Information Technology Reference
In-Depth Information
Ta b l e 4 . 4
Overall means, group means for three uncentred
and centred variables of the
Ocotea
data, together with the
canonical means calculated for the weighted and unweighted
CVAs. The sample sizes are:
Obul
, 20;
Oken
,7;
Opor
, 10.
Overall means
Ve s D
Ve s L
F i b L
113.946
391.243
1221.811
Group means of original data matrix
Ve s D
Ve s L
F i b L
Obul
98.100
412.000
1185.400
Oken
137.286
401.714
1568.857
Opor
129.300
342.400
1051.700
Group means of centred data matrix
Ve s D
Ve s L
F i b L
Obul
−
15.846
20.757
−
36.411
Oken
23.340
10.471
347.046
Opor
15.354
−
48.843
−
170.111
Canonical group means: Weighted CVA
Dimension
1
2
3
Obul
−
0.074
−
0.186
0.000
Oken
0.551
0.110
0.000
Opor
−
0.237
0.294
0.000
Canonical group means: Unweighted
I
CVA
Dimension
1
2
3
Obul
−
0.078
−
0.184
0.000
Oken
0.553
0.099
0.000
Opor
−
0.232
0.299
0.000
Canonical group means: Unweighted cent. CVA
Dimension
1
2
3
Obul
−
0.063
−
0.190
0.000
Oken
0.543
0.141
0.000
Opor
−
0.254
0.280
0.000
Actually the call was also made by specifying
weightedCVA = "unweightedI"
and
weightedCVA = "unweightedCent"
. Recall that in Tables 4.4 - 4.11, reference to
dimension 3, say, does not refer just to the third dimension but to the cumulative effects
of all three dimensions.
Tables 4.4 - 4.11 are given mainly to show the kind of output available from our R
functions. In this example they are not particularly interesting because, with only three
groups, all group information is exact in two dimensions; hence the many 100% values
shown for dimensions 2 and 3. Furthermore, only three variables have been used, again
giving exact information on the variables in three dimensions. We have also given the