Premium
Assessing the relative contribution of variables in canonical discriminant analysis
Author(s) -
Tardif Bernard,
Hardy Jean
Publication year - 1995
Publication title -
taxon
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.819
H-Index - 81
eISSN - 1996-8175
pISSN - 0040-0262
DOI - 10.2307/1222679
Subject(s) - linear discriminant analysis , jackknife resampling , statistics , mathematics , canonical analysis , sample size determination , discriminant , econometrics , computer science , artificial intelligence , estimator
Summary Tardif, B. & Hardy, J.: Assessing the relative contribution of variables in canonical discriminant analysis. – Taxon 44: 69‐76. 1995. – ISSN 0040‐0262. A procedure by which subsets of variables can be selected and their relative importance assessed in canonical discriminant analysis (CDA) is discussed. It is apparent from a review of the literature that only the standardized coefficients can be used for this purpose. However, an example relative to morphological variation of chives ( Allium schœnoprasum L.) in eastern North America indicates that it would be hazardous, at least with small sample size, to select variables by the magnitude of their standardized coefficients alone, considering their exceeding instability when different sets of observations are included in the analysis. We argue that selection of variables, to be conducted with confidence, should be based upon t‐values associated with jackknife standardized coefficients.