Typically, the distribution of macroparasites over their host population is highly aggregated and empirically best described by the negative binomial distribution. For parasitologists, this poses a statistical problem, which is often tackled by log-transforming the parasite data prior to analysis by parametric tests. Here, Ken Wilson and Bryan Grenfell show that this method is particularly prone to type 1 errors, and highlight a much more powerful and flexible alternative: generalized linear modelling.