Tuesday 17 August 2010

Off the ‘Spirit Level’ [update 4]

The authors of the British book The Spirit Level have a political agenda, and they’ve got it talked about everywhere. Even here. The NZ Labour MPs’ blog Red Alert for example is so excited it even has a ‘Spirit Level’ “tag”, and breathless comments from the likes of Grant Robertson that “These people’s work can not be dismissed.” And Colin James, the commentator on the tired and the bleeding obvious, wonders if the 300-page tome might not become  “a sort of guidebook for the next Labour ministry,” should there be one.

So what’s their work, and why are Grant Robertson and his comrades so excited about it? It’s a “revolutionary” thesis overturning all previous research: that societies with more “equal” incomes do better than those that don’t.

So how did they do what no other researchers before them have managed to do? Simple, They fudged the figures.

Here, for example, is a graph that is central to their thesis, purporting to show how much better off our life expectancy would be if we all copied “the workers’ paradises of Scandinavia and the egalitarian nirvana of Japan.” See how the culturally egalitarian Japan and Sweden skew the left side up, and the culturally meritorian USA skews it down?

SpiritLevel01And here it is again, this time with the many elephants in the room they neglected to include because they inconveniently contradict their thesis, charted using figures from that bastion of inequality, the UN. See how the addition of Hong Kong alters the right-hand side, and the additions of the likes of South Korea and the Czech Republic give a more “realistic” level to the left-hand end.

SpiritLevel02

Go to the SPIRIT LEVEL DELUSION blog Adding in the countries that contradict their thesis shows that there really is no significant trend at all (maybe one month, or two?), which means their much-hyped thesis is basically bosh.  [Further clarification on the graphs below.]

As the author of The Spirit Level Delusion, Christopher Snowdon, explains at Spiked, (from whom I stole those graphs) this sort of statistical legerdemain exposed here cannot be unintentional. Which means, to put it bluntly, that the authors have lied—and if you have to lie to make your point, that probably means you haven’t got one. Nonetheless, it’s a lie perfectly calculated to get the chattering classes talking; so after tearing apart their “research,” Snowdon draws the only conclusion about their thesis and its widespread acceptance that you could:

_Quote The only real difference between ‘less equal’ and ‘more equal’ countries is the size of the government and the amount it takes in tax, rising from less than 15 per cent of gross domestic product in Singapore to almost 50 per cent in Denmark. The fact that Singapore outperforms Denmark under almost every measure of what makes a country ‘do better’ only serves to underline the folly of The Spirit Level and, by association, the futility of its political agenda.
    That this agenda takes the form of zero-growth economics and eco-authoritarianism perhaps explains why journalists at the New Statesman and the Guardian have been so willing to suspend disbelief when confronted with such an improbable explanation for the problems of all mankind. It seems not to have struck them as odd that two left-wing epidemiologists were able suddenly to unearth a ‘theory of everything’ which had eluded the world’s finest minds for generations.
    To The Spirit Level’s legion of admirers, this uncanny turn of events is only proof of Wilkinson and Pickett’s unique genius. A more prosaic explanation is that the grand unifying theory had not been unearthed because it was never there.

As Ayn Rand used to say,

_QuoteIf there were such a thing as a passion for equality (not equality de jure, but de facto), it would be obvious to its exponents that there are only two ways to achieve it: either by raising all men to the mountaintop—or by razing the mountains.

And naturally, the exponents of forced equality always end up advocating the latter. That the talk about this book and its recommendations at the Red Alert blog usually ends with a recommendation to soak the rich is proof once again that this thesis is still fundamentally correct. But don’t expect them to change that one big idea in their policy manual—because it’s the only “big idea” they’ve actually got.

UPDATE 1: Phil Sage has some complementary comments over at the No Minister blog, concluding,

_QuoteRedistribution … is useful for solving short term problems but ultimately does not make a community more cohesive or happier; it just drags back the wealth creators, to the detriment of all.

Phil links to two excellent reports on the central thesis of The Spirit Level, which I hope he won’t mind me linking here; first, from the UK Taxpayers’ Alliance:

_QuoteBefore policymakers rush to enforce the income equality that the authors suggest is so vital to improve public health and general wellbeing, it is important that we properly scrutinise its claims. 
    The
new report published today by the TaxPayers' Alliance does just that.  The Spirit Illusion looks at whether the most important correlations established in the book can be replicated.
The findings are stark.  On almost no measure does the central claim of the Spirit Level, that income inequality decreases life expectancy, stand up to scrutiny…
    [The report’s] main point is that the most important statistical correlation between countries that the authors claim to have established – the connection they point to between life expectancy and income inequality in different industrialised nations – is simply wrong…

I recommend the report. It’s free. And the UK Policy Exchange has produced its own report on the phoney tome, Beware False Prophets, in which “Wilkinson and Pickett’s empirical claims are critically re-examined using (a) their own data on 23 countries, (b) more up-to-date statistics on a larger sample of 44 countries, and (c) data on the US states. Very few of their empirical claims survive intact.”  The hardback costs you £10 + £3p&p—but you can download the PDF free.

UPDATE 2: If you think those “dots” have moved in the two charts above, you’re right.  Chris Snowdon clarifies in the comments:

_QuoteThe two graphs are from exactly the same UN source, but are from different years. The first is from the 2004 UN Human Development Report, the second is from the 2006 report.
The reason I mention this is that, if you own a copy of The Spirit Level, take a look at references 2 and 6. Reference 2 is the 2006 report and is for their graph showing no relationship between life expectancy and GDP. Reference 6 is the 2004 report, and that's used to show there IS a relationship between life expectancy and inequality.
    Why use two different data sets? We can only speculate, but I would speculate that its because, even if you exclude places like the Czech Republic, the 2004 data fits their hypothesis better than the 2006. This, from two researchers who insist they took their data "warts and all."

UPDATE 3: Dinther sums it all up perfectly. At the end of the day, even if the research were true …

_QuoteI’d rather live a shorter life in freedom … than a longer and no doubt boring life in captivity.

UPDATE 4Spirit Level authors Wilkinson and Pickett's responded to the Spirit Level Delusion author’s 20 Questions to them. Hong Kong was excluded because, apparently, it’s not “an older, rich, developed, market economy.” Presumably because it’s a young, poor, undeveloped, communist state like Cuba?  I guess this gives you a taste of Wilkinson & Picket’s acumen.
Snowdon’s response to them is here. It’s good.
Some of the graphs from The Spirit Level Delusion are here. And Snowdon’s Spirit Level Delusion blog is here.

18 comments:

twr said...

Looks nice, but the dots have actually moved, so the two graphs aren't strictly comparable because they obviously use different source data.

Falafulu Fisi said...

I am debating this topic of The Spirit Level at No Minister with a lefty, Judge Holden, who argues a lot on that thread but provide no researches to back up his arguments despite me challenging him/her to produce some.

Swimming said...

Wonder why the life expectancy has dramatically increased for so many countries? Including Belgium and New Zealand. Even though some countries it has lowered in comparison to other countries, e.g. in one chart Belgium has lower life expectancy than Austria, in the other one it is the reverse. Oranges being compared with mandarins. Different data set?

Tim Johnston said...

Even recently deceased historian Tony Judt used those graphs in an article here:

http://www.nybooks.com/articles/archives/2010/apr/29/ill-fares-the-land/?page=1

Peter Cresswell said...

Sorry, I thought the captions made it clear: Fig 2 uses the standard Un figures-- Fig 1 are those used by the authors of The Spirit Level -- and the explanation for the difference is the responsibility of those authors.

Peter Cresswell said...

@Tim Johnston: That would be the same Tony Judt of whom David Cunliffe says "Few writers have impacted me as much."

Of Judt's latest book, Cunliffe says, "it is not a robust peer reviewed academic treatise." He got that much right.

Owen McShane said...

Actually there is some evidence to support the theory.
The failure of the Soviet Block and North Korea.

They all had or have much greater income inequality than the Western democracies.

Peter Cresswell said...

@FF: Ah, thanks for that. I'd forgotten Phil's post.

And no, I've never seen Judge Holden back up his/her assertions either.

Falafulu Fisi said...

And Colin James, seemed to have taken the bait in shown in his comment here:

James:
------
That methodology has respect. It has identified a relationship between substances, lifestyles and other influences and the incidence of various cancers. And Wilkinson and Pickett ran a lot of cross-checks: they insist the relationships between inequality and health and social problems in the rich world "are too strong to be dismissed as chance findings".

Can anyone see the naivety of Jame's patronising comment above?

This is the problem with MPs such as Robertson, Twyford , Fenton, and including political commentators like James, Trotter is that they don't read original research to see where the arguments and counterarguments lead to. They only take one side based on what they read and not being made aware of other recent available counter-arguments.

The dangerous thing here is that most politicians/lawmakers tend to listen to what James, Trotter are saying.

Christopher Snowdon said...

Great blog. I've bookmarked it.

To give a fuller answer to TWR, the two graphs are from exactly the same UN source, but are from different years. The first is from the 2004 UN Human Development Report, the second is from the 2006 report.

The reason I mention this is that, if you own a copy of The Spirit Level, take a look at references 2 and 6. Reference 2 is the 2006 report and is for their graph showing no relationship between life expectancy and GDP. Reference 6 is the 2004 report, and that's used to show there IS a relationship between life expectancy and inequality.

Why use two different data sets? We can only speculate, but I would speculate that its because, even if you exclude places like the Czech Republic, the 2004 data fits their hypothesis better than the 2006. This, from two researchers who insist they took their data "warts and all".

BTW, am I alone in being surprised to see New Zealand classed amongst the 'unequal hellhole' category in The Spirit Level? Everyone I've ever met from NZ has told me its the greatest country in the world (ditto Australia, at the risk of starting a fight). Is this just fervent patriotism for the benefit of foreigners? If it is, it's very convincing fervent patriotism.

Chris Snowdon (author of The Spirit Level Delusion)

Dinther said...

Besides the false "facts",

I rather life a shorter life in freedom aspiring and driven to do better one day, than a longer and no doubt boring life in captivity.

Peter Cresswell said...

Hi Chris, and thanks for your clarification. We're honoured by your presence. :-)

I guess most people don't realise just how perfidious so-called researchers like this really are. They haven't learned statistics to find the truth, they simply employ it to tell flat lies.

Yes, you're right. New Zealand is the best country in the world, and we'd like to keep it that way. Unfortunately, our redistributionists here are as active as they are in the UK--despite the fact that NZ's "rich" wouldn't even be the UK's "moderately well off."

goofey said...

'm sorry, but the second graph is also uses cherry-picked data and is just as unreliable. Neither graph tells us anything, even if they weren't massive over-simplifications of a complex and almost untestable hypothesis.

Falafulu Fisi said...

I just made a comment on an article by the authors of "The Spirit Level" and hopefully they (or one of them) will engage with me:

In defence of equality

I hope that the authors will reply to my post because I intend to tear them into pieces on the topic of data-analysis. This is the problem that I usually encounter, be it climate science, economics, sociology or whatever domain that data analysis is being applied. The analysts usually come up with simple statistical analysis thus to show/prove some functional relationship only for them to realize (when being alerted) that what they thought was a solid analysis is simply misleading. I had debated with a warmist named Dr. James Annan who had been a contributor to RealClimate about 2 years ago on his blog, which I exposed his simple climate statistical analysis method and shown that he was wrong (well, I showed him the proper way of doing ARMAX type modelling). Dr Annan simply couldn't refute what I was pointing out to him.

The reason was that he is an excellent climatologist (no doubt about that), but he simply lacked the depth of knowledge in numerical modelling which is completely a different field altogether. Most people think that once you're climatologist, sociologist, economist, etc, then automatically, you're expert in numerical modelling, but that's not the case.

So, I hope that the authors of "The Spirit Level" (TSL) will engage with me in their defence of their work at Prospect online magazine, so that I can expose their methods.

Most commentators on the internet who have made comments about TSL have no understanding of deep data analytics and numerical modelling, which they tend to see if it is peer review done by some professors ( Richard Wilkinson and Kate Pickett ) then they automatically assume that their work must be thorough and proper.

Christopher Snowdon I just sent you an email, because I want to ask you some questions about the TSL book.

Peter Cresswell said...

@Goofey, the more dots are added the more obvious it becomes there is not trend-- and the less tenable the 'Spirit Level' hypothesis becomes. That's Snowdon's point.

Falafulu Fisi said...

I just made a comment on an article by the authors of "The Spirit Level" and hopefully they (or one of them) will engage with me:

In Defence of Equality.

I hope that the authors will reply to my post because I intend to tear them into pieces on the topic of data-analysis. This is the problem that I usually encounter, be it climate science, economics, sociology or whatever domain that data analysis is being applied. The analysts usually come up with simple statistical analysis thus to show/prove some functional relationship only for them to realize (when being alerted) that what they thought was a solid analysis is simply misleading. I had debated with a warmist named Dr. James Annan who had been a contributor to RealClimate about 2 years ago on his blog, which I exposed his simple climate statistical analysis method and shown that he was wrong (well, I showed him the proper way of doing ARMAX type modelling). Dr Annan simply couldn't refute what I was pointing out to him.

The reason was that he is an excellent climatologist (no doubt about that), but he simply lacked the depth of knowledge in numerical modelling which is completely a different field altogether. Most people think that once you're climatologist, sociologist, economist, etc, then automatically, you're expert in numerical modelling, but that's not the case.

So, I hope that the authors of "The Spirit Level" (TSL) will engage with me in their defence of their work at Prospect online magazine, so that I can expose their methods.

Most commentators on the internet who have made comments about TSL have no understanding of deep data analytics and numerical modelling, which they tend to see if it is peer review done by some professors ( Richard Wilkinson and Kate Pickett ) then they automatically assume that their work must be thorough and proper.

Christopher Snowdon I just sent you an email, because I want to ask you some questions about the TSL book.

The Anthropology of Sound said...

The idea that statistics can be used and abused is not new. As Nietzsche said there are no facts only interpretations. What this article seems to argue is that statistics depending on one's agenda can be distorted and contorted to underpin ones own ideological position.

For the wise and humane it seems a truism to say that the more equal a society, the more content are its members. Hasn't this been proven through Western civil rights and universal suffrage movements. And isn't this the underpinning philosophy of any truly democratic society?

To argue against the underlying thesis of the Spirit level seems like a justification of authoritarian forms of capitalism that dehumanise the social conscience of people and their communities while also undermining social well being, mutual trust and sustainability.

How would being in a society where 1 percent of the population controls 80 percent of the wealth be good for anyone except for that 1 percent. Apart from by projecting the illusion that you too can be in that 1 percent if you just put your head down and work hard. Isn't that why they call it the American dream, because you have to be asleep to believe it.

Yet we are still so keen to bow down and pray to such an obviously flawed system which does nothing to better the plight of the majority of people around the world. Its a system of exploitation, without the crassness of overt slavery. Anyone who looks at international trade and the role of Transnational Corporations's in the third world knows this.

Tim Johnston said...

Quite a few myths in there AoS, but I'm hoping PC can demolish them more eloquently than I. However, to challege one particular point, assuming that that 1% want to do anything with their wealth other than play golf or sit and plot how to take over the remaining 20% they don't 'control', they have to make life at least tolerable for everybody else. They have no vested interest in making the 99% miserable, after all, who will buy the goods they produce if there are only landless peasants?

such extremes are more characteristic of medieval societies - and some third world ones - than a truly Capitalist society. If you're implying that such a system is somehow unnatural, then you have to account historically for how societies evolved in the main, which was largely authoritarian and hierarchical.

In a capitalist society, in contrast, were 1% of the population in control of 80% of the wealth, one would have to investigate corruption, nepotism and the political structure to find the causes of such 'inequality', and not the economic system itself.