+353-1-416-8900REST OF WORLD
+44-20-3973-8888REST OF WORLD
1-917-300-0470EAST COAST U.S
1-800-526-8630U.S. (TOLL FREE)


Introduction to Design of Experiments - Webinar

  • ID: 4061752
  • Webinar
  • March 2017
  • Region: Global
  • 60 Minutes
  • NetZealous LLC
25 % OFF
until Nov 30th 2021
The purpose of a designed experiment is to distinguish between random chance (luck) and genuine effects of an experimental treatment versus a control, or genuine differences between choices of materials, machines,methods, or other factors from the familiar cause and effect diagram. That is, if an improvement team identifies the choice of material,e.g. from one of three suppliers,as a possible factor that affects the outgoing quality,DOE can prove beyond a quantifiable reasonable doubt that it does. This reasonable doubt is known as the significance level in hypothesis testing.

Hypothesis testing is the foundation not only of DOE but also statistical process control (SPC) and acceptance sampling. This takeaway from the presentation will therefore equip the attendee to understand not only DOE but also SPC and quality acceptance sampling plans. The null or initial hypothesis-'null' means 'nothing'-is that there is no difference between the control and the experiment, or that choice of level (e.g. material supplier A, B,or C) in a factor (material) has no effect on the response variable, which is often the CTQ characteristic. The null hypothesis corresponds to the presumption of innocence in a criminal trial.

The alternate hypothesis,which must be proven beyond a reasonable doubt, is that there is an effect.We can, in statistics, quantify this reasonable doubt as the Type I risk or significance level, which is typically but not necessarily 5%. It is the risk of concluding wrongly that there is an effect when there is none,i.e.'the boy who cried wolf.' It is also vital to exclude extraneous variation sources from the experiment, which leads to the concepts of randomization and blocking. More data are always better, which adds the concept of replication. DOE recognizes, however, that data are not free, so experiments are always designed to get the most information out of as few data as possible.

There is a huge difference between a well-designed and a poorly-designed experiment. Frederick Winslow Taylor described a metal-cutting experiment that took about 20 years and cost millions of dollars in today's money because industrial statistics had yet to be invented when this experiment was performed. On the other hand, a modern pharmaceutical firm performed a far more complex experiment in roughly 4 weeks, and time to market is very important in this industry.

Why should you Attend?: A scientifically designed experiment economizes on time and material resources, and it returns actionable results in terms of root cause analysis. That is, DOE can identify the root cause of a problem to support corrective and preventive action (CAPA).It can also play a central role in process improvement by identifying and optimizing the factors that influence the critical to quality (CTQ) product characteristic.

While a one hour webinar cannot take the place of a college-level statistics course, it can equip the attendee to understand the basics of experimental design including selection of an adequate sample size, and exclusion of extraneous variation sources from the experiment. The attendee will also learn how to interpret an experiment's results in terms of its significance level, or the chance that observed differences between the control and the experiment are due solely to random variation or luck. The attendee will therefore be well equipped to work with Six Sigma Green and Black Belts, industrial statisticians, quality engineers,and similar subject matter experts to understand an experiment's design and its results.

Areas Covered in the Session:

Economic benefits of DOE
Hypothesis testing: the foundation of DOE,SPC,and acceptance sampling
Null and alternate hypothesis
Type I or alpha risk of concluding wrongly that the experiment differs from the control (or that a process is out of control, or that an acceptable production lot should be rejected)
Type II or beta risk of not detecting a difference between the control and the experiment, not detecting an out of control condition, and accepting a production lot that should be rejected
Factors, levels, and interactions
Interaction = 'the whole is greater or less than the sum of its parts.' One variable at a time experiments cannot detect interactions.

Randomization and blocking exclude extraneous variation sources from the experiment.

Replication means taking multiple measurements to increase the experiment's power.

Interpret the experiment's results in terms of the significance level, or quantifiable 'reasonable doubt' that the experiment differs from the control.
Note: Product cover images may vary from those shown
  • William A. Levinson William A. Levinson,
    Principal Consultant ,
    Levinson Productivity Systems

    William A. Levinson, P.E., is the principal of Levinson Productivity Systems, P.C. He is an ASQ Fellow, Certified Quality Engineer, Quality Auditor, Quality Manager, Reliability Engineer, and Six Sigma Black Belt. He is also the author of several books on quality, productivity, and management, of which the most recent is The Expanded and Annotated My Life and Work: Henry Ford's Universal Code for World-Class Success.

Note: Product cover images may vary from those shown
  • Manufacturing Engineers
  • Managers
  • Technicians
Note: Product cover images may vary from those shown