The ‘hockey stick’ affair

The use and abuse of a single graph to justify action on climate change shows the need for healthy scepticism.

AW Montford

Topics Science & Tech

From the moment it appeared in 1999, it was clear the ‘hockey stick’ graph was going to be very, very important. The graph, which appeared in a paper by US climatologist Michael Mann and others published in 1999, is a reconstruction of global temperatures over the past thousand years. Since for most of that period there were no weather stations monitoring temperature, a variety of proxy temperature measures, like tree rings, needed to be used.

Two things are striking about the graph. Firstly, the period from the year 1000 right through to the mid-nineteenth century shows relatively steady temperatures, despite the widespread belief that there was a ‘medieval warm period’ from around about 950 to 1250 AD. Secondly, the temperatures in Mann’s graph lurch sharply upwards – hence the ‘hockey stick’ nickname – particularly during the twentieth century, suggesting that the world had been getting sharply warmer and would continue to do so.

The ‘hockey stick’ graph

Within a week of the graph’s publication there was an article about it in the New York Times. This was pretty amazing considering Michael Mann, the lead author of the paper, had only received his PhD a few months before. A couple of years later, it turned up in the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), appearing five or six times, full size, full colour. It was fairly clear that the hockey-stick graph was important.

There was a BBC report some years later in which the reporter explained that it was really quite hard to overestimate how important the graph has been. In Canada, for example, the government even sent a leaflet out to every home in the country showing the conclusions of the graph: that current warming temperatures were historically unprecedented. That indicates how important, how influential this piece of research was. Indeed, it has been cited more than possibly any other paper in the field.

But then, in 2002/2003, a Canadian geologist called Stephen McIntyre came on to the scene. Having been a recipient of one of the Canadian government’s leaflets, he just thought the graph looked a bit, well, odd. So he went through the original research, and because of its rather dramatic shape showing steady temperatures for centuries and centuries and then a sudden lurch upwards in the twentieth century, McIntyre thought this just all seemed a bit dodgy.

This was partly due to McIntyre’s professional background in mining: in mining, the hockey-stick graph is a familiar phenomenon. It is a way for mining companies to encourage people to invest in them, so it probably set his alarm bells ringing.

Now McIntyre was to find two things wrong with Mann’s hockey-stick graph. The first was that the data behind the graph was inappropriate. Most temperature reconstructions use a very small number of tree-ring series. Mann’s hockey stick was rather different in that he used a much larger number, but the important tree-ring series within it were from one particular kind of tree called a bristlecone pine which grows for hundreds, possibly thousands of years and can be found in western America.

The problem, however, was that it was known that this kind of tree showed a growth spurt in the twentieth century, which meant that the pattern of the tree rings effectively had a hockey-stick shape. This was awkward since it was widely acknowledged that this spurt was not being driven by climate. Remarkably, one of Mann’s co-authors had even admitted this in a later paper, stating that the twentieth-century growth spurt was a mystery, but it was not climatic. So these tree rings were known to be problematic.

Following further research in 2004, McIntyre discovered a fragment of code from Mann’s statistical method – principal components analysis – on one of Mann’s own websites. There was an error in it, the effect of which was to overemphasise the bristlecone pines. So, while you had hundreds of tree-ring series, the only ones that mattered were the ones that were hockey-stick shaped.

This meant that it didn’t matter what data you put into Mann’s algorithm, if there was one series within it that had a hockey-stick shape there is a strong chance that, depending on the number of other series, a hockey-stick graph would emerge as the result. The algorithm was heavily weighted in favour of hockey sticks. It effectively disregarded any data that conflicted with, or contradicted, the hockey-stick finding.

It is now agreed – and expert panels have since looked at this – that the bristlecone pines are inappropriate as a proxy measure of temperature and that the statistical methodology used was biased.

Still, the argument that is now given as to why the hockey-stick graph was okay is that the other temperature reconstructions that have been created give broadly the same shape. It is debatable just how similar that shape is – many of them aren’t hockey-stick shaped at all, showing very high temperatures in the medieval period. They are more U-shaped, if you like. But the other factor here is that all these other temperature reconstructions use the bristlecone pines as well. Now, if you’ve got bristlecones in amongst a small number of tree-sing series, then you will invariably get a hockey-stick result.

So, to summarise: Mann’s method used a very large number of tree-ring series, but his statistical method ignored anything that was not hockey-stick shaped. In other reconstructions, fewer sets of data are used, but as a result the dubious bristlecone pine data has a greater impact on the final result.

McIntyre put his case against the hockey-stick graph down on paper in 2005, leading to quite a furore. The argument continues to this day. The IPCC is still trying to stand by Mann’s work, probably because it oversold the hockey-stick graph in the past and now finds it quite difficult to step back from it. It is even included in the Fourth Assessment Report of 2007. The IPCC is maintaining this argument that yes, there may have been problems with the statistics and with the data, but it gives broadly the same answer as the other temperature reconstructions. My impression is that the IPCC would like to drop it gently now without ever admitting what it got wrong.

After Climategate

In the wake of Climategate in late 2009, when a large number of emails and data from the Climate Research Unit (CRU) at the University of East Anglia found their way into the public domain – leading to some fairly embarrassing revelations about the working methods and attitudes of both CRU researchers and their collaborators elsewhere – the climatology community is in a bit of a state of flux. There are people within that community who are maybe just shifting position a little and trying to put a bit of distance between themselves and the more advocacy-based scientists. Whether that leads to an eventual disowning of the hockey-stick graph itself remains to be seen.

I think from this point on we will hear far less about scientific consensus from the climatology community and far more about uncertainty and dissenting views. This is not to say we have heard the end of global warming – there’s too much money floating around for people just to drop it. Climatology has had huge amounts of money flung at it while the big energy companies have investments in renewable technology based on farming enormous subsidies.

These financial pressures are key to distorting the debate about climate. The problem for many climatologists is that they cannot come out and say that global warming isn’t a problem anymore because so much of their funding depends upon it. This is also one of the problems that the IPCC has: the very people who can answer the question on the future direction of climate also depend on the answer to that being that ‘yes, it’s a problem’ to continue receiving funding. It is a difficult problem, and it is one that you get when you have government-funded science.

As for McIntyre, although he is demonised as a Big Oil-funded troublemaker by green activists and advocacy-based scientists, he’s proven very difficult to attack because of his position: he does believe global warming is real, and he does think it is a problem. At the same time, he’s just the type of guy that is going to go out and find the truth no matter where it lies. There is a story that upon hearing the argument that if the hockey-stick graph is wrong, then the planet could be much more sensitive to rises in CO2 levels than previously thought, and therefore global warming could be even worse, McIntyre just shrugged his shoulders and said: ‘If that’s the answer, then that’s the answer and we’ll just have to find it out.’ His integrity as someone who will pursue the truth wherever it may lie is very hard to question.

In the aftermath of McIntyre’s work and Climategate, there is a growing middle ground where people have decided that everyone bashing one another just isn’t helping to clarify matters. What we have to do is follow the science. And that means that we have to be open, we have to be questioning, and scientists have to engage with people outside their own fields.

This article is based on an interview by Tim Black.

AW Montford maintains the Bishop Hill blog and is the author of The Hockey Stick Illusion. (Buy this book from Amazon(UK).)

To enquire about republishing spiked’s content, a right to reply or to request a correction, please contact the managing editor, Viv Regan.

Topics Science & Tech


Want to join the conversation?

Only spiked supporters and patrons, who donate regularly to us, can comment on our articles.

Join today