You are here

Completely randomized designs

The elements of a research design are:

  • Research hypothesis
  • Treatment design
  • Experiment or observational study design

Examples of research hypotheses (see [Kuehl]):

  • A microbiologist hypothesizes that the activity of soil microbes depends on the level of moisture in the soil. Treatments with different amount of soil mosture may be set up to measure the microbe activity at different levels of soil moisture to evaluate the hypothesis.
  • A traffic engineer hypothesizes that traffic speed is related to the width of streat lanes. Lanes of different width are selected by the engineer, and traffic speed is measured at each lane width to evaluate the hypothesis

The treatment design is a part of an experiment design. In the design phase, one decides

  • what is the experimental unit;
  • how many replications are required;
  • how to assign a treatment to each experimental units.

The assignment of experimental units should avoid an experimentor bias, and thus it involves random choices. It is immaterial whether a lottery-style drawing is used, or a random number generator is used, as long as we ensure random assignment of experimental units to different treatment groups.

No additional criteria are used in assigning experimental units to treatment groups.