<p>Given all the research that has been conducted it seems extremely unlikely that significant numbers of people are moving from SAT-M scores of 700 to 800 through preparation. Earlier in this thread, I posted some links about this but they don’t seem to have had much effect on this. Here is some more evidence [Information</a> Processing: Test preparation and SAT scores](<a href=“http://infoproc.blogspot.com/2012/02/test-preparation-and-sat-scores.html]Information”>Information Processing: Test preparation and SAT scores). On the same blog, we find this paper that uses SAT-M scores to predict success in math and physics at the University of Oregon [other subjects were predicted to a significantly lesser degree]. Given that data it seems extremely implausible that there is no difference in math and physics performance based on SAT-M scores. However, the SAT-M ceiling is sufficiently low that the majority of math/physics majors will either have 800s or near-perfect scores so the predictive power will be much less.</p>
<p>
Does the College Board have data on inter-tester reliability? I guess it might be difficult, since each instance of taking the test presumably makes the student better at taking the test.</p>
<p>I think it’s also interesting to ask whether we consider a first-time-2400/second-time-2200 scorer the same or different from a first-time-2200/second-time-2400 scorer. Are both of those students “true” 2400 scorers, or is a 2200 more reflective of the actual ability of one (or both)?</p>
<p>EDIT: And re: 700M vs 800M scorers at MIT – since the 25th percentile for SAT math at MIT is a 740, I suspect the pool of 700M scorers is very small indeed.</p>
<p>I would also argue that “doing well in advanced math and physics courses” is not the only measure of success at MIT. I got a 690 in SAT math, and took the easiest math and physics courses at MIT (18.01/18.02, 8.01x/8.02x). But my lack of world-class ability in math and physics hasn’t stopped me from being a world-class neurobiologist.</p>
<p>
</p>
<p>I found this <a href=“http://media.collegeboard.com/digitalServices/pdf/research/Test-Characteristics-of%20-SAT-2012.pdf[/url]”>http://media.collegeboard.com/digitalServices/pdf/research/Test-Characteristics-of%20-SAT-2012.pdf</a>. On each section, there appears to a 68% chance an student will score within 30-35 points of the original score when retaking the exam. Assuming errors are not correlated across sections, that should translate into 50-60 points out of 2400. Assuming that errors act the same way at the top end of the spectrum this would make it incredibly unlikely that someone scoring a 2200 would outscore someone scoring a 2400 if they both retook the exam. </p>
<p>EDIT: The paper also says that a ~70 point difference on a section would happen 10% of the time between two test-takers of equal ability.</p>
<p>
</p>
<p>Sure, I would definitely agree. But previous claims have been more like SAT scores don’t matter past a certain point instead of students with poor SAT-M scores can still do really well in not heavily-quantitative fields. Also, I forgot the link in my last post that has more information on this sort of stuff [Information</a> Processing: Psychometric thresholds for physics and mathematics](<a href=“http://infoproc.blogspot.com/2010/05/psychometric-thresholds-for-physics-and.html]Information”>Information Processing: Psychometric thresholds for physics and mathematics).</p>
<p>speaking as a high school student, maybe there is no “intelligence” difference between some 2200 and 2400 students, just test prep, but that indicates maybe that the 2400 student was more willing to work hard and prepare, which is not such a bad thing.
and personally, I got a 2390 on the SAT (took it once) without any $500 prep courses or whatever. I did the SAT question of the day for a while, reviewed some vocab, did one or two practice tests and that’s it. I don’t think I would have gotten 2200 if retested, and you can’t just automatically assume that there is no difference between 2200 and 2400</p>
<p>Another thing to mention: We don’t all take prep classes…in fact, im fairly certain you can’t prep someone to a 2400…</p>
<p>^yes, there has to be the natural intelligence and potential there already. Test prep classes only take you so far. maybe 90% or even more of your own personal potential, but not everyone is capable of scoring a 2400 no matter what. everyone saying the SAT is just a formula and not an intelligence test, why do only ~400 students get a perfect score every year? that’s all the kids willing to put work into studying? I don’t think so…</p>
<p>
I think the claim has been that, past a certain point, SAT scores are not predictive of success at MIT.</p>
<p>Also, for a 700M we’re still talking about a score that’s ~95th percentile, so it’s not clear to me that this is a “poor” SAT math score.</p>
<p>College Board thinks that 60 points (per section, I presume) is a reliable differential between scores, in the sense that if one person scores 60 points higher than the other there is some level of confidence (95%? that wasn’t spelled out on the page I saw) that the higher scorer would continue to outscore the other (with the usual caveats about time lapse, illness, etc.)</p>
<p>Biological subjects demand their own types of intellectual talent, which is probably not well measured by anything on the SAT. I think the ability to organize and inter-relate complex data is needed in that field. Also, interest level will determine the amount of time that someone is willing to spend mastering the field. So I would not be surprised to find a person with a 700-ish SAT M score who is very successful in biology.</p>
<p>That is different from saying that there is no difference in performance of the students at MIT, though. To me it means that different students can be equally successful at MIT as long as the chosen majors are appropriate to each student’s interests and abilities.</p>
<p>For math and physical science, I think that the SAT M scores are meaningful differentiators, at least to some extent. As above, I am would say that 2 misses on any component of the SAT does not mean much; and there are USAMO competitors who didn’t score 800 on the SAT M, as I understand it. Beyond two mistakes, I would start to wonder, though. It seems to indicate a problem of attention, accuracy, or understanding.</p>
<p>If the scores are low9 to MIT standards) do they look at the comparison of the student with his country?? or is it oh…2050…lets throw it away and not care who he is ??</p>
<p>If yes … I don’t think I’ll apply! I hope what matt says is true…anyways!</p>
<p>To clarify the post above, by “it” in the second sentence of the third paragraph, I mean the fact that students with 700-ish SAT M scores can do well at MIT–but only as long as the major choice is the right fit, and not (in my opinion) in all majors.</p>
<p>In other threads, we (and I use the broad “we” here) have recently come to the conclusion that students who come to MIT without prior work in mathematics are unlikely to succeed as math majors. </p>
<p>I don’t think all human beings are the same, or that there is no distinction to be made among people who are adequately smart. But I do think it represents shrugging one’s shoulders at the status quo, and a failure of the department in question to educate its students, if the requirement for success in the MIT math department is a 750+ SAT score in math and substantial prior work in mathematics.</p>
<p>@mollie: I don’t think that’s any fault of the math department. It’s just, it’s hard to get mathematical maturity that fast! It’d be like me trying to major in Swahili, with no prior knowledge of the culture, grammar, or even common word roots (which, in math, would correspond to common proof methods or themes). It’s probably possible, but would take way too much grief. </p>
<p>From what I’ve seen on OCW, the lectures are very good, and the professors never assume the students know side stuff that hasn’t been covered yet. It’s just going to be hard, just like Spanish class is hell for me. Are language programs flawed if I can’t major and succeed in them with my background?</p>
<p>@everybody: Does anybody have any argument for CR/W score past 700 being meaningful? Flipping through my Blue Book, I see grammar questions on the difference between he/him and who/whom. These are simple rules that just aren’t covered in school, that take less than an hour to learn. That boosted my score by a question or two per test (20-40 points), and I don’t feel any smarter after learning that. Similarly, there are vocab lists out there to be memorized (and subsequently never used in conversation).</p>
<p>I also have heard claims of 200, 300, or even 400 point increases from tutoring centers, with real testimony backing it up. This ain’t an IQ test!</p>
<p>
</p>
<p>Although I definitely agree with this, it seems less clear that the admissions office does.</p>
<p>
</p>
<p>I agree that MIT’s math department does not do a very good job at catching up students with decent math backgrounds (I’m thinking motivated students with a 5 on Calc BC and maybe some basic linear algebra/multivariable calculus knowledge). At say Harvard, the very best math students will take 55 and the next best will take 25. I think Yale, Princeton, and Stanford have classes similar to 25 although the difficulty may vary slightly. In all of these cases, someone with a decent math background can get caught up reasonably quickly by working hard. HYPS also provide clear advising on these options.</p>
<p>At MIT the situation is very different. The very best students probably take some graduate classes. The next best (probably around 60 students per year) start off in advanced math classes with 18.701-2 being the most common. Obviously, a pretty serious background in proofs is required to do this. The bright student with a decent math background has a choice between 18.014, 18.022, and possibly 18.03, 18.06, and 18.700 if they passed the 18.02 ASE. Although these are all fine classes none of them will quickly prepare students for upper division course work. Furthermore, the advising on this is haphazard at best. AFAIK there is no good formal advising on this and students have to rely on upperclassmen which works will if they know upperclassmen math majors but doesn’t work so well if they don’t.</p>
<p>On the flip side I think MIT’s system works much better for people with substantial backgrounds in math. The way MIT does things you don’t have to repeat classes and can specialize much earlier. It’s less clear if there is some way to preserve the system for students with substantial math backgrounds while also making it easier for other students to catch up.</p>
<p>
</p>
<p>This isn’t always true. 18.152 this semester basically assumed familiarity with the basic results of measure theory which aren’t covered in 18.100. However, by the time he started using them everyone who hadn’t seen that stuff had already dropped the class.</p>
<p>
</p>
<p>I would imagine that higher CR/W scores correlate to some degree with a being a better reader and writer. I’m not sure how relevant that is to success at MIT though. I would not be surprised if there was a decent correlation between CR/W scores and HASS grades at MIT [excluding econ]. Whether the scores have any other predictive power at MIT is less clear.</p>
<p>molliebatmit, #91: I don’t see it as a failure of the Department of Mathematics, if they cannot bring a student who starts out 2 or 3 years behind up to the level of other strong students in a 4-year period. Presumably the students who started out “ahead” didn’t stop growing intellectually once they entered MIT.</p>
<p>I have a suggested solution to this, but it would require something like the Gates Foundation to provide the funding for it. I think that one should be honest upfront with a student who has high potential, but has been poorly served by pre-college education. I think the student should plan on a 6 or 7 year B.S. with the first 2 or 3 years funded by the Gates Foundation (to which unfortunately I have no connection) or some other foundation (ditto).</p>
<p>At some point, if a person wants a career as a mathematician, the student has to be caught up to the rest of the field by the completion of the Ph.D. or post-doc period. I think the catch-up extended B.S. would be the best solution to this.</p>
<p>As mentioned elsewhere, I am a faculty member at a large public research university. The Chair of our Department of Mathematics commented to a faculty committee that the Department tried to hold the number of domestic students at 50% of the entering Ph.D. class, although they had to adjust their standards downward for the Americans, in order to accomplish that. The persistence of this circumstance cannot be good.</p>
<p>Sorry this is absolutely ridiculous. 18.701/2 are junior/senior level classes, and the fact that the many people take the class as freshman does not change that. No one is behind if they don’t take 18.701/2 until their sophomore or junior years, and in fact at least one of the recent winners of the Jon A. Bucsela Prize in Mathematics, and award given to the best graduating math major each year, did exactly this.</p>
<p>I have no idea where this 6-7 year figure came from. People who take classes like 18.701 and 18.100 their freshman year are on track to take graduate level classes their junior, and even sophomore years. Shockingly enough, it is entirely appropriate for graduate students to be taking graduate classes. So someone who does not take 18.701 until they are a junior does not need to be an undergrad for 2 or 3 more years to take the graduate classes they would be taking if they were in graduate school anyways. And this is completely ignoring the fact that many undergraduate courses at MIT would be graduate courses elsewhere.</p>
<p>
</p>
<p>Why do you think this? In my experience, having taken 18.700 actually made 18.701 kind of easy, but it’s possible there were other factors in this or that I’m remembering incorrectly.</p>
<p>
Higher CR scores do correlate with full-scale IQ more tightly than math scores do. And surely what we’re really talking about here is intelligence, rather than aptitude at SAT math per se?</p>
<p>I’m willing to concede the point that, in many cases, 700M scorers at MIT would be well-served by being part of the ~85% of the undergraduate population not majoring in math or physics. But I am not so dismissive of the quantitative or technical potential of these students.</p>
<p>
Sure, but Americans do not make up 50% of the population of the world, so the same problem occurs whether or not Americans are good at math, or interested in pursuing math PhDs. </p>
<p>And, at least if math is anything like I know biology is, the top programs have the luxury of choosing among the best Americans. My PhD program is typically about 85% domestic students and 15% international, but this is clearly not true of biomedical sciences PhD programs at all research universities.</p>
<p>I’m not pretending to know the MIT course numbers–I have no idea what’s what, in terms of the course numbers. However, I am essentially certain that students do enter MIT 2 or 3 years “behind” a block of prospective math majors–possibly even further behind.</p>
<p>I think the only way to level the playing field in math is for the student who has high potential but who is not as advanced to have extra time to catch up. Then the students might have some chance of leaving for a Ph.D. program in math on an equal footing.</p>
<p>I would really like to see my university do the same thing, for students whose math placement indicates that they are 2 or 3 years behind the majority of the students majoring in math or the physical sciences here. One can’t expect the students to bear the tuition burden of what their public high schools did not permit them to accomplish. On the other hand, if they graduate in 4 years, having taken only what the majority of the majors take in the first year, their future employment/grad school prospects are somewhat limited (even with the math department bending the standards for Americans).</p>
<p>Your numbers are typical for Harvard across the fields that I am familiar with, molliebatmit, and probably for the majority of the top grad programs. It’s a bit further down where the numbers of foreign graduate students start to grow.</p>
<p>I don’t think our math department is looking for a string of publications by a prospective grad student–just the ability to deal with the entering grad courses in mathematics (which most math majors at MIT will have taken as juniors or seniors), with some possibility of success.</p>
<p>The public universities are tax-payer supported, and do have as part of their mission to serve the people of the state in which they are located. This is unlike MIT. In the past, I have been responsible for pushing my department to admit more foreign students–because I am sympathetic to someone who is very talented, but grew up elsewhere. Still, I think the set of strong universities is shifting over time. There are now some excellent research groups with strong financial support in research universities in China. In fact, some of the top Chinese grad students are now choosing to return to faculty positions in China. </p>
<p>So, although one wouldn’t expect Americans to account for more than their population share of the top young mathematicians (at the entering Ph.D. student level), one would no longer expect all of the world’s top math grad students to come here, plus a few universities in Europe, plus a few other universities.</p>