CVAPLOT procedure
Plots the mean and unit scores from a canonical variates analysis (D.A. Murray).
Options
Parameters
Description
CVAPLOT plots information from a canonical variates analysis. The type of graph to be displayed is controlled by the PLOT option with settings meanscores to draw mean scores, unitscores to display the unit scores and confidenceregion to display the 95% confidence regions about the means or the tolerance region for a population. The CREGION option specifies the type of confidence region that is drawn. The setting mean will draw the confidence region about the population means, and population plots the tolerance region for the populations. By default a 95% confidence region is calculated, but this can be changed by setting the CIPROBABILITY option to the required value (between 0 and 1).
Matrices containing the mean scores and units scores can be supplied directly, using options MSCORES and USCORES respectively, and option GROUPS can supply a factor defining the groupings of the units in the canonical variates analysis. Alternatively, you can supply a within-group SSPM and the scores will be calculated within the procedure, using the CVA directive and the CVASCORES procedure, and the groups will be accessed from within the SSPM.
The YDIMENSION and XDIMENSION parameters specify which dimensions are to be plotted in the y and x directions; by default these are dimensions 1 and 2 respectively. The WINDOW parameter indicates the window to be used for each plot (default 1), the TITLE parameter provides a title for each plot, and the SCREEN parameter indicates whether existing plots on the screen are to be kept or cleared each time (the default being to clear the screen).
Options: PLOT, GROUPS, MSCORES, USCORES, WSSPM, CREGION, CIPROBABILITY.
Parameters: YDIMENSION, XDIMENSION, TITLE, WINDOW, SCREEN.
Method
The CVA directive and the CVASCORES procedure are used to calculate the scores if necessary. A two dimensional representation of the results of the CVA is then plotted on the current high resolution graphics device. The 95% confidence region of the group means is calculated by circles of radius
SQRT( EDCHISQUARE( CIPROBABILITY; 2 ) / n )
about the means and the tolerance region of the populations is calculated by
SQRT( EDCHISQUARE( CIPROBABILITY; 2 ) )
(see Krzanowski 1988, page 374).
Reference
Krzanowski, W.J. (1988). Principles of Multivariate Analysis. Oxford University Press, Oxford.