The methods of statistical design of experiments (DOE) are used to make data-driven decisions in an efficient, cost-effective manner. DOE is used to support product and process design efforts, as well as kaizen and continuous improvement projects. The successful incorporation of DOE enables a maximum amount of information to be derived from a minimum number of well-designed experiments. Successful application is assured when individuals using DOE are properly trained and when appropriate DOE software is implemented. DOE's effectiveness is further enhanced when an organization establishes procedures to assure that all experiments across the organization are conducted in an efficient manner through the common platform of DOE.

The workshop will start by introducing the language of DOE and describing the benefits of well-planned statistical experiments. Hands-on experiments will be used to generate data that will be analyzed during the sessions. The workshop will incorporate of MINITAB software, which is especially effective for the design of experiments and the analysis of data resulting from experimental efforts. By relieving the participants from a concentration on mathematical formulas, the ability to understand important fundamentals of analysis is enhanced. The focus of the coverage will include the interpretation of graphical results, such as the meaning of main (independent) effects and interaction effects among factors analyzed. The development and use of mathematical models will be highlighted and the structured use of DOE to drive the sequential experimentation process will be described. A working knowledge of basic statistics is a prerequisite for the workshop.

TOPICS:

• DOE in the context of contemporary quality management (Deming's Management Methods, Six Sigma) and the need for statistical experimentation (Box, Taguchi).
• Types of data (attribute, variable, categorical) and data collection concepts (factors, levels, responses).
• Review of basic data analysis displays (run chart; histogram; probability plot) and numerical summaries (average; standard deviation).
• One-factor experimentation, including comparison of central tendency & variation (t test, F Test) and type I and II errors.
• Full factorial 2k experimental designs, including graphs of main & interaction effects, and calculation and interpretation of main and interaction effects.
• Modeling with DOE including the development and use of 2k mathematical model and the analysis of residuals.
• The design and analysis of follow-up experiments such as center point designs to confirm linearity and confirmation experiments
• Introduction to fractional factorial 2k-p experiments and the concepts of confounding and aliasing.