Monday, February 12, 2007

Nobel data

Some interesting Nobel data found here. The author of the study is a professor of evolutionary psychology in the UK.

To improve the analysis I propose we remove the economics prize and replace it with the Fields Medal, and also normalize the number of winners to the size of the country/institution :-) It seems that US dominance is actually *increasing* (at least over 20 year timescales; perhaps we're past the peak by now). See related posts here.

Note that Stanford + Berkeley (the bay area, not even counting UCSF, which won 3) beats any other country over the last 20 years.

Why there should be more science Nobel Prizes – and why proportionate credit should be awarded to institutions

Charlton BG
Medical Hypotheses 2007; 68: 471-3 – doi:10.1016/j.mehy.2006.11.003


The four science Nobel prizes (physics, chemistry, medicine/ physiology and economics) have performed extremely well as a method of recognizing the highest level of achievement. The prizes exist primarily to honour individuals but also have a very important function in science generally. In particular, the institutions and nations which have educated, nurtured or supported many laureates can be identified as elite in world science. However, the limited range of subjects and a maximum of 12 laureates per year means that many major scientific achievements remain un-recognized; and relatively few universities can gather sufficient Nobel-credits to enable a precise estimate of their different level of quality. I advocate that the Nobel committee should expand the number of Nobel laureates and Prize categories as a service to world science. 1. There is a large surplus of very high quality prize candidates deserving of recognition. 2. There has been a vast expansion of research with a proliferation of major sub-disciplines in the existing categories. 3. Especially, the massive growth of the bio-medical sciences has created a shortage of Nobel recognition in this area. 4. Whole new fields of major science have emerged. I therefore suggest that the maximum of three laureates per year in the categories of physics, chemistry and economics should always be awarded, even when these prizes are for diverse and un-related achievements; that the number of laureates in the ‘biology’ category of physiology or medicine should be increased to six or preferably nine per year; and two new Prize categories should be introduced to recognize achievements in mathematics and computing science. Together, these measures would increase the science laureates from a maximum of 12 to a minimum of 24, and increase their coverage. The Nobel Prize committee should also officially allocate proportionate credit to institutions for each laureate - both in retrospect for past prizes, and in the future.

...Nobel laureates nations and research institutions were measured between 1947-2006 in 20 year segments. The minimum threshold for inclusion was 3 Nobel prizes. Credit was allocated to each laureate's institution and nation of residence at the time of award. Over 60 years, the USA has 19 institutions which won three-plus Nobel prizes in 20 years, the UK has 4, France has 2 and Sweden and USSR 1 each. Four US institutions won 3 or more prizes in all 20 year segments: Harvard, Stanford, Berkeley and Caltech. The most successful institution in the past 20 years was MIT, with 11 prizes followed by Stanford (9), Columbia and Chicago (7).

Table 1 – Number of Nobel laureates by Nation – twenty year segments from 1947-2006. A minimum of three prizes in one time segment is required for inclusion.

Nation - 1947-66 - 1967-86 - 1987-2006

USA -50 - 88 - 126
UK - 20 - 25 - 9
Germany - 8 - 7 - 9
Switzerland - 3 - 7 - 7
Sweden - 3 - 7 - 1
Japan - 2 - 1 - 3
USSR/Russia - 7 - 2 - 2
France - 4 - 3 - 5

Table 2 - Number of United States Nobel laureates by Institution – twenty year segments from 1947-2006. A minimum of three prizes in one time segment is required for inclusion.

Institution - 1947-66 - 1967-86 - 1987-2006


Harvard - 9 - 13 - 5
Univ. California Berkeley - 7 - 3 - 4
Stanford - 4 - 5 - 9
Caltech - 4 - 4 - 5
Columbia - 4 - 1 - 7
Rockefeller Inst. & Univ. - 3 - 6 - 3
Chicago - 2 - 4 - 7
Princeton - 1 - 2 - 6
MIT - 1 - 5 - 11
Cornell - 1 - 4 - 2


Dave Bacon said...

Heh: The implication may be that Harvard is evolving towards being a ‘normal science’ university – albeit unusually large and successful.

BTW you might add the Turing Award as well as the Fields Medal.

Steve Hsu said...

Good point. Charlton (the author) actually does this if you look further down his blog page.

Unknown said...

What's wrong with the economics prize?

Steve Hsu said...

I don't think that economics is as well developed as the other Nobel-eligible sciences: biology/chemistry/physics.

Feynman once defined science as answering the question: "If I do this (makes generic motion with fist), what happens next?" I think if you look carefully, you will find the predictive power of economic theory to be quite limited. Further, the subject is heavily contaminated by ideology (always a problem when data is messy and conclusions have social ramifications).

You can even find serious academic books with titles like Economics is not yet a science. I wouldn't go that far, but it clearly isn't quite at the level of the other three disciplines yet. See the link below for a representative discussion on Harvard professor Greg Mankiw's blog.

From the comments:
"Firstly, as Larry Summers long ago pointed out, empirical rejection of a theory in economics is never a reason to reject that theory. This hardly resonates of scientifism. Anything but. In economics, if a theory is rejected by data, that empirical rejection simply becomes a "puzzle"; frequently that data is dismissed as being "wrong". The theory itself lives on. The equity premium puzzle, for instance, casts serious doubt on conventional theories of asset pricing and, indeed, the neoclassical growth model and standard utility functions. You yourself had vehemently criticised the New Keynesian Phillips curve as being empirically invalid; yet it forms the basis of an abundance of current reaserch work. In international economics, the theories of PPP and interest parity have been faced devestating rejection in the data; again both still form the corpus of much work in this area. In a serious science, something with a lack of empirical validity would surely be quickly disavowed.

Further, theres little consensus on anything in economics. Even at the most fundamental level, economists question the validity of such tenets as the slope of demand curves. Many support a higher minimum wage (Card said it would raise employment; labor demand curves slope up); many dont (labor demand curves slope down). I think if any body of knowledge aspires to scientifism, then there must be broad agreement on the very fundmentals of that discipline. In economics there isnt.

Also, and this is a very significant point. In economics you can establish a theory to prove *anything* you want. You can use data to prove *anything* you want. In this sense, any conclusion attained by an economist is necessarily suspect, since theres a lack of fundamental rigor in the first instance."

Anonymous said...

Economics *as taught* in U.S. universities isn't a science because it's a watered-down, politically correct version.

Economics theory as may be found in journals and elsewhere *is* -- and not less than any hard science.

Seth said...

I like the idea of somehow crediting the institutions where the prize-winning work was actually carried out. A lot of the big name universities are aggressive about hiring rising talent without necessarily playing much of a role in actively cultivating talent.

Of course measuring the value of an institution by weighing their N-prizes is pretty literal minded to begin with.

Unknown said...

Related to what STS says, What institution rightfully claims the Nobel laureates? The institution where he/she was educated and received his PhD, or where the prize-winning work was completed, or the institution where he/she is currently employed as a professor? I wonder if the data presented in the table is "clean", or consistent in how they've been attributed...and how the three separate charts would compare if you went through for each prize winner, made note of all three of those institutions, and made sure the charts are consistent for all three of the possible institutions the winners can be affiliated with?

AMIT said...

You have a good post and a point in your post.But sometimes Nobel prize is given wrong also.

Massachusetts Drug Rehab

Blog Archive