Before you conduct the experiment, review the following guidelines and complete the appropriate activities.
- Train individuals involved in the experiment
- Because errors in the experimental procedures can invalidate the results of an experiment, you should carefully document all procedures and train individuals on those procedures. Include the following
- Specify how to measure the response and note any special techniques that may be required.
- Describe how to set factor levels. Make sure everyone understands how to set the factors at each level.
- Explain how to set up the equipment for runs. For example, each time you change settings on a machine, you may want to run the machine at the new settings until it stabilizes before collecting the measurements for the experiment.
- Develop plans for troubleshooting. Communicate how to handle potential problems, such as missing measurements.
- Specify how to record special circumstances. Explain how to track any changes in conditions that may occur while the data is being collected.
- Validate measurement system
To trust your experimental results, you need to verify that your measurement system is accurate. You should verify the measurement systems that are used both to measure the response and to set the factor levels.
If the experiments are part of a larger improvement project, such as a six sigma project, the measurement system for the response should have been validated previously. Make sure that you verify the measurement system for the factors as well. For example, if you set an oven at 350° F, you need to be certain that the temperature is actually 350° F.
- Check all design combinations
- After you create your design, review the actual combinations of factor settings for each experimental run to make sure they are feasible. Consider alternatives to runs with combinations of factor settings that are potentially dangerous or are impossible.
- Perform trial runs
- Performing trial runs before you run an experiment is useful if you have the time and the budget. Trial runs allow you to:
- Assess the consistency of materials in the experiment.
- Check the measurement systems for the experiment.
- Test the experimental procedures and ensure that operators perform them correctly.
- Check that the different combinations of factor levels can be run safely.
- Obtain preliminary estimates of variation.