Comparing Dynamic Consumer Choice in Real and Computer-simulated Environments

Abstract

Actual supermarket purchases made by consumers over a seven-month period are compared with the choice decisions collected in one sitting in a laboratory simulation. The laboratory simulation was designed to mimic the original market environment. Although there were systematic biases in predictions of true purchase behavior from the simulated data, the “compressed simulations” were reasonably valid in predicting market shares and promotion sensitivity for brands across consumers.