We respond to the lies and distortions of mainstream economic research by Wilkinson and Picket.
I typically don't approve of appeals to authority, but since Wilkinson and Picket are masters of playing this game, and falsely market themselves to the media and broader public as representatives of mainstream research, I have no choice but give them a dose of their own medicine.
Some highlights:
• Wilkinson and Picket, pressed into a corner, now suddenly claim that Nobel prize winner James Heckman’s research supports their conclusions! When I asked professor Heckman about this, sending him the link to Wilkinson and Picket’s text, James Heckman answered bluntly:
“This is a misreprentation of my work”.
How dare these two charlatans casually distort the work of one of the leading economist in the world and put the authority of his name and research behind their unsubstantiated claims?
If they are willing to lie about even James Heckman himself, are we supposed to trust them regarding the "200 studies" that they assert demonstrate a causal link between income inequality and health?
• Wilkinson and Picket further claim that the Princeton professor Angus Deaton is not an expert and that, consequently, his review study in the Journal of Economic Literature, which arrives at the opposite conclusion to that of Wilkinson and Picket, is not worth citing. This is despite the fact that this paper appeared in a much more prestigious journal, with 3 times the impact factor, and has been cited twice as often by other scholars as Wilkinson and Picket's own work.
Leading economists consider Deaton enough of an expert to ask him to write the review article in the flagship review journal of the profession. According to Ideas, of tens of thousands active economists, Deaton is the 56th highest ranked economist on the planet.
Wilkinson and Picket (from the universities of Nottingham and York), give themselves the right to dismiss mainstream economic research when it fails to support their conclusions, not even acknowledging the existence of this more rigorous and objective work. Wilkinson and Picket are thereby failing to comply with the basic rules of scientific conduct. The naïve readers that trust Wilkinson and Picket are being fooled about the state of mainstream economic research.
• I am a blogger. When I wanted to measure the correlation between patents and inequality, I went to the homepage of the World Intellectual Property Organization and downloaded the data. The correlation between per capita patents and inequality was not statistically significant, and went in the opposite direction of what Wilkinson and Picket wrote. Wilkinson and Picket claim to represent serious science. It now turns out that their source for patents is the amateur internet site Nationmaster, not the original source as they initially claimed!
Wilkinson and Picket, who claim to present science, have lower data standards in their research than a blogger they are debating with does when writing a blog entry about their work.
• When defending their claims that income inequality caused higher mortality, they over and over claimed that there are "200" (or sometimes 300) studies that prove this causal link. If you look at their actual review study, you will notice that the source of this claim of external validity are 2 review articles. One of the two is by the paragons of scientific virtue Wilkinson and Picket themselves.
• The second study (Kondo et al. 2009) of 28 papers explicitly makes clear that is not looking at causation, but “association”. In economic jargon you write "association" when you do not feel confident that you have established a causal link. Not all 28 articles in the Kondo et al. study even find a statistically significant link, with the authors warning that the results should be interpreted with
"caution given the heterogeneity between studies, as well as the attenuation of the risk estimates in analyses that attempted to control for the unmeasured characteristics of areas with high levels of income inequality."
• The Wilkinson and Picket paper is not about mortality at all. They themselves write “We compiled a list of 155 published reports of research on the relation between income distribution and population health.”
Of the 155 papers, they find “A tally of numbers showed 88 wholly supportive Analyses” and 44 which they call partially supportive (mixed evidence).
They look at 155 papers, but even they actually only find clear results for 88. Yet they keep saying they have “200” papers, leaving the reader to believe 200 papers support their claims. In the Wall Street Journal they went further, and gave the impression they had “300” supporting articles. The 300 figure is the number of citations they have in their book, which is obviously very far from 300 papers all supporting their claims...
• Here is the killer: Read the Wilkinson and Picket criteria for selecting a paper that they later present as having proven a causal link:
“We classified them as wholly supportive if they reported only statistically significant associations between greater income inequality and poorer population health”
A statistical “association” is not evidence of causality. Association just means that two variables are statistically related, not that one is causing the other. This is what scientists write when they do not have enough evidence to establish causation convincingly.
Thus the Wikipedia page for association for example writes:
“In quantitative research, the term "association" is often used to emphasize that a relationship being discussed is not necessarily causal”
Adding controls does not magically make something causal, unless your model specification is correct and you are actually control for all determinants of the dependent variable that are correlated with the independent variable of interest. This is very difficult with complex variables such as health and inequality.
So Wilkinson and Picket are using double standards. In their published research they picked studies that did not have (or even claim to have) established a causal link, but in their representation of this research to the broader public they pretend that there is evidence for causation between inequality and poor health.
• When we asked for a paper showing a causal link between life expectancy and inequality,, they linked to study from China. The study looks at the statistical association between the Gini coefficient and “self-reported health” in some Chinese provinces.
First of all, we asked about life expectancy, but you if you look carefully you will see that this study is about self-reported health. The dishonesty never ends with these two people.
Second, Wilkinson and Picket understand too little about modern methodology in social sciences to realize that this is also a correlational study!
I will again repeat the standard criteria in modern social science: When you have variables related in complex ways, such as inequality and health, running a regression is just not enough. You either have to have to model the exact relationship between health and inequality, including all possible confounding variables, or use experimental or at least quasi-experimental methods to try to tease causal patterns out of the observational data.
Wilkison and Picket go from “200” studies to failing to cite EVEN ONE study that established a causal link between income inequality and life expectancy.
The market for ideas like other markets operates by the laws of supply and demand. There is clearly a large demand for the claims of Wilkinson and Picket, that equality is not only morally preferable (which I agree with), but scientifically proven to cause all sorts of good outcomes.
Because of this demand from politicians and journalists, Wilkinson and Picket were tempted to write their unscientific book. This book ignores more serious work in economics, distorts evidence, cherry picks data, relies on correlation rather than causation, prefers data from amateur sites on the internet to the original source data, and reports relationships which turn out not to be statistically significant.
Journalists, leftist, and Sweden’s Social Democratic party leader Mona Sahlin, trusted Wilkinson and Picket. One reason is because they wanted to believe that their message was true. The other, more problematic reason is that people have been taught to trust scientists, and Wilkinson and Picket have falsely marketed their shoddy ideological work as science.
Wilkinson and Picket are abusing the trust of the public in academia, which is their perhaps most serious crime, far worse than the offense of simply being wrong. The lesson is, I believe, that journalists and politicians need to be more careful of false prophets, especially if they market themselves as "scientists".
Friday, July 30, 2010
Tuesday, July 20, 2010
Japan's 2010 House of Councillors election: another "twisted Diet" coming up?
Guest Post by Manuel Alvarez RiveraElection Resources On The InternetWithout doubt, the Democratic Party of Japan (DPJ) has had a decidedly rough ride in power in the ten months since its historic House of Representatives election victory, which brought to an end more than half-a-century of nearly uninterrupted Liberal Democratic Party (LDP) rule. Last month the Japanese public was treated to the
Thursday, July 15, 2010
Japan's turn to China as a primary export market
According to the Financial Times, "China replaced the US as Japan’s biggest export market last year(2009)".(1) Here's a chart from RIETI showing the relative shares of Japan's total exports:It's remarkable that the proportion of exports to the USA has practically halved in ten years. Figures from Japan's Ministry of Finance show that Japan actually had a trade deficit with China in 2009,
Monday, July 12, 2010
Wilkinson and Pickett misrepresent research again
I want to point to an exchange from The Spirit Level authors Richard Wilkinson and Kate Pickett to again illustrate the lack of scientific rigor from these two authors.
The Spirit Level claims that inequality within American states causes lower life expectancy.
When Princeton Professor Angus Deaton and UIUC professor Darren Lubotsky controlled for racial composition, this effect vanished.
University of Massachusetts Amherst scholars Ash and Robinson wrote a reply, saying that they could find an effect, with a certain specification.
Deaton and Lubotsky responded:
“Ash and Robinson… consider alternative weighting schemes and show that in one of our specifications, in one data period, and with one of their alternative weighting schemes, income inequality is estimated to be a risk factor. All of our other specifications, as well as their own preferred specification, replicate our original result, which is supported by the weight of the evidence. Conditional on fraction black, there is no evidence for an effect of income inequality on mortality.”
Angus Deaton is not just any guy. Unlike Wilkinson and Pickett, he is a scientific heavy-weight. When he came to the University of Chicago the last time to speak about econometric methodology, I saw at least 3 Nobel Prize winners in the audience.
Furthermore, as any economist knows, with multiple variables and no clear model, you can data-mine any result you want in 1-2 specifications. If you have to use correlations, what matters is the weight of the evidence.
So how do Wilkinson and Pickett present this exchange to their non-academic , trusting audience?
They simply write that Ash and Robinson find a result, neglecting to even mention the Deaton and Lubotsky study - the one actual researchers take seriously, the one that has 145 citations in google scholar. The Ash and Robinson study in comparison has 1 (sic) citation (although in fairness it is 6 years newer).
If you only read and trust Wilkinson and Pickett, you would not even know the Deaton and Lubotsky study even existed!!!
Wilkinson and Pickett have weak evidence in terms of causation, so they rely a lot on falsely claiming scientific consensus. But if you read the articles, the truth is that the field of research they are engaged in is best described by a consensus that the inequality and health have not been convincingly established to be causally connected, a lack of clean identification, and with the weight of the evidence going against Wilkinson and Pickett.
What academics know, and the broad public does not, is that if you lack a clear model, lack clean experiments, and have multiple variables, have unobservable variables, and when what you are studying can be suspected to be endogenously related (such as health, race and income inequality), you can almost ALWAYS establish a result in some specification, through sufficient data-mining.
In this situation it is very important to rely on the best studies, rely on the weight of the evidence and model averaging, and if you theoretically have reason to belive some varible may be important (such as race), to include the control if possible.
But the scientific standard that Wilkinson and Pickett use is that if someone somewhere finds a correlation in one specification (that they like), that is the end of the story! Proof! Scientific Evidence!
They can ignore all the other research, even if written by much more senior researchers, in more prestigious journals, with better methodology, and go straight to the naïve public and claim that science has proven Wilkinson and Pickett’s ideology true.
Journalists should know this the next time they read Wilkinson and Pickett claim to objectively present scientific evidence.
The Spirit Level claims that inequality within American states causes lower life expectancy.
When Princeton Professor Angus Deaton and UIUC professor Darren Lubotsky controlled for racial composition, this effect vanished.
University of Massachusetts Amherst scholars Ash and Robinson wrote a reply, saying that they could find an effect, with a certain specification.
Deaton and Lubotsky responded:
“Ash and Robinson… consider alternative weighting schemes and show that in one of our specifications, in one data period, and with one of their alternative weighting schemes, income inequality is estimated to be a risk factor. All of our other specifications, as well as their own preferred specification, replicate our original result, which is supported by the weight of the evidence. Conditional on fraction black, there is no evidence for an effect of income inequality on mortality.”
Angus Deaton is not just any guy. Unlike Wilkinson and Pickett, he is a scientific heavy-weight. When he came to the University of Chicago the last time to speak about econometric methodology, I saw at least 3 Nobel Prize winners in the audience.
Furthermore, as any economist knows, with multiple variables and no clear model, you can data-mine any result you want in 1-2 specifications. If you have to use correlations, what matters is the weight of the evidence.
So how do Wilkinson and Pickett present this exchange to their non-academic , trusting audience?
They simply write that Ash and Robinson find a result, neglecting to even mention the Deaton and Lubotsky study - the one actual researchers take seriously, the one that has 145 citations in google scholar. The Ash and Robinson study in comparison has 1 (sic) citation (although in fairness it is 6 years newer).
If you only read and trust Wilkinson and Pickett, you would not even know the Deaton and Lubotsky study even existed!!!
Wilkinson and Pickett have weak evidence in terms of causation, so they rely a lot on falsely claiming scientific consensus. But if you read the articles, the truth is that the field of research they are engaged in is best described by a consensus that the inequality and health have not been convincingly established to be causally connected, a lack of clean identification, and with the weight of the evidence going against Wilkinson and Pickett.
What academics know, and the broad public does not, is that if you lack a clear model, lack clean experiments, and have multiple variables, have unobservable variables, and when what you are studying can be suspected to be endogenously related (such as health, race and income inequality), you can almost ALWAYS establish a result in some specification, through sufficient data-mining.
In this situation it is very important to rely on the best studies, rely on the weight of the evidence and model averaging, and if you theoretically have reason to belive some varible may be important (such as race), to include the control if possible.
But the scientific standard that Wilkinson and Pickett use is that if someone somewhere finds a correlation in one specification (that they like), that is the end of the story! Proof! Scientific Evidence!
They can ignore all the other research, even if written by much more senior researchers, in more prestigious journals, with better methodology, and go straight to the naïve public and claim that science has proven Wilkinson and Pickett’s ideology true.
Journalists should know this the next time they read Wilkinson and Pickett claim to objectively present scientific evidence.
Sunday, July 11, 2010
Wall Street Journal article
The day before yesterday the Wall Street Journal published an article co-written by me about the book Spirit Level. It is my brothers third, my first, in the journal.
I will be visiting my parents in western Sweden, so postings will continue to be light.
I will be visiting my parents in western Sweden, so postings will continue to be light.
Saturday, July 3, 2010
The Fiscal Connection
A recent poll by Pew shows that the public (who wants to reduce deficits) are against both tax increases and against most spending cuts.
This does not surprise me. In the public finance literature, it has been known for a long time that the public by large does not make the “fiscal connection”. This means they do not seem to be fully aware that taxes, spending and deficits are virtually a zero sum game, unless you very clearly remind them. Most people are not trained economists.
Of course, it is also possible in this particular poll that there are other spending cuts not specified in the poll that people actually support (such as cuts in public employee benefits). Also tax collection can be made more efficient, through lower rates and a more broad tax base.
But more generally, the implication of the lack of a fiscal connection is that polls that only ask about one part of the fiscal equation cannot be trusted. The sum of people who want tax cuts, spending increases and deficit reduction is far above 100%!
Even in Sweden, support for tax cuts regularly poll above 60-70%. Some individuals interpret the question narrowly as ‘would you like your personal taxes to go down’? Since we have self-serving bias, most of us, including those who want a larger public sector, think we are over-taxed relative to others.
But naïve Economic Conservatives that run on this agenda lose, because the necessary implication of the tax cut is reduction in spending, or larger deficits, alternatives not stated in the poll.
Meanwhile, leftist groups can regularly show support for new programs by simply asking “do you support more money for X”?
One solution is to compare answers over time and across demographical groups, but giving little weight to the raw percentage itself. Another way around this is to force people to choose, such as the question “Generally speaking, would you say you favor smaller government with fewer services, or larger government with more services?”
When asked that way, when asked this way 58% of Americans currently choose smaller government and 38% more services. But even here deficit reduction is not offered as an alternative.
Don’t trust any poll results about taxes, spending and deficits that do not force people to make the fiscal connection.
This does not surprise me. In the public finance literature, it has been known for a long time that the public by large does not make the “fiscal connection”. This means they do not seem to be fully aware that taxes, spending and deficits are virtually a zero sum game, unless you very clearly remind them. Most people are not trained economists.
Of course, it is also possible in this particular poll that there are other spending cuts not specified in the poll that people actually support (such as cuts in public employee benefits). Also tax collection can be made more efficient, through lower rates and a more broad tax base.
But more generally, the implication of the lack of a fiscal connection is that polls that only ask about one part of the fiscal equation cannot be trusted. The sum of people who want tax cuts, spending increases and deficit reduction is far above 100%!
Even in Sweden, support for tax cuts regularly poll above 60-70%. Some individuals interpret the question narrowly as ‘would you like your personal taxes to go down’? Since we have self-serving bias, most of us, including those who want a larger public sector, think we are over-taxed relative to others.
But naïve Economic Conservatives that run on this agenda lose, because the necessary implication of the tax cut is reduction in spending, or larger deficits, alternatives not stated in the poll.
Meanwhile, leftist groups can regularly show support for new programs by simply asking “do you support more money for X”?
One solution is to compare answers over time and across demographical groups, but giving little weight to the raw percentage itself. Another way around this is to force people to choose, such as the question “Generally speaking, would you say you favor smaller government with fewer services, or larger government with more services?”
When asked that way, when asked this way 58% of Americans currently choose smaller government and 38% more services. But even here deficit reduction is not offered as an alternative.
Don’t trust any poll results about taxes, spending and deficits that do not force people to make the fiscal connection.
Subscribe to:
Posts (Atom)