Page:Sm all cc.pdf/100



"An Experiment is a question which Science poses to Nature, and a measurement is the recording of Nature's Answer." [Planck, 1949] Experimental design determines whether a research report is read or ignored, whether a result is accepted or rejected, and whether a scientist is judged superior or inferior. Most scientists and many technicians can carry out an experiment successfully. An experiment’s value, however, depends not only on outcome but also on the skill with which it is designed. Fortunately, this skill, which develops with experience, also can be fostered deliberately. This chapter provides a variety of experimental tips, culled from the experiences of many scientists.

Like military planning, research planning has three levels [Beveridge, 1955]:
 * tactics, the small-scale and relatively short-term planning of an individual experiment. The key question is ‘how’. Tactics must be considered in the context of
 * strategy, the broader approach to the research problem, which involves an extensive suite of experiments. A strategy is most efficient if it is considered in the context of
 * policy, the determination made by funding agencies and committees concerning which general problems are the most crucial in a science.

Like business planning, research planning should involve the following:
 * risk analysis. What is the chance of success? What could go wrong and what would its impact be?
 * risk management. How can I improve the chances of success? How can I avoid possible factors that would make the final result ambiguous or misleading?
 * time analysis. How much time will it take for each step? How will potential problems affect this estimate?
 * time management. How much control do I have over the amount of time required for each step? Where can I streamline the procedure without weakening the integrity of the experiment?