Excerpts from the report Introduction: This report summarizes the results of a study of consumer purchases of seven food items that was conducted jointly by the Department of Agriculture and the Bureau of the Census in May 1955. The study was primarily methodological and was designed to measure the variations in response when questions about the purchases of selected food items were asked in different ways. More specifically, the study attempted to measure the magnitude and the direction of the variation in response associated with (a) varying the period of recall of the purchase of selected food items, and (b) changing the question wording. Most research people working on problems of collecting primary data about food or other product disappearance have long been aware of the existence of response errors. Accurate measurement of the consumers’ actions has been needed for many years; with the expansion in research about the consumer market, the problem has become more pressing. Attempts to measure response errors, however, have been primarily on an ex post facto basis. The present study represents one of the first large-scale attempts to measure response errors in a survey of food purchases with the use of an experimental design for evaluation as the primary objective of the overall project.