|
Jackknife estimates of assignment performance. |
|||||
| % left out in Jack – knife (500 replicates) |
1 |
10 |
50 |
80 |
CVA model |
|
|
|||||
| % correct |
93.6 |
92,7 |
90.3 |
72.1 |
Based on parental groups |
| % correct ns. |
0.5 |
0.5 |
1.3 |
10.5 |
|
| % false |
5.9 |
6.8 |
8.3 |
14.8 |
|
| % false ns. |
0.0 |
0.0 |
0.1 |
2.6 |
|
| % correct |
84.7 |
84.3 |
80.4 |
65.3 |
Including parental groups and hybrids |
| % correct ns. |
0.1 |
0.0 |
0.4 |
4.2 |
|
| % false |
15.3 |
15.6 |
19.2 |
28.9 |
|
| % false ns. |
0.0 |
0.0 |
0.1 |
1.6 |
|
|
Jackknife test of assignment. Percent correct and false assignments when fractions of 1% to 80% of the specimens are left out in the CVA procedure and then assigned to groups in the remaining dataset. Large fractions of the data can be removed without loss of the discriminatory power of CVA axes. | |||||
Nolte and Sheets Frontiers in Zoology 2005 2:11 doi:10.1186/1742-9994-2-11 |
|||||