The Consumer Price Index is arguably one of the most important statistics produced by the U. S. Government. With some 260 strata, the CPI represents a complex data collection problem. The Bureau of Labor Statistics is undertaking an experiment to make the CPI more accurate by incorporating point of purchase (POP) scanner data. The experiment began several years ago with scanner data on breakfast cereals. In this talk, we will review a little bit of theory on price indices and continue with an exploratory data analysis of the cereal data indicating how scanner data might adversely affect the CPI.