Searching for and Analyzing Designs with Two-Level Factors - Analyze Design Tab

Select the Analyze design tab of the Searching for and Analyzing Designs with Two-Level Factors dialog box to access the options described here.

Variables

Click the Variables button to display a standard variable selection dialog box, in which you specify the dependent variables of interest, the independent variable list (list of factors in the design), and an optional blocking variable. After selecting the variables for the analysis, STATISTICA reads the data and determines whether a valid two-level factorial design is specified in the selected variables. Specifically, the program checks 1) whether the selected list of factors contains only two or three distinct values, and 2) if three distinct values are encountered, whether the center point values occur only in those runs where all other continuous factors (those with more than 2 levels) are also set at their center point value. Note that if your design is not a typical two-level factorial design (with or without center points), then select the Central composite, non-factorial, surface designs option from the Design and Analysis of Experiments Startup Panel; there, no assumptions are made about the nature of the design, and any combination of factor settings, with any number of distinct levels, can be analyzed.

Variable selection for botched designs

When the Design contains botched runs check box is selected, an optional fourth variable can be specified to identify center points in the experiments (center points can be used to test for curvature, in the respective results dialog box, e.g., the 2(k-p) Analysis Results dialog box). The variable specified as the Centerpoint variable should contain values equal to 0  for all runs that are to be considered centerpoint runs (and which will be contrasted against all other points to test the curvature hypothesis). Selection of a Centerpoint variable is optional, and if none is selected, no runs in the experiment (the current data file) will be interpreted as center points, and no tests for curvature will be available later on the results dialog.

Note: Multiple dependent variables and missing data. When more than one dependent variable is specified, when reading the data, STATISTICA performs casewise deletion of missing data. Thus, a case or run is excluded from the analysis if it has missing data for any of the dependent variables specified for the analysis.

Design contains botched runs; factor low/high values not exact. Select this check box to analyze a botched design where the factor levels for all runs were not set precisely at their intended values, i.e., the factor levels are not consistent with the design as it was generated.

To recode factor values (levels) use. The options in this box are only available if 1) the Design contains botched runs check box is selected, and 2) variables have been selected for the analysis. You can select the User-defined high/low factor values option to specify values that are to be used to "scale" the values for the factors. For example, for some runs for a factor A the intended High setting might have been 100, but when running the experiment, the actual setting that was used was 90. In that case, 90 will be scaled to the range from Low to High, i.e., interpreted as less than 100; however, if the factor high value for factor A was specified as 90, then that value will be considered the standard High value (rescaled to +1 for the computation of Effects coefficients), and 100 will be considered a botched setting (greater than 90, and greater than +1 when computing the Effects coefficients). Additional details regarding the scaling of the effect coefficients are also provided in the topic Main Effects and Interactions for Experiments with Two-Level Factors.

By default, if the user selects the Automatically determine factor levels from file option button, then the program will take the observed minimum and maximum values found in the data to be the factor's low and high settings, respectively. If the user selects the User-defined high/low factor levels option button, then a general user entry spreadsheet will be displayed where the user can specify the Low and High values for each factor.