Journal of Research in Music
Education
http://jrm.sagepub.com/

Standardization of the Gordon Primary Measures of Music Audiation in
Greece
Lelouda Stamou, Charles P. Schmidt and Jere T. Humphreys
Journal of Research in Music Education 2010 58: 75
DOI: 10.1177/0022429409360574
The online version of this article can be found at:
http://jrm.sagepub.com/content/58/1/75

Published by:
http://www.sagepublications.com

On behalf of:

National Association for Music Education

Additional services and information for Journal of Research in Music Education can be found at:
Email Alerts: http://jrm.sagepub.com/cgi/alerts
Subscriptions: http://jrm.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://jrm.sagepub.com/content/58/1/75.refs.html

>> Version of Record - Apr 22, 2010
What is This?
Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

Standardization of the
Gordon Primary Measures of
Music Audiation in Greece

Journal of Research in Music Education
58(1) 75­–89
© 2010 MENC: The National
Association for Music Education
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0022429409360574
http://jrme.sagepub.com

Lelouda Stamou1, Charles P. Schmidt2,
and Jere T. Humphreys3

Abstract
The purpose of this study was to standardize the Primary Measures of Music Audiation
in Greece (N = 1,188). Split-halves reliability was acceptable across grade levels
(K through 3) for the Tonal and Rhythm subtests, but test–retest reliability was
generally unacceptable, especially for the Rhythm subtest. Concurrent validity was
mixed, with teacher ratings of musical achievement generally significantly correlated
with Tonal but not Rhythm subtest scores. Composite test means were significantly
higher for suburban and urban samples than for rural samples and were significantly
higher for higher grade levels. Item difficulty coefficients were significantly correlated
across grade levels. The Greek and U.S. composite means were similar except for
a significantly higher U.S. mean for grade 1. However, when the rural subgroup was
removed from the Greek sample to equate with the U.S. norming sample, there were
nonsignificant differences for grades K through 1, but significant differences in favor
of the Greek sample for grades 3 and 4.
Keywords
music aptitude tests, audiation, standardization, Greece, cross-cultural
The measurement of music aptitude has been of interest to researchers since the beginning of the 20th century. Emanating from the psychological research of Francis Galton
(1822–1911) and James Cattell (1860–1944), interest in quantifying individual differences and devising discrete measures of sensory acuity led to Carl Seashore’s music
1

University of Macedonia, Thessaloniki, Greece
Indiana University, Bloomington, IN, USA
3
Arizona State University, Tempe, AZ, USA
2

Corresponding Author:
Jere T. Humphreys, School of Music, Arizona State University, Tempe, AZ 85287-0405
Email: Jere.Humphreys@asu.edu

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

76		

Journal of Research in Music Education 58(1)

aptitude tests, published in 1919 (Humphreys, 1993, 1998). In the second half of the
20th century, based on the work of Seashore (1919), his followers, and others (Bentley,
1966; Drake, 1933, 1954; Gaston, 1957; Kwalwasser, 1953, 1955; Kwalwasser &
Dykema, 1930; Tilson, 1941; Whistler & Thorpe, 1950; Wing, 1958), Edwin Gordon
conducted research that led to the development of four music aptitude tests: Primary
Measures of Music Audiation (PMMA) for grades K through 3 (Gordon, 1979), Intermediate Measures of Music Audiation (IMMA) for children in grades 1 through 4 who
score above the 80th percentile on the PMMA (Gordon, 1982), Musical Aptitude Profile (MAP) for grades 4 through 12 (Gordon, 1965), and Advanced Measures of Music
Audiation (AMMA) for college-age students (Gordon, 1989).
These tests have been used in numerous research studies not only in the United States,
where all four tests were normed, but also in other countries. However, LeBlanc, Jin,
Stamou, and McCrary (1999) concluded that “research findings based upon one culture
cannot safely be assumed to apply to other cultures” (p. 76). More specifically, FeayShaw (2000) questioned the lack of concern shown for differences in culture, ethnicity,
and socioeconomic status in several research studies involving tests of music aptitude,
especially in how participant groups were described and how results were generalized
and otherwise interpreted. Later, on the basis of results of a study of the effects of beat
strength on music listening preference, LeBlanc et al. (2002) concluded that a variable
called “country” or “culture” should be added to the LeBlanc (1980) preference model.
Several researchers have investigated how Gordon’s music aptitude tests functioned in various countries or cultures. Chung (2002) reported comparable reliability
coefficients for the IMMA with Korean students in Korea and Korean American students (grades 1 through 4). Schoenoff (1972) found no significant differences between
MAP scores of German participants and the published norms (grades 4 through 12),
but Sell (1976) and Chuang (1997) reported significantly higher scores than the MAP
norms for Finnish and Taiwanese students, respectively, as well as high reliability and
concurrent validity coefficients. Jung (1992) conducted a 2-year predictive validity
study of the MAP for use in Korea and found that MAP scores correlated significantly
with a measure of Western music achievement (r = .33 to r = .66). In contrast, MAP
scores had low or nonsignificant correlations with a measure of achievement in Korean
traditional music (r = .01 to r = .28). Özeke and Humphreys (2000) and Wang (2007)
reported significantly higher scores on the AMMA than the published norms for Turkish
and Taiwanese students, respectively. Gordon (1991) validated the AMMA specifically
with and for German students.
In foreign applications of the PMMA, Yang (2002) reported that first-grade students in Taiwan scored significantly lower than the published norms on the Tonal
subtest and nonsignificantly lower on the Rhythm subtest. Yang reported split-halves
reliability coefficients ranging between .84 and .91. Examining the PMMA with a
sample of English children, Holahan and Thomson (1981) found adequate split-halves
reliability (r = .86 to r = .93) for the Tonal subtest by age group. In contrast, split-halves
reliability coefficients for the Rhythm subtest ranged from unacceptable (r = .45) to
marginally acceptable (r = .71). Gouzouasis (1993) found significant differences in

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

77

Stamou et al.

PMMA Tonal subtest scores, but not in Rhythm subtest scores, among groups of
5-year-old Canadian children of Chinese, East Indian, and Western European ethnic
backgrounds. Gouzouasis reported high split-halves reliability (r > .90) for both Tonal
and Rhythm subtests for each subgroup. However, the author concluded that content
validity for the Tonal subtest “may be called into question for use with East Indian
children” (p. 75), who scored lower than those in the other groups. The current investigation focused on use of the PMMA in Greece. One study (Pollatou, Karidimou, &
Gerodimos, 2005) involved use of the PMMA in Greece with a sample of 5- and
6-year olds (N = 95). The authors reported descriptive results for the PMMA as follows: Tonal subtest (M = 22.68, SD = 5.78), Rhythm subtest (M = 22.62, SD = 4.64).
No reliability or validity data were reported.
When standardized music tests are applied to populations other than the ones for
which the tests were constructed and normed, the crucial issue of validity emerges.
Because the validity and reliability of measurement instruments cannot be assumed
to transfer across populations, empirical data are needed to determine the viability of
measurement instruments, particularly when used cross-culturally. Therefore, the purpose of this study was to standardize a well-known music aptitude test for use in
Greece: the PMMA by Edwin E. Gordon (1979). The two main research objectives
were to (a) develop PMMA norms appropriate for use in Greece and (b) determine the
reliability and concurrent validity of each subtest at each grade level. Other research
objectives were to (a) compare results by grade (K through 3) and type of school setting (i.e., urban, suburban, rural) and (b) compare results from Greece with those from
Gordon’s original norming study.

Method
The written instructions for the PMMA Tonal and Rhythm subtests were translated
from the original English into Greek by the primary author of this study, a native Greek
with fluency in English. She also replaced the English-language instructions on the
audio test CDs with Greek translations.
The main standardization sample for this study included nearly equal numbers of
students from 12 urban schools (n = 397, 33.4%), 14 suburban schools (n = 391,
32.9%), and 16 rural schools (n = 399, 33.6%), all from Greece (N = 1,188). Participating
schools were classified as urban, suburban, or rural according to an official demographic
map published by the Greek government (General Secretariat of National Statistical
Service of Greece, 2008). This was a nonrandom sample determined by accessibility
(i.e., teachers’ interest and willingness to participate) to the three types of schools and
adequate class sizes. Every attempt was made to obtain a geographically diverse
(including Greek Islands) and demographically varied sample. Class size ranged from
10 to 26 students. The male:female ratios were also nearly equal for grade 1 (n = 86:100),
grade 2 (n = 173:161), and grade 3 (n = 293:271); however, the kindergarten sample
consisted of 38 males (36.9%) and 65 females (63.1%). The overall usable sample
(grades K through 3) consisted of 591 males (49.75%) and 597 females (50.25%). The

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

78		

Journal of Research in Music Education 58(1)

aptitude test was administered by the regular music teachers, all of whom had been
trained in music aptitude test administration as part of a 20-clock-hour seminar in quantitative research and music aptitude testing organized and taught by the three present
authors (Stamou, Humphreys, & Schmidt, 2006).
As a parallel to Gordon’s (1979) methodology, we employed split-halves and retest
reliability procedures. Samples employed for the retest reliability calculations for each
grade and subtest or composite ranged from n = 31 for the grade 1 composite (i.e.,
Tonal and Rhythm subtests combined) to n = 112 for the grade 2 Rhythm subtest. The
time interval between test administrations ranged from 10 days to 4 weeks.
For purposes of assessing concurrent criterion validity, we asked music teachers to
provide ratings of individual students’ music aptitude and music achievement based
on their observations of individual students across time. The researchers provided
teachers with written and verbal explanations concerning the definitions of music aptitude (i.e., perceived potential for future success in music) and music achievement (i.e.,
current musical competence or behavior) and were asked to rate each student on a
7-point scale (1 = low, 7 = high) for each category. The practice of correlating teacher
ratings with test scores is similar to methodology employed by Gordon (1965) in his
validation of the MAP. However, this differed from Gordon’s (1979) validation procedures for the PMMA, as he assessed validity by correlating scores with standardized
measures of academic achievement for each grade level and by correlating PMMA
and MAP scores for a sample of fourth graders.

Results
Descriptive statistics and reliability coefficients for each subtest and composite are
presented by grade level in Table 1. Split-halves reliability coefficients for the
Tonal subtest ranged from moderate for the kindergarten sample (r = .77) to moderately high for grades 1, 2, and 3 (r = .86 to .88). Split-halves coefficients for the
Rhythm subtest ranged from unacceptable levels for kindergarten (r = .42) and
grade 1 (r = .50) to marginally acceptable levels for grade 2 (r = .66) and grade 3
(r = .69). Split-halves reliability coefficients for the composite measure were
acceptable, ranging from r = .73 for kindergarten to r = .86 for grade 3 (Nunnally,
1970). However, the relatively low reliability of the Rhythm subtest accounts for
the unusual finding that for each grade level, the 40-item Tonal subtest was more
reliable than the 80-item composite (Table 1).
Retest reliability results for the PMMA Tonal, Rhythm, and composite measures varied
markedly by grade level and by subtest but were generally unacceptable, with Rhythm
subtest scores being particularly unstable across time (Table 1). We excluded the kindergarten retest data because of inadequate sample sizes (retest n = 11 to 15). The highest
retest reliability was found for grade 1 Tonal (r = .54), Rhythm (r = .45), and composite
(r = .64). There was a modest degree of stability in scores for grade 2 Tonal (r = .42) and
composite (r = .50) but very low stability for grade 2 Rhythm (r = .19). Grade 3 reliability
results were even lower, ranging from r = –.11 for Rhythm to r = .10 for the composite test.

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

79

Stamou et al.

Table 1. Primary Measures of Music Audiation Means, Standard Deviations, and Reliabilities
Current study

Gordon (1979)

Reliability
		
Grade and measure
M (SD)

Split			
halves
Retest
M (SD)

Reliability
Split
halves Retest

Kindergarten						
Tonal (n = 126)
22.95 (5.33) .77
—
24.70 (5.28)
.88
Rhythm (n = 129)
23.46 (4.22) .42
—
22.30 (3.74)
.72
Composite (n = 103) 46.56 (7.94) .73
—
47.00 (7.65)
.90
Grade 1						
Tonal (n = 224)
27.68 (5.50) .86
.54 (n = 52) 29.80 (5.03)
.89
Rhythm (n = 222)
25.94 (4.41) .50
.45 (n = 37) 25.80 (4.34)
.85
Composite (n = 186) 53.63 (8.81) .84
.64 (n = 31) 55.60 (8.25)
.92
Grade 2						
Tonal (n = 358)
31.22 (5.41) .80
.42 (n = 99) 32.00 (4.75)
.89
Rhythm (n = 398)
29.23 (4.76) .66
.19 (n = 112) 27.70 (4.55)
.86
Composite (n = 334) 60.87 (8.50) .77
.50 (n = 83) 59.70 (8.35)
.92
Grade 3						
Tonal (n = 620)
32.90 (5.04) .88
–.01 (n = 64) 34.60 (3.35)
.85
Rhythm (n = 652)
31.05 (4.58) .69
–.11 (n = 104) 29.40 (3.99)
.86
Composite (n = 564) 64.05 (8.45) .86
.10 (n = 50) 64.00 (6.29)
.90

.73
.60
.74
.70
.66
.75
.70
.73
.76
.68
.66
.73

Split-halves reliability coefficients were calculated on even and odd items with Spearman-Brown
correction. Retest reliability coefficients for kindergarten are not shown because of small (retest) sample
sizes.

As a measure of concurrent validity, teachers’ ratings of individual students’
musical achievement were correlated with student scores on the PMMA subtests
and composite for grades 1 through 3 (Table 2). There were significant correlations
for the Tonal subtest for grades 1 and 3 and for the composite test for grade 1 (rho =
.39 to .55). By contrast, teachers’ achievement ratings and Rhythm subtest scores
were not significantly correlated at any grade level (rho = .08 to .21). An additional
measure of concurrent validity, teachers’ ratings of students’ musical aptitude, was
not significantly correlated with subtest or composite test scores at any grade level
(Table 2).
Intercorrelations between the Tonal and Rhythm subtests ranged from r = .43 for
kindergarten to r = .62 for grades 1 and 3 (Table 3). Correlations between the Tonal
and composite scores ranged from r = .89 to r = .92 and between Rhythm and composite scores from r = .80 for kindergarten to r = .89 for grades 2 and 3. All
correlations were statistically significant (p < .001) because of the large sample
sizes involved.
Given the high correlations between the PMMA subtests and composite, only the
composite scores were compared for school settings and grade levels. The assumption

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

80		

Journal of Research in Music Education 58(1)

Table 2. Spearman Correlations (rho) for Teachers’ Musical Achievement and Aptitude
Ratings With PMMA Tonal, Rhythm, and Composite Scores
Grade and measure

n

Musical achievement rating

Musical aptitude rating

Grade 1			
Tonal
21
.53*
Rhythm
21
.21
Composite
21
.55*
Grade 2			
Tonal
21
.33
Rhythm
21
.18
Composite
21
.21
Grade 3			
Tonal
74
.39***
Rhythm
65
.08
Composite
65
.25

.23
–.08
.24
.23
.02
.03
.17
.01
.11

PMMA = Primary Measures of Music Audiation (Gordon, 1979).
*p < .05. ***p < .001.

Table 3. PMMA Intercorrelations (r) Among Subtests and Composite
Current study
Subtest

Rhythm

Gordon (1979)

Composite

Kindergarten (n = 103)
Tonal
Rhythm

.43***
—

.89***
.80***

Rhythm

Kindergarten (n = 127)
.45***
—

Grade 1 (n = 186)
Tonal
Rhythm

.62***
—

.92***
.87***

.62***
—

.91***
.89***

.49***
—

.59***
—

.90***
.89***

.84***
.82***
Grade 2 (n = 280)

.51***
—

Grade 3 (n = 564)
Tonal
Rhythm

.76***
.57***
Grade 1 (n = 202)

Grade 2 (n = 334)
Tonal
Rhythm

Composite

.89***
.88***
Grade 3 (n = 264)

.49***
—

PMMA = Primary Measures of Music Audiation (Gordon, 1979).
***p < .001.

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

.85***
.89***

81

Stamou et al.

of homogeneity of variance was met for the composite test among the urban, suburban,
and rural school settings (Levene’s F = 1.74, p > .05). An ANOVA test revealed significant overall differences among school settings, F(2, 1187) = 9.78, p < .001, but the effect
size was small (h2 = .016). Post hoc (Bonferroni) tests revealed significantly higher
means for suburban (p < .004) and urban (p < .000) schools than for rural schools.
ANOVA results confirmed that the composite means were significantly related to
grade level, F(3, 1186) = 165.24, p < .001, h2 = .30. Because of lack of homogeneity
of variance between grade levels (Levene’s F = 3.63, p < .05), the Dunnett T3 multiple
range test was used for grade-level comparisons. These results confirmed significant
differences in composite means between all pairs of grade levels, with means increasingly higher for each grade level (p < .001).
We determined item difficulty and discrimination values for the Tonal and Rhythm
subtests by grade level. For both subtests, as would be expected, mean difficulty coefficients were higher for the Tonal (M = .57, .69, .78, .82) and Rhythm (M = .59, .64,
.73, .78) subtests for each grade, kindergarten and grades 1, 2, and 3, respectively.
Similarly, the ranges of the difficulty coefficient means generally were larger for the
Tonal (.65, .69, .78, .71) and Rhythm (.62, .72, .76, .81) subtests for grades K through 3,
respectively.
The mean item discrimination values for the Tonal subtest suggest modest, decreasing levels of discriminatory power between high and low scorers for grades K through 3,
respectively (r = .36, .33, .30, .24), with similarly decreasing ranges in the discrimination coefficients (.82, .76, .60, .43). The mean discrimination values for the Rhythm
subtest were considerably higher and positively associated with grade level (r = .59,
.64, .73, .78) for grades K through 3, respectively, with similar ranges among grade
levels (.65, .65, .65, .52). The Rhythm subtest items tended to differentiate well among
overall high and low scorers.
To address the degree to which individual items were consistent in their difficulty
across grade levels, we calculated Spearman correlations among item difficulty values
for the Tonal and Rhythm subtests by grade level. The difficulty of individual Tonal
subtest items tended to be consistent (p < .001) across grade levels, with correlations
(rho) ranging from .81, between kindergarten and grade 2 and between grades 1 and
2, to .93, between grades 2 and 3. The range of correlations for item difficulty values
for the Rhythm subtest was somewhat wider: .59 between kindergarten and grade 3
and .92 between kindergarten and grade 1. Overall, there was a tendency toward relatively comparable degrees of difficulty across grade levels for individual Tonal and
Rhythm items. That is, items that were relatively difficult to discriminate at one grade
level were also relatively difficult to discriminate at another grade level.

Comparison With American Norms
Norms for the PMMA developed in this study (see Appendix) were based on a large,
cross-sectional sample of schools, teachers, and students from urban, suburban, and
rural school settings in Greece, whereas the original norms were based on a somewhat

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

82		

Journal of Research in Music Education 58(1)

smaller group of participants from a single American suburb more than 30 years ago
(Gordon, 1979). The sample sizes by grade level in the current study approximate or
exceed those from the original standardization study (kindergarten, n = 103 vs. 127;
grade 1, n = 186 vs. 202; grade 2, n = 334 vs. 280; grade 3, n = 564 vs. 264; total, n = 1,188
vs. 873, respectively). Gordon did not provide data on the gender of the norming sample
participants, although he did indicate that the sample was demographically heterogeneous.
The PMMA composite means for the U.S. (Gordon, 1979) and Greek samples were
not significantly different for kindergarten (t = .42, df = 228, p > .05, h2 = .06), grade 2
(t = –1.71, df = 612, p > .05, h2 = .14), or grade 3 (t = –.08, df = 826, p > .05, h2 = .01).1
The U.S. composite mean was significantly higher for grade 1 (t = 2.27, df = 386, p < .01,
h2 = .24). Because the U.S. sample was drawn only from a suburban setting and
because the Greek samples differed significantly between urban and suburban samples on one hand and the rural sample on the other, the authors then compared the U.S.
composite means (Gordon, 1979) with the respective grade-level means from the
combined urban and suburban Greek samples only, leaving out the Greek rural sample.
For kindergarten, the mean for the truncated Greek sample was higher than the mean
for the total Greek sample but remained nonsignificantly lower than the U.S. sample
mean (t = .06, df = 210, p > .05, h2 = .01). For grade 1, the significant difference in
favor of the U.S. sample was reduced to a nonsignificant difference when data from
the Greek rural sample were left out of the analysis (t = .91, df = 308, p > .05, h2 = .11).
The nonsignificant differences for the Greek total sample changed to significant differences in favor of the truncated Greek sample for grade 2 (t = 2.18, df = 505, p < .05,
h2 = .20) and grade 3 (t = 2.72, df = 630, p < .05, h2 = .24). Rosenthal effect sizes
confirmed that the differences between the full and truncated Greek sample and the
American sample were negligible for all four grade levels (≤ .30). Because the effect
sizes were small, the significant differences for all cross-cultural comparisons were
likely attributable to the large sample sizes.
Split-halves reliability coefficients for the Tonal subtest in the Greek sample, although
lower, are similar to those reported by Gordon (1979). The greatest difference was observed
between the Greek (r = .77) and American (r = .88) kindergarten samples. Notably,
reliability coefficients for the Rhythm subtest and composite test in the Greek sample
were substantially lower than those Gordon reported for each grade level. Moreover,
disparities among reliability coefficients for the Tonal, Rhythm, and composite means
were far greater for the Greek sample than for the original norming sample. For example,
with the Greek sample, the split-halves reliability coefficients for the composite test
ranged from r = .73 to r = .86 for kindergarten through grade 3 compared to r = .90 to
r = .92 for the U.S. sample (Table 1). Test–retest reliability coefficients were uniformly much lower for the Greek sample (r = –.11 to .64) than for the U.S. sample
(r = .60 to .76). Gordon did not report the number of participants who took the PMMA
twice, so perhaps his entire sample did so. In that case, the much smaller numbers of
Greek participants involved in the retest portion of the study could explain some of the
disparity in retest reliability between samples but probably not the entire disparity.
Moreover, the test–retest results probably cannot be explained by developmental fluctuations in music aptitude in young children as Gordon (1979) claims, because grade 1

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

83

Stamou et al.

participants’ scores were the most stable, whereas grade 3 scores were the least stable.
Similarly, the relative lack of reliability in the Greek sample cannot be attributed to a
lack of variance in the scores relative to the American sample (Table 1). Because the retest
samples in the present study were relatively small, additional research is warranted.
The intercorrelations between the Tonal, Rhythm, and composite measures for the
Greek sample were higher than those for the U.S. sample (Gordon, 1979) for each
grade level, except for a slightly higher U.S. correlation coefficient for the Tonal and
Rhythm subtests for the kindergarten samples and identical coefficients for the
Rhythm subtest and composite test for grade 3. These higher correlations for the Greek
sample may have resulted from the larger standard deviations in most subtests in the
current study, likely attributable in part to the more heterogeneous Greek sample.
Mean item difficulty levels on the Tonal subtest for the Greek sample were slightly
lower than those reported by Gordon (1979; kindergarten, .57 vs. .64; grade 1, .69 vs. .75;
grade 2, .78 vs. .81; grade 3, .82 vs. .86, respectively). However, correlations for the
difficulty levels between the Greek and U.S. samples were statistically significant and
positive, ranging from rho = .75 to rho = .87 for grades 1 and 2, respectively (Table
4), indicating a moderately high degree of consistency in the difficulty level of the
Tonal subtest items across the two samples. Item discrimination values were also similar on the Tonal subtest between the two samples, except for a lower value for grade
3 of the Greek sample compared to the U.S. sample (kindergarten, .36 vs. .37; grade 1,
.33 vs. .38; grade 2, .30 vs. .38; grade 3, .24 vs. .36, respectively). All correlations for
tonal item discrimination were also positive, and half were statistically significant
(Table 4), but the coefficients were lower than those for tonal item difficulty because
of the relatively restricted range of the item discrimination values. The results for both
the Greek and U.S. samples suggest that as a whole, the Tonal subtest items have a
modest level of discriminatory power. That is, the items do not tend to have great
power to differentiate overall high versus low achievers on the tonal subtest.
Difficulty levels were nearly identical on the Rhythm subtest for the Greek and U.S.
samples (kindergarten, .59 vs. .58; grade 1, .64 vs. .66; grade 2, .73 vs. .71; grade 3,
.78 vs. .75, respectively). This similarity is reflected in the positive and statistically significant correlations for the Rhythm subtest between the two samples, ranging from
moderate for grade 2 (rho = .39) to high for kindergarten (rho = .81) (Table 4). Greek
discrimination values were somewhat lower than the U.S. sample for the Rhythm
subtest (kindergarten, .25 vs. .28; grade 1, .25 vs. .31; grade 2, .28 vs. .33; grade 3, .23 vs. .34,
respectively). As was the case for the Tonal subtest, correlations for rhythm item discrimination were lower than for rhythm item difficulty, with a range of rho = .38 to
rho = .64. All correlations for item difficulty and item discrimination were positive
between the Greek and U.S. samples, and most were statistically significant (Table 4).

Discussion
This study demonstrated the viability of a version of the PMMA adapted for use in
Greece but with certain reservations. In most important respects, the Greek version
of the test functioned as intended in that the overall results paralleled the American

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

84		

Journal of Research in Music Education 58(1)

Table 4. Spearman Correlations (rho) Between the Greek Standardization and Gordon
(1979) Samples for Item Difficulty and Discrimination Values (r)
Tonal
Grade level
K
1
2
3

Rhythm

Difficulty

Discrimination

Difficulty

Discrimination

.81***
.75***
.87***
.78***

.28
.23
.36*
.53***

.81***
.53***
.39*
.49***

.59***
.64***
.42**
.38*

*p < .05. **p < .01. ***p < .001.

norms. Specifically, means by grade level were comparable to the American norms,
higher group means were associated with higher grade levels, group means tended to
be higher for the Tonal than for the Rhythm subtest, the internal consistency of the
measure was adequate, and the observed trend for difficulty and discrimination values
by grade level was as expected. Additionally, descriptive data for a sample of Greek
children ages 5 and 6 years reported by Pollatou et al. (2005) are very much in line
with the Greek norms determined for that age group in the current study.
Although the overall pattern of results was promising, a number of issues should be
considered by test users and future researchers. The validity of the PMMA for the
Greek population remains largely an open question, as the current methodology afforded
a very limited assessment of validity. Although teachers’ ratings of achievement correlated significantly with PMMA Tonal scores, they did not correlate with Rhythm
scores. Furthermore, teachers’ ratings of music aptitude were not correlated with PMMA
subtests or the composite for any grade level. These results might be attributed to the
noted low reliability of the Rhythm subtest in particular, possible lack of reliability in
the teachers’ ratings, or variation among teachers in their definitions and observations
of children’s music aptitude and achievement. Future research on the validity of the
Greek version of the PMMA should include additional criterion measures of musical
and nonmusical behaviors.
The differences noted between the Greek results and the original American norms
can be viewed in light of general cultural differences, cultural differences in elementary music curricula, and the fact that the American norms were published in 1979. It
is indeed open to question whether the original U.S. norms would be replicated in the
United States 30 years later. As for the current Greek results, composite means (based
on the total Greek sample) were significantly different from the American norms for
grade 1, which suggests that test results in Greece should be interpreted according to
the new Greek norms, not the original published norms. The somewhat larger standard
deviations for the Greek sample should be considered also. Both the differences in
means and especially the larger standard deviations can be explained by the fact that
the Greek sample was obtained from a large cross-section of schools, classrooms, and
school settings and thus was probably more heterogeneous than the original norming

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

85

Stamou et al.

sample, which consisted of a single suburban school district in the northeastern United
States. Standard deviations for the Tonal and Rhythm subtests reported by Pollatou et al.
(2005) for their Greek sample also exceeded the American norms. Statistical tests comparing the U.S. and Greek samples without the rural data revealed that the Greek urban
and suburban composite means were significantly higher than the respective U.S. means
for grades 2 and 3 but were not significantly different for kindergarten or grade 1.
Split-halves reliability coefficients for the Tonal subtest were similar for the Greek
and American samples, except for notably lower coefficients for the kindergarten portion of the Greek sample. These coefficients, however, were considerably lower at all
grade levels in the Greek sample for the Rhythm subtest, which had a similar effect on
the composite coefficients. For reasons of reliability, the Tonal subtest of the Greek
version of the PMMA developed for this study should be used with caution at the
kindergarten level. The same can be said for the Rhythm subtest at the kindergarten
and grade 1 levels.
Taken together, the internal and retest reliability and concurrent validity results for
the Greek version of the PMMA are mixed. The coefficients for internal consistency
for the Tonal and composite measures were within an acceptable range and were in
line with those reported in previous research (e.g., Gordon, 1979; Gouzouasis, 1993;
Holahan & Thomson, 1981). However, the internal consistency for the Rhythm subtest ranged from unacceptable to marginally acceptable and corroborated the results of
Holahan and Thomson (1981). Internal reliability results for the Rhythm subtest
should be viewed with caution, especially for kindergarten and grade 1 students. Similarly, as noted earlier, concurrent validity results based on teachers’ ratings of students’
music aptitude and achievement were poor for the Rhythm subtest at all grade levels.
It appears that Greek music educators and researchers can place more confidence
in results from the Tonal than from the Rhythm subtest. However, the retest reliability
results suggest that for all four grade levels, including older (i.e., grade 3) students,
developmental music aptitude as measured by the Greek version of the PMMA was
not stable across even a short period. The results also confirm that psychometric criteria for measurement instruments should not be assumed to be inherent qualities but,
rather, should be continuously reassessed within every test sample.
Beyond the principal objective of cross-cultural issues examined here, the current
findings suggest that future research should examine the stability of the PMMA, especially the Rhythm subtest. Although numerous studies, including those cited earlier,
have reported results for internal consistency of the PMMA, relatively few have reported
data concerning the stability of scores. Although some fluctuation in children’s developmental music aptitude is to be expected (e.g., Gordon, 1979), the poor stability of scores
in the current sample remains a major issue. As has been pointed out by Amos and
Humes (1998), the lack of reliability concerns in widely used standardized measures of
auditory perception (or any test instrument) may undermine the integrity of the interpretation of test scores and associated research conclusions and recommendations.
Attaining high retest reliability in tests of auditory perception with this age group
does appear to be possible. For example, in a study of the reliability of the Slosson
Auditory Perception Skills Screener, Erford and Luce (2005) found retest reliability of

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

86		

Journal of Research in Music Education 58(1)

r = .82 for a sample of 58 children ages 6 through 9. Notably, this 75-item measure
was administered individually (in accordance with the test manual). This suggests that
music education researchers should consider reexamining factors such as the administration condition, test length, and item characteristics that might influence the stability
of the PMMA. Individual administration was used by Amos and Humes (1998) in
their study of auditory discrimination in first and third graders. Interestingly, they
found that 40-item auditory subtests (equal in length to the PMMA subtests) yielded
poor retest reliability coefficients (r =.24 to r =.33), whereas 100-item and 180-item
auditory subtests yielded retest reliability coefficients of .70 to .78, respectively.
Because the PMMA is a frequently used test, it is important that researchers continue to examine its reliability and validity over time and with different populations.
The fact that three decades have passed since the first publication of the test points to the
need for further investigation of its technical characteristics. Furthermore, the PMMA
was intended to be used in the United States, and it was normed on a subsample of that
population. Unfortunately, too often the test continues to be used uncritically with
populations very different from the one it was intended to serve. Given the results of the
current study, additional research on the reliability and validity of the PMMA, both
within and between cultures, seems warranted.

Appendix
PMMA Tonal and Rhythm Percentile Norms for the Greek Sample
Tonal test
Raw
score
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24

Rhythm test

Grade
K

Grade Grade Grade
1
2
3

99
98
97
96
94
87
85
82
75
70
58

99
98
96
92
85
73
61
50
39
33
26
21
17
14
12
10

99
98
96
93
83
74
68
63
55
53
46
40
36
32

99
98
95
88
79
68
57
44
35
26
21
16
13
12
10
9
7

Raw
score
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24

Grade K

Grade 1 Grade 2 Grade 3

99
97
93
89
86
83
78
70
63

99
96
92
84
75
61
55
44
36
29
23
19
15

99
98
96
93
90
86
79
70
62
52
44
39

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

99
96
89
77
66
56
45
33
28
22
18
14
11
9

87

Stamou et al.

Appendix (continued)
Tonal test
Raw
score

Grade
K

23
22
21
20
19
18
17
16
15
14
13
12
11

53
48
44
35
25
19
13
10
8
5
3
2
1

Rhythm test

Grade Grade Grade
1
2
3
28
22
19
13
6
4
1

9
7
6
5
4
3
2
2
1

Raw
score

6
5
5
4
3
2
2
1

23
22
21
20
19
18
17
16
15
14
13
12
11

Grade K
50
42
37
28
16
11
6
5
3
2
1

Grade 1 Grade 2 Grade 3
31
24
15
11
5
4
3
3
2
1

11
8
7
5
4
3
2
2
1

6
5
4
3
2
2
1

PMMA = Primary Measures of Music Audiation (Gordon, 1979).

Declaration of Conflicting Interests
The authors declared no conflicts of interest with respect to the authorship and/or publication of
this article.

Funding
Pythagoras II: Funding of research groups in the University of Macedonia, Priority Action
2.2.3e, Action 2.2.3, Measure 2.2, Operational Programme for Education and Initial Vocational
Training, co-financed by the European Union (3rd Community Support Framework, 75% by the
European Social Fund, 25% national Greek resources).

Note
1. In the absence of raw data from the original norming study, the authors computed a t test
for each grade by hand after computing the respective variance coefficients for the U.S.
samples (Gordon, 1979) via the following formula: SD2(N) = Sx2. The Bonferroni adjustment for multiple tests was applied for each t test (∞/4): p < .05 = .0125.

References
Amos, N. E., & Humes, L. E. (1998). SCAN test-retest reliability for first and third graders.
Journal of Speech, Language, and Hearing Research, 41, 834–845.
Bentley, A. (1966). Measures of musical abilities. London: Harrap Audio-Visual.
Chuang, W. C. J. (1997). An investigation of the use of the Musical Aptitude Profile with Taiwanese
students in grades four to twelve. (Unpublished doctoral dissertation). Michigan State University,
East Lansing.

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

88		

Journal of Research in Music Education 58(1)

Chung, H. (2002). An investigation of the nature and characteristics of music aptitude of
Korean-American students living in Pennsylvania. (Unpublished doctoral dissertation). Temple
University, Philadelphia.
Drake, R. (1933). Four new tests of musical talent. Journal of Applied Psychology, 17, 136–147.
Drake, R. (1954). Drake Musical Aptitude Tests. Chicago: Science Research Associates.
Erford, B. T., & Luce, C. L. (2005). Reliability and validity of the Slosson Auditory Perception
Skills Screener. Perceptual and Motor Skills, 101, 891–897.
Feay-Shaw, S. (2000). Multicultural perspectives on research in music education. Bulletin of the
Council for Research in Music Education, 145, 15–26.
Gaston, E. T. (1957). Tests of Musicality. Lawrence, KS: Odell’s Instrumental Service.
General Secretariat of National Statistical Service of Greece. (2008). De facto population: Area
and population by range, density and elevation zones. Retrieved December 26, 2008, from
http://www.statistics.gr
Gordon, E. E. (1965). Musical Aptitude Profile. Chicago: GIA.
Gordon, E. E. (1979). Primary Measures of Music Audiation. Chicago: GIA.
Gordon, E. E. (1982). Intermediate Measures of Music Audiation. Chicago: GIA.
Gordon, E. E. (1989). Advanced Measures of Music Audiation. Chicago: GIA.
Gordon, E. E. (1991). The Advanced Measures of Music Audiation and the Instrument Timbre
Test: Three research studies. Chicago: GIA.
Gouzouasis, P. (1993). Music audiation: A comparison of the music abilities of kindergarten
children of various ethnic backgrounds. Quarterly Journal of Music Teaching and Learning,
4(2), 70–76.
Holahan, J. M., & Thomson, S. W. (1981). An investigation of the suitability of the Primary
Measures of Music Audiation for use in England. Psychology of Music, 9, 63–68.
Humphreys, J. T. (1993). Precursors of musical aptitude testing: From the Greeks through the
work of Francis Galton. Journal of Research in Music Education, 41, 315–327.
Humphreys, J. T. (1998). Musical aptitude testing: From James McKeen Cattell to Carl Emil
Seashore. Research Studies in Music Education, 10, 42–53.
Jung, K. H. (1992). A two-year predictive validity study of the Musical Aptitude Profile for use
in Korea. Doctoral dissertation, Temple University, Philadelphia.
Kwalwasser, J. (1953). Kwalwasser Music Talent Test. New York: Mills Music.
Kwalwasser, J. (1955). Exploring the musical mind. New York: Coleman-Ross.
Kwalwasser, J., & Dykema, P. W. (1930). Kwalwasser-Dykema Music Tests. New York: Carl
Fischer.
LeBlanc, A. (1980). Outline of a proposed model of sources of variation in musical taste. Bulletin of the Council for Research in Music Education, 61, 29–34.
LeBlanc, A., Fung, C. V., Boal-Palheiros, G. M., Burt-Rider, A. J., Ogawa, Y., Oliviera, A. J.,
et al. (2002). Effect of strength of rhythmic beat on preferences of young music listeners in
Brazil, Greece, Japan, Portugal, and the United States. Bulletin of the Council for Research
in Music Education, 153, 36–41.
LeBlanc, A., Jin, Y. C., Stamou, L., & McCrary, J. (1999). Effect of age, country, and gender
on music listening preferences. Bulletin of the Council for Research in Music Education,
141, 72–76.

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012

89

Stamou et al.

Nunnally, J. C. (1970). Introduction to psychological measurement. Toronto, Ontario, Canada:
McGraw-Hill.
Özeke, S., & Humphreys, J. T. (2000). Music teacher education in Turkey and in the USA: Musical aptitude and attitude toward teaching. In D. A. Tjeldvoll, T. Thune, & M. Sneve Olsen
(Eds.), Higher education, quality and evaluation in comparative and international perspectives (Studies in Comparative and International Education Vol. 4, Report No. 7, pp. 67–80).
Oslo, Norway: University of Oslo Institute for Educational Research.
Pollatou, E., Karidimou, K., & Gerodimos, V. (2005). Gender differences in musical aptitude,
rhythmic ability, and motor performance in preschool children. Early Childhood Development and Care, 175, 361–369.
Schoenoff, A. W. (1972). An investigation of the comparability of American and German norms
for the Musical Aptitude Profile. (Unpublished doctoral dissertation). University of Iowa,
Iowa City.
Seashore, C. E. (1919). Seashore Measures of Musical Talents. New York: Psychological Corporation.
Sell, V. H. (1976). The musical aptitude of Finnish students: An investigative study in comparative
music education. (Unpublished doctoral dissertation). University of Wisconsin–Madison.
Stamou, L., Humphreys, J. T., & Schmidt, C. P. (2006). The effects of instruction on selfassessed research knowledge, ability, and interest among Greek music educators. Music
Education Research, 8, 175–189.
Tilson, L. (1941). A study of the prognostic value of the Tilson-Gretsch Test for Musical Aptitude.
Teachers College Journal, 12, 110–112.
Wang, J. C. (2007). A comparative study of college students’ musical aptitudes and musical
preferences in the U.S. and Taiwan. (Unpublished doctoral dissertation). Arizona State
University, Tempe.
Whistler, H., & Thorpe, L. (1950). Musical Aptitude Tests. Los Angeles: California Test Bureau.
Wing, H. (1958). Standardized Tests of Musical Intelligence. Sheffield, UK: City of Sheffield
Training College.
Yang, T. B. (2002). The comparative effects of the traditional Taiwanese curriculum and a
curriculum based on music learning theory on the developmental music aptitudes and singing
performance of first grade students in Taiwan. (Unpublished doctoral dissertation). Michigan
State University, East Lansing.

Bios
Lelouda Stamou is an assistant professor of music education at the University of Macedonia in
Thessaloniki, Greece. Her research interests include music aptitude development, study of musical behaviors, music and flow experience, and teacher training.
Charles P. Schmidt is a professor emeritus at Indiana University. His research interests include
social psychology of music and instrumental music education.
Jere T. Humphreys is a professor of music education at Arizona State University. His research
interests include historical and empirical aspects of scientific thinking in music education.
Submitted May 15, 2009; Acceptance October 22, 2009.

Downloaded from jrm.sagepub.com at ARIZONA STATE UNIV on June 9, 2012