Experimental Design - Summary for Variables (Factors)

Factor values (e.g., Low value, High value, Center value) vary with each type of design. If a particular factor value is only available for specific designs, that is indicated in the descriptions below.

Factor Name

Enter factor names in the Factor Name column. Note that to view these names in the resulting spreadsheet, select the appropriate option in the Denote factors group box.

Low Value

In the Low Value column, enter the value for the low setting of the respective factor.

Low Label

In the Low Label column, enter the corresponding text label that you want to use for the low setting of the respective factor. Note this is not applicable for Mixture Experiments.

Center Value

In the Center Value column, enter the value for the center setting of the respective factor (available for Central Composite Designs and Experiments with Three-Level Factors as well as for the Three-Level factors in Experiments with Two- and Three-Level Factors). Note that in Central Composite designs, the center point values are used to "anchor" the values for the star points (see Anchoring of star points).

Center Label

In the Center Label column, enter the corresponding text label that you want to use for the center setting of the respective factor.

High Value

In the High Value column, enter the value for the high setting of the respective factor.

High Label

In the High Label column, enter the corresponding text label that you want to use for the high setting of the respective factor. Note this is not applicable for Mixture Experiments.

C/Q: Continuous or qualitative factor. (Experiments with Two-Levels/Screening Designs).

In the last column of the spreadsheet (for Experiments with Two-Level Factors and Screening Designs) enter either a C (or c) or Q (or q) to indicate the nature of the factors (Continuous or Qualitative). Continuous factors are those where the factor settings can be changed continuously over the range from the respective low to high settings (e.g., temperature). Qualitative factors are those where the factor settings are categorical in nature, and which, therefore, cannot be continuously adjusted (e.g., Machine 1 vs. Machine 2; one cannot set this factor to Machine 1½). The identification of factors as continuous or qualitative is important if you want to add center points to the design (see option Center points per block). Center points cannot be added for qualitative factors; hence, to balance the design, when you request center points, STATISTICA will construct full factorial designs for all qualitative factors at each center point, for all continuous factors. For an example, refer to the description of option Center points per block. This is available for Two-Level Factor designs and Screening Designs.

Star low/high label (Central Composite Designs)

In the last two columns of the spreadsheet for Central Composite Designs enter the text labels that you want to use to label the respective star points in the design.

Lower constraints (Mixture Experiments)

The standard simplex-lattice and simplex-centroid designs require vertex points, that is, pure blends consisting of only one ingredient. In practice, those points may often not be valid, that is, pure blends cannot be produced because of cost or other constraints. For example, suppose in a 3-component example, component B is constrained so that xB >= 200. The total of the 3-component mixture must be equal to 800. This constraint can be visualized in a triangular graph by a line at the triangular coordinate for B=200, that is, a line that is parallel to the side opposite to the B vertex point. Now, to construct the standard design in the sub-triangle, the maximum values for components A and C must be adjusted. (Later, in the analysis, you can review the parameter estimates for the so-called pseudo-components, i.e., treating the constrained triangle as if it were a full triangle.)

After you enter the lower and upper values for the three components (0 <= xA <= 800; 200 <= xB <= 800; 0 <= xC <= 800, Total = 800) into the Specifications for Mixture Variables spreadsheet, and then click the OK button, the following message will be displayed:

The selected specifications (highs, lows, and mixture total) are inconsistent and have to be adjusted to produce a standard design; do you want to adjust the factor highs?

If you click the Yes button, then the Specifications for Mixture Variables spreadsheet will be displayed again with the adjusted values that permit STATISTICA to construct the standard design in a sub-triangle (or sub-simplex). Specifically, when there are lower constraints for one or more of the components, the program will adjust the High Value for the other components, so that:

x'high,i = Total-Li

Here, x'high,i stands for the adjusted High Value for the i'th component, Total stands for the mixture total, and Li stands for the sum of the lower constraint (limit) for all components excluding component i.

Upper- and lower-bound restrictions (Mixture Experiments)

When there are both upper and lower constraints (as is often the case in experiments involving mixtures), then the standard simplex-lattice and simplex-centroid designs can no longer be constructed, because the subregion defined by the constraints is no longer a triangle. In that case, use the option Designs for constrained surfaces and mixtures on the Quick tab of the Startup Panel to find the vertex and centroid points for the constrained region in the full simplex.

OK

Click the OK button to accept the entries made and return to the previous dialog box.

Cancel

Click the Cancel button to return to the previous dialog box. Any entries made will be disregarded.

Contents

Index

Search Results

Experimental Design - Summary for Variables (Factors)