Bad experimental design

If the treatments are allocated to experimental units in a way that is associated with their naturally varying characteristics, the apparent relationship between the treatments and the response can be distorted.

This is similar to the effect of lurking variables in observational studies.

Good experimental design can avoid the potential effect of lurking variables.