[R] multiple-test correlation
Robert A LaBudde
ral at lcfltd.com
Sun Feb 14 20:04:49 CET 2010
At 07:35 AM 2/14/2010, Manuel Jesús López Rodríguez wrote:
>I am trying to study the correlation between one
>"independent" variable ("V1") and several others
>dependent among them ("V2","V3","V4" and "V5").
>For doing so, I would like to analyze my data by
>multiple-test (applying the Bonferroni´s
>correction or other similar), but I do not find
>the proper command in R. What I want to do is to
>calculate Kendall´s correlation between "V1" and
>the others variables (i.e. "V1" vs "V2", "V1" vs
>"V3", etc.) and to correct the p values by
>Bonferroni or other. I have found
>"outlier.test", but I do not know if this is
>what I need (also, I would prefer to use a less
>conservative method than Bonferroni´s, if possible).
>Thank you very much in advance!
One approach might be to first test for any
correlations via a likelihood ratio test:
Ho: P = I (no correlations) or covariances are diagonal
T = -a ln V ~ chi-square [p(p-1)/2]
V = det(R)
a = N -1 - (2 p +5)/6 N = # data
p = # variables
Reject Ho if T > X^2 (alpha, p(p-1)/2)
Then do the pairwise tests without familywise
error control. I.e., this is similar to doing the
F test in ANOVA before doing LSD testing.
Robert A. LaBudde, PhD, PAS, Dpl. ACAFS e-mail: ral at lcfltd.com
Least Cost Formulations, Ltd. URL: http://lcfltd.com/
824 Timberlake Drive Tel: 757-467-0954
Virginia Beach, VA 23464-3239 Fax: 757-467-2947
"Vere scire est per causas scire"
More information about the R-help