The follow-up segment has Stephen Colbert interviewing Thomas Herndon, the UMass econ grad student whose term paper on replicating R&R eventually led to the once-influential paper's unraveling.
Friday, April 26, 2013
Monday, April 22, 2013
The Data Stories blog has a great interactive infographic on a grim, but all too real, phenomenon that goes by the name "missing women". The post is triggered by a recent paper in EPW by Siwan Anderson and Debraj Ray who went beyond sex-selective abortion and infanticide to examine data on women's "excess mortality" across all age groups in India [The authors have presented summary at the Ideas for India site, which I linked to sometime ago].
Here's a short description of the infographic:
The paper calculated the numbers for 2003. I’ve redone the numbers to calculate excess female deaths per thousand across age-groups for India, China, and a number of Indian states [specifically, Haryana, UP, Kerala, Bengal, Bihar, and "Others"] for 2010. Each bar shows excess mortality per 1000 women in that age-group only.
Saturday, April 20, 2013
A great story from Reuters: How a student took on eminent economists on debt issue - and won. The student is Thomas Herndon, in the UMass grad program in economics. This thing is buried deep inside the report, but worth highlighting:
Herndon's paper began life as a replication exercise for a term paper in a graduate econometrics class.
A couple of fun links:
Kieran Healy: New Tools for Reproducible Research [a nifty infographic]
In the comments section, we find a link to EuSpRIG, The European Spreadsheet Risks Interest Group -- with its own annual conference. Its website has a page devoted to spreadsheet horror stories. The RR paper hasn't made it to this page so far, though.
Thursday, April 18, 2013
In 2010, Reinhart and Rogoff published a paper that, by many accounts, turned out to be influential in framing economic policy options back then. After looking at the dataset, economists at the University of Massachusetts have poked multiple holes in R&R's methodology, including an error in their Excel spreadsheet program. Start with Andrew Gelman's advice to R&R: "it’s best to admit your errors and go on from there," and look at the links for the backstory.
This episode has triggered a flurry of posts on similar disasters in economics literature: Catherine Rampell has a post on A History of Oopsies in Economic Studies. See also: Dean Baker on In History of Economic Errors, Martin Feldstein Deserves Mention, Paul Krugman on another disastrous Excel error.
See also: Felix Salmon on the R&R fiasco.
Tuesday, April 16, 2013
In his WSJ opinion piece, Wilson gives a blunt answer by way of a "professional secret": "Many of the most successful scientists in the world today are mathematically no more than semiliterate." He does make the rather unexceptionable point that "if your level of mathematical competence is low, plan to raise it," but also adds that "you can do outstanding scientific work with what you have." Of course, if your level of math is not all that high, he also suggests you avoid "most of physics and chemistry, as well as a few specialties in molecular biology."
His article has led to tons of responses: Paul Krugman agrees, but with a caveat: "at least in the areas I work in, you do need some mathematical intuition, even if you don’t necessarily need to know a lot of formal theorems."
A couple of other responses. In Science's Cult of Calculation, Jag Bhalla talks about the rivalry in science between "math-monks" vs. "pluralist reasoners". Terry McGlynn places Wilson's piece within the context of Tribalism in the sciences: empiricists vs. theoreticians.
The tribe of theoretical ecologists appears to have got very upset with Wilson's piece for reasons that lie within the internal politics of that field -- this post explains some of it, and has tons of links.
Saturday, April 13, 2013
Blog Discovery of the Year: Data Stories (Tag line: "... On India one chart at a time"). It's a personal blog maintained by "a Delhi based journalist" -- that's all we know about the blogger behind this wonderful site. Here's the first post. Here's a post with a catchy title -- We are the 5 %:
I leave you to draw your own conclusions about what it means to be ‘privileged’ in this country. I also leave you with this question: If the census takers had asked each one of these households, what ‘class’ of society they thought they belonged to, or where they fit in within the income distribution, what do you think their response would have been ( and by ‘their’, I also mean ‘our’)?
Amy Freitag at Southern Fried Science: I’m a scientist. A social scientist. Please opine on the validity of my discipline.
Jenny Rohn at Occam's Corner: Show me the money: is grant writing taking over science?
Tuesday, April 09, 2013
... have become so numerous that even the king of the MSM jungle -- I mean, NYTimes -- has an article about it [I thank Prof. Dheeraj Sanghi for the e-mail alert]. The bottomline is that since their numbers have become large, and they have become better at deceiving their audience, even the smart and the savvy have fallen prey to them. In other words, the list of victims goes beyond the clueless and/or the greedy.
The article does a good job of identifying the problem and its variants (both conferences and journals) with plenty of illustrative examples. Where it goes horribly wrong is when it indulges in a bit of drive-by shooting by conflating the problem (scam journals and conferences) with open access models of scientific publishing.
I mean, while it is true that there has been a huge increase in the number of journals and conferences that take money from authors and publish their papers with little or no peer review scrutiny, it is also something that would have happened irrespective of whether these papers are placed in the public domain. The drivers for this trend are easy to enumerate, and include:
the huge expansion in scientific research (especially in the non-OECD countries) in recent years.
the ever-increasing pressure on researchers to publish, publish, publish.
the global university ranking business that puts a premium on research publications (and now, citations and other scientometric indicators), which in turn adds to (2).
Since the number of papers has grown (but the number of journals has not), someone is just going to come along and provide an outlet for all these new papers. And since a growing fraction of these extra papers is junk, the said someone is just going to take money to publish them. Insert your favorite cliche here -- Win-Win, the Invisible Hand, Magic of the Market, Nature abhors unpublished papers, whatever.
And none of this has anything to do with open access.
It is a fact that almost all of these "journals" are online-only operations. But open access has nothing to do with this business model, except perhaps as an unintended consequence.
* * *
I became aware of this problem quite sometime ago -- in November of 2009 -- when someone perpetrated a scam by "organizing" a scamference in an auditorium in the IISc campus! See this, this, and this (and see this as well!).
Since then I have also been alerted about all kinds of online journals. At least one of them used Google Sites (a free website builder) for its operations; that is, it didn't even bother to have, and pay for, its own domain name!
Filed under: Stuff Indian Government Says
Amitabh Sinha reports in The Indian Express that Dr. T. Ramasami has expressed his desire to bring to an end his 7-year tenure as Secretary, Department of Science and Technology. A committee has recommended a set of people to succeed him, but two of The Chosen "have expressed their unwillingness for the job."
At this point, it's all mildly interesting, but the blog-worthy awesomeness lies in the Groucho-Marx*-in-reverse quality of what comes next:
One member of the committee, however, said that the reluctance of scientists to join the ministry was understandable. "We would be slightly worried if candidates are too eager to take the post. Most of the scientists like to keep doing their scientific work," he said. [Bold emphasis added]
* * *
*Just in case the Groucho Marx's quip didn't spring to mind immediately, here it is: "I don't want to belong to any club that will accept people like me as a member."
... I don’t leave comments on other blogs, I don’t have another Twitter account, Facebook account, Google Plus account, or any of that. Hey, I am trying to have a life.
-- Paul Krugman on his blog.
Monday, April 08, 2013
Over at Kafila: The Delhi University Four Year Structure -- Myths and Reality.
The Telegraph reports from Assam: Pay-cut plan for off-campus doctorates. What in hell is an 'off-campus doctorate'?
The Deccan Herald published a news story about the founder of ISI, Pakistan's spy agency; but it used a picture pf Prof. Mahalanobils, the founder of ISI, the Indian Statistical Institute [Hat tip to Kaneenika Sinha, who has a screen-grab of the online version].
The story has been available online since 4 April, and at least four people have pointed out the horrible error; and the DH has still not corrected the article. WTF, DH?
Through him we learn of the (non) workings of hearing aids, what is Lombard reflex and how deafness saves one from that, why TV is a good companion (close captioned) for the deaf than the movie halls, of 'quiet coaches' in England trains, of Beethoven's despair and how his reclusive character is a put-on to cover-up his growing deafness (as he explains in his letters), how Francis Goya's deafness could have enhanced his concentration to do better paintings in his later years and several such interesting deaf stuff.
These sections are peppered with word play and dead-pan humour
One of the strongest curses in English language is 'Damn your eyes!' (much stronger than 'Fuck you!' and definitely more satisfying) [...] "Damn your ears!" doesn't cut it. Or imagine if the poet had written, 'Drink to me only with thine ear...; It's actually no more illogical than saying drink with thine eyes [...] Nor would 'Smoke gets in your ears' be a very catchy refrain for a song [...] 'There's more in this than meets the ear' is something Inspector Clouseau might say, not Poirot.
Thursday, April 04, 2013
Myth 8: The Novartis decision undermines the global search for new medicines.Read more...
Fact: Of all the canards, this is probably the most ludicrous. Big Pharma makes the vast majority of its profits on sales to rich patients in rich countries. Nearly 75% of global drugs sales by dollar volume in 2011 was in Europe, North America, and Japan. Indian sales comprised less that 2% of global sales. Drug giants do not make R&D decisions or shut down promising drug candidates because they didn't squeeze a little extra profit out of small market.
To the contrary, drug companies waste a lot of research dollars now trying to evergreen existing medicines instead of focusing on truly innovative medicines. They spend nearly 2 1/2 times on marketing and administration as they spend on R&D. Despite the "risks" of R&D they still retain more in profits than they actually spend each year on R&D.
[Thanks Prof. Guhan Jayaraman for email share]
Wednesday, April 03, 2013
Tarun Jain (Indian School of Business) at Ideas for India: Should Bribe Givers Be Let Off?
In 2011, Economist Kaushik Basu argued that for a class of bribes, the law should not punish the bribe-giver. This column presents results of experiments conducted to test this idea and provides insights for anti-corruption efforts.
Dirk Matten in Globe and Mail: India’s generics drug ruling will help, not hinder, innovation. He reiterates what I consider to be central to the Supreme Court verdict; Indian news outlets that I have seen seem to miss this point, though:
The crucial point here is whether the version of Glivec for which Novartis was claiming patent protection is actually a ‘ new ‘ drug. What the Indian supreme court in fact ruled was not that Novartis should not enjoy patent protection on their new drugs; it mainly concluded that the new edition of Glivec, for which the company applied for protection, was in fact not sufficiently ‘new’ – not different enough from the old version of Glivec, for which the patent had expired.
This points to a well know strategy of the pharmaceutical industry. Rather than fighting generic companies, ‘originator’ companies such as Novartis just marginally change the chemical formula of an existing drug whose patent is about to expire and then pretend to having come up with an entirely new one, for which of course they should enjoy full patent protection.
Brad Plumer at the Wonk Blog: Expensive batteries are holding back electric cars. Can that change? A part of the argument hinges on the lack of a Morre's Law in energy storage technology!
Tuesday, April 02, 2013
T.V. Padma in Scientific American: India Court Ruling Upholds Access to Cheaper, Generic Drugs. It's about yesterday's landmark Supreme Court verdict that denied Novartis a patent for its evergreened version of its anti-cancer drug.
Henry R. Bourne (UC-San Francisco) in eLife: The Writing on the Wall. Here's the abstract:
The biomedical research enterprise in the US has become unsustainable and urgent action is needed to address a variety of problems, including a lack of innovation, an over-reliance on soft money for faculty salaries, the use of graduate students as a source of cheap labour, and a ‘holding tank’ full of talented postdocs with limited career opportunities.
Kristina Lerman in ACM SIGMOD Blog: Stop Publishing So Much Already!
Rebecca Skloot in NYTimes: The Immortal Life of Henrietta Lacks, the Sequel:
The Lackses’ experiences over the last 60 years foretold nearly every major ethical issue raised by research on human tissues and genetic material. Now they’re raising a new round of ethical questions for science: though their consent is not (yet) required for publishing private genetic information from HeLa, should it be? Should we require consent before anyone’s genome is sequenced and published? And what control should gene-sharing family members have?
SMBC has a "mathy" variant of "Talk dirty to me" -- with probably the first ever appearance of the Frobenius method in a comic strip.
Monday, April 01, 2013
Akshat Rathi has a great article: Aakash is no silver bullet (it appeared -- without hyperlinks -- as an op-ed in The Hindu). Citing the failure of the OLPC project to live up to all the hype about how it would revolutionize education, he pours a lot of cold water on the idea that Aakash, somehow, is the magic gadget Indian kids have been waiting for.
Even if the government somehow, however difficult it may seem, is able to get access to cheap tablets, they are not going to help achieve its aims. Can a laptop overcome the negative impact of a bad teacher or poor school? Can it make children smarter despite the lack of electricity, water, toilets or playgrounds? Can it overcome the limitations of stunted growth among the malnourished? Can Aakash increase productivity of the workforce to counterbalance the money invested in it?
There is no evidence that it can do any of these things. [...]
He also has a follow-up post where he responds to comments.
* * *
While I have nothing against the R&D project on Aakash (especially when it is coupled with small-scale experiments on technology-enabled learning), I do have a problem with the vast, massive social experiment that the government plans to build around it -- all in the name of education. This obsession with treating gadgets as magic wands just doesn't make sense -- especially when studies have shown that laptops for school children are not such a great idea even in rich countries (and if you want links to studies on OLPC in other countries, go to Rathi's post). Why then are Indian states like Tamil Nadu and Uttar Pradesh spending huge sums of money to give a laptop to their students?
I shared some of these and other similar thoughts with Samanth Subramanian who has a report on the turbulent ride the Aakash project has had in recent days.
* * *
Interestingly, Satish Jha, an official OLPC cheerleader in India, has an article trashing the Aakash project. It's a bit rich, isn't it, that a man working for a hi-tech huckster admonishes his fellow citizens for not learning from the masters about how to develop low cost gadget. What is he going to do next -- sell snake oil? Oh, wait!