<p>owned… </p>
<p>I know 5 people last year who turned down Ivies (2 turned down Harvard and Yale) for Amherst and Williams…</p>
<p>Re: #59
The data I presented earlier was actual hard data. The actual number of PhDs earned by, for example, Swarthmore graduates over a ten year period (from the NSF PhD completions database) divided by the actual number of Swarthmore graduates from the ten year period offset five years earlier.</p>
<p>The fields are from the PhDs – the NSF data base categorizes all PhDs earned in the United States by several dozen academic fields. </p>
<p>I’ve got it for all 1350 or colleges that graduated at least one future PhD during the ten year period.</p>
<p>BTW, since I can tell you might be just starting out in exploring these statistics, here’s a hint. Don’t look a single year. Look at longer intervals (like a ten year period). Otherwise, you’ll get caught out by significant year to year fluctuations.</p>
<p>For example, you cherry picked a year in which Swarthmore had its highest ever number of math majors (25) from a range of 4 to 25 since 1981. And, you just happened to pick a year in which Swarthmore had the smallest number of math PhDs (from a range of 1 to 5 per year) since 1981. So, you would draw very inaccurate conclusions from a single year sample; ignoring the fact that 2006 graduates won’t begin earning PhDs until 2011 or so. If you pick a longer time frame, the year to year fluctuations tend to even out.</p>
<p>Also, you picked a really tough field to try to match up college majors and PhDs. Computer geeks study in three different departments: math, computer science, and engineering – and get PhDs in several different fields, including engineering. Plus, many math majors are getting advanced degrees in Economics and statistics based social sciences.</p>
<p>Oh, and by the way, only 5,111 of the 270,807 PhDs granted in the United States between 1994 and 2003 were Math PhDs. 4,150 more were granted in Computer Science. To give you an idea of how small the numbers are, CalTech grads only went on to get 56 Math PhDs or an average of 5.6 per year.</p>
<p>
</p>
<p>Such chortlings, however revealing, are beside the point, as I did not purport to use “PhD completion, in and of itself” as a ranking system, and have no interest in producing such a ranking. You posted a ranked list and that drew some comment. That’s it. </p>
<p>
</p>
<p>As I said, that number is not uninformative. It’s not conclusive, either.
A more relevant figure, per your posted proxy for “disadvantage” or lack thereof, is not the raw number or proportion of PhD’s, but the number of PhD’s in a given field (or related fields) relative to the number of undergrad majors. At Harvard and MIT, for instance, math and physics majors will get PhD’s in cognate fields at a much higher rate than you reported. Biology majors will get into top med schools at a certain rate, and so on. The fact that English, government and sociology majors will get PhD’s at a very low rate drives the numbers down at Harvard, doesn’t influence the numbers at MIT, and drives up the numbers at Swarthmore (i.e. the humanities majors are more likely to get PhD’s coming from LAC’s than from HYP).</p>
<p>
</p>
<p>If by “cherry picking” you mean that I took a look at the first year of recent data that came up on the screen and left it at that, yes. I believe I referred to that as a “quick look”, which it was. I did not go back to 2004 much less 1981 and pick the nicest data point, no. If you have some more extensive analysis of the Swarthmore data since 1981 showing that their per-major math/physics/etc PhD rate hits the Harvard/MIT levels, or similar chances of admission to the best graduate programs, by all means, post it.</p>
<p>
</p>
<p>Actually, Harvard is #5 in per graduate Econ PhDs and MIT is #7.</p>
<p>Of course, certain schools tend to produce more PhDs per graduate in some fields, fewer than others. That’s true of Harvard. It’s true of MIT. It’s true of Swarthmore. Actually, one of the most useful things from the PhD data is to actually look at a particular school and see where it ranks in PhD production across a number of fields. It’s a little dicey because some fields have such small numbers nationally. However, it can give you a good sense of where a school’s specialties lie. Some are balanced. Some are obviously one-sided (like the tech schools). Sometimes you see some surprises.</p>
<p>However, it is very difficult to be at the very top of per graduate PhD production without a lot of science, math, and engineering PhDs. That jumps out from the data immediately when you see two tech schools (CalTech and Harvey Mudd) producing the largest number of PhDs per graduate of any schools in the country.</p>
<p>For the most part, the LACs (and universities) at the top of the PhD production charts all produce significant numbers of math, science, and engineering PhDs. For example, here are the percentage of total PhDs from each school in the math, science, and engineering fields from six of the top PhD producers in the country:</p>
<p>53% Princeton
51% Reed
45% Harvard
44% Carleton
39% Swarthmore
36% Yale</p>
<p>These are all very high numbers as I doubt that a third of students at any these schools major in those fields.</p>
<p>BTW, your grouping of Harvard and MIT makes no sense whatsoever. They aren’t even in the same universe in term of PhD production. MIT, of course, is a tech school. Harvard is near the top of per capita PhD production because it produces pretty solid numbers across the board – or as you would seem to put it, because it produces a lot of cheap PhDs in social sciences and humanities compared to a real school like MIT! </p>
<p>Reed and Swarthmore are consistently in the top 5 per capita PhD producers because they are strong across all fields, too. Reed tilts a little more towards math and science. Swarthmore is dominant in social science PhDs while still being a respectable #6 in math/science/engineering. But, you would expect that. It’s always been a very “politically active” school and its always been a premed factory.</p>
<p>Harvard is also pretty balanced across the board. It’s really hard to find a field where Harvard doesn’t produce a lot of PhDs. Chemistry is one of the lowest for both Swarthmore and Harvard – only 26th in the country for Swat, 35th in the country for Harvard. I don’t think that’s a reflection on the chemistry departments as much as the recent trend towards bio-chem related PhDs.</p>
<p>
</p>
<p>What you have to do is get the actual number of majors for the ten year period from 1994 - 2003. Or, at the very least, look at a five or ten year period to come up an average number of majors per year. Otherwise, you will get caught by year to year variability. I’ve done it for a few departments. It’s a royal pain in the ass.</p>
<p>Plus, it doesn’t tell you anything that you don’t get from the field-specific PhD numbers. For example, if a college is #1 in the country in per graduate Economics majors, you already know that Economics attracts students at that school and that Economics majors go on to get PhDs at higher than average rates. I’m not sure what more useful data you could learn beyond that. I mean…you’d have to be a total idiot as a high school senior to sit there and say, “hmmm, if I major in Widgets at Framus University, I have a 32.187564% chance of getting a Widgets PhD eleven years from now!”</p>
<p>
</p>
<p>Yes, which is why “economics” was edited to “government” shortly after the original posting. Never mind: the point was the same in both cases, which is that there is a lopsided distribution of PhD rate among fields within each of these schools, and this influences the numbers in noticeable ways. It’s getting late, so I will try and answer the rest later.</p>
<p>From the Harvard Crimson a few years back</p>
<p>Opinion
No Intellectuals Need Apply
Published On 12/9/1999 12:00:00 AM
By ADAM I. ARENSON</p>
<p>It’s time we, the student body, write a collective letter to our friends at Williams or Swarthmore, Wesleyan or Amherst. It doesn’t have to be long, just enough to admit the truth: Liberal-arts colleges, you win. You possess the nation’s most innovative minds, the most intellectual student body. You are the stomping grounds for the great thinkers of the next millennium. We, Harvard, will stop trying to lord over you, stop saying that we are better or smarter, because it just isn’t true. You can out-think us any day of the week.</p>
<p>There. Once that is said, we can all go on with being more honestly what we are–not intellectuals. </p>
<p>Don’t get me wrong: Harvard students can be, and often are, great. They just aren’t intellectuals. Sure, that first night the entire dorm gathered in somebody’s common room and shared a bunch of ideas about what college was supposed to be, about where they were from and what they thought was really important, but the minute placement tests came along, let alone classes and extracurriculars, everyone was holed up in their room, hard at work or hard at play, and the great intellectual college conversations you had dreamed of became the thing of nostalgia and viewbooks. </p>
<p><a href=“http://www.thecrimson.com/article.aspx?ref=98869[/url]”>http://www.thecrimson.com/article.aspx?ref=98869</a></p>
<p>I don’t buy Arenson’s argument entirely. The statistics suggest that Harvard is actually one of the more “intellectual” universities. It is consistently one of the top per capita PhD producers, which seems to correlate with “intellectual” campus cultures. I don’t think Harvard is a “will I have to know this for the test?” kind of place.</p>
<p>I do think Harvard is structured to foster more individual pursuits and less community interaction or group learning. Some of that is the physical layout of the university. Some of that is the highly decentralized management/faculty structure. Some of it is just a function of size.</p>
<p>see link for information on grad school acceptance the Ivies and the LACs</p>
<p><a href=“http://www.wsjclassroomedition.com/college/feederschools.htm[/url]”>http://www.wsjclassroomedition.com/college/feederschools.htm</a></p>
<p>TommyBill - that link is a great resource - however, it focuses on Professional schools (law, business, medicine) rather than graduate school (science, english, political science)</p>
<p>law, business and med schools ARE grad schools.</p>
<p>not only that, but they are the most commonly attended grad schools.</p>
<p>because they are the most lucrative.</p>
<p>Harvard for sure isn’t particularly intellectual. This can be seen from the fact that 14 out of 20 USA 1st team All - Academic students chose Harvard for next year as opposed to 20 out of 20. We know that these students are only going to Harvard for prestige not academics or perhaps USA Today is just wowed by the prestige factor when they chose these students?</p>
<p>The other six are going to Boston College, Yale, Princeton, Brown, Stanford, and Caltech.</p>
<br>
<br>
<p>Hmmm…not a single LAC among them.</p>
<p>
</p>
<p>But that is also an argument for not making much of the PhD production numbers in the first place. Any highly selective, academically oriented school will attract a lot of students who go on to PhD’s. If some universities had an astonishing disproportion not explained by obvious factors, such as art schools not producing engineering PhD’s, we would know about it. Differences that are seen (e.g., the Wall Street Journal figures) are mostly due to the profile of incoming students, not the school. Grad school admissions committees at least attempt to reckon the objective level of qualification and if done properly this should reduce much of the supposed “disadvantage” associated with any school; if the AdComs did the job perfectly it would be no difference at all where a student goes as long as the level of preparation were the same.</p>
<p>In other words, there should be relatively little “treatment effect” of the school on one’s chances of anything. I imagine that what effect there is moderately favors LAC because of the (probably oversold) higher degree of nurturance per student, and a bigger-fish-in-smaller-pond effect that the grad school admissions process doesn’t fully neutralize. The brand-name of top 3-5 research schools is probably also given higher weight than due, but overall I expect some minor advantage for LAC at any given capability range of students.</p>
<p>
</p>
<p>Actually, the exact opposite of what you say is revealed by the PhD production rates and that’s what makes the data so interesting.</p>
<p>There are plentiful examples of colleges and universities with super high admissions selectivity that don’t produce particularly large percentages of PhDs. For example, Duke and Dartmouth and UPenn have extraordinarily high median SAT scores, but do not produce PhDs. at high rates: Duke only 41st per capita; Dartmouth at 54th, Penn at 65th. Meanwhile, there are some schools with much lower admissions stats that easily outperform that (Kalamazoo, Earlham, Beloit. etc.).</p>
<p>Or, compare the two elite Chicago area universities, both with very high admissions stats. UChicago at #9 in PhD production versus Northwestern at #68. These differences cannot be explained by SAT scores. Instead, the numbers of indicative in fundamental differences in the students each school attracts and the learning environment.</p>
<p>Note that nothing was said about SAT: “differences… mostly due to the profile of incoming students”. That sounds about the same as your “fundamental differences in the students each school attracts”. It is unclear how strong the effect of the learning environment is on a student’s grad school plans and options.</p>
<p>A LAC with a self-selected pool of scholastically inclined students, and without a lot of dud admits to drag down the numbers (athletes, idiot children of the wealthy, etc) will bolster its PhD numbers. Duke adjusted for major and type of student or even with just the athletes subtracted, will not be 41st, and similar if less extreme rises would occur for the Ivy league schools. Duke ranked very high in the Wall Street Journal data, which is closer to a meaningful measure than schoolwide PhD rate.</p>
<p>
There are clearly two factors, the student and the undergrad school. However, for an undergrad school to rise to the top of per-capita future PhD production in a field, it is likely (certain?) that the school is providing top-quality preparation in the field. This is necessary, but not sufficient (the student is the other necessary part) to be admitted to a PhD program and complete it.</p>
<p>Granted, it is possible that a school at the top of future PhD production is sending its students to low-quality PhD programs, but is there evidence of this?</p>
<p>
</p>
<p>Actually, the Wall Street Journal methodology was so flawed that it is basically useless. They looked at too few grad schools, with too much northeast bias, and made the critical mistake of looking at a single year’s data.</p>
<p>
</p>
<p>So, what you are saying is that, if Duke had a different student body, it would produce a higher percentage of PhDs? I totally agree. That’s the whole point. If Duke had a different student body, it wouldn’t be Duke. The difference between UChicago and Northwestern is the student body. One produces a lot of PhDs. The other doesn’t. </p>
<p>Understanding those differences and how they relate to an individual’s preferences is what selecting a college is all about.</p>
<p>
</p>
<p>Someone who really wanted to know the answer would not have to do all this.</p>
<p>They could follow ID’s instructions to create a list of top PhD producing colleges, say the top 30 or so. Then use the NSF database to find how many grads from each school received their PhD from a “top” university. Here the careful way would be to use the NRC rankings for the mid 90’s. These are perhaps somewhat dated now, but they were current when people who received their doctorate degrees over the last few years were applying to grad school. Pick some representative fields across science, social science and humanities, and run the numbers.</p>
<p>ID is certainly right that trying to do this for every field and every college would be a lot of work. However, the question was whether the few top LAC’s send students to grad schools that are comparable to those attended by graduates of a few top universities. This is a much more manageable task- 30 colleges instead of 4000. Cutting the number of fields down to a handful then further limits the scale of the project.</p>
<p>Results would be fascinating.</p>