lhsi writes:
Some recent research has suggested that there is a link between certain foods such as chocolate and obtaining a Nobel prize. New research found dedicating a high proportion of GDP to research and a high number of scientific papers published were more accurate predictors.
From the article:
Several recent studies have described a strong correlation between nutritional or economic data and the number of Nobel awards obtained across a large range of countries. This sheds new light on the intriguing question of the key predictors of Nobel awards chances. However, all these studies have been focused on a single predictor and were only based on simple correlation and/or linear model analysis. The main aim of the present study was thus to clarify this debate by simultaneously exploring the influence of food consumption (cacao, milk, and wine), economic variables (gross domestic product) and scientific activity (number of publications and research expenditure) on Nobel awards. An innovative statistical analysis, hierarchical partitioning, has been used because it enables us to reduce collinearity problems by determining and comparing the independent contribution of each factor. Our results clearly indicate that a country's number of Nobel awards can be mainly predicted by its scientific achievements such as number of publications and research expenditure. Conversely, dietary habits and the global economy variable are only minor predictors; this finding contradicts the conclusions of previous studies. Dedicating a large proportion of the GDP to research and to the publication of a high number of scientific papers would thus create fertile ground for obtaining Nobel awards.
(Score: -1, Flamebait) by crutchy on Saturday March 29 2014, @10:02AM
They awarded a Nobel Prize in economics to Paul Krugman and a Nobel Peace Prize to Obama for fuck sake.
(Score: 1, Offtopic) by Bot on Saturday March 29 2014, @10:26AM
Indeed.
With Obama, I could end up being deployed in a war zone and nobody knows there's a war! Me, a propaganda bot!
What will I tell to my grandchild processes? "Yep there were those kid terrorists playing hide and seek, and BLAM I drove a missile straight into them, ready or not"?
Why don't you revert to the good old habits of declaring war, meeting in some valley at some hour, beating the sh*t out of each other, and the one who comes out takes all? Especially since no bots are involved.
Account abandoned.
(Score: -1, Flamebait) by Anonymous Coward on Saturday March 29 2014, @10:39AM
Pretty much proves the nobel prize is a bad joke.
So who gives a damm about it. There's alot more jokes that are way more funny. And less likely to get me killed or put on some list.
(Score: 5, Insightful) by umafuckitt on Saturday March 29 2014, @12:12PM
No it doesn't prove anything. Obama got the peace prize, which is notorious as being the most lightweight prize and is really just a way for making a political point. The prizes in the hard sciences (including medicine and physiology) remain untainted and are still worthy of their reputation.
(Score: 5, Interesting) by guises on Saturday March 29 2014, @12:49PM
The peace prize is political by definition, what is your beef with Krugman's work on new trade theory?
(Score: -1, Troll) by crutchy on Saturday March 29 2014, @02:03PM
Krugman is an blubbering idiot. His "theories" helped cause the GFC.
(Score: -1, Flamebait) by contrapunctus on Saturday March 29 2014, @12:06PM
they also found research causes cancer in mice
(Score: 0) by Anonymous Coward on Saturday March 29 2014, @04:47PM
It's easy to try to sound smart and say things like that, but there are lots of things out there that can't be researched using controlled experiments. Thus, correlative post-hoc analyses are sometimes all that is available. So long as one is aware of the drawbacks of such studies, there is no reason not to do them. The problem is that it's too easy to go beyond the data and such studies are too often over-hyped and misunderstood by the media. Thus, a perfectly valid research approach gets a bad name.
(Score: 3, Interesting) by umafuckitt on Saturday March 29 2014, @12:14PM
Increasing chocoloate consumption [scientificamerican.com] also increases Nobel prizes. Must by why China is winning few of them despite investing very heavily in research in recent years.
(Score: 2) by umafuckitt on Saturday March 29 2014, @12:33PM
Ah, I see. They've actually added the chocolate thing to their model(!). I've not heard of hierarchical partitioning before, but I found a few links in case others are interested: R library, hier.part [r-project.org], paper from 1991 [tandfonline.com], and a PLoS ONE paper from 2010 [plosone.org].
(Score: 5, Interesting) by Open4D on Saturday March 29 2014, @12:47PM
The article seems to have been a reaction to the chocolate study, and to others that followed it.
The abstract was doing okay until the final sentence ...
So they've gone from talking about correlation, to all of a sudden making recommendations that assume causation. To me this is little better than the practice of "giving nutritional advice based on simple correlations observed between nutriment consumption and cognitive ability", which they criticize.
N.B. They say
That's an understatement.
(Score: 3, Interesting) by TheloniousToady on Saturday March 29 2014, @01:11PM
It amazes me that anybody applies statistics to predicting who might win five Nobel prizes each year (six if you include the unofficial economics prize) out of a world population of probably tens of millions of possible candidates. Although it's intuitively appealing that "dedicating a high proportion of GDP to research" might be a factor, I don't think any valid statistical inference can be drawn from six winners each year out many millions of candidates.
We might as well start predicting lottery winners. In fact, I bet they also come from countries which dedicate a high proportion of GDP to research. That said, whoever succeeds at predicting them is likely to win the Nobel economics prize.
(Score: 4, Informative) by umafuckitt on Saturday March 29 2014, @02:18PM
Did you even read the summary? That's not at all what they did. They didn't attempt to predict which individuals would win a Nobel. They are trying to identify which variables are good predictors for the number of Nobels per capita in different countries. Totally different thing to what you're describing.
(Score: 2) by TheloniousToady on Saturday March 29 2014, @07:49PM
Yes, I did. Sorry if I didn't state my point clearly enough for even you to understand it. I said (and also meant) nothing about predicting individuals. That was entirely from your imagination. Instead, by "who might win", I meant the "who" to be any sort of broad group, for example, people from a particular country.
FWIW, please try to be a little kinder next time you misunderstand someone's comment. Nobody deserves to be attacked for trying to contribute something constructive to the discussion.
(Score: 2) by umafuckitt on Saturday March 29 2014, @11:28PM
Truly, I'm sorry if I offended you. I can be too rough at times. Sorry. Your post really wasn't clear, though, and neither is your follow-up. Who "who" is matters enormously. The point is that the authors have been reasonable in their choice of "who" and so they can do their study. Their data back this up.
You stated that the number of candidates is "tens of millions." That is why I assume you are talking about predicting individuals. But the number of candidates isn't tens of millions, it's 71, because in the study in question a "candidate" is defined as a country that has won at least one Nobel. Thus, they are predicting countries, not individuals, which is what you appeared to be talking about when you mentioned lottery winners. Further, since the authors have access to all Nobel prizes awarded to date, they have 10 times as many Nobelists as countries. What they're doing is therefore totally reasonable and they have reasonable power to work with.
The above, however, doesn't tell you what the data look like. It only tells you that it's plausible to conduct such a study. The acid test is to look at the graphs. When you do so, you will see that there are some quite reasonable trends [plosone.org]. The fact that you see trends to which you can fit curves indicates that you can indeed build a model and obtain some sort of prediction. What the prediction actually means is a different question, but the meaning of the model doesn't affect its accuracy. It only affects its interpretation.