[R] OT: single measure of (group) correlation with more than two vectors?

John Kane jrkrideau at yahoo.ca
Mon Jan 21 15:13:32 CET 2008

What you may want is a measure of inter-rater
reliablity though what you are discussing is not the
way it is normally used. 
  Try googling "inter-rater reliability" or have a
look at a text book like  Introduction to classical
and modern test theory  Linda Crocker and James
New York : Holt, Rinehart, and Winston, 1986.

--- context grey <mobygeek at yahoo.com> wrote:

> Hello, 
> I have a survey in which a number of people rated a
> set of items on a 
> 1..5 scale.  I believe it would be desirable to
> argue
> that the
> people's responses are correlated, and thus that the
> rating task
> makes sense to people.
> Is there a standard approach to this?   With only 2
> people,
> the correlation coefficient between their responses
> would
> be an interpretable number, (though probably there
> is
> some
> stronger way to assess whether the results are
> _significantly_ correlated).
> With N>2 people, there is the correlation matrix,
> but
> it does not
> give a nice single number.  Thinking, the
> determinant
> of
> the matrix of response vectors might be a
> possibility,
> since it will be low if the rows are correlated.
> Though, it should be normalized somehow to be
> interpretable.
> Is there a standard approach to this problem?
> Thanks for any advice
> Be a better friend, newshound, and
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained,
> reproducible code.

More information about the R-help mailing list