Decomposition behavior in aggregated data sets

One way to aggregate data is to combine several sets with the same structure, but no overlap in their ranges of values — for instance, aggregating prices before and after a period of hyperinflation. Looking at nonparametric tests on three ‘items’, we compute the relation of the decomposition of the underlying voting profiles of such aggregated sets to those for the original data. We focus on the Basic components, including examples of ‘pure Basic’ sets, computed using Sage. This yields several interesting results about consistency of nonparametric tests with respect to this kind of aggregation, and suggests types of non-uniformity which are not detected by standard tests.