DATA:      Fisherís Iris data, e.g. in Mardia, Kent & Bibby (1979) or Minitab
           example files
DIRECTORY: \CANOCO\SAMPLES\METHODS\CVA
PROBLEM:   How do three Iris species differ in flower morphology?
ILLUSTRATION OF:
+ a standard CVA with triplot showing group overlap and group means
+ permutation test of differences between groups using CVA
+ forward selection in CVA
+ how to obtain means and correlations of the variables (for all data or 
  per group)
+ how to obtain a pooled within-group correlation matrix
_______________________________________________________________________________
FILE TYPE    |  FILE NAME    |         DESCRIPTION
-------------+---------------+-------------------------------------------------
 SPECIES     | iris_spe.dta  | Indicators to which species (Iris setosa, 
             |               |  I. versicolor and I.virginica) each of the 150
             |               |  specimens of Iris belongs (50 per group)
ENVIRONMENT  | iris_flw.dta  | 4 measurements on the flower of each specimen 
             |               | (length and width of the sepal and of the petal)
 DERIVED     | stmeans.dta   | standardized group means (spec_env.tab 
             |               |  of the CVA)
             | iris_fl2.dta  | copy of iris_flw.dta for the trick to obtain
             |               |  means and within group covariance matrix
 PROJECTS    | cva.con       | default CVA on the Iris data
             | cva_forw.con  | forward selection of variables 
             | means_v.con   | project giving means and correlations for
             |               |  I.versicolor only (using a standard trick)
             | withinco.con  | project giving the pooled within correlation
             |               |  matrix by specifying the groups as covariables
-------------+---------------+-------------------------------------------------