Orthogonal Array Design

The Orthogonal Array Design dialog creates an orthogonal array design. Typically a researcher begins a designed experiment by generating a design, which is a data frame indicating the combinations of experimental variable levels at which to take observations. The researcher then measures some outcome for the indicated experimental variable combinations, and records this by adding a new column to the design data frame. The data may then be analyzed using ANOVA or other techniques.

Orthogonal array designs are essentially very sparse fractional factorial designs constructed such that inferences may be made regarding main (first-order) effects. Level combinations necessary for estimating second and higher order effects are excluded in the interest of requiring as few measurements as possible.

To generate an orthogonal array design

Choose Data __image\arrow5.gif Design __image\arrow5.gif Orthogonal Array. The dialog shown below appears.

The Orthogonal Array Design dialog requires some knowledge of S-PLUS language syntax. An example is illustrated in the dialog shown below, followed by the resulting data set.

__image\orthog.gif

The Orthogonal Array Design dialog has the following options:

Levels

Enter a vector of the number of levels for the factors in the design.

Minimal Residual DF

Optionally, specify the minimum residual degrees of freedom requested for a main-effects-only model. The default value is 0, unless the number of levels in the factors are all equal in which case the default is 3.

Factor Names

Optionally, specify names for the factors. In the example, you are creating a design with three variables, with two levels of the first variable and three levels for the second and third variables.

Randomization

Randomize Row Order Select to randomize the order of the rows in the design.

Restricted Factors Optionally, specify a vector (either numeric or character) naming some factors (columns) in the design which shouldn't be scrambled.

Save In

Enter the name for the object in which to save the results of the analysis. If an object with this name already exists, its contents are overwritten.

Related programming language functions

oa.design, randomize