What is Blocking?
Blocking is where you control sources of variation (“nuisance variables“) in your experimental results by creating blocks (homogeneous groups). Treatments are then assigned to different units within each block. The term blocking comes from agriculture, where different pesticides or growing techniques were used on different plots (or blocks) of land.
What is a Blocking Factor?
A blocking factor is a factor used to create blocks. It is some variable that has an effect on an experimental outcome, but is itself of no interest.
Blocking factors vary wildly depending on the experiment. For example: in human studies age or gender are often used as blocking factors. In medical studies, institution type might be used to remove effects caused by size of institution or population types served. In experiments involving microarrays, experimenters, microarray batches, or print-pins might be used as blocking factors (Wit & McClure, 2004). Other possibilities for blocking factors:
- Consumption of certain foods.
- Over the counter food supplements.
- Adherence to dosing regimen.
- Genetic differences in metabolism.
- Coexistence of other diseases or disorders.
- Other medications used.
Types of Blocking Design
Many different types of blocking designs exist, including:
- Randomized block design: In this design type, the researcher divides experimental subjects into homogeneous blocks. Treatments are then randomly assigned to the blocks.
- Matched pairs design: is a special case of randomized block design. In this design, two treatments are assigned to blocks of subjects. The goal is to maximize homogeneity in each pair. In other words, you want the pairs to be as similar as possible.
- Latin square designs: These are based on the Latin square,
an ancient puzzle where you try to figure out how many ways Latin letters can be arranged in a set number of rows and columns (a matrix); each symbol appears only once in each row and column. The Latin square design ensures that each letter is evenly followed by another letter, to protect against order effects.
Each of these designs is less flexible and requires much more complex analysis than a completely randomized design. However, they offer a significant increase in efficiency.
References
Wit, E. & McClure, J.. (2004). Statistics for Microarrays: Design, Analysis and Inference. John Wiley & Sons.