Notes on designing an experiment

What is the purpose of the experiment?

If the answer is something vague like

  • `to investigate new varieties of sunflower' or
  • `to find out about the nutritional effects of margarine',
    there may be a perfectly worthwhile exploratory experiment to be done but a statistician is unlikely to be able to help you.

    Usually it is better to state a specific question that is to be answered, such as

  • to estimate how much better drug A is than drug B in reducing inflammation
    (of course, the two drugs would need to named more precisely)
  • to test the hypothesis that there is no effective difference between organic and inorganic nitrogen as a source of fertilizer
  • to fit a model for how much sucrose is absorbed at different distances from the surface of the liver.

    protocol purpose treatments experimental units method observational units
    measurements design justification randomization plan analysis

    Page maintained by R. A. Bailey

    Page updated 17/11/01