Meisel, Robert (1991, ASQC) Eastman Kodak Company, Rochester, NY
Experiments help companies control process variables and implement process improvements. However, even statistically well-designed experiments sometimes produce unclear results. This paper explains that there is more to planning an experiment than statistical design. A thorough experiment planning process includes four phases: (1) Preparation, (2) Execution, (3) Analysis, and (4) Recommendations.
The method described in this paper is based on the Plan-Do-Check-Act cycle. The steps involved in the preparing for an experiment include: (1) Define the problem; (2) Define the objectives; (3) Design the experiment: state the objective statistically, define the factors, define the responses, and determine the statistical design; (4) State the assumptions; (5) Plan the analysis; (6) Check the expectations; (7) Estimate the complexity and cost; and (8) Plan for execution.
After executing the experiment, the analysis and follow-up activities include: (1) Conduct a post-experiment meeting; (2) Screen the data; (3) Analyze the data; (4) Develop conclusions; and (5) Evaluate the experiment. The last step in the Plan-Do-Check-Act cycle is to make recommendations that will help experimenters reach meaningful results.
Experimenters who follow the guidelines in this article can conduct experiments in an organized, formal manner and avoid failure due to deficiencies in the planning process.
Manufacturing,Plan-Do-Check-Act (PDCA) cycle,Process improvement,Statistics,Experiments