Bad and good experimental design

The conclusions that can be drawn from an experiment depend on how it was designed. In a badly designed experiment, it is often impossible to tell how the treatment affects the response.

Weight gain in calves — a badly designed experiment

This diagram illustrates how the effect of a factor can be incorrectly estimated when the experimental design is poor.

Good experimental design avoids major differences between the two groups of experimental units, and some strategies are described in later pages of this section.