How to lie with statistics

Environmentalists often draw ludicrous conclusions from their beloved statistical models. A recent prominent example is the estimate from the Global Humanitarian Forum, an organisation led by Kofi Annan (a former secretary-general of the United Nations), that global warming is causing more than 300,000 deaths and about $125 billion (£77 billion) in economic losses each year. The report is also endorsed by, among others, Jeffrey Sachs.

But the estimates are described as a “methodological embarrassment” by Roger Pielke, a political scientist at the University of Colorado, Boulder, who specialises in disasters, in a recent blog post. He points to several flaws in the model including:

• The stochastic nature of extreme weather events. In other words it is impossible to say for sure that an extreme weather event, such as a hurricane, is the result of climate change. It may be that climate change makes more events more likely but they would probably happen in any case without it.

• A shortage of good quality data. For sweeping conclusions to be justified they must be based on better data than is generally available.

• The role of various other potential factors that act in parallel and interact. For example, with economic development it may be that there are more buildings to destroy in a hurricane. But it does not follow that the physical force of hurricanes has necessarily become more destructive than in the past. .According to Piekle: “the increase in disasters observed worldwide can be entirely attributed to socio-economic changes. This is what has been extensively documented in the peer reviewed literature, and yet — none of this literature is cited in this [Global Humanitarian Forum] report. None of it!”

Piekle has also written a critique of similar methodological flaws (PDF) in the Stern Review on the economics of climate change.