Why is Baillie/Pilcher corr coff not same as BaPi corr?
Posted: 05 Mar 2009, 01:03
Question: Are the indicated fields above signs of an error in CDendro? Why these differences?
In this case, the normalized data of the reference has been calculated as mean values from normalized data of all members of a collection.
The sample 6680NY has then been compared with that reference curve and the best matches displayed in the four leftmost columns of the table above.
As a complement to the table, a number of other correlation indicators can be displayed. E.g. GLK and Skeletonvalues, but also the Baillie/Pilcher (BaPi) value as shown in column number 5 above. Though when the Baillie/Pilcher values in the "BaPi" column were calculated they were made from a normalization of the mean value ring width curve of the reference. This difference in the way data was calculated explains the differences in the table.
When the reference is based on a single member or on a .wid-file, then this type of differences will not show up!
Note: The table is the result of a testing case. Normally you would never use the Baillie/Pilcher algorithm as your main normalization method and also ask for a BaPi column as above.