Choose a Design

• Consider cost of a run, error estimates, precision, interactions required
• Consider deleting runs from a design to make it more affordable
• Consider acceptability of mde and avc
• Consider adding error runs to improve mde or avc

Modify the Design Selection

• Iterate on choices, consider trade-off
• Specify Dispersion Analysis if required: Either use design replicates or a subsidiary design.
• Specify Error runs if desired

Convert from Standardized to Analysis Design

• Transform Factor-Levels to experimental values, if necessary
• Randomize run order
• Generate data file, Analysis data sheet

Do the Experiment

• Record the response(s) and comments concerning any special circumstances or events
• Record recognized casual factors: Date, Time, Environment, People, Equipment, Unplanned factors

Prepare data for Analysis

• Retrieve the Analysis Data
• Calculate any composite responses

• Establish interactions for analysis; Retrieve design interactions; Consider deletions or additions
• Incorporate Casual Factors

Do an Analysis Analysis is an iterative procedure For each recorded or computed response

• For qualitative factors, consider graphical presentation of the data
• For quantitative factors, consider both graphical presentation and regression analysis; Consider a preliminary analysis on normalized data; Consider transforms for the factors and response

Evaluate the results of a regression analysis

• Error, Lack of Fit
• Significance of the fitted parameters
• Plots of the results
• Consider changing the factors and interactions in the model.
• If applicable, do the response surface calculations
• Review principal axis components
• Examine contour and 3-d plots

Validate the Conclusions

• Use Fold-over or repeated design for significant factors to validate interactions and main effects
• Use a small local design to validate an optimal response

Evolutionary Designs

• Move a small design in the direction of increasing performance, with caution.

Reset the Process

• Monitor for expected results.

Learn more about the DOE tools for designed experiments in Six Sigma Demystified (2011, McGraw-Hill) by Paul Keller, in his online Intro. to DOE short course (only $99) or online Advanced Topics in DOE short course (only $139), or his online Black Belt certification training course ($875).