1. Phillip M. Hash Calvin College Grand Rapids, Michigan pmh3@calvin.edu
Development and Validation of a Music Self-Concept Inventory for College Students
Phillip M. Hash
Calvin College
Grand Rapids, Michigan

2. Purpose
Develop a brief music self-concept inventory for college students
Easy to score
Reliable
Tests 3-factor model
Use
My interest
Develop a brief music self-concept inventory for college students
easy to administer
tests a three-factor model defined by Austin (1990)
demonstrates acceptable internal reliability of α ≥ .80
Useful to researchers and educators who want to measure music self-concept among students

3. Definition
Self-concept: “An individual’s relatively stable and organized perception and evaluation of themselves within socially important domains that are formed through experience with and interpretations of one’s environments.” (Marsh, Parker, & Craven, 2015, p. 116; also see Shavelson, Hubner, & Stanton, 1976)
Term sometimes used interchangeably w/ self-esteem
Self-concept is hierarchical. Individuals hold both global self-concept and more specific self-concepts about different areas in their lives.
The hierarchy can progressively narrow from beliefs regarding academic and non-academic aspects into more discreet types of self-concepts such as those related to specific academic areas (e.g., language arts, math), social relationships (e.g., family, peers), or physical traits (e.g., height, complexion) (Pajares & Schunk, 2001).

4. Self-Concept Model (Shavelson, Hubner, & Stanton, 1976)
Language Math

5. Self-Concept Model Including Artistic Domain Vispoel (1995)

6. Music Self-Concept (Vispoel, 1994)

7. Three- Factor Model of Music Self-Concept (Schmitt, SEMA, 1979; Austin, 1990)
Support/ Recognition from Others
Personal Interest/ Desire
MUSIC SELF CONCEPT
Model of Global Music Self Concept
3 factors: Influence of Others; Interest & Desire; Perceptions of Ability
This model comes from a study by Schmidt (1979) who developed the Self-Esteem of Music Ability scale in a dissertation.
This scale for adolescents years old
Contained 43 items
Factor analysis indicated the subcales, which Austin (1990) further refined.
Perception of Music Ability

8. Need for Measuring Music Self-Concept
Self-Concept Can Influence
Decisions to participate & persist
Achievement & enjoyment
(Marsh, Craven, Mclnerney, 2003)
Reciprocal Effects Model (Marsh & Martin, 2011)
Develop both Self-Concept & Achievement
MSCI a tool to help
ID instructional strategies
ID populations who might benefit from curricular revision or intervention
Relationship of self-concept & achievement
Achievement and self-concept are causes and effects of each other (reciprocal effects model)
Need to work towards both in education (Marsh & Martin, 2011)
Self Concept influences
participation
persistence
enjoyment
Music self-concept can
influence decisions to participate and persist
affect music achievement and enjoyment
Value of measuring music self-concept:
Identify instructional strategies that prove effective for increasing this characteristic (e.g., Randels, 2010; Sanders, 2000)
Identify students who might benefit from
curricular revision (Gumm, 1990)
various forms of intervention (e.g., Austin & Vispoel, 1998; Sichivitsa, 2004)

9. Instrument Development
Examined Self-Esteem of Music Ability (SEMA) (Schmidt, 1979)
Children/Adolescents age 10-15
Used recently by Draves (2008); Randels (2010); & Kruse (2012)
43 positively & negatively stated items
Factor Analysis indicated 3 subscales
Also VanderArk (n.d.) & Svengalis (1978)
Created New Items
Based on SEMA (reworded & stated positively)
Some items original
Examined Self-Esteem of Music Ability (SEMA) (Schmidt, 1979)
Used recently by Draves (2008); Randels (2010); & Kruse (2012)
43 positively & negatively stated items
Created new items
Based on SEMA (reworded & stated positively)
Some items original

10. Instrument Development
Demographic Information
15 Statements/3 Equal Subscales (Others, Interest, Ability)
Sent to 3 music education faculty (face validity)
Conducted Pilot Test
Revise Ability Subscale
Proceeded to Main Study
Demographic Information
Gender
Age/college year
Music Experience in and out of school
Fifteen statements in three equal subscales (Others, Interest, Ability)
Five-point Likert Scale (1 = strongly disagree, 5 = strongly agree)
Sent to three music education faculty (face validity)
Conducted Pilot Test
Indicated need to revise some statements in the ability subscale due to low reliability
Proceeded to Main Study
As a measure of construct validity, a subset of respondents (n = 55) also completed a modified version of the SEMA in order to determine the correlation between total and subscale scores on this instrument and the MSCI

11. Administration Small LA College in Midwest
Completed by Intact Classes for Non Music Majors (N = 237)
Administered by Researcher
Scripted Instructions
To measure construct validity, a subset (23%) also completed a modified version of the SEMA in order to determine the correlation between total and subscale scores on this instrument and the MSCI

12. Data Analysis – Reliability & Validity
Internal reliability (α)
Construct Validity (SEMA vs. MSCI)
Criterion Validity (music participation & gender)
Discriminant Validity (factor r)
Factor Analysis
Exploratory (eigenvalue = 1 criteria)
Confirmatory (constrained to 3 factors)
Principal Axix w/ Promax Rotation
Descriptive Statistics
Cronbach’s alpha (internal reliability of scale and subscales)
Correlation (factors, MSCI vs. SEMA)
Factor Analysis
Principal Axis w/ Kaiser Normalization
Promax rotation (correlated items)
Exploratory (eigenvalue = 1 criteria)
Confirmatory (constrained to 3 factors)

13. Participants (N = 237) Male = 41%; Female = 58%; Unspecified = 1%
18 to 48 years of age (M = 19.87, SD = 2.77)
Freshman (n = 27%), sophomore (n = 30%), junior (n = 25%), and senior (n = 19%) classes
White (n = 73%) and non-White (n = 27%)
Determined by photographs & instructor knowledge of students
In-school music participation: (M = 5.86, SD = 3.82) years
Out of school music participation: (M = 6.10, SD = 4.66)
Combined participation: (M = 12.06, SD = 7.24) years
Male = 40.9%; Female = 57.8%; Unspecified = 1.3%
18 to 48 years of age (M = 19.87, SD = 2.77)
Freshman (n = 26.6%), sophomore (n = 29.5%), junior (n = 24.9%), and senior (n = 19.4%) classes
White (n = 73.4%) and non-White (n = 26.6%)
Determined by photographs & instructor knowledge of students
In-school music participation: (M = 5.86, SD = 3.82) years
Elective Ensembles & classes
Out of school music participation: (M = 6.10, SD = 4.66)
Community or church groups and private lessons
Combined participation: (M = 12.06, SD = 7.24) years

14. Initial Draft - Results
Reliability - Total Scale: α = .95; Subscales: α =
Bartlett’s test (χ2 = , p < .001) = correlations appropriate for factor analysis
KMO measure (.94) = adequate sample size
Subject-to-variable ratio = 15.8:1.
MAIN STUDY

15. Initial Draft - Factor Analysis
Exploratory (59.5% variance)
2 factor solution
(a) perception of abilities with influence from others and (b) personal interest or desire
Confirmatory (63.1% variance)
8.57 (others), 1.18 (interest), 0.84 (abilities)
Items intended to constitute each factor loaded highest under their respective columns.
“I am a capable singer or instrumentalist” & “I can be creative with music” were deleted due to cross-loadings.

16. Final Version
13 items
Reliability - Total scale: α = .94; Subscales α =
Factor analysis explained 63.6% of the variance
All but one of the rotated factor loadings exceeded .50
Eigenvalues = 7.37 (others), 1.17 (interest), and (abilities)
Subscale scores (r = ) Subscales vs. total scale (r = )
13 items
Reliability - Total scale: α = .94; Subscales α =
Factor analysis explained 63.6% of the variance
Influence of others (54.1%), interest (5.7%), and ability (3.7%)
All but one of the rotated factor loadings exceeded .50
Eigenvalues = 7.37 (others), 1.17 (interest), and 0.84 (abilities)
Subscale scores moderately correlated with each other (r = ) and highly correlated to the total scale (r = )
Interfactor correlations (r = ) met the ≤ .85 cutoff that generally serves as the criterion for discriminant validity in applied research (Brown, 2015)
High correlation w/ SEMA: total .94; subscales

17. Validity
MSCI & SEMA: total (r = .94; subscales r = ) – Construct
Interfactor correlations (r = ) – Discriminant
In School (r = .51), Out-of- School (r = .58), & Total (r = .64) Participation - Criterion
MSCI scores males vs. females non-significant (ANOVA)
ANCOVA controlling for total participation found males significantly outscored females
Supported by some literature (e.g. Simpkins, Vest, & Becnel, 2010)

18. Pattern Matrix for Principal Factor Analysis with Promax Rotation of the MSCI (final version)
Factors
Item I
Others II
Interest
III
Abilities
My family encouraged me to participate in music.
.95
I have received praise or recognition for my musical abilities.
.92
Teachers have told me I have musical potential.
.85
My friends think I have musical talent.
.60
.32
Other people like to make music with me.
.52
.39
I like to sing or play music for my own enjoyment.
Music is an important part of my life.
.74
I want to improve my musical skills.
.71
I enjoy singing or playing music in a group.
.62
I like to sing or play music for other people.
.56
I can hear subtle differences or changes in musical sounds.
.84
I have a good sense of rhythm.
.76
Learning new musical skills would be easy for me.
.45

19. Conclusions
The MSCI effectively measures music self-concept as described by 3- factor model
Use the MSCI to (a) assess change or development in music self- concept, (b) identify differences in various aspects of music self- concept, (c) compare music self-concept among different populations, or (d) examine relationships between music self-concept and other variables
Use MSCI scores to (a) assess students’ attitudes towards their musical potential and accomplishment and (b) identify students who might require extra support

20. Conclusions
The MSCI might be useful for students below the college level.
Future studies should test the MSCI among people of varying ages and backgrounds, and examine additional variables, models, and theories that might explain this construct.
The MSCI might be useful for students below the college level.
The structure of music self-concept likely does not change between junior high school and college (Vispoel, 2003).
Flesch-Kincaid Reading Ease score of 67.3.
Average grade level reading score of 7.0 (Readability-Score.com).
MSCI might be effective with students as young as middle school.
Future studies should test the MSCI among people of varying ages and backgrounds, and examine additional variables, models, and theories that might explain this construct (e.g., Schnare, MacIntyre, & Doucett, 2012).