Latin square design

A statistical experimental design in which two noninteracting extraneous variables in addition to the independent variable may be controlled. For example, to observe the effect on sales (the dependent variable) of three different alternative-package designs in three different types of retail stores in three different economic conditions.