Pearson r is exceptional in that it is the descriptive statistic, the
point estimate of the parameter, the test statistic, and the standardized
effect size estimate, all in one. In the dark ages, stats texts included a
table of "critical values of r given n," so there was no need to compute t or
F. Today many stats programs give you r and p without t or F.
I advise my students to indicate the sample size, in one of these ways:
r(n = 96) = .37, p = .xxx
r = .37, t(94) = xxx, p = .xxx
r = .37, F(1, 94) = xxx, p = .xxx
and then report a confidence interval for rho.
Cheers,
Karl W.
-----Original Message-----
From: Rick Froman [mailto:[email protected]]
Sent: Tuesday, July 21, 2009 5:32 PM
To: Teaching in the Psychological Sciences (TIPS)
Subject: RE:[tips] Reporting Correlations in APA Style
I wish this was clearer because there are all kinds of variations in the
reporting of correlations in the literature. It seems that most often they are
reported in tables because you seldom see a research project with a single or
even a few correlations. Usually there are a number of correlations reported.
When there is just one, I sometimes see the N reported instead of the df. If
you are going to just report N instead of the df, I believe it should be
something like r(N=25)=.75, p=.02. However, since the calculation of the p
requires the use of df, I believe it should be r(23)=.75, p=.02. None of the
examples in the 6th edition involve Pearson r but the relevant passage says on
p. 34, "For inferential statistical tests (e.g., t, F, and chi square tests),
include the obtained value or magnitude of the test statistic, the degrees of
freedom, the probability of obtaining a value as extreme or more extreme than
the one obtained (the exact p value), and the size and direction of the
effect." This always leaves me in doubt because the t-distribution is actually
used to determine the p-value for the Pearson r correlation so it seems as if
the t result might go in there somewhere but I have seldom seen that done. The
manual also goes on to suggest reporting effect sizes and confidence intervals.
However, the basic format for reporting inferential results doesn't seem to
have changed from the 5th edition.
Rick
Dr. Rick Froman, Chair
Division of Humanities and Social Sciences
John Brown University
Siloam Springs, AR 72761
[email protected]
________________________________________
From: Paul C Bernhardt [[email protected]]
Sent: Tuesday, July 21, 2009 3:03 PM
To: Teaching in the Psychological Sciences (TIPS)
Subject: [tips] Reporting Correlations in APA Style
What is the proper way to report a single correlation within the text of a
paper? Do you report degrees of freedom or N? There is little consistency in
what I find searching on the internet for university APA help sites. I wasn't
able to find the answer at APAStyle.org.
Particularly considering there is a new 6th edition, maybe those of you who
have obtained it can comment if it has better information. I've looked through
the 5th edition and come up empty. The example paper in the 5th edition that
has a correlation for a study with a sample of 60 showed r(59) = .87, p < .01.
Curious. Either there was a lost participant, not mentioned because of how they
laid out the paper in the book, or they thought the degrees of freedom for a
correlation was N-1, which is incorrect (it is N-2).
Paul C. Bernhardt
Department of Psychology
Frostburg State University
Frostburg, Maryland
---
To make changes to your subscription contact:
Bill Southerly ([email protected])
---
To make changes to your subscription contact:
Bill Southerly ([email protected])
