Design of a Screening (Plackett-Burman) Experiment - Add to Design Tab

Replicating the Design

Select the Add to Design tab of the Design of a Screening (Plackett-Burman) Experiment dialog to access the options described here.

Add to the design

Use the options in the Add to the design group box to add replicates, points, or columns to the standard design or to select foldover. Note that the settings in this group box only affect the displayed design (via option Summary: Display design on the Quick tab or Display design tab). The changes do not affect the results (e.g., the design resolution) displayed in the Summary box.

Number of genuine replicates

Enter a number in the Number of genuine replicates box (or adjust the number using the microscrolls) to replicate the complete design the respective number of times. For example, if you enter a 1 into the box, then Statistica adds one replicate to the design, resulting in two identical parts of the design. If you want to replicate selected points from the design (e.g., so that you can later obtain an estimate of pure error; see Replicating the Design), first display the design (via the Summary: Display design option on the Quick tab or Display design tab), add rows to the spreadsheet (use option Insert - Add Cases), and then copy and paste the respective runs into those new rows.

Number of center points (per block)

Enter a number in the Number of center points (per block) box (or adjust the number using the microscrolls) to add the respective number of center points to the design. Center point runs will have factor values that are in-between the respective factor high and low settings for all continuous factors. The usefulness of center points is discussed in the Introductory Overview. In short, center points will allow you later in the analysis to perform explicit Statistical significance tests of curvature or non-linearity in the relationship between the factors and the dependent variable. To maintain a balanced and orthogonal design, if there is blocking in the current design, then Statistica will add the same number of center points to each block in the design. If the design includes qualitative factors (see option Change factor names, values, etc. on the Quick tab or Display design tab) then Statistica will construct for each requested center point a complete factorial design for the qualitative factors. For example, if your design has two qualitative factors, and you request 1 center point, then Statistica adds a total of 2 * 2 = 4 center points to the design (per block, if there is blocking). Those 4 center points will be arranged so that there will be a full factorial design for the 2 qualitative factors, with all continuous factors set at their center.

When generating (printing) two-level designs on the results dialogs for 2(k-p) designs or Screening designs (or design generated via the 2(k-p) maximally unconfounded and minimum aberration designs), when Centerpoints are requested, an additional column is added to the reported spreadsheet, with indicator values (0, 1) to identify the centerpoint values in the design. This center point indicator variable can be used when analyzing a "botched version" of this design, where center points cannot be unambiguously determined from the data.

Number of blank columns (dep. vars).

Enter a number in the Number of blank columns box (or adjust the number using the microscrolls) to add the respective number of blank columns to the displayed design (via the Summary: Display design option on the Quick tab or Display design tab). This option is useful if you want to print the spreadsheet with the extra columns to use as a convenient data entry form.

Foldover (enhance resolution)

One way in which any resolution III design can be enhanced and turned into a resolution IV design is via foldover (e.g., see Box and Draper, 1987, Deming and Morgan, 1993). If you select the Foldover (enhance resolution) check box then Statistica duplicates the points in the design, reversing the signs in the duplicated experimental runs (i.e., all high factor settings will be changed to low factor settings, and all low factor settings will be changed to high factor settings; see also the Introductory Overview. Note that the resultant increase in the design resolution will not be reflected in the summary displayed in the Summary box.