Design-Expert® 7.1 for Windows

What's New in Version 7.1

• Upfront power calculation for factorial designs: This mainstreams in the design-builder a ‘heads-up’ on the percent probability of seeing the desired difference in each response — the signal — based on the underlying variability — the noise.

• “Min-Run Res V” designs are now available for 6 to 50 factors: Resolve two-factor interactions (2FI's) in the least runs possible while maintaining a balance in low versus high levels.
  
  
• CCD’s are available that are based on the Min-Run Res V fractional-factorial core — now up to 50 factors: Take advantage of a much more efficient design for larger numbers of factors.
  
  
• Bookmarks for reports with a toolbox to facilitate selection: This will save you a lot of time scrolling through long statistical outputs such as the design evaluation and analysis of variance.
  
  
• Display grid lines on 3D-graph back-planes: This feature provides a better perspective on the varying height of a response surface.
  
  
• Simple ratio constraints, such as A/B>1, can be entered in directly: This sort of thing is fairly common, for example, A might be air pressure upstream of a check valve and B the pressure after, but it will work only when A exceeds B.
  
  
• Save graphs to files in enhanced Windows metafile (EMF), PNG, TIFF, GIF, BNP, JPEG, and encapsulated Postscript (EPS) formats: Many publications do their artwork in one of these file types .
  
  
• Fraction of design space (FDS) graph for design evaluation: This enhancement, suggested to us by DOE guru Douglas Montgomery, provides very helpful information on scaled prediction variance (SPV) for comparing alternative test matrices — simple enough that even non-statisticians can see differences at a glance and versatile enough for any type of experiment — mixture, process or combined.
  
  
• Design layout can now be modified via a right-click list with added columns for point type and other alternative attributes: Make your "recipe" sheet more informative.
  
  
• “Design model” choice added for statistical analysis: This is handy for data from experiments based on a computer-generated D-optimal design.
  
  
• More flexibility in handling various file types when opening files: Very helpful default that automatically recognizes any data in the Design-Ease (.de*) or Design-Expert (.dx*) format – including ones produced from older versions.
  
  
• From the Design node, display mixture constraints coded in actual, real, or pseudo values: An important distinction for understanding the experimental region of formulation.
  
  
• Keystroke option (Ctrl+/-) to move through alternate solutions from numerical optimization: This saves mousing around.
  
  
• Import and export text files to get responses: Something do-able by anybody
  
  
• Write transfer functions in format (.vta) readable by VarTran® software (Taylor Enterprises): This sets the stage for statistical tolerancing and sensitivity analysis leading to more robust designs.

New Design Creation

• Design-builder updates resolution of two-level fractional factorials when the number of blocks is changed: Immediately see how segmenting a design might reduce its ability to resolve effects.
• Block names are now entered during the design build: Identify how you will break up your experiment, for example by specific shift, material lot or the like.
• “Min-run Res IV plus two” option: Ask for two extra runs to make your experiment more robust to missing data.
• User-defined base factors for design generators: You have more flexibility to customize fractional factorial designs.
• Expanded D-optimal capabilities—impose balance penalty, force categoric balance: This feature helps users equalize the number of treatments.
• CCD’s offer new alpha choices of “Practical,” “Orthogonal Quadratic” and “Spherical”: Develop more control over where you put your ‘star’ points.
• Coordinate Exchange capability for D-optimal designs: Avoid the arbitrary nature of designs constructed from candidate point sets.
• In General or Factorial D-optimal designs, categorical factors can be specified as either nominal or ordinal (orthogonal polynomial contrasts): This affects the layout of analysis of variance (ANOVA).
• Simple ratios are allowed, for example: A/B, when specifying multilinear constraints: This is especially handy for mixtures with components that require
  certain stoichiometries.
• Specify the same amount for low and high in a mixture design: This is handy for keeping track of fixed component levels—these do not appear in the model.

• “Semifold”: In only half the runs needed by a normal foldover, augment Res IV designs to resolve specified 2FI's aliased in the original block of runs.
• Add center points, blocks and replicates without rebuilding the design: This will be a real time-saver.

• * From Alias List, Pareto Chart or Effects Plots views, right-click on effects to show aliases: Never lose sight of what really is being measured in fractional-factorial designs.
• * Select alternative aliased effects: Choose what you think makes most sense based on your subject-matter knowledge.
• * Backward stepwise regression is now applicable to factorial designs: This is useful for quickly analyzing general (categorical) factorials.
• * Means and standard deviations for all experimental inputs (factors) and outputs (responses) are added to the Design Summary screen: This provides a handy assessment of your system.
• * The user can define their preference for sums of squares calculations for both numeric and categoric factors to be sequential, classical, or partial: These distinctions are important for statisticians who want to do ANOVA in specific ways.
• * Cox model option for mixtures: May be more informative for formulators with a standard (reference) blend to which they’d like to compare more-optimal recipes.
  

• * DFFITS: Spot influential runs via this deletion diagnostic that measures difference in fits when any given response is removed from the dataset.
• * DFBETAS: See from this deletion diagnostic how model terms change due to an influential run.

• * Grid lines on contour plots: See more readily what the coordinates are at any given point.
• * Select the details printed on flags planted on contour plots: As a user you now can control this feature.
• * Confidence bands on one-factor plots: Get a good feel for the uncertainty in a predicted response as a function of the factor level.
• * Color-codes for positive versus negative effects: Assess plus or minus impacts on half-normal and Pareto plots.
• * Smart tic marks: Get more-reasonably rounded settings straight off.

• Export the graph to a file: Save the graph as an enhanced metafile (.emf) that can be inserted as a picture from file to Microsoft Word and the like.
• Set row status to normal, ignore or highlight: This allows users control over their design matrix.

• Annotation option on reports: This will be a boon to those who may be unfamiliar with all the esoteric statistics needed for design evaluation.
• Customizable design evaluation content and power levels: Use the OPTIONS button to select which statistics to display, specific power levels to report, and whether to display the standard error or variance on the graph (with the option to scale by N—the number of runs in the design).
• Specify model terms to ignore so they don’t display in the alias list: For example, don’t bother showing interactions of four or more factors.
• Evaluation can be done on either design or a particular response: Shows the effect when data is missing from a specific response, but not all responses.

• Tutorial movies: See Flash demo’s of features via Screen Tips—a very effective way to show how to navigate through the software.
• Internet links: These are helpful connections to further information.

• * XML (eXtensible Markup Language) capability: Export design files or reports in viewable format that can be manipulated for further processing (The XML tool also allows import of designs created externally).
• * Scripting capability: Run Design-Expert software in batch mode so it can be tied into more comprehensive lab-ware or used to cycle through massive quantities of data, for example from computer-based simulations.

Back to Top