Design of Experiments (DOE) is an important technique for root cause analysis (RCA) and process improvement. As an example, when potential trouble sources are identified from a cause and effect diagram, DOE can be used to determine which of the factors are likely to be important. DOE can also develop quantitative models of the nature y=f(x) (y is a function of x) where y is often a critical to quality characteristic.

While DOE is normally a subject for full-length college courses, the basics can be covered in a one-hour webinar. These fundamentals include hypothesis testing, which also carries over into acceptance sampling and statistical process control, as well as design of the experiment to exclude extraneous variation sources (randomization and blocking techniques).

**Objective of the webinar:**

Attendees will learn the fundamentals of DOE, some of which carry over into other industrial statistics applications such as acceptance sampling and statistical process control.

- Hypothesis testing is the foundation of almost everything we do with industrial statistics.
- The null hypothesis, or starting assumption, is that there is no difference between the experiment and the control, a production lot is acceptable, or a process is in control.
- The alternate hypothesis is that the experiment differs from the control (is better than the control in an improvement activity), a production lot should be rejected, or a process is out of control and needs adjustment.
- We must prove the alternate hypothesis beyond a quantitative reasonable doubt that is known as the Type I risk, alpha risk or, in acceptance sampling, the producer’s risk (of wrongly rejecting an acceptable lot).

- DOE can save an enormous amount of time and money, as shown by comparison of an experiment performed in the late 19th century (prior to the development of industrial statistics) and an even more complex one performed roughly 100 years later. This underscores the value of DOE in the language of money, i.e. the language of upper management.
- Understand the concepts of factors, levels, and interactions (the whole is greater or less than the sum of its parts). Factors such as machine or material are often identified from a cause and effect diagram during root cause analysis.
- Recognize the need to exclude extraneous variation sources from the experiment through randomization and blocking, and also the need to use a sufficiently large sample to get meaningful results (replication).
- Explain the results of an experiment in terms of its significance level or P value(chance that the observed results are due to random chance).

**Areas Covered in the Session :**

- Value of DOE in the language of time and money, as shown by comparison of an experiment performed by Frederick Winslow Taylor during the late 19th century, and an even more complicated one performed by a pharmaceutical company that sought FDA approval for a diagnostic test
- Hypothesis testing as the foundation of most industrial statistics applications including not only DOE but also statistical process control and acceptance sampling (e.g. ANSI/ASQ Z1.4 and ANSI/ASQ Z1.9)
- Interactions, or situations in which the whole is greater or less than the sum of its parts. Interactions cannot be detected by one variable at a time experimentation.
- Experimental design considerations including randomization, blocking, and replication.
- This webinar will provide a sufficient foundation for attendees to work effectively with industrial statisticians, Six Sigma Green and Black Belts, and similar subject matter experts.

**Who Should Attend:**

- Quality Departments
- Manufacturing Departments
- Engineering Departments
- Technicians, Supervisors and Managers

FDB2494