There is a new version of this course. Please see JMP® Software: Modern Screening Designs.

This advanced course presents strategies and methods for designing experiments to screen many factors in an optimal study, as well as several specialized analytical tools that respect the limited information available in such experiments. This course is designed to help scientists and engineers choose an appropriate technique for their particular situation.

Learn how to assess the complexity of the situation in terms of the number of factors and the number and nature of terms in the model as a guide to selecting a method to design the screening experiment and analyze the empirical evidence. You will be able to

• recognize situations that benefit from a screening experiment
• make a fractional factorial design or Plackett-Burman design
• make an orthogonal or near-orthogonal array or a definitive screening design
• make a Bayesian D-optimal design or split-plot design with custom design
• identify situations where each of the screening designs might be most useful
• identify likely effects in the response using the effect screening emphasis and tools in the Fit Least Squares platform or the Screening platform
• select a model using the Stepwise platform with forward selection or All Possible Models under the heredity restriction.

Who should attend

Advanced JMP analysts who need to screen many factors through experimentation

Formats available	Standard duration
Classroom:	2.0 days

Before attending this advanced course, you should complete the JMP® Software: Custom Design of Experiments or JMP Software: Classic Design of Experiments courses or have equivalent experience.

This course is not an introduction to design of experiments.

This course addresses JMP software.

Need for Screening

• learning the important principles in screening such as: the impact that choice of design has on the variance of estimates and bias in model prediction, inflation of variance in parameter estimates due to correlation, and estimation efficiency as measured by the estimate's confidence interval size
• planning experiments with classic screening solutions including fractional factorial and Plackett-Burman designs

Combinatorial Screening Designs

• planning experiments with orthogonal or near-orthogonal arrays
• planning experiments with definitive screening designs
• planning experiments with custom designs, like Bayesian D-optimal designs and split-plot structures for restricted randomization

Model Selection in Screening

• identifying active factors using the effect screening tools in the Fit Least Squares platform
• selecting models with the Screening platform
• selecting models with forward step-wise regression
• selecting models with all-subsets with heredity restrictions