Professional Documents
Culture Documents
Interpreting The Magnitudes of Correlation Coefficients - James F. Hemphill
Interpreting The Magnitudes of Correlation Coefficients - James F. Hemphill
Contents monomethod correlation coefficients, which ment (Lipsey & Wilson, 1993). I chose these
yield results that are artificially large com- two reviews because they serve as important
pared with associations found between real- summaries of the psychological literature
Hemphill .....................................................78
world, independently measured variables. included in meta-analytic reviews. Studies
Baer on Otto & Heilbrun .............................79
Cohen (1988) provided perhaps the most in Meyer et al. (2001) involving medical
Heilbrun & Otto reply ..................................80
widely known guidelines or “operational def- assessments were identified and removed
initions” (p. 79) that are more realistic than from my analyses. Effect sizes, reported us-
those above for interpreting the magnitude of ing Cohen’s d by Lipsey and Wilson (1993)
DOI: 10.1037/0003-066X.58.1.78 correlation coefficients typically found in the for 302 independent meta-analytic studies
behavioral sciences. These guidelines are (see their Table 1), were converted to Pear-
Interpreting the Magnitudes “offered as a convention . . . for use when son product–moment correlations (rs; see
of Correlation Coefficients no others suggest themselves” (Cohen, Rosenthal, 1991, Formula 2.20). Correla-
1988, p. 79). According to Cohen, correla- tion coefficients were sorted in ascending
tion coefficients in the order of .10 are order in terms of their magnitudes and divid-
James F. Hemphill “small,” those of .30 are “medium,” and ed into three groups with approximately equal
Simon Fraser University those of .50 are “large” in terms of magni- numbers of coefficients. This was done sep-
tude of effect sizes (see pp. 77–81). Cohen arately for the assessment, treatment, and
The American Psychological Association’s seems to have arrived at these guidelines combined meta-analytic studies.
(APA’s) Task Force on Statistical Inference largely on the basis of his considerable As indicated in Table 1, the values in
admonishes researchers to “always present ef- experience with effect sizes and correlation the lower, middle, and upper thirds were
fect sizes for primary outcomes” (Wilkinson coefficients. For this comment, an attempt similar for the sets of meta-analytic studies
& the APA Task Force on Statistical Infer- was made to extend Cohen’s benchmarks independently compiled by Meyer et al.
ence, 1999, p. 599), and the fifth edition of the by deriving empirical guidelines concerning (2001) and by Lipsey and Wilson (1993).
Publication Manual of the American Psycho- the magnitude of correlation coefficients The current analyses suggest that approxi-
logical Association states that “for the reader to found among psychological studies. mately one third of the correlation coeffi-
fully understand the importance of your find- cients presented in Table 1 are less than .20,
ings, it is almost always necessary to include
Empirical Guidelines one third fall between .20 and .30, and one
some index of effect size or strength of rela- third are more than .30 in magnitude (see the
tionship” (APA, 2001, p. 25). Given the sig-
I have examined and reanalyzed two large, rightmost column of Table 1). The large
nificance of these statements and the wide-
diverse, and impressive summaries of the number of research participants, the large
spread use of the correlation coefficient as a
research literature concerning psychological number of meta-analytic studies (i.e., 380)
key index of effect size (Meyer et al., 2001;
assessment (Meyer et al., 2001) and treat- on which these values are based, the simi-
Rosenthal, 1991), it is perhaps surprising that
empirical guidelines for interpreting the mag-
nitude of correlation coefficients typically found
among psychological studies are not widely Table 1
available. The Distribution of Correlation Coefficients Found Among Studies Included in
In the absence of empirical guidelines, Meta-Analytic Reviews, and Empirical Guidelines for Interpreting the Magnitude
investigators presumably will impose other of Correlation Coefficients
guidelines for interpreting the magnitude of
Distribution
correlation coefficients. Many of these other of correlation Assessment Treatment Combined Empirical
guidelines by which the magnitude of corre- coefficients reviewa reviewb reviewsc guidelinesd
lation coefficients are compared are unreal-
istically large and “inappropriate” (Meyer et Lower third .02 to .21 – .08 to .17 – .08 to .17 < .20
al., 2001, p. 132). Three of these bench- Middle third .21 to .33 .17 to .28 .18 to .29 20 to .30
marks that Meyer et al. (2001) discussed Upper third .35 to .78 .29 to .60 .30 to .78 > .30
include comparisons with (a) a perfect cor- a
78 meta-analytic studies concerning psychological assessment reviewed by Meyer et al. (2001).
relation, which is virtually never found in b
302 meta-analytic studies concerning psychological treatment reviewed by Lipsey and Wilson (1993).
applied psychological research; (b) reliabili- c
A total of 380 meta-analytic studies.
ty coefficients, which often greatly exceed d
Guidelines are based on all 380 meta-analytic studies.
the values of validity coefficients; and (c)