The purpose of this short note is to discuss the split sample method recently proposed by Decrouez and Hall (2014). The
split sample method reduces the oscillations of the coverage by splitting the sample in two. It is applicable to most existing
confidence intervals for parameters of lattice distributions, including the binomial and Poisson distributions. For simplicity,
we will in most of the remainder of the paper limit our discussion to the binomial setting, with the split sample method
being applied to the celebrated Wilson (1927) interval for the proportion p. Our conclusions are however equally valid for
other confidence intervals and distributions.