Bad Science

Read this book. Ben Goldacre is a doctor + blogger. This is his good-natured rant about the manipulative tricks of money grubbing charlatans who adopt the trappings of science. His targets include homeopaths (homeopathic drugs are no better than placebos), pharma companies (trials which show expensive drugs to be ineffective are not published), and the media (who publicise a fake health scare a week). Great fun.

I hereby proclaim that Moonballs from Planet Earth and Bad Science are kindred souls.

The trouble with bad science actually starts where the book leaves off, when one moves beyond pharmacology. There are many fields worthy of scientific enquiry, where placebo-controlled, double-blind, randomized trials are not possible.

For instance, Earth Sciences. It is worth knowing if we are making our planet uninhabitable. However, we can't find out by doing an experiment. We can't hold out a control sample of several dozen similar planets where the fossil fuels were never burnt, and compare the richness of life-forms observed a few thousand years later in the test and control. So scientists have to use models, which are intrinsically fallible.

Calling out the shortcomings of the models used is central to being an honest scientist. However, lists of model caveats don't make for good TV (or for good top-management presentations). So the media coverage of global warming is about as alarmist as the fake-health-scare-a-week stories that Ben Goldacre rants on about.

The guy who first called this non-science, was Bjorn Lomberg, in the Skeptical Environmentalist. It is not light reading, but it is also worth looking up, just to get a sense for how hard it really is to construct good science, with limited data, in the thick of an emotionally charged, politicized debate.

Slumdog Smarts

Slumdog Millionaire makes a good point about intelligence: the chaiwallah knows a lot of answers, if the questions relate to his life experiences.

This thought sits on a serious problem with IQ testing, or with most standardized tests like the SAT, GRE or CAT. They don't test native intelligence. They test familiarity with a specific way of thinking, which is embedded in culture.

This is the reason why cultural minorities tend to do poorly on standardized tests. A favourite example from this piece by Malcolm Gladwell illustrates this point: Kpelle tribals from Liberia naturally group knives with potatoes, because knives are used to cut potatoes. While this is quite logical, standardized tests generally expect knives to be grouped with other tools, and potates with other root vegetables.

Unfortunately, this unsurprising and well understood limitation of standardized testing has led to horribly complicated racial profiling for university admissions in the USA, and in explicit, even more divisive, quotas in India.

Surely the more creative route is in designing culture-neutral tests? And in validating these instruments sufficiently to bring them into mainstream use?

This blog is idealistic enough to believe that better technology can at least alleviate really knotty political problems.

Series = Home + Away

Why not define a test series to span a set of both home and away matches? It seems like the obvious best answer to me. Yet, almost nobody in the cricket establishment is talking about it.

As I write, England are trying to conjure up an unlikely result in Port of Spain to square the test series against the Windies, who seem determined to draw the game and clinch the series. The cricket would be a lot more fun if the Windies were trying to win...but the Wisden Trophy is at stake.

Yet, the Windies arrive in England next month to start a new series. If the series were defined to span games played in both locations, there would be less of a home advantage, there would be fewer dead games, and both sides would play more natural and attacking cricket for more of the time.

In 2008 India and Australia played 8 test matches, four in Australia and four in India. Wisden thinks Australia won one series 2-1 and India won the next series 2-0. I think India won the Border-Gavaskar trophy 3-2, a very fair score.

Think back to South Africa checking out at the Oval in 2008. Or Rahul Dravid not enforcing the follow on at the Oval in 2007. Most test series are already scheduled as home-away combos. Surely there is an obvious way to avoid silliness of this sort.

Gwyneth Paltrow's Hindu Haircut

Regular readers of this blog may not have been aware: Gwyneth Paltrow has had a haircut.

Gwyneth revealed that this was part of the healing process following the death of her father. "I was very very attached to my hair," she says. "I still had hair from when my father was alive. I made it a talisman. Then one day, on a shoot with Mario Testino, I suddenly said 'I need to cut it now'. It was almost as if it was part of the grieving process. I just had to let something go."

Gwyneth may not have known this, but she is a karma yogi going through the process of samskara.

A karma yogi fulfills her destiny, or achieves personal growth, or attains moksha, by facing up to the slings and arrows of outrageous fortune with integrity and dignity, in accord with her dharma. Rituals, like cutting one's hair when a father dies, are ways of ways of coping with grief and moving the soul along its natural journey.

Traditional Hindu doctrine might have preferred for the psyche to step into a new life-stage in front of the sacred fire rather than at a Mario Testino photo-shoot, but that surely is a minor detail.

Banker or Blogger (2)

A second innings then for this post...in response to Greg Pye's excellent comments.
____________________________
On the recent increase in US and UK Ginis, the OCED data does actually show a time series. The story doesn't really change if one looks back to the mid 80s. Here are the rankings from ~20 years ago from a sample of 24 (rather than 30) OECD countries:

1. Mexico 0.452
2. Turkey 0.434
4. United States 0.338
8. United Kingdom 0.325
12. Canada 0.287
15. New Zealand 0.271

- The biggest mover is New Zealand, which was a hot little economy in the 90s, and saw a sharp inequality rise then. Kiwi Ginis seem steady over the last decade

- The USA, and most of the OECD, have expectedly seen a steady rise in inequality
since the mid 70s, the earliest data on this table

- Inequality in Britain sems to have peaked 1990 and remained there or thereabouts ever since. The steep rise seems to have happened in Margaret Thatcher's time. Surprisingly, and maybe incorrectly, the last data point actually shows a small Gini decline in the UK
__________________________________
On the irrelevance of national statistics, I do think inequality is experienced most powerfully within a tight reference group. As H.L. Mencken once said, "a man's satisfaction with his salary depends on whether he makes more than his wife's sister's husband". But I do think there is a difference between the angst induced by a brother-in-law's Ferrari, and the unease triggered by the suffering of millions of decent people.

What I'm trying to get at is the impact that that suffering has on the psyche. Does a keenly felt awareness of suffering push the psyche towards work, towards seriousness?

I love the thought that sensitivity to suffering need not be limited by national boundaries. Absolutely. But the boundary could provide the psyche with a prop with which to make peace with the suffering of innocents.

The central point in the first post was that the psyche tends to deal with injustice by wrapping itself around work, around seriousness. I'm sure it sometimes does. But the weakest part of the argument is that the psyche could, and often does, cope by imposing an identity on the people suffering, turning them into the "other". Race, caste, class, religion, nation...any schism will do. Once that boundry has been established, the psyche is free to go ahead and have fun.
___________________________

Was this post triggered, at some subliminal level, by the hoop la around Slumdog Millionaire?

Blogger or banker?

Here's the theory on people who work for a bank and blog for fun. Do they think of themselves as bankers or bloggers?

It boils down to their Ginis. You see, some people have magic lamps inhabited by blue-suited banker Ginis, some people's magic lamps have sailor-suited Ginis... :)

Actually, a Gini coefficient is a measure of statistical dispersion, and is a standard measure of income inequality in a society. My take is that people from more unequal societies are more likely to define their identities in terms of who they are at work.

Why?

Imagine a relatively well-off person living in an obviously unequal country. He needs to make peace with the fact that he lives a comfortable life, but the people from the slum/ favela/ ghetto/ council estate/ inner city live obviously miserable, abysmal lives. The sheer presence of that abyss, the unspoken fear and guilt that that abyss evokes, pulls at the psyche of the comfortably-off like gravity. The psyche protects itself from that pull by believing that privilege and comfort are deserved, earned, by hard work, by education, by qualifications, by seriousness.

In this unequal context, it is hard to think of oneself in purely frivolous terms. It feels like being the surfer on the beach in Apocalypse Now. Its the reason why cricket in India or football in Mexico are not just silly games played for fun, they are about the redemption of national pride.

So what do I expect to observe in the data? I expect people from more unequal societies to wrap their identity ever more tightly around their professional selves.

Here are Ginis for some of the OECD-30. Their rankings are:

1. Mexico: .474
2. Turkey: .430
4. USA: .381
7. Great Britain: .335
8. New Zealand: .335
12. Canada: 0.317

The two most unequal OECD members are Mexico and Turkey. Fortunately, I have friends from Mexico and Turkey who tell me their compatriots unambiguously define who they are in terms of who they are at work.

Also, to my earlier observation, Britain's Gini is the same as New Zealand, and is a lot lower than the USA. Canada is even further away from the USA than is Britain. So if the theory holds, Canadians should be a less likely to derive their identity from work than either Americans or Britons, despite Canada's stereotypical cultural location somewhere in-between the USA and Britain.

India is not in the OECD. So I looked up the World Bank's Ginis metrics, which show that India is better (i.e. more equal) than the USA.

While that is flattering, and says something important about the world's only superpower, the World Bank might be systematically underestimating South Asian inequality. Pakistan looks really good on the same metrics, more equal than the Netherlands, Canada, France or Switzerland. That doesn't ring true. My hunch is that India really is in the mid - 40s pack, along with Mexico, China, Jamaica and Turkey.

A more classical theory, which involves no melodrama about the gravitational pull of the abyss, is the impact of marginal tax rates on labour supply. More equal societies have higher marginal tax rates. People therefore have less reason to work hard to earn money. They therefore invest more of their time, and identity, in leisure rather than labour.

I buy into the conventional theory, but it doesn't quite feel complete. Maybe that is because I remember an India with high marginal tax rates, in Indira Gandhi's time, when people still wrapped their identities around their work, even if they didn't work especially hard. The ways in which people construct their identities change more slowly than tax policy.

A first class map

Here's a cricketing puzzle that's got me stumped.

England has about 39 counties or shires, or maybe that should be 42, or maybe even 46, an interestingly vague number unlike the unimaginative precision of the 50 American states...but the English taste for vagueness is not what this post is about. However one chooses to count, 18 adds up to less then half the English counties. Yet, only 18 counties (the darker coloured counties on the map above) play first class cricket. How did this come to pass?

For instance, Shropshire, that fabled land where the Empress of Blandings covered herself with such glory, would surely have a deep rooted cricketing culture? Or consider Lincolnshire, a well-populated county wedged in between the cricketing powerhouses of Yorkshire and Nottinghamshire. Surely Lincolnshire would be able to muster a decent county team? Club cricket is played in both Shropshire and Lincolnshire. But both seem content to remain Minor Counties, without objecting to the condescension loaded into that term.

Even more mysterious, Durham, the current champions, are the only Minor County to have been promoted to the first class level in 80+ years. This was a one time "historic" event. There is no mechanism by which the top Minor Counties get promoted to the first class level, and the weakest first class counties get relegated to the minors.

This instinctively feels odd to me, as an Indian cricket fan, because the growth of cricket outside India's traditional metropolitan centers is one of the most fascinating changes in Indian cricket over the last 25 years.