Top producers of PhDs in physical sciences, comp sci, math

College Search & Selection
1

<p>This is a ranking of the top 20 baccalaureate schools of PhDs in physical sciences, computer science, and math. Physical sciences include chemistry, physics, astronomy/astrophysics, earth sciences, atmospheric sciences, and marine sciences. PhD number also includes math and comp sci.</p>

<p>This information is from the NSF Survey of Earned Doctorates 1995-2006 and from IPEDS. I divided the number of PhDs from a particular baccalaureate school by the number of bachelors degrees awarded by that school in the same fields (physical sciences, math, comp sci). I think this is a more valid way to "normalize" PhD production.</p>

<p>The number of bachelors degrees is the total over three years 2002-2004 in physical sciences, computer science, and math. These were the earliest available from the IPEDS website. I thought a three-year total would be more reliable than a one-year snapshot.</p>

<p>Other methods divide by total undergraduate enrollment. With this method, schools are penalized if they have large numbers of undergrads in other fields.</p>

<p>I limited the schools to roughly the top 100 US News universities and top 100 LACs plus the SUNY schools, some additional tech schools, and a few schools that are not top 100 in US News but which produce large numbers of PhDs.</p>

<p>school, PhDs produced in physical comp sci math 1995-2006, bachelors degrees granted 2002-2004 in same fields, ratio</p>

<pre><code> 1 PRINCETON UNIVERSITY 400 198 2.020
2 YALE UNIVERSITY 301 197 1.528
3 CALIFORNIA INSTITUTE OF TECHNOLOGY 434 288 1.507
4 REED COLLEGE 136 102 1.333
5 OBERLIN COLLEGE 157 136 1.154
6 RICE UNIVERSITY 275 240 1.146
7 CARLETON COLLEGE 240 240 1.000
8 FRANKLIN AND MARSHALL COLLEGE 104 105 0.990
9 HARVARD UNIVERSITY 627 649 0.966
10 HARVEY MUDD COLLEGE 238 249 0.956
11 UNIVERSITY OF CHICAGO 341 373 0.914
12 SWARTHMORE COLLEGE 104 117 0.889
13 OCCIDENTAL COLLEGE 71 90 0.789
14 SUNY COLL OF ENVIRON SCI & FORESTRY 18 23 0.783
15 UNIVERSITY OF PENNSYLVANIA 195 256 0.762
16 CORNELL UNIVERSITY 555 745 0.745
17 LAWRENCE UNIVERSITY 58 78 0.744
18 BOWDOIN COLLEGE 80 108 0.741
19 MASSACHUSETTS INST. OF TECHNOLOGY 821 1115 0.736
20 POMONA COLLEGE 84 116 0.724
</code></pre>

2

<p>The next 20:</p>

<pre><code> 21 BROWN UNIVERSITY 257 360 0.714
22 KENYON COLLEGE 40 57 0.702
23 HAVERFORD COLLEGE 84 123 0.683
24 AMHERST COLLEGE 90 132 0.682
25 FURMAN UNIVERSITY 100 151 0.662
26 DAVIDSON COLLEGE 44 67 0.657
27 WABASH COLLEGE 47 73 0.644
28 COLORADO COLLEGE 69 108 0.639
29 WILLIAMS COLLEGE 141 221 0.638
30 BELOIT COLLEGE 41 67 0.612
31 COLLEGE OF WOOSTER 95 156 0.609
32 PEPPERDINE UNIVERSITY 16 27 0.593
33 BATES COLLEGE 69 117 0.590
34 UNIVERSITY OF CALIFORNIA-BERKELEY 864 1467 0.589
35 RHODES COLLEGE 30 51 0.588
36 COLLEGE OF WILLIAM AND MARY 249 425 0.586
37 DREW UNIVERSITY 44 76 0.579
38 BRANDEIS UNIVERSITY 79 141 0.560
39 KNOX COLLEGE 44 79 0.557
40 WELLESLEY COLLEGE 65 118 0.551
</code></pre>

3

<p>The correlation between SAT scores and the PhD rate was moderately high, about +.6, so SATs account for about a third of the differences in PhD production. What accounts for the rest of the variability in PhD production?</p>

4

<p>I think your data size is WAY tooooo small. </p>

<p>Your errors are too large to rank them explicitly like you did.</p>

<p>For example:
14 SUNY COLL OF ENVIRON SCI & FORESTRY 18 23 0.783</p>

<p>would have data points, including errors, of: 18 +- 4.2 and 23 +- 4.8</p>

<p>which has quite the range of values for ratios.</p>

5

<p>next 20:</p>

<pre><code> 41 MIDDLEBURY COLLEGE 61 111 0.550
42 EARLHAM COLLEGE 30 55 0.545
43 UNIVERSITY OF NOTRE DAME 153 282 0.543
44 GRINNELL COLLEGE 93 172 0.541
45 NORTHWESTERN UNIVERSITY 181 336 0.539
46 WESLEYAN UNIVERSITY 91 171 0.532
47 SAINT OLAF COLLEGE 126 244 0.516
48 DUKE UNIVERSITY 185 365 0.507
49 BRYN MAWR COLLEGE 71 142 0.500
50 HOPE COLLEGE 82 168 0.488
51 COLGATE UNIVERSITY 80 169 0.473
52 DARTMOUTH COLLEGE 166 354 0.469
53 OHIO WESLEYAN UNIVERSITY 27 58 0.466
54 WHITMAN COLLEGE 66 143 0.462
55 CASE WESTERN RESERVE UNIVERSITY 145 321 0.452
56 VASSAR COLLEGE 57 128 0.445
57 COLLEGE OF SAINT BENEDICT 18 41 0.439
58 COLLEGE OF THE HOLY CROSS 67 153 0.438
59 ALLEGHENY COLLEGE 67 156 0.429
60 STANFORD UNIVERSITY 319 746 0.428
</code></pre>

6

<p>next 20</p>

<pre><code> 61 AUGUSTANA COLLEGE 40 94 0.426
62 CONNECTICUT COLLEGE 19 45 0.422
63 KALAMAZOO COLLEGE 55 131 0.420
64 UNIVERSITY OF WISCONSIN-MADISON 365 870 0.420
65 BUCKNELL UNIVERSITY 76 183 0.415
66 HENDRIX COLLEGE 28 68 0.412
67 WHEATON COLLEGE - ILLINOIS 49 119 0.412
68 UNIVERSITY OF THE SOUTH 24 59 0.407
69 GUSTAVUS ADOLPHUS COLLEGE 84 207 0.406
70 UNIVERSITY OF MICHIGAN-ANN ARBOR 411 1021 0.403
71 UNIVERSITY OF DELAWARE 144 359 0.401
72 UNIVERSITY OF CALIFORNIA-DAVIS 304 761 0.399
73 WAKE FOREST UNIVERSITY 65 166 0.392
74 MACALESTER COLLEGE 70 180 0.389
75 UNIVERSITY OF ROCHESTER 177 458 0.386
76 LAFAYETTE COLLEGE 44 114 0.386
77 UNIVERSITY OF VIRGINIA-MAIN CAMPUS 250 651 0.384
78 COLUMBIA UNIVERSITY IN NEW YORK 188 492 0.382
79 UNIVERSITY OF PUGET SOUND 60 161 0.373
80 AUBURN UNIVERSITY-MAIN CAMPUS 66 182 0.363
</code></pre>

7

<p>The ranking is flawed because it does not include engineering and because schools report undergrad majors differently.</p>

<p>Princeton for instance, reports zero undergrads in the IPEDS survey (or its CDS) classification for computer science while it reports all of them in engineering instead. A quick look at the Princeton CS department shows more undergrads in CS than in math, physics and chemistry combined. Including the actual CS students would more than double the denominator and cut by more than half the productivity shown for Princeton from 2 to 0.88.</p>

<p>Additionally, at many schools with large engineering departments, math, physics and chemistry majors double major in engineering and pursue PhDs in engineering rather than the natural sciences. The shown productivity is artifically reduced by eliminating these students from the numerator. </p>

<p>The only way to compensate for the different reporting methods would be to include engineering in the reporting and use the traditional STEM productivity (science, technology, engineering and math).</p>

8

<p>cellardwellar-
You are right about Princeton. The data for Princeton should read:
6 PRINCETON UNIVERSITY 400 301 1.329</p>

<p>Princeton had 103 comp sci/eng grads in the time frame 2002-2004 (198+103=301). They are demoted to 6th.</p>

<p>They don't offer "computer engineering" but their comp sci BSE majors are required to meet certain engineering course requirements.</p>

<p>I avoided the inclusion of engineering so all universities and LACs could be compared, including those without engineering.</p>

9

<p>next 20</p>

<pre><code> 81 CLARKSON UNIVERSITY 58 164 0.354
82 MOUNT HOLYOKE COLLEGE 42 119 0.353
83 JOHNS HOPKINS UNIVERSITY 133 378 0.352
84 UNION COLLEGE 41 117 0.350
85 RANDOLPH-MACON COLLEGE 14 40 0.350
86 CENTRE COLLEGE OF KENTUCKY 24 70 0.343
87 SMITH COLLEGE 61 178 0.343
88 SUNY COLLEGE AT CORTLAND 14 41 0.341
89 UNIVERSITY OF RICHMOND 54 161 0.335
90 ALBION COLLEGE 40 120 0.333
91 WOFFORD COLLEGE 22 66 0.333
92 UNIVERSITY OF COLORADO AT BOULDER 252 768 0.328
93 WILLAMETTE UNIVERSITY 36 110 0.327
94 SUNY COLLEGE AT GENESEO 73 224 0.326
95 UNIV. OF ILLINOIS AT URBANA-CHAMPAIGN 419 1289 0.325
96 URSINUS COLLEGE 33 102 0.324
97 BRIGHAM YOUNG UNIVERSITY 299 933 0.320
98 BAYLOR UNIVERSITY 70 220 0.318
99 DENISON UNIVERSITY 34 108 0.315
100 MARQUETTE UNIVERSITY 39 125 0.312
</code></pre>

10

<p>I can't think of a way in which these rankings are useful. Many of the great Computer Science schools like UIUC, Umich, stanford, and UWisc are going to be at the bottom because they are more prestigious then say, Reed. I mean, a PhD program can be anywhere on this list and it won't make it good or bad.-</p>

11

<p>Titan-
Not sure I understand. This is a ranking of undergrad schools according to how many of their undergrads go on to get PhDs in sciences, math, comp sci.</p>

12

<p>So, for example, you're dividing the number of PhDs in math granted in a year by the number of BSes granted in the same year. It seems to me like that would favor schools having few professors with very large research groups. Few professors means fewer classes taught, so less undergrad enrollment, and to get that ratio really high you just need to have research groups of 10-20+ people per professor.</p>

13

<p>RacinReaver-
This is the number of PhDs granted 1995-2006 to students who received their bachelors from a particular undergrad school divided by the number of bachelors granted by that school 2002-2004.</p>

14

<p>next 20</p>

<pre><code> 101 GETTYSBURG COLLEGE 25 81 0.309
102 TRINITY COLLEGE 29 95 0.305
103 COLBY COLLEGE 50 164 0.305
104 UNIVERSITY OF CONNECTICUT 88 292 0.301
105 SUNY AT BUFFALO 189 629 0.300
106 GEORGETOWN UNIVERSITY 43 144 0.299
107 TUFTS UNIVERSITY 66 222 0.297
108 SUNY COLLEGE AT POTSDAM 47 159 0.296
109 BIRMINGHAM SOUTHERN COLLEGE 18 61 0.295
110 UNIVERSITY OF CALIFORNIA-SAN DIEGO 354 1202 0.295
111 WASHINGTON UNIVERSITY IN ST. LOUIS 147 502 0.293
112 SAINT JOHNS UNIVERSITY 45 155 0.290
113 EMORY UNIVERSITY 65 224 0.290
114 UNIV OF NORTH CAROLINA AT CHAPEL HILL 193 670 0.288
115 UNIVERSITY OF FLORIDA 233 809 0.288
116 CARNEGIE MELLON UNIVERSITY 244 854 0.286
117 INDIANA UNIVERSITY-BLOOMINGTON 172 615 0.280
118 SUNY COLLEGE AT FREDONIA 41 147 0.279
119 HANOVER COLLEGE 17 61 0.279
120 PRINCIPIA COLLEGE 10 36 0.278
</code></pre>

15

<p>No surprise here. Of all the colleges in the U.S., Caltech and Yale offer by far the most resources and opportunities for undergraduate science and engineering majors (on a per student basis). This is generally true regardless of which way you look at the data - central campus federal grants per science major, class sizes, professors per science major, etc. Other schools are very good too, but you won’t necessarily receive as much personal attention.</p>

16

<p>The other problem with engineering, monydad, is that earning a doctorate in that area is quite different than in many of the other physical sciences. It’s far less frequently sought after and, from my understanding, is quite undesirable from an industry standpoint.</p>

<p>The MD production thing is interesting. I recently found some stats on the Brown page that showed about 20% of Brown students get their MD ten years out. The chemistry department sends about the same number of students to medical school as they do to graduate school or employment right away (and in fact, nearly all chem students get either a masters, PhD, or MD within 10 years). I have no idea if this is standard, but it certainly speaks to how changing motivations and environments can affect the numbers.</p>

<p>I’ve smacked posterX around on his professors per science major ratios time after time, but he’s not really interested in anything but crew and pumping up Yale.</p>

<p>That being said, only 5 or so of the top 20 surprise me, and the prominence of LACs on these style lists time and time again just further brings home the point that these style environments are extremely successful at producing top scientists.</p>

17

<p>Good work collegehelp. </p>

<p>There are still some limitations that must be considered in using this type of information. For example, grads may have better access to the workforce in CS , investment banking, etc, from some schools moreso than from some others, or the school may attract proportionally more students who have other postgraduate goals (MD, etc) as well. To pick a few, perhaps Stanford, MIT, CMU grads in these majors had unusually high access to exceptional jobs out of undergrad, and a number of them preferred that route. But that doesn’t at all mean other grads from the same college who want PhDs are in any way disadvantaged in pursuing those goals. In an ideal world, you’d want the denominator limited to those grads of the college who actually wanted to pursue a PhD in that major.</p>

<p>But a denominator equal to the number of majors is far preferable to the one that keeps popping up on CC, which uses a denominator of the entire university population, regardless of what all those people are doing at the university , whether they all are there to even study sciences, etc. in the first place, and whether a Phd is even the logical terminal degree for the area they actually are studying.</p>

<p>So clearly this is on the right track.</p>

<p>It would be interesting to see the same table for engineering. I tried to do this once myself, but then I realized it was flawed because some schools had materially expanded their undergrad engineering student enrollment following the relevant earlier period, and I only had current enrollment figures.</p>

<p>The data on the gross number of students who subesquently earned PhDs in these fields is also quite nice to see. Since that number is unpolluted by these various denominators- all of which are flawed since none of them actually count only those students who actually wanted to pursue PhDs in these fields…</p>

<p>Re: post #13, the period that the undergrad majors are counted up should precede, by an appropriate number of years, the period that the Phds received are counted up. Otherwise schools that expanded their enrollment/# majors in these fields some time after the Phd recipients attended would be improperly penalized.</p>

<p>note: sorry this should have been before above post; I got it out of order while editing.</p>

<p>Regarding engineering, I’d like to look at that separately not mixed in with the others. And your expressed concerns about it may also apply to CS, seems to me, which might also be stripped out.</p>

18

<p>Yeah, for some reason, I didn’t think of CS but it’s true-- there exist, especially coming out of stronger programs, many jobs around 6 figures for undergrads who come out with a B.S. in CS. Not only is there less incentive industry-wide, but it may be an area where there is significantly less incentive coming out of top programs relative to the medium programs.</p>

19

<p>"…the prominence of LACs on these style lists time and time again just further brings home the point that these style environments are extremely successful at producing top scientists. "</p>

<p>Assuming that what it is, in reality, suggestive of isn’t that these style environments are comparatively unsuccessful at helping their grads obtain top-flight actual jobs in the real world after college, due to a relative dearth of on-campus recruiting and/or relative lack of great reputations among employers, and/ or that their grads do comparatively poorly in med school admissions, due to lower test scores for one thing. The PhD may be “best” goal these grads can get, whereas at some other places maybe some consider it only their third 'best" option. Because they actually have superior access to the other two.</p>

<p>Perhaps.</p>

<p>Consider that possibly a student considering choice of MIT vs, Franklin & Marshall to eventually study PhD in Physics might want to choose MIT.</p>

20

<p>Something is COMPLETELY WRONG with your data, collegehelp. You are saying that with a graduating class of over a thousand students each year, Princeton only had 198 math and science majors in three years! I don’t think so.</p>

<hr>

<p>I know for a fact that your number of 107 math and science graduates in three years is completely wrong at Swarthmore. They average between 100 and 125 science and math division majors each year:</p>

<p><a href=“http://www.swarthmore.edu/Documents/administration/ir/degsmajorschart.pdf[/url]”>http://www.swarthmore.edu/Documents/administration/ir/degsmajorschart.pdf&lt;/a&gt;&lt;/p&gt;

<p>Your number of science and math PhDs is also wrong.</p>

<hr>

<p>Wait a minute? Are you trying to count all sciences except biology majors and PhDs? That won’t work with interconnections between sciences these days. Are you excluding biochem majors?</p>

<p>Do youself a favor. If you want to try to attempt this kind of analysis, at least start with the big picture (like all science, math, and engineering) to even see if you numbers make any sense. You have to allow for the fact that the undergrade major doesn’t precisely match the PhD field with that kind of precision.</p>