MDS Results - Advanced Tab

Select the Advanced tab of the Results dialog box to access the options described here.

Option Description
Summary: Final configuration Creates a spreadsheet with the coordinates of the final configuration.
D-hat values Creates a spreadsheet with the transformed input data values calculated according to the monotone regression procedure.
D-star values Creates a spreadsheet with the transformed input data values, calculated according to Guttman's rank image procedure.
Distance matrix Creates a spreadsheet with the matrix of reproduced distances, given the current number of dimensions and configuration of points.
Summary statistics Creates the reproduced distances along with the D-hat and D-star values. The elements in the spreadsheet are sorted by the transformed input data values (D-hat, D-star), and each element is referenced by its matrix subscript. For example, the element denoted as D(4,2) refers to the element in the fourth row and the second column of the similarity (or dissimilarity) matrix.
Graph final configuration, 2D Plots the final configuration of variables (objects) in two-dimensional format. If you have more than 2 dimensions in your analysis, the Select Two Dimensions for Scatterplot dialog box is displayed, in which you select the dimensions for the 2D scatterplot.
Graph final configuration, 3D Plots the final configuration of variables (objects) in three-dimensional format. If you have more than 3 dimensions in your analysis, the Select Three Dimensions for Scatterplot dialog box is displayed, in which you select the dimensions for the 3D scatterplot.
Graph D-hat vs. distances Produces a plot of the transformed input data values (D-hat) against the reproduced distances. The more closely the points in the plot cluster around the diagonal, the better is the fit of the respective model.
Graph D-star vs. distances Produces a plot of the transformed input data values (D-star) against the reproduced distances. The more closely the points in the plot cluster around the diagonal, the better is the fit of the respective model.
Shepard diagram Produces a scatterplot where each point represents a distance - data point pair (note that the original data are plotted against the horizontal x-axis in the plot). In addition, the D-hat values are indicated in the graph by a step function.

Statistica performs nonmetric multidimensional scaling, and the D-hat values are computed by a monotone (non-linear) transformation of the original data (dissimilarities).  Thus, the step function monotonically increases or decreases. The closer the fit of the step function to the data points plotted in the graph, the better the fit of the respective model, that is, the better is the reproduction of the distances (or, specifically, the rank order of distances) in the input data, given the respective number of dimensions.