<p>It would appear they are better at math on average. This assumes they have the same proportion of international students, who might score a bit lower due to language differences. I don’t know enough about computer science grad school admission, MIT or elsewhere, to tell you what they might be looking for. </p>
<p>Do you really think admission to Stanford undergrad and graduate school are even remotely similar? Yes, grad schools do overlook stats sometimes to get better grad students. For instance, if someone has a better publication record in a lab-based science such as synthetic chemistry, they might ignore some less than stellar grades in physical chemistry. </p>
<p>For undergrad, they might ignore superior stats and academic achievement in one candidate to pick a premed who had lots of community service. It’s apples and oranges.</p>
<p>To be honest, for undergraduate admission, I don’t understand why Stanford does not want to admit students with higher SAT scores. I believe Stanford can do it if it wants to do it. </p>
<p>But in graduate school admission, I am also puzzled why MIT doesn’t want to admit the students with higher GRE scores. Years ago, a friend of mine applied to MIT mechanical engineering graduate program. He had a GRE quant score 800, and a GRE subject score in math 990 out of 990, ranked #1 in his department. Plus, he had won some research rewards. But he was rejected by MIT and ended up in a public unversity. </p>
<p>I think the admission result is very hard to predict.</p>
<p>EECS at MIT does not consider GRE for admissions.</p>
<p>Regardless, neither of you believe the conclusions you’re reaching. Stanford CS grad students are not smarter than MIT grad students because of GRE scores, Caltech undergrads are not smarter than MIT undergrads because of their SAT scores, and MIT undergrads are not smarter than Stanford undergrads because of their SAT scores.</p>
<p>
</p>
<p>I’d like to see income breakdowns from Caltech or MIT if you know of any available. I saw some relatively detailed stats on Penn and Northwestern, which showed they are not working-class schools. While HYPS haven’t released detailed statistics, the data that is known about them is in line with Penn and Northwestern with regard to income distribution (although the Pell Grant % varies, the rest does not). I can’t imagine that Caltech and MIT would be any different (MIT and Stanford usually have about the same proportion of low-income students and first-gen students).</p>
<p>And if there were socioeconomic diversity within a portion of SAT score ranges, it is least likely to be at the high end of the range.</p>
<p>I think it starts mattering a lot less at the particularly high ranges, because there are abilities the SAT does not screen for which are very essential to future success, even on a purely academic level. At the level of 2300 vs 2350, we’re often just talking of how the test went that day or who does best on the SAT. </p>
<p>
</p>
<p>At a certain point, graduate schools just make a call. There are just a million factors contributing to what produces good thesis work, because you’re effectively working for years and years to produce something deep. Being very sharp with figuring tricky details out (effectively what a test asks you to do) helps, but being very skilled and having a knack for your specialty are equally important. Sometimes, a student has both, and sometimes a student has much more of one than the other.</p>
<p>Of course, the student you mention had a research award too. All I can say is probably the student will get very far from that public university if as strong as you claim.</p>
<p>
</p>
<p>Me too, aside from a few exceptions where the student is either a prodigy or in a field where very strong work plus connections to the right people can get you in with certainty.</p>
For the GRE general test in math, virtually all highly competitive technical students score 800. </p>
<p>About 10% of students who take the GRE score an 800 in math – the test is extremely easy. The math on the GRE is actually less advanced than the math on the SAT. It does not surprise me whatsoever that top technical graduate programs might not find the GRE general test particularly useful for predicting student success in their programs.</p>
<p>I think so too. To get a full score in math, it’s ok to make 2 or 3 mistakes usually. If you make 7 to 8 mistakes, you still get a score like 750. I think any students with a decent middle school math skill and having practiced seriously can do very well in this test.</p>
<p>But if you look at the people in MIT’s EECS PhD program, it is puzzling though. EECS is highly math intensive. Since MIT’s average score is a little over 770, I guess 25% of the MIT EECS students are below 750. If that is the case, I would say 25% of the people admitted to MIT EECS PhD program have a lousy math skill. We are talking about MIT here. What is wrong? Are all of them inventors or excellent researchers?</p>
<p>^ I don’t think you can conclude that scoring below a 750 means you have a “lousy math skill.” It could simply mean they cared less about the test since they knew the top CS grad schools don’t care much for it (the general rule of thumb is that the GRE/subject tests are used to gauge your abilities when the admissions committee is unfamiliar with the college you’re coming from). I do think that since MIT doesn’t care for the GRE (at least in EECS, not sure about its other departments), it will judge its candidates purely on their ability to conduct insightful/creative research. TBH, I wish Stanford and all the other top CS schools would follow the same philosophy.</p>
<p>I’ll add that the average verbal scores for many (if not most) top CS programs is a little embarrassing. It probably has more to do with the higher enrollment of international students who don’t have the English skills, but still an average of <600 is poor. Reading drafts of even native speakers’ papers is sometimes saddening.</p>
Presumably the average GRE score reported by the department is only for LGO (joint SM/MBA) applicants, as other applicants would not have even submitted GRE scores.</p>
<p>^ good point, I forgot that EECS does require the GRE for some. Still, even if the non-LGO students had a 770 Q average, I wouldn’t conclude that their math skills are inferior.</p>
<p>If you can still get a few wrong and get 800, then I would conclude that something is wrong with their math skills if they aren’t getting 800. For the average to be lower than 800 would be very weird, particularly for electrical engineering.</p>
<p>What Mollie said probably explains it though. Grad school and undegrad admissions are both holistic, but in grad school the holistic evaluation doesn’t extend beyond your promise/achievement in your field. That’s true at Stanford or MIT.<br>
In engineering, though, it’s worth noting that it’s easier to get into the PhD program in most engineering departments at MIT if you were an undergrad. Most people say that it is the opposite in the science departments–although Mollie says that there were plenty of bio undergrads who got into MIT. </p>
<p>In general, when you start to value non-academic factors in admission, the academic talent in your entering class becomes more heterogeneous–relatively weaker on average.</p>
<p>I would (I don’t mean to imply everyone should, but I think this is a fair suggestion) ignore all this altogether and see how they did in challenging math classes (even if not very advanced, depending on how much math they need), hopefully at a good school – if they have a consistently good record, they’ve proved their skills are up to snuff. If not, no 800 on the GRE will save them.</p>
<p>If they come from a less well known school, I’d say much more detail than either GRE scores or class results would be necessary to ascertain their skills are OK.</p>
Oh, I certainly wouldn’t, either. I mean, I feel like it’s stretching it to call the quantitative section “math”. By the time I took the GRE, I had to study for the math section because I was used to doing statistics and calculus, not geometry and algebra, and I would imagine that it’s even more disorienting for an engineer.</p>
<p>I got an 800 on the quantitative section, and my husband, who’s a very talented aerospace engineer and is quite comfortable with differential equations and lots of other math I don’t understand, got a 760. I can assure you categorically that he is more talented at math than I am.</p>
<p>
But there are weird tricks with the GRE, since it’s adaptive – you can get up to two(?) questions wrong on the test and still get an 800, but only if they’re considered difficult questions. You start out with questions of medium difficulty, and your correct/incorrect answers to the first few questions set the difficulty trajectory of the test as a whole. So if you make a stupid mistake early on, it will affect your score to a much greater degree than if you make a stupid mistake later in the test.</p>
<p>
Only starting my year, though. Prior to that, there was an official department policy against inbreeding.</p>
<p>I’m not sure how a comp. sci PhD program would admit people, because I’m not sure that grades/scores are that indicative of ability. </p>
<p>While you might not assume an individual with a score lower than 800 has a problem in math, significant differences in average scores (~40-50 points in a section) do tend to reflect some difference in academic quality or at least selection criteria.</p>
<p>Anyway, Stanford (even in their STEM applicants) tends to select more for non-academic qualities than MIT and certainly more than Caltech. This has to lower the academic talent level below what they could have had (and this has been my observation based on people they have admitted). And it’s kind of absurd to assume that Caltech doesn’t have smarter people on average than either MIT or Stanford. Their class size is ~200 per year, whereas MIT’s is 1000 (and I assume Stanford has significantly more STEM people than 200 per year.) So it’s easy for Caltech to take only the best. At Stanford, I think this affects the quality more in STEM disciplines that premeds tend to major in–bio, chem, and some of the engineering disciplines Stanford is not the best in (chem E, some other engineering disciplines–> probably not comp sci. or EE much at all).</p>
<p>^ I’m not going to judge people on SAT or GRE scores. Intelligence has far too many facets to it that IQ tests, SAT, etc. are all hugely inadequate. Perhaps it’s because I work in AI and spend most of my time reflecting on the complexities of intelligence, but I think that brilliance can be displayed in a huge variety of ways, so terms like “smarter” are just too difficult to pinpoint. Once you get to a certain level - such as Caltech and MIT - it becomes a wash, even accounting for size.</p>
<p>Similarly, IIT students try to say they’re “smarter” than MIT students because of the exams they have to take, but it’s mostly unfounded. In particular, they ignore the importance of creativity in intelligence (creativity and reasoning are very highly related), and I think that’s something MIT students have a lot of. In fact, I think there are only a tiny handful of schools that can match MIT students in terms of creativity within STEM disciplines.</p>
<p>I’ve thought about this in detail as well, and am familiar with many students who have gone to all three universities.</p>
<p>Caltech is holistic as well, and is unusual in that professors and undergrads are heavily involved in the admissions process. They aren’t impressed by high test scores either; it’s more that math/science subscores that are below ~750 are considered red flags (or 800 for the Math SATII, which has a generous curve.) Automaticity and being meticulous and precise are important in science; you can have all the creativity in the world, but if you are mistake-prone it’s going to limit you. If you are getting 700 on the SAT math, you are either (a) mistake- prone or (b) don’t understand the concepts. I will add that I think Caltech thinks it is a useful asset to be able to think like a physicist, even in fields which don’t actually use it. I agree with this. </p>
<p>If you aren’t advanced enough to be able to do well on these very remedial tests, then you probably will have a lot of trouble at Caltech. Even for very smart people, I wouldn’t recommend to use Apostol for your first time you see calc or Purcell for E & M.</p>
<p>
</p>
<p>I’ve read Gardner too, but the people at the top tend to display a very high level of most of the intelligences. And if they are weak in remedial math for instance as evidenced by SAT score, it’s unlikely they are really going to be at the top of their scientific field (again, assuming we are taking people with decent preparation.) </p>
<p>Anyway, I don’t see Stanford trying to find different types of intelligence, unless you are talking about emotional/social intelligences.</p>
<p>What is your interpretation when a STEM candidate is admitted with B’s in calculus, something that I’ve seen Stanford do? In my view, the only other assets these people had were non-academic EC’s to make up for that. That demonstrates organizational skill, ambition, and ability to get along with others–not intelligence.</p>