Specifically, the more aggregated the data, the more normal the analytical relationships tend to be, thus increasing statistical power in the analysis. However, normality comes at the price of missed detections at the detail level. So, the choice of aggregation levels has to be made on a case-by-case basis taking into account the inherent characteristics and risk level of the underlying transactional data. These challenges introduced by the aggregation become more evident in a continuous audit of Big Data, and they present topics for
much future research.