|Top Previous Next|
This window can show four different plots, which are each selected using the radio buttons to the left of the graph in the MDS Options panel.
Samples (2D) - This shows the plot of the final positions for the samples or objects (columns).
Samples (3D) - If you selected more than 2 dimensions for the MDS, this will plot axes 1 to 3.
Stress vs iterations - This shows the plot of the stress against iteration number for a two-dimensional non-metric multidimensional scaling model. A successful analysis should show an approximately asymptotic decline in stress with iteration number.
Stress vs Dimension - This shows the plot of the final stress value plotted against the dimension of the model. The maximum dimension on the plot is the maximum number of dimensions selected during setup. This plot can give an idea of the true dimensionality of the data set as stress will reach a minimum value once the necessary number of dimensions to fully express the distances between sites is reached. It gives an impression of the suitability of a two-dimensional plot to show between-site relationships. If a two-dimensional representation is satisfactory, then an increase in dimension number will not greatly decrease stress.
Show labels - This tick box switches object names on and off.
Variable to plot - In its default setting, the plot will show all sites/samples at an equal size. If you select one of the variables from the drop-down, then the size of the marker representing each site/sample will be proportional to the occurrence of that variable in the site/sample. This can be useful if a distinct group of sites/samples appears in the MDS plot, to detect which variables account for the grouping on the plot. Select variables from the drop-down menu; if all the sites/samples in the group are shown with larger bubbles, then it is likely that the magnitude of that variable's presence in those sites/samples is responsible for the grouping together of those sites/samples (see the example plot below). To scroll rapidly through the variables in the drop-down menu, click into the menu, then use the Up and Down arrows on your keyboard to scroll through the list of variables.
Size of bubble - Controls the maximum size of the bubble markers in the plot.
Transparency of bubble - sets the level of transparency of overlapping bubbles in the plot.
The plot below shows an MDS analysis of the Dune species demo data set, and demonstrates the use of the bubble plot to highlight the variables responsible for sites' grouping on the plot. Sites 3, 4, 8, 9, 12 and 13, and sites 14, 15, 16 and 20, have formed two distinct groups on the MDS plot. Using the drop-down menu to select individual variables, we can show that the magnitude of the occurrence of species 'Agrsto' (the grass, Agrostis stolonifera) is in part responsible for the positioning of the two groups of sites. Scrolling through the other variables, you may determine which other species contribute to the position of the two groups, and which species distinguish between the two groups.
You can also draw a perimeter around groups: