correlation coefficients - upper case or lower case R?

[Roo]

Could someone tell me whether I should be using an upper case or lower
case "R" when reporting correlation coefficients for univariate
analysis? SPSS always uses upper case, but I've read comments elsewhere
that it should strictly speaking be lower case (although upper case "R"
for multivariate analysis). However, "r" seems to be used for Pearson
correlation coefficient, which I always thought was negative or
positive - and the SPSS "R" value is always positive (only the
standardised beta coefficient can be negative) - so may be I'm
inappropriately using "R" where I meant to use Pearson "r" coefficient
(despite the fact they're the same value)?????

Help - I'm so confused! RRRRRRRRrrrrrrrrrrrrgh! (or is that
rrrrrrrRRRRRRRgh?)

Many thanks!

Roo Payne

[Richard Ulrich]

Post by Roo
Could someone tell me whether I should be using an upper case or lower
case "R" when reporting correlation coefficients for univariate
analysis? SPSS always uses upper case, but I've read comments elsewhere
that it should strictly speaking be lower case (although upper case "R"
for multivariate analysis). However, "r" seems to be used for Pearson
correlation coefficient, which I always thought was negative or
positive - and the SPSS "R" value is always positive (only the
standardised beta coefficient can be negative) - so may be I'm
inappropriately using "R" where I meant to use Pearson "r" coefficient
(despite the fact they're the same value)?????
Help - I'm so confused! RRRRRRRRrrrrrrrrrrrrgh! (or is that
rrrrrrrRRRRRRRgh?)

It sounds to me like your intuitions are pretty sound, for
using r versus R in the same ways that I do. I don't know that
I have read any set of rules, but I know that you can always
use a small r for a Pearson r; and I think a small r is
questionable (or wrong) anywhere else.

I don't know which "SPSS 'R' value is always positive" -- but
that would, indeed, indicate that it is not (strictly) a Pearson r .
R-squared is generally in capitals, even though it could be
interpreted as a Pearson statistic. Multivariate R's are generally
in capitals, too - I suppose, because they are not univariate.

But, like I said, I don't know of any source of rules.
Until you find one, be consistent in what you write, and don't
argue too hard with an editor who wants it otherwise.