Evaluation - Model

Design Evaluation is used to evaluate the statistical properties of the design, before response data is collected. The model screen is used to choose the model that the design should be evaluated against. There are also customization options available.

Model Type: For factorial designs the available terms are limited to factorial interactions or polynomial terms. For mixture designs the choice is between a Scheffe and Slack. If Slack models are chosen, an additional box to pick the variable to slack will appear.

Order: Select the terms to appear in the model, i.e. main effects, 2FI, quadratic, etc. There will be one order per grouping (Mix 1, Mix 2, Process) on combined designs.

Response: The default is to evaluate the structure of the whole design. If data has been entered in a response, the pull-down list will allow it to be chosen. This is most often used to see the effect of a missing run on the capabilities of the design.

Ignore: Only available for Factorial models. Interactions at or above the order specified on the Ignore pull down are not displayed in the Results. It often simpler to read the evaluation report when the higher order interactions are ignored.

Options: Control what statistics will be displayed on the evaluation report and whether or not an estimate of fraction of design space (FDS) will be computed.

Power Options: Opens a dialog to establish the parameters for calculating power on the evaluation.

Add Term: A term can be typed rather than finding it in the list. Often used to add higher-order and complex terms to the model.

Terms List: Specific terms can be added or removed from consideration. Double-click on a term to toggle it into or exclude it from the model. Right-clicking on a term will also allow setting a term to be an error term . Select several terms and right-click on them to simultaneously change them.