# [R] Standard correlation

michael watson (IAH-C) michael.watson at bbsrc.ac.uk
Fri Sep 3 15:20:39 CEST 2004

```OK I better clarify what I mean as it appears it may not be a standard
test.

The pearson correlation coefficient, in laymans terms, uses the shape of
a curve around that curve's average to compare two curves.  The standard
correlation coefficient measures the shape of a curve around zero, and
uses that to compare the two curves.

Therefore a measure that starts at 1 and increases away from zero, and a
measure that starts at -4 and increases towards zero, will be deamed
similar via pearson's correlation coefficient, and dissimilar via the
standard correlation coefficient.  This is useful when "increase away
from zero" is very different behaviour from "increase towards zero".

There are some descriptions here:

http://ccgb.umn.edu/support/software/gspring/HelpPages/GSUM-120.html
http://www.optimaldesign.com/AMHelp/HowTo/HowToChooseClustParam.htm

-----Original Message-----
From: Stefan Drees [mailto:sdrees at sdrees.de]
Sent: 03 September 2004 14:03
To: r-help at stat.math.ethz.ch
Cc: michael watson (IAH-C); Stefan Drees
Subject: Re: [R] Standard correlation

On Fri, Sep 03, 2004 at 01:30:36PM +0100 - a wonderful day
- michael watson (IAH-C) wrote:
> Is there a function for computing the standard correlation
> coefficient (not pearson) in R?
help (cor) yields the following in my R 1.9.1 installation:
"""
...
cor(x, y = NULL, use = "all.obs",
method = c("pearson", "kendall", "spearman"))
...
"""

HTH,
Stefan.
--
.o. e-mail: stefan at drees.name, web: www.sdrees.org, +49 700 SDREESDE ..o
fingerprint = 516C C4EF 712A B26F 15C9  C7B7 5651 6964 D508 1B56 ooo
stefan drees  -  consulting and lecturing  -  problems to tasks

```