Statistics / Unit 5: Collecting Data
5.6 Random Sampling, Random Assignment and Blocking
Notes
Two kinds of randomness serve different purposes.
- Random sampling (selecting from the population) allows results to be generalized to the population.
- Random assignment (assigning to treatments) allows a cause-and-effect conclusion.
- A block is a group of similar experimental units. Randomly assign treatments within each block to control a known source of variation.
- A matched pairs design pairs similar units (or uses the same unit twice) and randomizes the order or assignment within each pair.
- Example: 40 volunteers are randomly assigned to two treatments. We can conclude cause and effect, but only for people like the volunteers.
Practice
Volunteers are randomly assigned to two treatments. What conclusion is justified if the treatment groups differ significantly?
Random assignment supports cause and effect. Volunteers are not a random sample, so generalizing is limited.
A researcher tests two fertilizers on fields, some sunny and some shady. She groups the fields by sunny or shady and randomly assigns the fertilizers within each group. This design is:
The groups of similar fields are blocks, with random assignment inside each block.
Each subject tries both of two headache treatments, in random order, on different days. This is a:
Each subject serves as their own pair, with the order randomized.
Ready for more? Take the Unit 5 test.