relationships between self-concept and achievement
TRANSCRIPT
JOURNAL OF RESEARCH IN PERSONALITY 19, 365-372 (1985)
Relationships between Self-Concept and Achievement
IN-SUB Sow3
Kong-Ju National Teachers’ University. Kong-Ju. Korea
AND
JOHN HATTIE
Centre for Behavioural Studies, University of New England, Armidale. New South Wales. Australia
This study is based on a multifaceted and hierarchical model of self-concept and explores the relationships between the various facets of self-concept and academic achievement. The model of self-concept has general self-concept at the apex, then academic self-concept, presentation of self, and social self-concept. There are 7 third-order factors: achievement, ability, and classroom self-concepts (academic), physical self-concept and confidence in self (presentation of self). and peer and family self-concepts (social). Further, there are four specific-subject self-concepts. The argument proposed in this study is that the relations between achievement and the facets of academic self-concept are greater than the relations between achievement and presentation of self and social self-concept. This argument is not rejected. Adolescents consistently evaluate themselves across various academic subjects, but there are no relationships between achievement and other aspects of the self. 0 1985 Academic Press. Inc.
There has been considerable interest in the relationship between self- concept and academic achievement. Unfortunately, these relationships have been confounded by the use of global and more specific self-concept scales (see Mintz & Muller, 1977). Some have argued that specific self- concepts, especiaJly subject-specific self concepts, are more highly related to self-concept than global self-concept scores (Wylie, 1979).
This argument presupposes a hierarchical, multifaceted model of self- concept, but few have specified such a model. Song and Hattie (1983) have demonstrated that there is much evidence for accepting a model of self-concept similar to that proposed by Shavelson, Hubner, and Stanton
This study was supported in part by a grant to the second author from the Australian Research Grants Commission. Send requests for reprints to Dr. John Hattie, Department of Behavioural Studies, University of New England, Armidale, New South Wales, Australia.
365 0092-6566185 $3.00
Copyright 0 1985 by Academic Press. Inc. All nghtc of reproduction in any form reserved.
366 SONG AND HATTIE
(1976). General self-concept is at the apex of the hierarchy with academic self-concept, social self-concept, and presentation of self as second-order factors. Academic self-concept can be further subdivided into classroom, achievement, and ability self-concepts. Social self-concept can be divided into peer and family self-concepts and presentation of self can be divided into confidence of self and physical self-concepts. Further divisions can be made at the base of the hierarchy. For example, ability self-concept can be divided into subject-specific self-concepts. Figure 1 presents this model. (See Song & Hattie, 1983, for further details and evidence for this model.)
There have been many studies that have reported low correlations between general self-concept measures and academic achievement. For example, Perkins and Shannon (1965) found the correlations between self-concept, as measured by a modified version of the Coopersmith Self- Esteem Inventory, and achievement scores on the Metropolitan Achieve- ment Test were not significant. This finding was replicated by Williams (1973). Hansford and Hattie (1982) in a meta-analysis found that self- concept correlated .18 with achievement and self-esteem correlated .22 with achievement. The correlations were based on 728 correlations (71 studies) for self-concept and 187 correlations (3 1 studies) for self-esteem.
A major problem with such studies relates to the nature of the “general” measures. In some cases what is meant by general self-concept are nonacademic self-concepts such as social self-concept or presentation of self, whereas in other cases what is meant are combinations of all the second-order factors in Fig. 1.
Concerning the relationship between specific-subject self-concepts and academic achievement, a large number of studies have been presented by Brookover and associates (Brookover, Erickson, & Joiner, 1967; Brookover, Paterson, & Thomas, 1962; Brookover, Thomas, & Paterson, 1964). For example, Brookover et al. (1964) using data from over 1000
I I I
I I I Mathematics Language Social Studies Natural science
FIGURE 1.
SELF-CONCEPT AND ACHIEVEMENT 367
urban seventh-grade students, found correlations between grade point average and self-concept of ability of 57 for both males and females. The correlation between specific-subject self-concepts and the associated achievement scores were of a similar magnitude (e.g., social studies r = 56 for boys, 58 for girls, and science r = .61 for boys and .51 for girls).
Marsh, Relich, and Smith (1983) reported that grades in English, math- ematics, and science were more highly correlated with each of the cor- responding self-concepts than with general self-concept. In a further study, Marsh, Smith, Barnes, and Butler (1983) further supported the argument that academic ability was more highly correlated with self- concepts in the same area, less correlated with other academic self- concepts, and least correlated with nonacademic self-concepts. They also found support for separated academic self-concepts with verbal and quan- titative components. Marsh and O’Neill (1984) noted that the verbal and quantitative self-concepts were uncorrelated with each other even though achievements in the corresponding areas are substantially correlated.
In Hansford and Hattie’s (1982) meta-analysis, the correlation between self-concept of ability and achievement was .42. This was based on 130 correlations from 20 studies. Most of the studies used the Brookover test.
Two major problems with studies such as Brookover et al. are that they are based on a test that relates more to self-estimates of ability. and the test mixes items that relate to self-concept of ability and self- concept of achievement (see Song, 1982, and Wylie, 1979, for more details on the differences between self-concept of ability and self-concept of achievement). This is not to say that self-estimates of ability have no part in self-concept of ability, but they are not the only part.
To overcome these problems a set of items was constructed by Song (1982) to assess self-concept of ability, self-concept of achievement, classroom self-concept, and specific-subject self-concepts. Typical items in the self-concept of ability scale include “I think that I have the ability to get good grades in school work,” and “I think that I am capable of getting the results I would like to obtain in school work.” Items in the self-concept of achievement scale include “I am proud of my report card,” and “I am satisfied with my school work.” For classroom self- concept, items include “I feel left out of things in class,” “Most of my teachers do not understand me,” and “I feel worthless in class.” For specific-subject self-concept, items include “I feel good about my ability in -,” “My ability in - is better than that of my close friends,” “I am satisfied with my ability to do school work in -, ” and “I think that I am capable of getting the results I would like to obtain in -.”
The purpose of this study is to assess the relationships between various facets of self-concept and academic achievement. It is predicted that the
368 SONG AND HATTIE
correlations between self-concept and achievement are greatest between specific-subject self-concepts, then between classroom, achievement, and ability self-concepts, and are minimal with presentation of self and social self-concepts.
METHOD
Sample
The study was carried out in Seoul, Korea. It was decided to use 14- to IS-year-olds for two reasons. First, research into the development of self-concept points to its stability by this age (e.g., Bloom, 1964; Sanford, 1962). Second, students at this age have developed adequate verbal facility to comprehend written and spoken instructions. The initial sample consisted of 2530 students but, data from 233 students were excluded because they omitted at least one test. Thus, the final sample size was 2297.
Instruments
Scales were developed to measure 11 facets of self-concept: classroom, achievement, ability, peer, family, confidence in self, physical, and four subject-matter self-concepts- English, mathematics, social studies, and natural science. Items for each of the scales were self-referential statements gleaned from a range of other self-concept instruments and from items written by the authors. All items were pilot tested on a group of 87 Australian adolescents and as a consequence some refinement of the items took place.
Academic achievement was estimated by the grade point average for mathematics, language, social studies, and science. These were obtained from school records. In an attempt to minimize the GPA unreliability, the GPAs over four achievement tests, administered during a 2-month period, were averaged.
RESULTS
The estimates of reliability ranged from .76 to .94 (Table 1). The items within each scale were factor analyzed using maximum likelihood factor
TABLE 1 NUMBER OF ITEMS, ESTIMATE OF RELIABILITY (a). AND SUMMARY INFORMATION FROM A
FACTOR ANALYSIS SPECIFYING ONE FACTOR FOR THE SELF-CONCEPT TESTS
Self-concept tests
Classroom 17 Achievement 15 Ability 16 Mathematics 16 Language 16 Social studies 16 Natural science 16 Peer 17 Family 23 Confidence in self 14 Physical 14
No. of (Y
.83
.82
.89
.94
.93
.92
.94
.81
.91
.79
.76
No. of items > .30 on one factor
16 15 16 16 16 16 16 16 23 14 14
Average off-diagonal
absolute residuals
.05
.04
.04
.03
.03
.03
.03
.0.5
.02
.OS
.OS
SELF-CONCEPT AND ACHIEVEMENT 369
analysis (McDonald & Leong, 1974). For each scale, one factor was specified and from an inspection of the number of items that have loadings greater than .30 and the average absolute off-diagonal residuals, it is possible to conclude that the scales are unifactorial
The correlation between the achievement scores and the self-concept scores are presented in Table 2. The correlations between the four achievement scores are, not surprisingly, all high, and there are indications of one general factor. The correlations between the subject-matter self- concepts and achievement indicate that there is discrimination between the scales. For example, the correlation between mathematics achievement and mathematics self-concept can be compared with correlations between mathematics and the other specific-subject self-concepts. The value of .38 is greater than the other three correlations. Similarly, the comparisons can be made for the other subjects. There are indications that there is discrimination between the scales, but as expected, there is much overlap.
The correlations between achievement scores and achievement self- concept and ability self-concept are very similar. It should be noted that there is 50% of the variance in common between achievement and ability self-concept. Thus, these two facets of self-concept do contribute some unique variance, but they have much in common. As predicted, the correlations with the academic self-concepts are much greater than with the nonacademic self-concepts.
Conclusions
The facet of self-concept that has the greatest relationship to academic achievement is, not surprisingly, academic self-concept. The relationships with the other second-order factors of self-concept-presentation of self and social self-concept-are extremely low. These findings provide evidence supporting the construct validity of the model of self-concept. As expected, from theoretical concerns the correlations between specific-subject self- concepts and the respective achievement scores are larger than the cor- relations between sepcific-subject self-concepts and other achievement scores. Both these sets of correlations are larger than the correlations between nonacademic self-concept scales and achievement. The more general academic self-concepts relate positively to the specific-subject self-concepts; particularly achievement self-concept.
Unlike the results reported by Marsh and O’Neill(1984), the correlations between verbal and quantitative self-concepts are reasonably similar, and closely parallel the correlations between achievement scales. In this study, the self-concepts in specific subjects are as differentiated from each other as are achievements in those subjects.
Studies that attempt to investigate the relationships between self-concept and academic achievement can lead to different conclusions if the mul- tifaceted aspects of self-concept are not taken into account. An example
TABL
E 2
RELA
TION
SHIP
S BE
TWEE
N SE
LF-C
ONCE
PT
SCAL
ES
AND
ACHI
EVEM
ENT
Self-
conc
ept
scal
es
Cla
ssro
om
100
Achi
evem
ent
59
100
Abilit
y 49
71
10
0 M
athe
mat
ics
36
52
61
100
Lang
uage
37
51
58
54
10
0 So
cial
stud
ies
34
43
50
43
48
100
Scie
nce
34
49
57
59
47
56
100
Peer
50
25
28
19
21
21
19
10
0 Fa
mily
45
37
33
26
23
22
22
38
10
0 C
onfid
ence
50
35
37
25
26
28
28
51
48
10
0 Ph
ysic
al
28
19
19
13
15
15
15
30
29
46
100
Achi
evem
ent
scal
es
Mat
hem
atics
15
36
27
38
22
07
12
-0
1 15
01
-0
7 10
0 La
ngua
ge
17
33
27
22
34
13
14
01
12
05
-04
62
100
Socia
l st
udie
s IS
34
30
23
22
28
20
01
12
04
-0
4 so
,5
1 10
0 Sc
ienc
e 16
33
28
24
23
18
26
00
IS
05
-0
4 58
53
6.
5 10
0
SELF-CONCEPT AND ACHIEVEMENT 371
of a statement that can be very misleading is provided by Purkey (1970): “there is a persistent and significant relationship between the self-concept and academic achievement at each grade level, and that change in one seems to be associated with change in the other” (p. 27). Academic achievement can be related to some but not all aspects of self-concept.
This study is based on a clear and defensible model of self-concept. Future studies, especially those investigating the causal nature of self- concept and academic achievement, could well profit from being based on a model such as the one outlined in Fig. I. Research concerning causality may become less equivocal if the second-order model rather than the general self-concept is used. As Cook and Campbell (1980) have argued, a prerequisite of causal claims is that there must first be some covariation between various facets. If, as Hansford and Hattie (1982) and the results from this study confirm, general self-concept has extremely low relationships with achievement, then it is not surprising that there have been contradictory findings as to the direction of causal relationships. If the various facets of self-concept are used, however, different and less equivocal findings may result.
REFERENCES
Bloom, B. S. (1964). Stability and change in human characteristics New York: McGraw- Hill.
Brookover, W. B.. Erickson, E. L., & Joiner, L. M. (1967). Se/f-concept of abiliry and
school achievement: III. Relationship of self-concept to achievement in high school.
USOE, Cooperative Research Project 2381. East Lansing: Michigan State University. Brookover, W. B., Paterson, A., & Thomas, S. (1962). Self-concept of ability and school
achievement. USOE. Cooperative Research Report, Project 845. East Lansing: Michigan State University.
Brookover, W. B., Thomas, S., & Paterson, A. (1964). Self-concept of ability and school achievement. Sociology of Education, 37, 271-278.
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design nnd analysis ,fbr
field settings. Chicago: Rand McNally. Hansford, B. C., & Hattie, J. A. (1982). Relationships between self and achieve-
ment/performance measures. Review ofEducational Research, 52, 123-142. Marsh, H. W., & O’Neill, R. (1984). Self description Questionnaire III: The construct
validity of multidimensional self-concept ratings by late adolescents. Journal of Ed-
ucational Measurement, 21, 153-174. Marsh, H. W., Relich, J. D., & Smith, I. D. (1983). Self-concept: The construct validity
of interpretations based upon the SDQ Journal of Personality and Social Psychology,
45, 173-187. Marsh, H. W., Smith, I. D., Barnes, .I., & Butler, S. (1983). Self-concept: reliability,
stability, dimensionality, validity, and the measurement of change. Journal ofEducational
Psychology, 75, 772-790.
McDonald, R. P., & Leong, K. S. (1974). COFA: A Fortran program for common factor
analysis. OISE: University of Toronto. Mintz, R., & Muller, D. (1977). Academic achievement as a function as specific and global
measure of self-concept. The Journal of Psychology. 97, 53-V.
Perkins, C. W., & Shannon, D. T. (1965). Three techniques for obtaining self-perceptions in preadolescent boys. Journal of Personality and Social Psychology, 2, 443-447.
372 SONG AND HATTIE
Purkey, W. W. (1970). Self-concept and school achievement, Englewood Cliffs, NJ: Prentice- Hall.
Stanford, R. N. (Ed.). (1962). The American college. New York: Wiley. Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Self-concept: Validaiton of
construct-interpretation. Review of Educational Research, 46, 407-441. Song, I. S. (1982). The dimensionaiity and relationships between home environment, self-
concept and academic achievement. Unpublished Ph.D. thesis, University of New England.
Song, I. S., 62 Hattie, J. A. (1983). The dimensionaliry ofse&oncept. Manuscript submitted for publication.
Williams, J. M. (1973). The relationship of self-concept and reading achievement in first- grade children. Journal of Educational Research, 66, 376-380.
Wylie, R. C. (1979). The self-concept: The theory and research on selected topics (Vol. 2, rev. ed.). Lincoln: Univ. of Nebraska Press.