OPINION20 August 2010

The difficulty of proving your point

The debate over the bestselling book The Spirit Level is a reminder of how much people’s view of (or interest in) the evidence can be influenced by what they already think.

In the preface to their 2009 book The Spirit Level, Richard Wilkinson and Kate Pickett use data from the World Bank, WHO, the UN and the OECD to argue that more equal societies do better in almost every respect than more unequal ones. The prevalence of problems as diverse as infant mortality, imprisonment rates, teenage pregnancy, murder and obesity is, they say, statistically linked to income inequality, and the book is filled with graphs illustrating these correlations.

But their conclusions have come under attack from right-wing think tank the Policy Exchange, as well as the Taxpayer’s Alliance. In an article for the Guardian the Policy Exchange’s Natalie Evans says the claims they make are not supported by the statistics. Some of the conclusions, she says, rely on one or two extreme cases to make a case, while others result from “clusters” of countries that tend to show similarities, such as the Scandinavian or English-speaking nations.

Of course, a heated debate has ensued, not least on the comments thread of Evans’ article. What’s interesting is how little the discussion focuses on the facts. The problem with topics like this is that some people are simply inclined to believe one thing or the other, regardless of evidence. Many people (particularly Guardian readers, we suspect) hold the book’s central assertion that inequality is bad for us all to be a self-evident truth. They will continue to believe it with or without statistical analysis to back it up, and they are happy to say so. Similarly, one suspects that the likes of the Policy Exchange and the Taxpayer’s Alliance are inclined to believe that a society where people are free to enrich themselves is better because, well, that’s what they believe.

If research is about making an argument or telling a story based on evidence, it’s worth considering just how much people’s view of (or interest in) the evidence can be influenced by what they already think.