The Spirit Level Debated

Thanks to Etienne for sending the video link.

I read the Spirit Level by Richard Wilkinson and Kate Pickett while I was doing an internship in London over the winter. The debate about this book is very interesting both from an ideological and statistical point of view, but it is particularly interesting when ideology and statistics interact. Below is a debate between the Spirit Level authors and Christopher Snowdon author of The Spirit Level Delusion and Peter Saunders author of Beware False Prophets.

The debate is interesting because neither side are statisticians, rather highly knowledgeable applied users, and both are using statistical arguments to refute or support their claims. On the anti-spirit level side, there is indeed a principle in statistics that one should remove outliers because they can do funny things to your data and cause model assumptions to fail. On the pro-spirit level side, there is a ‘real world’ principle which suggests that you should do your best to represent the real world and leave outliers in if they are plausible. In my interactions with statisticians I have found them generally to be quite philosophical about the nature statistics and indeed their work. I wonder if there is a required philosophy of statistics course for all PhDs? Probably. Anyway, the point is that to the non statistician the rules that both sides talk about seem hard and fast, and indeed they are presented as if they were requirements to be met before anything could be inferred from the data. However, many of these rules are not hard and fast and in the end choices are made about including outliers, variables and representing the real world that may or may not be agreed upon by other researchers. It’s a constant debate. I do however agree with Richard Wilkinson who suggests that the point of the book is not too show robust evidence about inequalities but rather to present a general theory of inequalities. The published literature is more robust statistically than a book written for a much broader audience so better to critique that literature than a few simply models. That being said they could have presented their evidence and the most robust evidence in the Spirit Level. Watch the video and tell me what you think.

Submit a Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>