DO DISABLED ADOLESCENTS VIEW THEMSELVES DIFFERENTLY FROM OTHER YOUNG PEOPLE?

Viewing physical disability as a stigma, quite a few articles have over the years dealt with issues relating to the self-concept of adolescents with physical disabilities. Using the political guidelines stressing social integration and participation for people with disabilities as a starting point this article focuses on how indicators of social integration can explain variation in the self-concept among adolescents aged 13-20. The analysis shows that there are small differences in selfconcept between disabled and non-disabled adolescents. The analysis reveals however that parental behaviour is of significant greater importance for the selfconcept of disabled than for non-disabled adolescents. Towards the end of the article the provoking question is asked if parents may serve as "gatekeepers" for social participation for adolescents with disabilities thus contributing to continuing segregation and stigmatisation. At least for the last two decades, the main guideline for Norwegian policy towards people with disabilities has been to prevent discrimination and segregation through policies and reforms aimed at increasing the level of social integration and participation in different social arenas. These national policies echo internationally based interest in the conditions for disabled people, starting with UN's "International year of disabled persons " in 1981. It should however be added that this interest can be traced back to the work of disabled activists in the late sixties and early seventies. One can with some assurance postulate that today's adolescents with disabilities represent a first generation grown up in an "inclusive society". Following this line of thought one could expect that adolescents growing up in an inclusive society would not experience segregation and discrimination to the same extent as earlier generations of disabled adolescents. Instead, growing up together with non-disabled children, sharing at least to some extent a common social world, could result in similarities also in "inner" representations like the self-concept. Two questions are asked in this article. First, is there a difference between the SJDR Volume 2, No. 1 2000 39 LARS GRUE AND ARVID HEffiERG self-concept of disabled and nondisabled adolescents and secondly, what significance do contextual conditions (reflecting social participation) have in explaining variations in self-concept for adolescents with disabilities? The main objective of the study has been to make it comparative in the sense that the findings will be related to findings from a sample of non-disabled adolescents. The term disability is a complex one and therefore one that has been extensively discussed by academics and activists (Oliver 1990; 1996, Søder 1990, Booth 1991, Crow 1996; Pinder 1996, Barnes 1996, Shakespeare 1996), just to mention some of the important contributors in this debate. Drawing on a social model of disability we are aware that the problems disabled adolescents experience will be connected to structural and institutional arrangements and practices in the surrounding society. As a consequence adolescents with different disabilities may also experience their disability differently. On the other hand their experience of being different may be the same since all groups represent minorities in the larger society. To trace possible differences between adolescents with different disabilities we decided to let the study include one group with physical disabilities, primarily affecting the ability to walk, and one group with hearing impairments. As it turned out in the preliminary analysis of the data (Grue 1998), the difference between the two main groups were minor and in the analysis presented in this article we have decided to treat them as one group We want however to stress that this study is based on a questionnaire. This may cause problems to some adolescents with a hearing-impairment since they use sign-language as their "first language" and may not be as comfortable with written texts as other adolescents. Some adolescents with a severe hearing impairment will also, to a degree, attend special schools or classes and as a consequence view the hearing-impaired as a cultural minority. To what extent these facts may have "coloured" the answers given by the hearing-impaired adolescents in this study is impossible to assess, but our uncertainty should be noted.

At least for the last two decades, the main guideline for Norwegian policy towards people with disabilities has been to prevent discrimination and segregation through policies and reforms aimed at increasing the level of social integration and participation in different social arenas. These national policies echo internationally based interest in the conditions for disabled people, starting with UN's "International year of disabled persons " in 1981. It should however be added that this interest can be traced back to the work of disabled activists in the late sixties and early seventies. One can with some assurance postulate that today's adolescents with disabilities represent a first generation grown up in an "inclusive society". Following this line of thought one could expect that adolescents growing up in an inclusive society would not experience segregation and discrimination to the same extent as earlier generations of disabled adolescents. Instead, growing up together with non-disabled children, sharing at least to some extent a common social world, could result in similarities also in "inner" representations like the self-concept.
Two questions are asked in this article. First, is there a difference between the SJDR -Volume 2, No. 1 -2000 self-concept of disabled and nondisabled adolescents and secondly, what significance do contextual conditions (reflecting social participation) have in explaining variations in self-concept for adolescents with disabilities? The main objective of the study has been to make it comparative in the sense that the findings will be related to findings from a sample of non-disabled adolescents.
The term disability is a complex one and therefore one that has been extensively discussed by academics and activists (Oliver 1990;, Søder 1990, Booth 1991, Crow 1996Pinder 1996, Barnes 1996, Shakespeare 1996, just to mention some of the important contributors in this debate. Drawing on a social model of disability we are aware that the problems disabled adolescents experience will be connected to structural and institutional arrangements and practices in the surrounding society. As a consequence adolescents with different disabilities may also experience their disability differently. On the other hand their experience of being different may be the same since all groups represent minorities in the larger society. To trace possible differences between adolescents with different disabilities we decided to let the study include one group with physical disabilities, primarily affecting the ability to walk, and one group with hearing impairments. As it turned out in the preliminary analysis of the data (Grue 1998), the difference between the two main groups were minor and in the analysis presented in this article we have decided to treat them as one group We want however to stress that this study is based on a questionnaire. This may cause problems to some adolescents with a hearing-impairment since they use sign-language as their "first language" and may not be as comfortable with written texts as other adolescents. Some adolescents with a severe hearing impairment will also, to a degree, attend special schools or classes and as a consequence view the hearing-impaired as a cultural minority.
To what extent these facts may have "coloured" the answers given by the hearing-impaired adolescents in this study is impossible to assess, but our uncertainty should be noted.

Self concept
Historically there is a dividing line between viewing self-concept as a onedimensional construct (Coppersmith 1967;Piers-Harris 1969) or as a multidimensional set of domain-specific selfconcepts (Mullener & Laird 1971;Marsh 1987). Rosenberg (1979) has developed a model to incorporate both perspectives. This understanding has been further developed by Harter (1985;1986; and Harter et al (1997Harter et al ( , 1998. Some caution is required in establishing a clear distinction between an understanding which primarily emphasises the merits of domain-specific self-concept and an understanding whose primary focus is on a global self-concept. This caution underlies Harter's model (1988) and is supported by, among others, Bracken (1996). He asserts that the self-concept can best be described and understood along a continuum from domainspecific to global self-concepts, and not necessarily in such a way that emphasis on one perspective excludes other perspectives. Self-concept is hardly a constant variable, but can change in the course of a lifetime.
Harter's model (Self-Perception Profile for Adolescents (1986,1988) forms the basis for analysis of self-concept in this article. The model represents an approach that both emphasises the significance of domain-specific selfconcepts and the significance of a global self-concept standing above domain-specific self-concepts. Hence the model permits illumination of the relation between domain-specific selfconcepts and a global self-concept. Harter belongs to a cognitive tradition within social psychology, but her approach incorporates viewpoints anchored in a behaviourist tradition among others Bracken (1996).
Harter's Self-Perception Profile for Adolescents (1988) contains in all nine sub-scales: Whereas the various sub-scales measure domain-specific self-worth, according to Harter's reasoning the "global selfworth" 8 scale provides a general measure of the individual's degree of satisfaction with himself or herself when account is taken of the fact that different people attach different importance to the individual sub-domains on which self-perception is based.
In this study we have chosen a model that reflects an understanding in which the self-concept is viewed as an empirical correlate to the extent and the quality of social affiliation in various domains. The self-concept develops in interaction with key social actors in life domains that are important for the individual. In adolescence these actors are parents, peers and school.
There is a growing amount of literature on the self-concept of physically disabled young people, but much of the existing literature is confined to pointing out that disabilities influence or do not influence the self-concept. Less attention has been paid to identify those factors that can help explaining variation in the self-concept among disabled adolescents (Specht et al. 1998). Research carried out, particularly some years ago, documents low self-concept (and a high incidence of depression) among physically disabled young people (Schechter 1961;Glaser, Harrison & Lynn 1964). More recent research (Harvey & Greenway 1984;Varni et al 1989;Lawrence 1991) also finds that this group of young people have a poorer self-concept than other young people. On the other hand, researchers such as Eckhart (1988), Sherill (1990), Arnold & Chapman 1992, King et al. 1993, Appelton et al. (1994, Stevens et al. (1996), and Specht et al (1998), find that on the whole disabled adolescents do not have lower self-concept than other adolescents.

The samples
The following analysis builds on two sets of data based on the same quetionnaire. The first set, Young in Norway (YiN) comprises a selection of 9,680 young people in the age range 13-20 years. The respondents are pupils from 67 schools where each grade is equally represented. Using a register in which all schools in Norway are represented, a cluster sample was made with the school as a unit. This sample was then stratified in relation to geographical area and school size. In Norway 98.5 per cent of all young people in the age range 13 to 16 are in the ordinary school system (primary and lower-secondary school). After primary and lower-secondary school the vast majority go on to upper-secondary school, and 80 per cent of 18 year olds are still at school. The sole criterion for exclusion from the survey was major reading and writing difficulties. This was because data collection was based on personal completion of the questionnaire in school hours. The reply rate in this survey was 97 per cent.
The second data set comprises 311 disabled young people in the age range 13-20 (referred to as DY in the survey) who either have a hearing impairment or physical disability. The group of hearingimpaired comprises sign-language users as well as non-sign-language users. The group of physically-disabled comprises young people with health conditions such as osteogenesis imperfecta, meningo-myelocele, juvenile rheumatoid arthrites, cystic fibrosis and congenital heart defects. The group of hearingimpaired was contacted through D0ves Forlag (a publishing company for people with hearing impairments), fylkesaudiopedagogene (professionals responsible at county level for education of the hearing-impaired) and Briskeby kompetansesenter (a national competence centre for the hearing-impaired). The group of physically disabled was contacted through Frambu, a national center for people with rare impairments. Of the total of 251 forms dispatched through Frambu Health Centre, 176 or 69 per cent were returned (response rate). Of these, we decided to reject 19, mainly due to incomplete entries. The final sample comprises 157 persons in the age range 13-20, corresponding to 62 per cent of the total number of forms administered. The group of hearingimpaired were sent 280 forms. The share of returned forms was 185 or 66 per cent. Of these, 31 were rejected due to poor completion. The final sample of this group comprises 154 persons corresponding to 55 per cent of the total number of forms administered. It is difficult to make firm statements on the effects of the dropout rate on the empirical analysis that followed. What we can say is that the 19 forms from the physically disabled group, and the 31 forms from the hearing-impaired group that we decided to exclude from the analysis gave no indication to a systematic dropout on central variables like type or degree of disability. These forms were left out because many of the major questions were left unanswered. There is however a possibility that not answering some questions could indicate some sort of personal problems with relevance to one's self-concept. One the other hand one can not postulate that such possible problems would be related to being disabled. Also in the YiN study, forms were excluded due to poor completion. The major problems relate to the real "dropouts", the nonresponders. In the physically disabled sample 31 per cent and in the hearingimpaired sample 34 per cent of the forms were not returned. Unfortunately we had no possibility to administer an analysis of the "real dropouts". These adolescents could very well represent a group experiencing major problems directly affecting their self-concept in a negative way. On the other hand it could just as easy be argued that this group of "dropouts" could represent adolescents not looking upon themselves as disabled at all. On receiving the form from one of the agents administering the forms, they could then decide not to participate just because they did not feel that they belonged to the group to be studied.
We do not know with complete certainty what the samples and analysis would look like if all or most of "the dropouts" were included. Accordingly we can not state that the sample and analysis definitely are representative for a population of "physically disabled and hearing impaired youth". As a consequence we will ask the reader to show some caution when it comes to making vast generalisations from the study and the data presented.

The Scale
To adapt Harter's SPPA as far as possible to Norwegian reality, the original measure was somewhat altered. The sub-scales Job Competence and Behavioural Conduct were omitted. Job Competence was omitted primarily because combining work with schooling is far less common among young Norwegians than among young Americans. Behavioural Conduct was omitted because a number of surveys have shown this scale to be unreliable (Wickstr0m 1995).
Moreover, the items on the Romantic Appeal scale are adapted to Norwegian youth culture, and the wording of questions was altered somewhat in relation to the original set-up. Analysis of the revised SPPA shows that revision did not weaken the original SPPA, either with regard to reliability or validity (Wickstr0m 1995). The items included in "the global self-worth scale" are two negative items ("I am often disappointed with myself and "I do not like the way I'm leading my life" and three positive items "I'm happy with myself most of the time" and "I like the kind of person I am" and "I'm very pleased with myself).

The model
To answer our first question, if there's a difference between the self-concept of disabled and non disabled adolescents, we have chosen to present the mean and standard deviation for the different scales in Harter's SSPA for both samples.
To answer the second question, what significance do contextual conditions have in explaining variations in the selfconcept, we have constructed a multivariate model. In the model, different measures of contextual conditions reflecting social participation and inclusion are introduced in blocks. The first block deals with loneliness, or the experienced degree of lack of social affiliation. The measure used is a short version of the UCLA Loneliness Scale (Franzoi & Davis 1985), an indexed measure initially developed by Russel et al (1980). Experienced loneliness is viewed as an indication of low level of overall social participation. The short version of UCLA consists of four items, two positively worded items ("I feel in tune with people around me" and "I can find companionship when I want it") and two negatively worded items ("No one really knows me well" and "People are around me but not with me"). A preliminary analysis (Grue 1998:163) shows almost no difference between the two samples on this index. On a scale from 1-4, (where 4 represents a high level of felt loneliness) the mean values are 1.91 (standard deviation .59) for DY and 1.83 (standard deviation .54) for YiN.
Block 2 comprises three formative indices that measure attitudes to school. The indices, formative in nature, may however reflect social participation and inclusion in school life and activities. As a consequence the indices can be useful in an analysis of the connection between the school as a social arena and adolescents' self-concept. These measures encompass both the social and knowledge-related aspects of everyday school life (Grue 1998). Some examples of items included in the indices ("Good grades are important" and "We learn interesting things at school") measuring positive attitudes, ("Going to school is a waste of time" and "There is too much theory and not enough practical work at school") measuring negative attitudes and ("I have had a fight with one of the teachers" and " I have been sent out of the classroom") measuring opposition to school. A preliminary analysis (Grue 1998:124-125,136) showed no significant differences between the two samples on either of these tree indices).
Block 3 comprises four formative indices that measure degrees of participation in leisure time activities. Index 1 measures the degree of participation in purposive activities (e.g. a job after school hours), index 2 measures participation in adult-controlled activities (e.g. pursued something like a hobby together with my mother or father), index 3 measures peer-oriented activities (e.g. been to a youth club) and index 4 measures participation in what has been defined passive use of time (watching television and video). As for the indices in block 2 a preliminary analysis of the indices that form block 3 (Grue 1998: 156-160) showed no significant differences in the overall activity level for the two samples.
Block 4 comprises three reflective indices that measure parental behaviour. The first two, care and over-protection, are taken from Parental Bonding Instrument (Parker et al 1979). The care dimension covers the following positive items (My Parents) ("Were affectionate to me" and "Appeared to understand my problems and worries) and negative items ("Did not talk with me much" and "Did not help me as much as I needed") and "Did not seem to understand what I needed or wanted"). The protective dimension covers the following positive items. (My parents) ("Liked me to make my own decisions" and "Let me decide things for myself) and negative items ("Were overprotective of me" and "Did not want me to grow up" and 'Tried to control everything I did"). The third index comprises items that measure parental involvement in the adolescent's friends ("My parents usually know where I am and what I do during weekends" and "My parents know pretty well whom I have as friends" and "My parents know most of my friends" and "My parents like most of my friends" and "It's important for my parents to know where I am and what I do in my free time").
We would argue that the three indices can be used as valid indicators on how parental behaviour is experienced by adolescents. The assumtion that follows is that parental behaviour also will be of some importance for adolescents' opportunities to participate in social activities. This then implies that the indices should be included in a multivariate analysis of factors of importance for adolescents' self-concept. A preliminary analysis (Grue 1998:110-114) showed no significant differences between the two samples on the care and protective dimension indicating that both groups experienced their parents' behaviour similarly. The index measuring parents' involvement in friends showed a small but significant difference (p<.05) between the two samples. Disabled youth experienced that their parents were slightly less involved in their friends than the non-disabled group. However, both groups reported a high level of parental involvement with their friends.
As a fifth block in the model we have decided to include the variables age and gender.

Results
Analyses show overall small, but some significant, differences between disabled and non-disabled young peoples' domainspecific and global self-concepts when measures of central tendency and variance are employed (table I). has shown that values below the mean young people's self-concept, but that only rarely occur in any of the sub-disabled young people have somewhat scales (Wickstr0m 1995). Table I lower self-esteem in the domains shows generally large similarities athletic competence, practical skills and between disabled and non-disabled romantic appeal. Furthermore, a correlation analysis (Pearson's r) showed relatively minor differences in the strength of bivariate correlations between the various categories of domain-specific selfconcepts and between domain-specific self-concepts and global self-concept for the two groups ( Table 2). The analysis shows a close correlation between global self-concept and physical appearance. However, the strength of this correlation is the same for both groups. Block 1 of the model presented in table  3 shows that for both groups the experience of loneliness has a bearing on "global self-worth". With this variable the explained variance in the variable "global self-worth" is 16 per cent for YiN and 18 per cent for DY.
Block 2 of the model shows that for both groups negative attitudes to school are associated with lower "global selfworth", while a positive attitude is associated with improved "global selfworth". The variables in block 2 contribute to an increase in the explained variance of 4 per cent for YiN and 7 per cent for DY.
Block 3 of the model shows that the extent and location of leisure activities has no significance for "global selfworth".
Block 4 of the model shows that young people's experience of parental behaviour is of central significance for "global self-worth". Increased care has positive significance for "global selfworth" for both groups. For disabled adolescents the degree of parental overprotection and "parental involvement in friends" is also of significance. Increased over-protectiveness makes for poorer "global self-worth". Reduced parental involvement in the adolescent's friends has the same effect. The analysis further shows that experiencing parents as caring or over-protective affects the significance of loneliness for "global self-worth". Introducing the variables one by one shows that the "care" variable is significant for reducing the effect of the variable "loneliness". Hence, overall, block 4 of the model shows that parental behaviour is of importance for young people's self-concept, an observation also documented in earlier surveys (May 1967, Rosenberg 1979, Harter et al 1996,1997, but more so for disabled than for non-disabled adolescents. For YiN the variables in block 5 raise the explained variance by 5 per cent and for DY by 6 per cent. Block 5 of the table, which encompasses the variables "gender" and "age", shows that "gender" is a central variable. The analysis shows that boys (value 2) have higher "global selfworth" than girls (value 1) and that this applies both to disabled and nondisabled adolescents. Whether one is a boy or a girl is also significant for the relation between loneliness and "global self-worth", but only in the case of disabled adolescents. For disabled boys the significance of loneliness for "global self-worth" is reduced. The explained variance for the two variables included in block 5 is 5 per cent for YiN and 3 per cent for DY. The explained variance for the entire model is 34 per cent for the DY sample and 30 per cent for the YiN sample.
The most important difference between YiN and DY, when all variables are included in the analysis, refers to parental behaviour. Administering a two tailed t-test (Paternoster et al. 1998), the following t values as regards the size of the difference in regression coefficients were produced, "Over-protection" t=2.50, "Involvement" t =2.01.

Discussion
The overall impression and main finding from the analysis is that there are small differences in the self-concept between adolescents with and without disabilities (table 1). For both groups the experience of loneliness clearly and unequivocally has a negative correlation with self-worth (table 3). From the data we do however not know the exact reason behind the experience of loneliness. The reason could either be a lack of friendships (having few friends), or a feeling of having less intimate and close friends that one would have wanted. However, the main point is that for most people, the experience of loneliness prompts them to ask why they are feeling lonely. Is it because others do not wish to have one a friend, or as a close friend? Are you not attractive as a friend? Understood in this way, the experience of loneliness may reflect insecurity as regards one's own attractiveness and it is this insecurity that is reflected in a low selfconcept. That the experience of loneliness has such a central significance for the self-worth as found in the analysis, underlines the significance of social affiliation in adolescence. The close connection between global selfworth and the domain-specific selfconcepts of "appearance", "social acceptance" and "romantic appeal" (table 2), underscores the importance of being "liked" in adolescence.
Leisure time activities and most of all the educational system represent social arenas where policies and reforms aimed at increasing social participation for disabled children and adolescents have been visible. The school represents a social arena where all young people attain the same social status. They become pupils which may constitute a feeling of oneness in relation to the teachers and school administration. Further more age is the most institutionalised form of social grouping in schools. Children and adolescents are divided into year groups and not grouped by any other characteristic they may have like sex, colour of the skin or disability (we are aware that exceptions can be found). In this sense the school and educational system represent an inclusive social arena where disability may be of lesser importance than other characteristics. Returning to the analysis (table 3) one finds that negative attitudes to school are associated with a negative self-concept, while positive attitudes are associated with a positive self-concept. A reasonable interpretation of this finding is that it reflects the degree of social acceptance and social affiliation with one's fellow students. Positive attitudes towards school can follow as a consequence of sharing attitudes and evaluations with one's fellow students. On the whole adolescents do have positive attitudes towards going to school and to the process of learning. Adolescents not sharing this common attitude, will to an extent be outsiders or at least different from the majority of their fellow students. Whether their negative attitudes are a response to the educational regime or to a lack of social involvement is hard to tell. And maybe not that important either since the result will be the same -a feeling of being different from the others. Since there is no significant difference between the two samples we feel confident to argue that this observation supports our initial assumptions regarding the consequences of inclusive policies in this sector of adolescents' lives. Drawing a Foucaultian perspective of identity formation, Priestley (1999) points out that on the one hand we become known to others through a variety of external disciplines and discourses often institutionally embedded. On the other hand we make ourselves known through selfknowledge and by speaking about ourselves, making ourselves social subjects. He studied disabled children and adolescents in two mainstream schools and discovered on the one hand that other pupils and staff made them known by using stereotype categories and discourses. On the other hand disabled pupils made themselves known in a variety of ways. He further observed that age and gender could be more important for group formation than disability. The study showed that the disabled pupils, like other children and adolescents, spoke about unpleasant incidents at school and they were not passive in the construction of their identities within the school context. He stated that: "These are not passive vulnerable children in the Dickensian novel or the socio-medical research literature " (Priestley 199:98).
In an other study, Allan (1996) points out that disabled children in school did not adopt fixed disability identities or standardised strategies for resistance towards being categorised. Their identities shifted in an ongoing and dynamic process. This is a point also stressed by Harter (1988) who further argues that adolescents may put different weight on different domains. The point we want to make in relation to the analysis presented above is that the inclusion of disabled children and adolescents in mainstream schools can be understood as an inclusion in an opportunity structure. The social field available for negotiating identities and domain specific self-concepts is more open and less fixed than in a segregated school. Disabled adolescents may certainly experience problems in relation to fellow pupils and the meanings structures legitimated by the school and as a consequence develop a low self-concept on some domains. On the other hand the school may open up for resistance strategies and for opportunities to develop positive selfconcepts on other domains resulting in an overall high score on "global selfworth".
Turning now to the observation that changes in levels of participation in various types of activities are not found to have any significance for change in self-concept, either for disabled or nondisabled young people. This was an observation that initially surprised us since we had suspected a rather close correlation between participation in both organised and unorganised leisure time activities. One possible explanation for this lack of correlation between participation and self-concept could of course well be that the activities included in the questionnaire, and therefore in the analysis, are not reflecting affiliation and participation of substantial importance in relation to adolescents' self-concept. The important point to stress in this respect, is the fact that during the last few decades our society has turned into a more open society than before. This leading to the observation that today a broader spectre of activities and opportunities are available not only for non-disabled, but also for disabled persons during childhood and adolescence. Following this line of argument one could argue that this spectre of opportunities are difficult to cover in a questionnaire. At least in such a way that one can identify activities of importance for developing domain specific self-concepts that on the whole will contribute to an overall positive "global self-worth".
On the other hand we would like to add that there could be a different explanation. From the preliminary analysis of the two samples we know that participation in various activities listed, at the outset is generally fairly high for most young people in both groups. Hence a change in the level of activity will only have marginal effects, also in relation to the self-concept.
As stated above both the educational system and leisure time activities represent social arenas where political initiatives aimed at increasing the level of participation for disabled adolescents can be traced. What happens within the family context is still pretty much dependent on family resources in the widest meaning of the term and on parental behaviour.
Caring parents may be both attentive and involved in the young person's life. They take problems and questions the adolescent is grappling with seriously and respond and give feedback that helps to confirm and strengthen the young person's evaluations and decisions. Care shown by parents may also be expressed in active parental encouragement to establish contacts and friendships outside the family. In this way they give the adolescent a feeling of confidence in his or her personal choices and decisions, which in turn has a positive effect on his or her selfconcept. Parental significance for the self-concept formation has also been thoroughly documented by Harter et al, (1997;. For both groups, care shown by parents is associated with "global self-worth" in such a way that an increase in care contributes to a more positive "global self-worth". Adolescents who experience their parents as caring and participating have a better "global self-worth". Parental care helps to reduce the significance of loneliness for "global self-worth". For disabled adolescents increased over-protectiveness is associated with poorer "global self-worth" -an observation also made by other authors (Blumetal, 1991;Lie, 1993). The importance for adolescents' selfconcept of parents who are not overprotective, but supportive and caring, has also been documented in other studies (Blum et al, 1991;Resnick et al, 1992). Moreover, Wolman & Basco (1994) in a study of young people with spina bifida have shown that the factors that made the biggest positive contribution to young people's selfconcept were age-adapted parental behaviour which allows increasing scope (with age) for social interaction with peers.
For non-disabled young people the degree of parental involvement in their friends has no bearing on the selfconcept. For disabled young people, on the other hand, this is not the case. The explanation may be that experiencing a lack of parental interest in one's friends may reflect a basic parental lack of interest in the life the adolescent lives or wants to live outside the family. That this is only significant for disabled young people's self-concept probably merely reflects the fact that parents in general are more significant in the lives of disabled young people. For this group of young people it is important that parents are interested in their life outside the family and concerned with what they are doing and with whom. In respect to the interpretations above, we would however like to make the point that the nature of the variables in the analysis makes it difficult to have clear cut opinions on the direction of causality in the correlation between parental behaviour and self-concept.
To be more precise and to give an illustration. For example we find that parental care is clearly correlated with "global self-worth" and within the model (table 3) we interpret this causally with parental care as the independent variable. We are however aware of the fact that experiencing ones parents as caring could be an expression of a positive appreciation of ones selfconcept. As a more general comment it could be said that many models in the social sciences are constructed in a way that implies that the scientist proceeds on an "as if manner in relation to the status of the different variables in the analysis. Causality is often assumed and used in model building as a means for obtaining new knowledge and insights (Asher 1976).
Several surveys have documented how gender-specific socialisation in the family has material consequences for boys' and girls' perception of personal abilities and opportunities. Regard for personal abilities is highest among boys. A possible explanation for this observation is that parents may tend to have lower expectations of girls' intellectual endeavour. Hence they contribute (unintentionally) to girls' internalising a picture of themselves as weaker than boys in intellectual contexts (Philips & Zimmerman 1990). A survey by Spence and Helmreich (1978) problematises the gender-role issue, showing that gender-role identity appears to be the key factor in understanding the relation between selfperception and personal abilities. Girls with a feminine identity more frequently held their personal abilities and potentials in low esteem compared with boys and compared with girls having what was termed "an androgynous self-perception". Educational research, particularly research targeted at illuminating what happens in the classroom, has also uncovered some of the mechanisms underlying and possibly explaining the reason for girls' and boys' differing perceptions of personal abilities (see for example Bjerrum Nielsen & Rudberg, 1989;Entwhistle 1993).
The analysis showed that for disabled boys the significance of loneliness was not quite as important for "global selfworth" as for disabled girls. One possible explanation for this difference, which is small but noticeable, could be that disabled girls on the whole are more oriented socially towards friends than are boys. This indicating that parents, in this respect, may play a more important role for disabled boys than for disabled girls of which the analysis in table 3 gives some indication.

Conclusion
The analysis shows that disabled youth as a group do not have lower "global self-worth" than non-disabled youth. The results underline the importance that social participation on the same social arenas as others and the feeling of being socially integrated have for one's self-concept in adolescence. In this respect the results indicate that the political guideline stressing social participation for disabled people, has been successful when adolescents are focused in relation to self-concept. It is however important to highlight the significance of parental behaviour. Care-dominated parental behaviour is associated with high self-concept for both groups, while over-protective parental behaviour is primarily of importance in terms of weakening the self-concept of disabled adolescents. One possible explanation for this observation may be that parents can serve as "gatekeepers" to young people's social participation and thus reduce the possibilities of social participation made possible in society as a whole. If young people with disabilities, due to parental "protection" are cut off from, or experience a low degree of social participation, they are in danger of experiencing social isolation and stigmatisation that in turn may surface as a lower self-concept. We hope this article may be of some importance in triggering research that may shed more light on the connection between social integration and selfconcept in adolescence.
Notes: 1 Susan Harter, in her Self Perception Scale for adolescents, uses the terms "global self-worth" and domain specific self-worth. We are using these terms when we are directly referring to her scales and to the tables. As noticed we use the expression "self-concept" elsewhere in this article. There is however no difference of meaning between "self-concept" and "global self-worth" as the concepts are used in this article.