a joke but there is nevertheless a certain craftsmanship to the construction of the problems. If the above problem is typical, ETS is not doing a bad job given the limitations on material that can be included.</p>
<p>^^What I’m saying is that the exercise is not too difficult for a high school student, and especially, an MIT applicant, to complete. I’m not criticizing the ETS.</p>
<p>
</li>
</ol>
<p>Here is my comment. I think different tests have different functions. The SAT is intended as a mass check for applicants for basic reading, writing and reasoning/calculation skills. This is fine for the masses applying to arbitrary – probably not very useful for MIT beyond a very small check.</p>
<p>I would say it is not the SAT, but the SAT IIs and APs which need reworking. The SAT, I think, does what it needs to do under its limitations. Nobody should reward a 2380 significantly over a 2300, definitely doesn’t make sense – the purpose of the SAT is not to gauge excellence, but to check off some basics. </p>
<p>The SAT IIs, on the other hand, seem to fundamentally miss the point of testing subjects, as do aspects of the APs in claiming they really examine a college level of mastery. A lot of the things they test are just ETS catch-phrase questions that you can easily learn of without really understanding the subject. My impression is that the SAT IIs need be replaced with something like the APs in place today, and the APs should be replaced with something legitimately college level, taught in the proper sense.</p>
<p>The AP Physics using Griffiths makes sense. The AP Calculus using a no-nonsense calculus book is a good idea. Every question need not be a proof or super theoretical, as this would miss the point – it should be possible to pass without having a total mastery of the subject. But right now, people who barely have basic knowledge of calculus get 5’s. Certainly MIT is correct in not using these as end-all indicators. I have no obsession with standardized testing, but the fact of the matter is standardized testing often sets the standards. This means that if a school in today’s day and age is to maintain appeal to incoming high schoolers for its calculus program, it just needs to show that 70+ percent get 5’s. That’s laughably easy to do, even teaching at a deplorable level. Similar remarks for the AP Physics.
To get a 5 on calculus, one should be able to do the basics very well, and show at least fairly good understanding of the theory. If the standards are set higher by the national exams, then the colleges will automatically take the exams more seriously, if we’re realistic. And high schools wanting to align to this will have to raise their standards. There’s no use yelling at teachers – they simply are teaching to the standards the students would be willing to learn at, given what they have to do to achieve good scores and get into college.</p>
<p>Further, faraday correctly notes that expense should not be a problem. Nobody says you have to take an AP class to take an AP exam. The books should be available dirt cheap, if not free online for those who want them. </p>
<p>My principle is always that one must ask if an exam is serving a real purpose, and if not, align it to the purpose.</p>
<p>Yes, I agree with with AP calculus thing. It causes calc classes to be focused more on rushing through material and memorizing procedures instead of building a stronger understanding of the concepts.</p>
<p>
</p>
<p>I actually agree with everything you wrote (*), and was addressing the ETS bashers. As you said, below 800 on the math usually indicates not being all that strong (since several mistakes are allowed before the score goes lower), and 700-750 already starts to indicate problems. In the absence of a national curriculum the SAT does a good job of cognitive discernment in the low to significantly above-average range using basic material. At the high end other metrics are more meaningful.</p>
<p>(*) except that there is no difference between ratios and fractions!</p>
<p>
Depends on the curve, but almost never true.</p>
<p>
MIT admissions doesn’t seem to think so, and they have a ton of data to base that on.</p>
<p>
</p>
<p>We are talking about the Math SATII, or at least I am. A score of 780 on the Math SATII is like a score of 680 on the Math SATI, in terms of the number of mistakes you can make and get that score.</p>
<p>
</p>
<p>I’d have to see the data, but their conclusions may be skewed by the fact that the first year was pass/fail (where people are forced to take math-related subjects) and the fact people who have trouble with math will end up in non-mathematical majors (e.g., bio).</p>
<p>Even if they looked at the first year “hidden grades”, I’m not sure what that would mean because a lot of people weren’t trying.</p>
<p>
</p>
<p>According to ETS, below 800 on the post-1995 math SAT I is equivalent to below 780 on the old math SAT, so yes, some mistakes (or nonanswers) are allowed. I don’t know how many mistakes were allowed for 800’s on the old exam. The SAT-II (College Board Achievement Test) math scale is far more generous.</p>
<p>[SAT</a> I Individual Score Equivalents](<a href=“http://professionals.collegeboard.com/data-reports-research/sat/equivalence-tables/sat-score]SAT”>http://professionals.collegeboard.com/data-reports-research/sat/equivalence-tables/sat-score)</p>
<p>
</p>
<p>750 math SAT will make it very clear to MIT admission that you are not a math genius. Whether they admit you for other reasons is a separate matter. The admission rates approximately double every +50 points on the math SAT, even at the upper range. See the current thread on “reconstruction of 2009 domestic admission rates” for some data.
Also, admission of below 700 (and possibly below 750) scorers has to be heavily stacked toward nonacademic priority categories, such as minorities and children of MIT benefactors. Women may or may not be over-represented in those ranges.</p>
<p>
They’ve said that if your score is above a low 700/high 600, they don’t consider it a factor in their decision at all.</p>
<p>
Correlation doesn’t prove causation, especially in this case. If the rest of your application is really strong, you are likely to study hard for the SATs, so that your scores don’t disqualify you.</p>
<p>
‘Children of MIT benefactors’ don’t get a leg up in admissions. And “The mean SAT scores of [MIT’s] minority undergraduate students are higher than the mean SAT’s for all of the students enrolled in the Ivy League.”</p>
<p>
</p>
<p>That equivalency is probably based on percentiles, not the number wrong; maybe the top chunk of people started performing better. I know for a fact that 1 wrong in 1995 (after the change) got you a 780 on the Math SATI. Two wrong was 760. It was 20-30 points off per question. So recentering did not give you any leeway on the math SATI, at least the year after it changed.</p>