We present the results of a 3 x 4 x 4 hierarchical experiment using high-temperature gel permeation chromatography (GPC) to analyze the molecular weight distribution (MWD) of replicate ethylene polymerizations with metallocene/methylalumoxane (MAO) catalysts. Our aim was to decompose the variance imparted to the MWD by a) GPC analysis itself (sigma(T)(2)), b) heterogeneity in the polymerization reactor (sigma(S)(2)), and c) random fluctuations in conditions between polymerizations (sigma(P)(2)). This variance component analysis was performed on several measures used to characterize the centre, breadth and skewness of MWDs. Using analysis of variance tables, significance tests and contrast tests, we were able to reveal statistically significant differences in the MWD of polyethylene synthesized in replicate polymerizations at identical conditions, as well as differences between samples drawn from the same polymerization. By comparing the results of this analysis using different measures of MWD, we were able to draw conclusions regarding the suitability of various indicators conventionally used to characterize the MWD. The techniques discussed are very general and applicable to almost any experimental scenario.