SS's critique centres on two of the most pressing issues in quantitative methods: sample and analysis. In short, the sample critique is that WP 'cherry picked' certain countries which fitted their hypothesis. Snowdon, for example, points out that they should and could have included South Korea, Slovenia and the Czech Republic in their analysis, yet didn't. The method critique is that WP rely on simplistic bivariate techniques (i.e. between 1 outcome variable (e.g. homicide) and 1 explanatory variable (inequality)). As my own stats teacher told us, social scientists very quickly move on from two-variable analyses to more complex, multivariate analyses which can explore the impact of multiple explanatory variables.
Thus, to an extent, I think that The Spirit Level is vulnerable to these accusations and that SS's criticisms are reasonably founded.
But - and this is a big but - I would then argue that the whole Spirit Level debate completely misses the point. As WP say in the RSA video, their book is more of a precis of hundreds of papers and decades worth of research which, time and again, has demonstrated a significant relationship between inequality and social ills. This is where the real evidence of the inequality effect lies and, in their attempt to communicate this to a much wider, non-epidemiological audience, WP have perhaps relied on shaky statistical ground.
The main challenge for WP's antagonists - and something I haven't seen them do - is to take on all the other evidence which suggests that inequality makes societies worse. As a book on its own, The Spirit Level is an easy target for conservatives and liberals who are ideologically comfortable with high social inequalities. What I would really like to see from them - and what is a much bigger challenge - is to engage with the much wider, and much more powerful, arguments which back up WP's more contentious piece of work.