Bias in consumer panel and store audit data

Abstract

Sample survey data such as consumer panel data and store audit data may deviate remarkably from each other and from internal secondary data. These deviations are called data biases. In this study we use information regarding some brands of a fast moving nondurable consumer good to show that the data bias may lead to a large bias in parameter estimation, implying wrong indications for the use of marketing decision variables. A formal analysis explaining the coverage factor - measuring data bias - by a number of exogeneous variables can lead to a reduction in the case of consumer panel data. A similar reduction of the data bias in store audit data can be reached through the use of a reweighing procedure.