Blocking in statistics is used in the Design of Experiments. This article will explain the difference between blocking and randomizing, as well as when and why to use it and best practices.

What is blocking in statistics?

A DOE is used to assess the effect of input variables or factors on your output (or response). First, identify any factors that could have a significant impact on your response variable. You will use DOE to test these factors to determine if they can affect your response variable to change.

You want to get the most from your experiment by minimizing the influence or noise of other factors in your environment that are not relevant to your research.

You want to know the effects of temperature, machine speed and pressure on the lamination integrity of a sheet of glass. You should measure your glass sheet under the same experimental conditions to minimize other factors. You will need to perform your experiment in each of the three shifts that the plant operates.

Blocking allows you to control the sources of variation (” non-essential variables”) by creating blocks (homogeneous group). The treatment is then assigned to the different blocks. blocking is a term that comes from agriculture. Different pesticides and growing techniques were used on different blocks (or plots) of land.

A common method to reduce the nuisance variable effect is blocking. This involves dividing up people in an experiment that relies on the value of a nuisance variable.

What is a blocking factor in statistics?

blocking element is a factor that can be used to create blocks. It is a variable that has an effect upon an experimental outcome but is not itself of interest.

The experiment will determine the blockers. In human studies, gender and age are common blocking factors. For medical research, the institution type could be used to eliminate effects due to size or populations served. As blocking factors in experiments involving microarrays experimenters, microarray batches or print-pins could be used (Wit & Mclure, 2004). There are other options for blocking factors

  • Certain foods are not allowed to be consumed.
  • Over-the-counter food supplements.
  • Respect the dosage regimen.
  • There are genetic differences in metabolism.
  • Other diseases and disorders can coexist.
  • Other medications may also be used.

Different types of blocking design in statistics

There are many types of blocking designs available, including:

  • Randomized Block Design: This design type allows the researcher to divide experimental subjects into homogeneous blocks. The blocks are randomly assigned treatments.
  • Matched pair design: a special case for randomized block design. This design assigns two treatments to each block of subjects. Each pair should be homogeneous. This means that you want the pairs as similar as possible.
  • Latin Square Designs These are based upon the Latin square.
    This ancient puzzle asks you to find out how many Latin letters can be placed in a certain number of columns and rows (a matrix). Each symbol is only one-percentile in each column and row. To protect against order effects, the Latin square design makes sure that every letter is equally followed by another letter.