The Main Idea

A Blocking Variable is a variable that a researcher knows is a nuisance factor. Variables like age, gender, income, and education level are controllable and are often accounted for in a study via blocking.

First block, then randomly assign. Blocking occurs when a researcher first divides a random sample into homogeneous groups before randomly assigning the individuals in each group to treatment and control groups.

Keep blocking to a minimum. We should block unwanted variables that can be measured and might influence the outcome, but we should keep blocking to a minimum. Just block the most important nuisance variables and let randomization handle the rest.  

Use a randomized block design when a known nuisance factor is measurable and controllable or when the sample size is not large enough to ensure equal groups in a random sample. Otherwise, allow randomization to handle the confounding factors.