A COMPARISON OF THE SELF-CONCEPTS OF GIFTED PRE-ADOLESCENTS AND A NON-GIFTED PEER GROUP

This project investigated the multidimensional self-concepts of gifted and regular education fifth grade children. Potential differences in the ratings of Global, Physical, Social and Academic self-concept domains were investigated. Ratings of the importance of the specific domains were also examined. Participants consisted of 37 integrated gifted and 251 regular education classmates enrolled m the fifth grade of a suburban school district. Participants completed four selfreport measures: a variation of Kuhn and McPartland's "Who Am I?" task, an "Importance Scale" which examined the perceived importance of three self-concept domains, Marsh's (1981) SelfDescription Questionnaire, and a "Pie Chart" of the relative perceived importance of the three self-concept areas. Academic grouping and sex differences in the rating and importance of the self-concept domains were examined. Results indicated significant differences m the participants' ratings of their global and domain-specific selfconcepts and the importance of the specific domains. In accordance with previous research, gifted participants demonstrated Academic self-concepts that were significantly higher than those of the regular education sample. Significant differences were also found in the perceived importance of the domains for the gifted and regular education samples. Significant sex differences in the direction of traditional gender roles were found on all instruments. Results are discussed in terms of the influence of various socialization factors.

The study of a child's self-concept is important in both psychology and education. Self-concept has been seen as the core of personality (Lecky, 1945) and the emergence of a positive selfconcept has been viewed as one of the objectives of the school system (Purkey, 1970). The study of the self-concept of gifted populations is particularly important in investigating the emergence and maintenance of gifted behaviors. Bailey ( 1971) has suggested that the self-concept affects all areas of the personality and either restricts or enhances the person's capacity to fulfill his or her native potential (cited in Maddux, Scheiber, & Bass, 1982). Tannenbaum proposed that intelligence and self-concept work in a complementary fashion to produce gifted behaviors. A feedback loop is formed: if the individual thinks of him or herself as gifted, he or she will act in line with this self image and these high achievements will then further enhance the self-concept . Thus the self-concept of gifted children is important to study since it may be a substantial factor in influencing how well gifted children achieve their potential (Schneider, 1987).
However, in order to study self-concept, one must determine its definition.
In a review of the research,  reveals that no clear, concise, universally accepted operational definition of selfconcept exists.
However, there appears to be wide acceptance of some aspects of self-concept by researchers. Researchers agree that in general terms, self-concept is an individual's perception of him or herself. This implies that self-concept is global and consistent across situations.
In contrast, self-concept can also be viewed in a more specific manner implying that self-concept is made up of distinct 2 categories.  proposed that these categories include academic, physical and social dimensions. In more specific terms, self-concept involves the attitudes, feelings, and knowledge of an individual's abilities, skills, appearance and social acceptability Jersild, 1965;Labenne & Greene, 1969;.
Also important in the study of self-concept is the conceptualization of the construct within a theoretical framework.  state that this corresponds to defining the nomological network of the construct. Nomological research involves internal and external examinations of the construct.
Internal analyses or within network studies involve an examination of the relationships among the particular facets of the construct. For example, the relationship between academic and physical dimensions of self-concept might be examined. External analyses or between network studies involve examining the relation of the construct under study with another construct, which has been proposed to be mutually exclusive. Examining the relationship between academic self-concept and academic achievement would be an example of an external analysis (cited in . In the present study, an internal, within network analysis of the dimensions of self-concept will . be conducted.

Models of Self-Concept
Many models of self-concept exist. One of the oldest is the model proposed by William James in 1890. He suggested that the selfconcept is composed of two distinct, yet conceptually related dimensions: the I-self and the Me-self. The I-self is the processor or "knower," whereas the Me-self is the product or what is known about oneself . More recently, Byrne (1984, p. 430) has proposed four possible models of self-concept, which differ in terms of their definition, situation-specificity, and interrelatedness of their dimensions: 1. Nomothetic -Self-concept is seen as a unidimensional phenomenon, where a global, general self-concept can explain behavior in a large number of settings and situations.
2. Hierarchical -Self-concept is seen as being comprised of multiple, separate facets which can be ranked in a hierarchical fashion. Situation-specific dimensions are located at the base of the model, whereas more situation-stable, general self-concepts are at the apex.
3. Taxonomic -Self-concept is viewed as having many facets which are separate and situation-specific.

Compensatory
-Self-concept is seen as contammg several facets, which are related inversely.
For example, a low physical self-concept might be compensated for by a higher academic or social selfconcept.
As the amount of research has increased, and the variety and robustness of instruments used in self-concept research has expanded, an increasing amount of support has been found for multidimensional models of self-concept (see Marsh & Gouvernet, 1989, for an overview).
Although Marsh  It is possible that when the self-concepts of exceptional children (gifted, handicapped or learning disabled) are examined, support for one of the other models of self-concept may be found. Rosher and Howell (1978) suggest that the significance of the impact of disability status on Academic self-concept might best be explained as an overcompensation in one field when there is a weakness m another (in . Support for the Compensatory model has also been found when the self-concepts of learning disabled, gifted and normal students are compared . Finally, additional support for the Compensatory model has come from canonical correlational studies. Marx and Winne found bipolar relations among facets of self-concept, and interpreted the findings to mean that self-concept is composed of compensatory components (in . In regards to gifted children, an overall positive self-concept may be obtained by balancing a poor social or physical self-concept with a more positive performance in academic areas.

Instrumentation and Methodology Issues
The variety of instrumentation and methodology used in prev10us studies of self-concept make it problematic to draw general conclusions from the research literature. Most existing measures of self-concept view self-concept in a global manner. In addition, existing measures of self-concept have been widely criticized for their limited norms, lack of reliability, poor validity, and poor theoretical bases Lakey, cited in Schneider, 1987;. Two of the most widely-used measures of self-concept with children are the Coopersmith Scale and the Piers-Harris Children's Self-Concept Scale (Schneider, 1987).  reviewed the psychometric properties of these two scales. The Coopersmith inventory showed high test-retest and split-half reliability, but the validity was questioned, since the scale was constructed by reworking a version of an adult self-concept scale.
The Piers-Harris Scale was considered an improvement, since it derived its questions from a pool of statements generated by children.
However, the Piers-Harris mainly examines global selfconcept. Factor analyses of the Piers-Harris have derived several factors, but they have not been shown to be stable across studies .

5
Even when examining instruments that are designed to measure multi-dimensional aspects of self-concept, results have not been consistent. Marx and Winne (1978) and  used a multitrait-multimethod study in order to examine the construct interpretations of three self-concept inventories: the Factor analysis studies of various instruments also provided uneven support for the multidimensionality of self-concept. In general, the studies found evidence for more than one factor, but the derived factors were often difficult to interpret, inconsistent across different samples, or were not clearly related to the scales that were intended (Marsh & Smith, 1982). However, more recent research (see Marsh & Gouvemet, 1989, p A multitrait-multimethod analysis of these two instruments has demonstrated their convergent and discriminant validity (Marsh & Gouvernet, 1989).

Sex Differences in Self-Concept
Also of interest in the study of self-concept is whether sex differences exist, and, if so, in what domains of the construct. Wylie (1968) reviewed the literature, which consisted of primarily American studies and found that girls between the ages of 8 and 13 had more positive self-concepts than boys. In contrast, research which had used Australian subjects showed that boys had higher general self-concepts than girls (Connell, Stroobant, Sinclair, Connell, & Rogers, 1975;and Smith, 1975, 1978, cited in Marsh, Relich, & Smith, 1983. However, these studies did not control for the researchers' possible use of different models of self-concept, or instrumentation.
It is possible that differences may have been lost when a global scale or score was used. Marsh and his colleagues have continued to examine these sex differences in more detail.

7
Their work with pre-adolescent Australian populations has found some reliable findings across sexes specific to particular components of self-concept. Utilizing Marsh's Self-Description Questionnaire, which is based on a multifactor model of self-concept, they found that boys had a higher physical self-concept than girls. Within the academic domain, boys had a higher math self-concept, whereas girls had a higher reading self-concept (Marsh, 1985;Marsh, Barnes, Cairns, & Tidman, 1984;Marsh, Relich, & Smith, 1983 ).
Work has also been done focusing on the educational setting of the subjects and whether an interaction with sex exists. Marsh, Relich, and Smith (1983) found that sex differences in the academic domains previously mentioned were influenced by the whether the subjects attended single-sex classrooms or integrated classes. They found that sex differences were smaller for students attending single-sex classes. Marsh, Relich and Smith suggest that selfconcepts are formed relative to the reference group, so that when the group contains both boys and girls, the differences in self-concept are accentuated in the direction of traditional sexual stereotypes.
Conversely, it follows that if the reference group is of the same sex, then differences would not be as large.
The idea that one's reference group influences one's selfperceptions is particularly relevant in the study of exceptional children's self-concepts. Schneider ( 1987) reports that analyses of sex differences have been reported in very few studies of gifted 8 children's self-concepts. No sex differences were found by Karnes and Wherry ( 1981 ), but their self-concept measures were rather global. Kelly and Colangelo ( 1984) found that boys enrolled in a gifted program had higher global self-perceptions than boys enrolled in regular education classes. These regular education boys had higher self-perceptions than special learning needs males. There was no significant effect of class type for girls. However, in this study, the degree of integration that the exceptional children had with their regular classmates was not mentioned.  examined the self-concepts of gifted and non-gifted children in grades 5, 8 and 10. No sex differences were found at the 5th grade level. At grades 8 and 10, boys displayed higher general and physical self-concepts than girls, regardless of their gifted or nongifted group status. No sex differences were found in the academic or social self-concept domains.  again examined the self-concepts of gifted children as a function of age, sex, and school program. Gifted children in grade 5 who were integrated with their non-gifted peers for most of the school week had higher academic self-concepts than gifted children who were enrolled in self-contained classes. No sex differences in self-concept were found at grade 5. In grades 8 and 10, integrated gifted children had higher academic self-concepts. In addition, boys had higher global and physical self-concepts than girls. No multivariate interaction effects reached significance. Thus, age, sex and school program may be important variables in self-concept research.
Gifted Children's Self-Concepts One group which has been a focus of the self-concept research is the gifted. However, this group has been difficult to define since no universal definitions of giftedness have been adopted. Some researchers, such as , have defined the gifted in terms of high scores on tests of intelligence, yielding quotients of 130 and above.
Other researchers prefer definitions which additionally include areas of high achievement, as in above-average creativity and task commitment (Renzulli, 1978). Congress proposed a general definition used in Public Law (PL) 97-35, the Education Consolidation and Improvement Act, in 1981. According to this definition, the gifted are "children who give evidence of high performance capability in areas such as intellectual, creative, artistic, leadership capacity, or specific academic fields, and who require services or activities not ordinarily provided by the school in order to fully develop such capabilities (in ." The particular definition of giftedness utilized by the researcher is important to note when considering the literature on this population . Early studies of gifted childrens' self-concepts found that, generally, gifted children have a higher global self-concept than their non-gifted peers (see Schneider, 1987, for a review). Several researchers have found that when gifted children are compared with non-gifted children, the gifted subjects have higher scores across different measures of self-concept and self-esteem (Lehman & Erdmins, 1981;Maddux, Schneider, & Bass, 1982;Ketchman & Snyder, 1977, in Olszewski-Kubilius, 1989O'Such, Twyla, & Havertape, 1979). However, other researchers have reported no significant differences . The inconsistent findings may be due to several variables. Two such variables are the normative reference group used (if one was utilized) and the instrument's theoretical view of self-concept.  maintains that gifted children's self-concepts are greatly affected by how others perceive and react to special abilities. It is possible that the reference group affects gifted children's self-appraisals, so that when gifted students are homogeneously grouped and compare themselves to other high functioning children, the resulting self-appraisals will be lower than if they compared themselves to a heterogeneous mixture including children of lower abilities.
Addressing the second variable of the instrument's theoretical underpinnings, not all self-concept instruments divide self-concept into specific domains. Therefore, some of the inconsistent findings may be due to early studies utilizing instruments which were constructed viewing self-concept as a global concept rather than divided into distinct domains. It is possible that when discrete portions of the self-concept are examined, differences between gifted and non-gifted children may be found. This has been the case when academic self-concept has been examined. Researchers have found that gifted children have higher academic self-concepts, compared to their social or physical self-concepts (Ross & Parker, 1980). When gifted children have been compared to their non-gifted peers, it has been found that gifted children possess higher academic selfconcepts than non-gifted children (Kelly & Colangelo, 1984;. However, the research does not seem as clear cut when other portions of the self-concept are examined (see 1 1 Schneider, 1987, for a review). Kelly and Colangelo (1984) reported that gifted children also possessed higher social self-concepts than general students, although this finding was not confirmed when the control subjects were learning disabled . Thus, some research has proposed that gifted children have higher academic rather than other areas of self-concept, specifically the social self-concept. One may wonder whether children who possess high academic self-concepts are compensating for lower selfperceptions in other areas (Schneider, 1987). These findings may point in the direction that a compensatory process exists in the formulation of self-concept for all children.

Importance of Self-Concept Domains
In studying the self-perceptions of children, very little research has addressed the importance of the specific domains of self-concept to the child. However Harter borrowed from the Jamesian self- She then formulated a discrepancy score by subtracting the importance score from the competence score. Harter found that the results supported James' formulation since there was a systematic linear trend between the children's discrepancy scores and the self-worth scores. Children with low self-concept had the highest discrepancy scores, reflecting their view that the importance of the areas of the self-concept exceeded their perceived competence. As self-worth increased, the discrepancy score systematically decreased, so that children with high levels of self-worth had nearly a zero discrepancy score ). Harter interpreted these findings to mean that a child with low self-worth, or a low general self-concept, can not discount the importance of areas where he or she feels inadequate. Two domains were found to be particularly difficult to discount: behavioral conduct and scholastic competence. Harter points out that these two domains have particularly high normative importance ratings, and are ones that the children felt were most important to their parents.

Present Study
The purpose of this study 1s to address several issues. The first purpose · is to explore the global and domain-specific self-concepts of gifted children. As previously demonstrated, research in this area is not abundant. It is possible that gifted children have different selfconcepts than their normal ability peers. To address this possibility, self-concepts of gifted subjects will be compared with those of normal children. Based on Harter's preliminary research, it is hypothesized that gifted children will have higher academic selfconcepts than social or physical self-concepts, since the academic area is one that is difficult to discount and is probably additionally reinforced by teachers, parents and other gifted peers. Non-gifted children will not be expected to have the same self-concept pattern.
Second, like Harter and James, I believe that it is not enough to measure the self-concepts of children. The relative importance of the particular aspects of self-concept must also be determined. It Representatives from the program contact teachers at the end of the first and second grades for orientation regarding the selection procedure of children to enter the program in grade three.
Identification occurs through a combination of teacher recommendation, standardized test data, and parent inventory.
Participating students are bussed to a central school for a half day of activities once every other week. Instruction consists of a combination of small and large group activities which center around a common theme. Individual work relating to the unit of study is also assigned. Program goals include acquiring of advanced skills in critical thinking and cognitive synthesis, developing interpersonal skills, enhancing self-concept, and producing high-quality products. ALAP students are grouped by age. Fifth and sixth graders from the same school participate in activities together. District-wide, the total number of fifth grade students participating in the program was 60.
A total of 288 fifth graders participated in the study. Of this number, 34 were ALAP participants. The ALAP group's mean age was 10.5 years, with a range from 10 to 11 years of age. The regular education group's mean age was 10.7 years, with a range from 9 to 12 years of age. Sample sizes differed slightly for each analysis since incomplete profiles were dropped from the analysis.

Instruments
Each participant completed a packet of four instruments.
Instructions for each of the instruments were read aloud to the participants at the start of the task. Items on the longer checklists were also read aloud and the child was asked to follow at the examiner's pace.
1 5 The first instrument administered was a variation of Kuhn and McPartland's "Who Am I?" task (1954). This instrument requires the child to self-generate ten descriptions of him or herself by responding to the question, "Who Am I?" Kuhn and McPartland's version of the test was scored by coding the proportion of selfderogating statements and scoring for strength of self-derogation or self-positivity. Poor convergent validity with three additional selfconcept measures was found by . Data on discriminative and predictive validity was not reported. This type of scoring did not suite the purposes of this project, and another method was used. Validity and reliability data of this scoring system is not known. However, the instrument appeared to have face validity and followed from the purpose of the study, and therefore was included. This measure was chosen to be administered first because of its open-ended approach. It was believed that the participants would therefore have no preconceptions of reporting particular areas of self-concept.
The second instrument was developed for the purposes of the study. The "Importance Scale" consisted of ten statements which participants were asked to respond to on a 5-point Likert-type scale.
Responses ranged from "Not At All" (1) to "Very" (5) important. Each 1 6 question corresponded to each of the seven specific self-concept domains of Marsh, Parker and Smith's (1981) Self Description Questionnaire (SDQ), as well as to three larger self-concept domains (Academic, Physical and Social). The "Importance Scale" was administered second because it covered similar material to the SDQ, but in a more abstract manner.
The third instrument administered was Marsh 's Sill Description Questionnaire (Marsh, Parker, and Smith, 1981), an eightscale, 72-item instrument intended to measure seven aspects of the self-concepts of children between the ages of seven and thirteen , as well as a general global self-concept measure. Each item was responded to on a five-point Likert-type scale from "false" (1) to "true" (5). The areas of the SDQ include measures of Academic (Reading, Mathematics, and General-School subjects), Physical (Physical Abilities/sports and Physical Appearance), and Social (Peer Relations and Parent Relations) self-concepts. A measure of global self-concept is also available, but was not focused on in this study.  reviewed the SDQ and reported internal consistency alpha coefficients between .82 and .93. Test-retest coefficients were moderate to high values, depending on the test-retest interval.
Construct validity assessed through factor analysis yielded seven factors plus a global self-concept factor, as was expected by a priori analyses . The SDQ has since been revised (1988) and now incorporates slightly different items which correspond to the same eight factors of self-concept.
The final measure was a "Pie Chart" of self-concept which was constructed for the purposes of the study . This task required the I 7 participant to fill in the areas of Physical, Social and Academic selfconcept in order to describe how important each of these areas is in proportion to his or her overall self-concept. This measure was designed to be an alternative measure of the importance of the selfconcept areas, as well as a simple and interesting task to enhance the participant's motivation.

Procedures
Classroom teachers gave all fifth grade students at the participating schools a passive consent form to be read by their parents. Parents were asked to return the signed form only if they wished their child not to participate in the study (see Appendix A for a copy of the consent letter). Non-participating children were provided with alternate activities by their classroom teacher.
The examiner administered the inventories to all participants m their regular education classrooms. Participants were asked to read along with the examiner and complete the items at that rate.
Instruments were administered in the aforementioned order: the Who Am I?, the Importance Scale, the Self-Description Questionnaire,  A Principle Components Analysis of the ten statements comprising the Importance Scale was performed to assess the underlying structure of this measure. Three factors emerged, which accounted for 61.9 percent of the variance. Factor 1 corresponded to Academic self-concept, Factor 2 to Physical self-concept, and Factor 3 to Social self-concept. Table 2 shows the three factor structure and the factor loadings of each item. Factor 1 -Academic Self-Concept =--F=ac=t=o"'--r--=L=o=ad=i ... & 1. It is important to me to do well in Reading.
. 8 3 8. It is important to me to do well in Math.
. 81 5. It is important to me to do well in all my school subjects.
. 79 7. It is important to me to be a good student. .64 Factor 2 -Physical Self-Concept 9. It is important to me to be good-looking . 3. It is important to me to be healthy, sporty, and good-looking. 6. It is important to me to do well in sports.
Factor 3 -Social Self-Concept   scale of the negatively worded items was not utilized. Table 5 lists the scales and the particular items comprising each subscale.    The means are shown in Table 9. Simple Effects tests examining the differences between the academic groupings for each of the three domains were also performed. The results indicated that ALAP and regular education participants differed significantly in their ratings of the Social and Academic domains only.   Note: Each item was scored on a scale from 1 (false) to 5 (true).
In addition, a significant Sex by Subscale interaction was found, F(6, 1368)=5.19, p<.05. The two groups appeared to be rating their specific abilities quite differently. Females rated each self-concept area lower than males, except for the Reading and General-School domains. Simple effects tests were performed to determine which specific areas differed significantly from the others. Significant differences between males and females were found in the Physical Abilities, Physical Appearance, Peer Relations, Reading, and General-School areas. Figure 1 shows a graph of the seven domain scores for the two groups. Simple effects tests for domain differences within each sex group were also performed. Overall, these results mirror those found in the earlier three domain analysis. Table 12 shows the pairwise differences for males, and Table 13 shows · the pa1rw1se differences for females.  General-School abilities. A graph of the two group's scores for each specific area is shown in Figure 2. Simple effects tests for domain differences within each academic grouping were also performed.
Significant differences in the ratings of the seven domains were found within both the ALAP and Regular Education groups. The  Table 14 shows the pairwise differences for the ALAP group and Table 15 shows the pairwise differences for the Regular Education group.  Participants were asked to complete the instrument by designating and labeling slices of the pie chart in terms of the importance of each of the three self-concept domains to them.
Complete protocols were scored by ranking the pie slices from largest (1) to smallest (3 ). It was possible to achieve tied ranks. In terms of Academic status, a total of 37 ALAP and 246 Regular Education protocols were analyzed. In terms of gender, a total of 140 male and 132 female protocols were analyzed. Each of the self-concept domains were correlated with the others. The correlation between the Social and Physical domains was r=-.39; for the Academic and Physical domain, r=-.62; for the Academic and Social domains, r=-.47.
Inter-rater reliability .coefficients were calculated by having the faculty advisor rescore the random sample of fifty protocols. Kendall's W was also calculated for the male and female groups. The mean ranks of the three pie chart domains for each of the groups were plotted. Different patterns existed for the ALAP group , whose ranking of the areas from most important to least important was Academic, Physical, and Social, and the regular education students whose ranking of the self-concept domains was Physical, Social, and Academic. Figure 3 shows the rankings of the pie chart areas by ALAP grouping.
Differences also existed in the rankings of the areas for males and females.
Males ranked the Physical domain as most important.
Social and Academic self-concept was less important. Females M e a n R a n k

Discussion
Appraisal of General and Specific Self-Concepts 43 Significant differences were found in the participants' ratings of their global and domain-specific self-concepts. Sex differences were found on the global self-concept measure, with males reporting higher overall self-concepts than the females. This finding is in concordance with Marsh's research with Australian participants (Marsh, Relich, & Smith, 1983).
Significant differences were also found in the appraisal of three domains of self-concept. The participants were able to differentiate seven domains of self-concept . These areas were also rated differently, with the fifth-grade sample rating their relations with their parents and their physical abilities the highest. The overall sample rated their abilities in general school subjects the lowest.
Group differences were found in the ratings of self-concept domains. Significant Academic grouping by Domain interactions were found on both the three and seven domain analyses. As with the research of Kelly and Colangelo ( 1984 ), and , the ALAP participants demonstrated academic self-concepts that were significantly higher than that of their normal-ability peers. Results from the Who Am I? scale indicated that the ALAP and regular education groups differed in the of number of sentences they utilized in each self-concept category.
However, there were no significant differences between the two groups m the ranking of these categories.
When differences between the self-concept domains within the regular education and ALAP groups were examined, the results become more complex. Significant differences were found in the ratings of the three domains by the regular education sample. The Social domain was rated the highest, followed by the Physical 44 domain, and finally the Academic domain. However, the ALAP group did not perceive differences between the three self-concept areas.
Clearer sex differences were found in the ratings of selfconcept.
As with the general self-perception rating, males were found to have a higher composite self-concept score. Differences between males and females were also found of the ratings of the specific self-concept domains. Differences between the two groups were found primarily on the Physical domain, with males ranking the area the highest, and females, the lowest. In the seven-domain analyses, females reported lower abilities in all areas except Reading and General School subjects. This finding supports earlier research by Marsh, who found that females consistently reported higher Reading self-concepts than males.
Significant differences between ratings of the self-concept domains were also found within the male and female samples. On the three domain analysis, males rated the Physical and Social domains significantly higher than the Academic domain. Females rated the Social domain the highest, whereas the Physical and Academic domains were rated significantly lower. On the seven domain analysis, many significant differences were found within the groups. However, as with earlier research findings, females rated their reading abilities significantly higher than their math abilities.
In addition, males rated their math skills significantly higher than their reading skills. 45 Differences in the Importance of Self-Concept Significant differences were found in the overall rating by the entire sample of the importance of the three self-concept domains.
The Social area was rated as most important, the Academic area of second importance, and the Physical area as least important.
Group differences were also found in the importance of the This pattern must be examined with caution, however, as a low level of agreement was found within the groups in their rankings of the importance of the domains.
As with the appraisal of the self-concept areas, clearer sex differences were found in the importance of the self-concept domains. Males rated the Physical domain as significantly higher m importance than their female peers. Males also showed a higher level of concordance in their rankings on the Pie Chart Instrument.
Females rated the Academic, Social and Physical domains of similar importance.

Implications
The previous results offer confirming evidence for the concept of a multidimensional self-concept. Significant differences were found on the seven self-concept domains of the SDQ. This result was to be expected, since the SDQ was created through work with 46 Shavelson's hierarchical model of self-concept. Domain differences were also found on the Importance Scale, an instrument based on the specific self-concept areas of the SDQ. Differences in the rating of the importance of Physical, Academic and Social areas provide evidence for the multidimensionality of the importance of these domains.
An interaction of academic grouping and domain of selfconcept was found on the three and seven domain analyses of the SDQ. In concordance with the previous findings of Kelly and Colangelo (1984) and , the gifted sample rated the Academic area of self-concept higher than did their regular education peers. Within the social domain, significant differences were found in perceptions of peer relationships, with gifted children viewing their abilities in this area much lower.
Interestingly, no significant differences were found in the gifted sample's ratings of the importance of the various domains. Thus , although gifted children rated their social abilities with peers as the lowest domain, this area may not impact as profoundly on their overall self-concept.
When differences m the rating of the self-concept areas within the academic groupings were examined , results were inconsistent with previous research. Contrary to Ross and Parker ( 1980), the gifted sample did not rate the Academic domain significantly higher than the others. Instead, the ALAP group viewed the three areas relatively equally in terms of describing themselves. Differences were found in the importance of the three dimensions, pointing to a need for further examination of these areas for each of the groups m the future. Furthermore, because no significant differences were found in the ALAP sample's rating of the three domains, the research of   Future research may attempt to take into account these definitional and sampling problems. Research using definitional criteria as an additional variable may be conducted.  have utilized a similar approach, although those researchers did not use an instrument designed to address multiple dimensions of self-concept.
An · additional area for further research is to determine a way to link the appraisals of the self-concept domains with their relative perceived importance. The use of Harter's discrepancy formula model was not considered to be an appropriate means to analyze the instruments utilized in this research. However future research may provide a weighting method to more fully connect these aspects of self-concept and determine their influence on the global self-concept.  It is i u1portant to me to get along we ll with my parents.
It is important t o me to be healthy, sport y , and goodlooking .
It is import ant to me to have good friends 1&nd be liked by 111y friends.
It is important to me to do well in all my s c:hool subjects.
It is impor-tant to me to do well in sports .
It is ir.pcu·tant to me to be a good student.
It is i:n_, portant to me to do well in Kath.
It i• i,.nportant to =e to be good-lo ,oking.
It is i::porta n t to me to b• lik•d and gt1L along well with people .