Campus News

Published by Communications and Public Affairs (519) 824-4120, Ext. 56982 or 53338

News Release

January 27, 2005

U of G prof says climate fears based on flawed statistics

A University of Guelph professor is part of a team that has found a computer programming error in the temperature history graph that’s been used to demonstrate global warming and the need for the Kyoto Protocol, which comes into effect Feb. 16.

Economics professor Ross McKitrick and his coauthor, Stephen McIntyre, have uncovered fundamental flaws in the “hockey stick graph” used by the Intergovernmental Panel on Climate Change to argue that the 1990s were the warmest decade of the millennium. Their research is appearing in Geophysical Research Letters and the interdisciplinary science journal Environment and Energy.

The widely used hockey stick graph, which shows average temperatures since AD 1000, got its name because it looks like a hockey stick lying on its side; average temperatures remain steady for 900 years and then shoot up in the 1900s, creating the blade of the hockey stick. Since the graph was published in Nature and Geophysical Research Letters by University of Virginia professor Michael Mann in the late 1990s, it’s been the basis for showing the effects of global warming.

“We uncovered several flaws in the hockey stick study, notably a computer programming error that overemphasized ‘hockey stick’ shapes in the climate data and downplayed everything else,” said McKitrick.

Because reliable temperature measurements have been available only since about 1850, Mann’s study used the dimensions of tree rings to calculate earlier average temperatures. But McKitrick and McIntyre found that each tree ring series was transformed incorrectly by subtracting the 1902 to 1980 mean, rather than the mean over the whole length of the series. “As a result, the variance of hockey stick-shaped tree ring data – which makes up only a small portion of the database – gets inflated, and a subsequent part of the program puts all the weight on it, effectively ignoring the rest of the data,” said McKitrick.

The researchers applied Mann’s method to lists of trendless random numbers with the same pattern of variability as tree ring data. “His method yielded a hockey stick-shaped pattern 99 per cent of the time,” said McKitrick. “This erroneously suggests that the data are dominated by hockey stick patterns, when, in fact, the data we fed in are simply machine-generated noise with no underlying trend.”

When McKitrick and McIntyre corrected the analysis and recalculated average temperatures using the same tree ring data used by Mann, they found that the 20th century still experienced an increase in average temperature, but the change is well within the bounds of natural variability.

Ross McKitrick

Department of Economics
(519) 824- 4120, Ext. 52532, or

Stephen McIntyre
(647) 287-5425

For media questions, contact Communications and Public Affairs: Lori Bona Hunt (519) 824-4120, Ext. 53338, or Rachelle Cooper, Ext. 56982.

Email this entry to:

Message (optional):