Completely Randomized Block Design
In a completely randomized block design, the experimental units (participants) are divided into homogenous (similar) groups called blocks. The blocks help to minimize the potential effects of a nuisance factor–a factor the researcher is not interested it, but one that can cause variation in the response factor. Some well-known nuisance factors that are often controlled for include environmental factors like temperature, light, time of day, or physiological factors in living subjects like weight, sex, or age.
To use a block design, the researcher will chose a factor to control, such as age. All subjects exhibiting a similar characteristic of this factor (say they are close in age for example, or are all of a particular gender) are grouped together in similar sized blocks. Then they are randomly assigned to the treatment groups. This will ensure that any variability due to the nuisance factor is evenly distributed across the treatments, and any variability in the response variable attributed to the nuisance factor is isolated and minimized.
Video Placement
[insert video of an animation that shows blocking and random assignment. Like this one, which is a good start but can be improved on. I’m thinking an interaction of some kind would be amazing here, one that lets students do the dragging and dropping themselves. Could that be programmed?]
To summarize, in a completely randomized block design, the experimental units are first grouped into homogenous (similar) blocks, then randomly assigned to treatments within each block.
Now you try answering Question 4, about nuisance factors.
question 4
Consider the Salk vaccine experiment again. Suppose there was previous research to suggest that the male children in the experiment would respond more favorably to the Salk vaccine injections than the female children.
Part A: Would gender be considered a nuisance factor? Explain.
Part B: What are some other possible factors in the experiment, other than the factor of interest, that would be considered nuisance factors?
Video Placement
[worked example Video: a 3-instructor worked example in the style of Question 5, demonstrating how to identify if an experiment has used completely randomized block design.]
Now you try it. Assess your understanding of blocking by answering Question 5.
question 5
A marketing research firm wants to determine the most effective method of advertising: print, radio, or television. They recruit 300 volunteers in the study. The chief researcher believes that level of education plays a role in the effectiveness of advertising, so she segments the volunteers by level of education. Of the 300 volunteers, 120 have a high school education, 120 have college degrees, and 60 have advanced degrees. The 120 volunteers with high school diplomas are randomly assigned to either the print advertising group, the radio group, or the television group. The same procedure is followed for the college graduates and advanced degree volunteers. Each group is exposed to the advertising. After one hour, a recall exam is given and the number of correct answers is recorded.
Part A: Identify the factor of interest.
Part B: Does this experiment have blocks?
- a) No, there are no blocks within the experiment.
- b) Yes, the method of advertising and the level of education both serve as the block.
- c) Yes, the method of advertising serves as the block.
- d) Yes, the level of education serves as the block.
Part C: Can a completely randomized block design be used to help the marketing firm to determine the most effective method of advertising?
- a) No, because there are no blocks within this design.
- b) Yes, because the treatments are randomly assigned to the experimental units within each block.
- c) No, because there needs to be more than two treatments.
- d) Yes, because the treatments are not randomly assigned to the units within each block.
