Most young people in Norway attend upper secondary school. In this country more than 95% of those leaving lower secondary school in the spring of a particular year start upper secondary education in the autumn. This is also the case for most vulnerable youth, those at risk for later adverse development, e.g. students with special educational needs. However, the flow through upper secondary education is poor. Statistics Norway (2009) reports that only two-thirds of the 2003 admission cohort had completed upper secondary school after five years. This rate has been fairly constant over the past 10 years. For students in the highest decile of marks when they started in upper secondary education, more than 90% succeed, whereas the corresponding rate for the lowest decile is less than 15%. Because most students with special educational needs belong to the lowest admission level, these data indicate that their success in upper secondary education is very modest. Adolescents with special educational needs are often considered to be vulnerable, for example in On your own without a net (Foster et al. 2005), a book that discusses the transition to adulthood for six vulnerable populations. Among them are young people with learning disabilities, emotional disturbances and behavior disorders, mental disability, and social problems. These are conditions that qualify for special educational provision.
This article–which presents longitudinal data from a large-scale longitudinal project–examines the transition to four domains of adult life for former students with special educational needs. When this research started in 1996, these adolescents were 17 to 18-years-old, and at that age they had different kinds of disabilities. Approximately 45% were diagnosed with general learning difficulties. More than half of the students struggled with basic skills, such as reading, writing, and arithmetic. A large number of them also displayed psychosocial problems, such as difficulties with human interaction, challenging behaviour, lack of care, and drug problems. Approximately 50% of the adolescents suffered from at least one of these problems. However, it should be noted that most of these difficulties were characterised as minor ones. Precisely defined functional difficulties related to eyesight, hearing, movement, and coordination affected just a small percentage. However, comorbidity was high; most students were diagnosed with more than one functional difficulty. For more details, see the Appendices in Båtevik, Kvalsund, and Myklebust (1997). It should be observed that specific functional difficulties are not used in the subsequent analyses, but an additive index called functional level.
This research project has so far primarily analysed transitional process among these young people from the age of 17 to their early twenties, a phase of life often referred to as emerging adulthood (cf. Arnett 2000). These analyses include important domains like competence attainment (Myklebust 2007), employment and economic independence (Båtevik and Myklebust 2006; Myklebust and Båtevik 2005), spare-time networks (Kvalsund and Bele 2010), pregnancy and reproduction (Myklebust and Solvang 2005), romantic relationships and independent living (Myklebust 2004). The analyses reveal the significance of a life course perspective, which emphasizes how prior conditions and transitions–in education, for example–influence present adaptation (cf. Pallas 2003).
Previously life course research frequently restricted itself to investigating one event or one transition at a time (Buchmann and Kriesi 2011: 484). This has also been the case in the present project, for example analyses of economic independence (cf. Myklebust and Båtevik 2005). The aim of the present article is to analyse transitions to four domains of adult life and thus, show a more complete picture of the life courses of vulnerable youth. In accordance with the ambitions of modern life course theory (cf. Mayer 2009), the goal is not only to study single transitions but also to analyse interrelated trajectories from adolescence to adulthood. A similar shift of focus towards a holistic, life course perspective is also evident in disability research (cf. Stewart et al. 2010: 10). Such analyses may then reveal patterns of cumulative advantage and disadvantage, for example how disabilities and educational outcomes influence subsequent adaptation to adult life.
This longitudinal study draws inspiration from two complementary fields of knowledge: life span psychology and life course sociology. The first of these includes the proximal context (e.g. family and social relationships), whereas the second approach includes more distal features (e.g. structural and cultural context). The combination of proximal and distal contexts ‘supplies a set of space-time coordinates, which differently shape the timing and form of the transitions through which the life course is constructed’ (Bynner 2008: 219).
The life course is a theoretical orientation anchored in a contextual perspective. A theoretical orientation is a framework for research that helps to identify and formulate research topics and guides the selection of methodological design. This has definitively been the case in the present research project that started in the mid 1990s (cf. Kvalsund and Myklebust 1996). Essentially, the term life course refers to the course of biological ageing. However, in the social sciences, life in society is pivotal. An individual's life course is affected by past events, is influenced by present opportunities or restrictions, and can often be comprehended on the basis of future prospects. The life course is frequently perceived as the sum of those pathways or trajectories (e.g. in education, work, and family life) that individuals and groups follow through historical, geographical, or structural contexts. An example of such a context is the present structure of special education in upper secondary schools in Norway.
Today, the life course perspective is increasingly applied to the study of social processes. Naturally, the life course manifests itself not only in the long term but also in the short term, thus making transition the most appropriate concept. This view is in accordance with ideas expressed by leading proponents of the life course paradigm (cf. Elder and Johnson 2003: 54). Examples of such transitions are when children leave the parental home or when adolescents with special educational needs obtain paid work.
Studies of trajectories and transitions require a longitudinal design, which is generally a very expensive approach. Presumably for this reason, most longitudinal studies of adolescents with disabilities have small samples with a restricted time span. However, this is generally the case in the study of educational trajectories (see Blossfeld, Schneider, and Doll 2009, for a discussion of German longitudinal studies).
Longitudinal data collected prospectively are generally considered best for studying processes and outcomes. This approach allows the researcher to determine the date of important events and transitions more precisely than is possible through retrospective registration. Using longitudinal data collected prospectively, we may register exactly when various problem conditions appear and what type of support each special needs student receives at that time. For these reasons, a considerable amount of time and money has recently been spent collecting longitudinal data on adolescents with special educational needs in Norway.
A research project titled Adult Life on Special Terms provides the data for this article. This research is funded by the Norwegian Council of Research and is an extension of a longitudinal study that started collecting information on 760 special needs students as early as the spring of 1996. This follow-up sample belongs to a population of more than 2000 special needs students in six counties in different parts of Norway. Two cohorts of special needs students were involved: (1) the admission cohort of upper secondary schools in the autumn of 1994 from three counties, Møre og Romsdal, Nord-Trøndelag, and Hedmark; and (2) the admission cohort of upper secondary schools in the autumn of 1995 from these same three counties, in addition to the counties of Finnmark, Rogaland, and Oslo. Taken together, these six counties are meant to represent Norwegian society in terms of its geographical, demographic, and occupational structure. These longitudinal studies have been approved by The Norwegian Social Science Data Services and The Norwegian Data Inspectorate. Ethical considerations have been taken, e.g. informed consent and protection of informants’ anonymity.
From the spring of 1996 to the winter of 2002, information was obtained about these adolescents once or twice a year. Class teachers and school counsellors filled in questionnaires, whereas the parents and the special needs students themselves were interviewed by telephone.
In the follow-up sample, data on all of the 760 students were obtained from the schools. During the winter of 2002, we succeeded in interviewing or obtaining information on 494 of the former special needs students, yielding a response rate of 65%. Compared with most other follow-up studies of adolescents with disabilities, a response rate above 60% is considered to be a rather good one (cf. Levine and Nourse 1998: 222). However, the respondent attrition makes the sample somewhat biased. Compared to the population, persons who attended regular classes, and persons who took academic courses are under-represented. Those who attended special classes and those who had unspecific courses are over-represented. On the other side, the proportion in the sample who attended vocational courses is almost identical to that registered in the population. The gender ratio is also similar (cf. Myklebust 2007: 218–19).
Crucial domains of adult life
The passage to adulthood is often complicated, especially for those with somatic, psychological and social problems. A recent international review concludes that vulnerable young people experience poor outcomes across the major domains that mark the transition to adult life (cf. Osgood, Foster, and Courtney 2010: 216). Similar conclusions are drawn by Myklebust (2010) in an evaluation of transition outcomes for Norwegian students with special educational needs.
In this article analyses of transitions to four interrelated domains of adult life will be presented. In this case, the discussion of previous research in these fields will be rather brief.
Independent living and romantic relationships
Leaving the parental home to establish an independent life is one of the most important decisions youth make in their transition to adulthood. In the US, the number of young adults living independently outside the parental home five years after leaving school is substantially lower amongst young people with disabilities (37%) than amongst their peers without disabilities (60%) (Blackorby and Wagner 1996: 408).
Until recently, research on romantic relationships among young people has been remarkably sparse (Giordano 2003: 268; Larson et al. 2002: 61–62; Karney et al. 2007: 5). When this is taken into consideration, it is no surprise that there are very few studies concentrating on romantic relationships among adolescents with special educational needs. However, an exception is a study by Grue (2001) demonstrating that 19-year-old women with physical disabilities in Norway have established romantic relationships to a lesser degree than their peers without disabilities. Along the same lines, a British study demonstrates that 50% of 33-year-old men with learning difficulties have neither experienced a romantic relationship nor lived independently from their parents. The corresponding percentage amongst peers without learning difficulties is 20% (Evans et al. 2001: 22).
The transition to parenthood is a pivotal event. This experience may be called a turning point, that is, a point at which the life takes a new course. Sometimes their lives move in a positive direction, but it also happens that some adolescents find themselves facing a more difficult road ahead of them.
A longitudinal study of New Zealand women support the notion that teenage pregnancy is a risk encountered by young women with ‘educational underachievement, conduct problems and attentional problems’ (Woodward, Fergusson, and Horwood, 2001: 1181). This conclusion is supported by a British study that indicates that young women with school problems are particularly at risk of untimely pregnancies (Kiernan 1997: 426). Early childbearing clearly reduces the chances of completing schooling, and given the importance of education in current society, these young women are definitely at a disadvantage (Hofferth, Reid, and Mott 2001). It should also be noted that the National Child Development Study in England documents that early parenthood – before the age of 23 – indicates rather bleak future prospects, and more negative for women than for men (Hobcraft and Kiernan 1999: 26, 35).
It is a moot point whether teenage births are as unfortunate occurrences as they were previously considered to be. One of the sceptics, Frank Furstenberg Jr., claims that early childbearing is not a cause of long term disadvantage among females who otherwise would have had bright prospects. His conclusion merits quotation: ‘For someone tottering on the brink of school dropout, with no real likelihood of going to college, an untimely birth, if not a salvation as some have claimed, is certainly not a disaster either’ (Furstenberg 2003: 34 ). This position is further substantiated in a recent review of early transitions to motherhood (Furstenberg 2009: 228–30)
Over the past decades, in many countries the labour market prospects for youth have declined, especially for those with low education (cf. Danziger and Ratner 2010 for a discussion of the situation in US). The rather poor employment for disabled people has been documented for fairly affluent countries in a review in Handbook of disability studies (Schriner 2001: 645). For those who do work, earnings are generally modest. Vulnerable young people are not included as a separate category in ordinary labour market statistics. However, it is an established fact that people with disabilities are often unemployed. In Norway, 45% of people with disabilities had a job in 2007, compared with 75% of the total population aged 16 to 66 (Fossestøl and Widding 2007: 17). Many studies discuss the circumstances that contribute to unemployment amongst youth with disabilities. However, little research exists of what factors assist vulnerable young people obtain jobs that provide decent pay. An exception is a study by Myklebust and Båtevik (2005). Their analyses reveal the importance of previous successful educational transitions. The pivotal role of education is also demonstrated by Vedeler and Mossige (2010) in a qualitative study of pathways into work for Norwegians with mobility disabilities.
In the subsequent analyses 51 students in post-secondary education are excluded since most of them are expected to be gainfully employed after their studies are finished. It would certainly be incorrect to include these students among those who have failed in their transition to work.
Figure 1 indicates to what extent the former students with special educational needs have experienced transitions into the domains of adult life briefly discussed above.
Figure 1 demonstrates that the greater part of the youth have for shorter or longer time lived outside the parental home–women to a greater degree than men. Most men and women have also experienced romantic relationships. Full time work is more common among men (about one-third) than women (less than a half). A quarter of the women have children. The same is the case for only one ninth of the men.
Figure 1 indicates the situation for these former special needs students when they are in their early twenties. However, the crucial question is this: Which antecedent variables have contributed to these outcomes? In accordance with life course perspectives we have to examine both proximal and distal factors, variables that often are intertwined in complicated ways. This is for example a challenge when analysing competence attainment: ‘Individuals typically are sorted into educational positions on the basis of characteristics that by themselves might determine subsequent educational success or failure’ (Pallas 2003: 168).
The critical test is, however, whether the effect of a particular covariate remains when we control for the effects of other independent variables. When the dependent variable is dichotomous–which is the case here–and there are many independent variables, logistic regression analysis is often an efficient approach. This type of analysis indicates the effect of a specific covariate on the dependent variable while adjusting for the effects of the other covariates. Figure 2 illustrates the assumed relationships between the variables in the model.
The dependent variables–the adult outcomes–in box 4 have been discussed previously. Now should be added some comments on the independent variables, covariates that are placed in boxes with varying time proximity to the dependent variables.
In box 1 there are three independent variables. Gender should always be included, not least when analysing outcomes like work and reproduction. Being a parent affects young females and males differently. The young mother is often responsible for the greater part of the care of the child, whereas the young father is expected to increase his efforts to achieve gainful employment. Included in this box is also a variable mapping whether the adolescents’ parents are divorced or not. This may be regarded as an indicator of the family's financial status, the degree of stability in the home environment, and the availability of adult supervision of schoolwork and support in critical juvenile transitions (cf. Chen and Kaplan 1999).
The variable functional level in box 1 merits special attention, because it pervasively influences the life of people with disabilities. The questionnaire that was circulated in the spring of 1996 contains 13 different indicators that record problems of a somatic, psychological, and social nature. For each indicator, counselors or class teachers classified each student into one of four categories–extending from no difficulties at all to very great difficulties. When these 13 problem indicators are added, an additive index for functional level is created. In the following analyses, this index is divided into quartiles.
The covariates in box 2 may be called school factors. The first variable maps the situation at the start of upper secondary education, indicating what type of class context each student belongs to. During the first school year 43% of the students received specially adapted teaching solely within the framework of regular classes. Almost a quarter of them were placed in classes of a reduced size, whilst 11% received all their teaching in small groups. The rest attended various types of classes during the school year or had other kinds of special arrangements. Placement in special classes is intended to help students with special needs, but it often has negative side effects, since it may have undesired effects, e.g. stigmatisation, a view expressed by Pallas (2003: 169) in this way: ‘Stigma is the extent to which a particular stratified location in the schooling system confers a devalued social identity on a student. Students in special education classes frequently are stigmatized, as are students retained in grade.’
Students with special educational needs often experience grade retention, in the model captured by the variable progress. This covariate reports whether the students are on time when they have spent about the half of their allotted time in upper secondary education. This variable is added in an attempt to overcome what Hagestad and Dannefer (2001: 7) call ‘the time 1 problem,’ that is, the pervasive tendency to consider social-structural characteristics only in the first phase of a longitudinal study.
The variable marks in box 2 is the average of the marks received in all subjects. This covariate is assumed to be a rather crude indicator of capabilities to learn and to succeed in school and society. The three variables in this box–class placement, progress and marks–are primarily assumed to have indirect effects on adult outcomes–especially work–because these variables influence competence attainment, a variable that usually affects adaptation to adult life (Myklebust 2007; Myklebust and Båtevik 2005).
Box 3 contains proximal variables, covariates measured just before the adult outcomes. Competence attainment is believed to be crucial. Adolescents who have successfully completed upper secondary school are assumed to have better chances in adult life than those with no vocational or academic competence. Another important factor is whether the adolescents have a driving licence or not. On the one hand, obtaining such a permit demonstrates some level of capability. On the other hand, the possession of a driving licence is a prerequisite for some types of work. We, therefore, assume that adolescents with such a licence succeed in adult life to a greater degree in comparison to those not allowed to drive a car.
Behind the label health behavior in box 3 are the independent variables smoking and drinking. These habits are often acquired in the adolescence; in this study most of the smokers had started before the age of 18. Smoking and drinking are included because they in many quarters are considered to indicate an adult life style. However, these are also ambiguous indicators of adulthood, since excessive smoking and drinking are evidenced to have adverse health effects, at least in the long run. Finally, found in box 3 is the variable health condition: This is a subjective evaluation of whether the respondents suffer from illnesses or disabilities that prevent them from doing what they want in different domains of life.
Results and discussion
When variables from this model are entered into the logsistic regression analyses, we discover how a particular covariate influences each of the four adult domains when the effects of the other covariates are controlled. Odds ratios above 1 indicate that the likelihood of tranisitons is higher for a given category compared to the reference category. Odds ratios below 1 indicate that the likelihood of transitions is lesser. Table 1 shows the results of the analyses.
|Lived independently N = 494||Had a romantic relationship N = 494||Had children N = 494||Achieved full time work N = 443+|
|Functional level at age 17|
|0. Fourth quartile|
|1. Third quartile||1.00||1.47||3.35**||1.07|
|2. Second quartile||0.63||3.01**||3.04**||1.60|
|3. First quartile||0.68||4.25**||2.51*||1.80|
|0. Not divorced|
|Type of class|
|0. Special class|
|Degree of progress|
|0. Not on schedule|
|1. On schedule||1.29||1.14||0.93||0.88|
|0. Below median level|
|1. Above median level||2.16**||1.00||1.29||0.77|
|0. No formal qualifications|
|1. Formal qualifications||1.94*||1.83*||1.04||2.64***|
|0. No driving licence|
|1. Driving licence||0.50*||2.26**||2.42**||1.27|
|0. Do not smoke|
|0. Do not drink|
|Present health condition|
|0. Health problems|
|1. No health problems||1.32||2.64***||0.70||2.33***|
|Note: +Students in postsecondary education are excluded from the analyses of full time work.|
|*p< 0,05; **p< 0,01; ***p< 0,001. For each independent variable the value 0 is the reference category.|
Gender influences the four adult domains differently. The likelihood of ever having lived outside the parental home is less than a half for men compared to women. The same is the case for romantic relationships. The likelihood of being a parent is less than a third for men compared to women. However, men have more than twice the likelihood than women to have full time work.
Functional level markedly affects the chances of ever having had a romantic relationship. The higher functional level, the higher the likelihood is of having had this experience. Those in the first quartile are, for example, more than four times more likely than those in the fourth quartile to ever had a romantic relationship. The likelihood of having children is more than threefold for those in the third and second quartile of functional level compared to those in the fourth level. For those in the first quartile the likelihood is two and a half greater. Functional level does not significantly affect independent living and full time work.
A parental divorce affects the children's adaptation to adult life in different ways. Young people with a parental split-up have twice the chance of having had a romantic relationship compared to those without such experiences. On the other hand, a rupture between the parents may indicate later occupational maladjustment for their children. Whether the parents are divorced or not have only insignificant effects on reproduction and independent living among their descendents.
The three variables called school factors–type of class, degree of progress and marks–do not significantly affect later adult adaption. There is, however, one exception: Those with marks above the median, are twice as likely to ever have lived independently compared to those with results below the median level.
Although the direct effect of the school factors seem to be negligible, their indirect effect through competence attainment may be important. Academic or vocational qualifications affect favourably three of the adult domains–independent living, romantic relationships and full time employment. It does not influence the likelihood of having children.
Possessing a driving licence positively influences romantic relationships and reproduction, but it negatively affects independent living. The positive effect on full time work is insignificant. Those who smoke are far more likely than those not using tobacco to have children, to have lived independently and to have experienced a romantic relationship. The effect on this last domain is enormous. Those smoking have nearly an eightfold greater chance of ever having such a relationship.
Alcohol consumption influences only one of the adult domains, romantic relationships. Compared to those who not use alcohol, young people who drink are almost three times more likely to be have been romantically involved. And finally, the present subjective health influences adaptation to adult life in this way: those with no health problems are more than two-and-a-half times more likely to have romantic involvement, and have more than twice the chance of having full time work compared to those with health problems.
Multivariate analyses have frequently demonstrated that covariates do not have the same effects within all subcategories. Therefore separate analyses for men and women have been carried for each of the four domains of adult life. Space limitations only permit presentation of data from the two domains where the differential effects are most pronounced–romantic relationships and reproduction.
Table 2 demonstrates the effect of each covariate when adjusting for the effects of the other independent variables in the model. The comments below will be restricted to effects at least significant at 0.05 level, a level of statistical significance often difficult to obtain in subcategory analysis with small numbers.
|Had a romantic relationship||Had children|
|Women N = 190||Men N= 304||Women N= 190||Men N=304|
|Functional level at age 17|
|0. Fourth quartile|
|1. Third quartile||1.38||1.65||9.49***||1.18|
|2. Second quartile||2.58||3.55**||7.06**||1.30|
|3. First quartile||12.17*||3.43*||9.11***||0.65|
|0. Not divorced|
|Type of class|
|0. Special class|
|1. Regular class||0.63||0.91||0.71||0.94|
|Degree of progress|
|0. Not on schedule|
|1. On schedule||1,04||1.10||1.16||0.76|
|0. Below median level|
|1. Above median level||1.44||0.75||1.88||1.10|
|0. No formal qualifications|
|1. Formal qualifications||2.34||1.53||0.63||1.75|
|0. No driving licence|
|1. Driving licence||1.73||2.66**||1.69||5.51*|
|0. Do not smoke|
|0. Do not drink|
|Present health condition|
|0. Health problems|
|1. No health problems||2.06||2.6**||0.91||0.50|
|Note: *p< 0,05; **p< 0,01; ***p< 0,001. For each independent variable the value 0 is the reference category.|
Briefly, the effects of the covariates on romantic relationships are:
- • Functional level positively affects the likelihood for ever having had romantic relations for men as well as women. The effect is especially dramatic for women in the first quartile.
- • Possessing a driving licence is advantageous in the romantic sphere, especially for men.
- • The effect of smoking is extreme. Men and women who smoke have seven to ten times more likelihood of ever having experienced romantic relationships than those not smoking.
- • Alcohol consumption also increases the likelihood of romantic relations, for women as well as men.
- • Those not experiencing health problems have higher chances of ever having had romantic relationships, men to a greater degree than women.
The covariates influence the likelihood of having children in this way:
- • The effect of functional level is strong for women. Females in the three highest quartiles have 7 to 9 times the chances of having children than those in the lowest quartile. The effect is negligible for men.
- • Those women who have experienced a parental divorce have a threefold likelihood of being mothers compared to those without divorced parents. Among men there is no such effect.
- • Men with a driving licence have a five times greater chance of having children than those without. This variable has no significant effect on women.
- • Both men and women who are smoking have children to a much greater extent than those not using tobacco. One may wonder whether early smoking and early reproduction are both indicators of an adult life style.
The other covariates in the model have only insignificant effects on romantic relationships and reproduction. The exception is, however, women with divorced parents who have three times the likelihood of ever having had romantic relationships compared to fellow women who never have experienced a parental disruption.
To what degree do vulnerable young people experience transitions to adult life? This question has been answered by presentation of longitudinal data from four spheres of life: finding full time work, acquiring an independent living, establishing romantic relationships, and having children. Longitudinal designs are considered to provide strong data for analysis of transitional processes. However, in this case some limitations should also be noted. The assessments of functional disabilities are made by many different experts who do not necessarily use the same criteria. This may be a threat to the validity of the variable functional level, one of the most central dimensions in the analyses. Another limitation is that the sample is somewhat biased due to respondent attrition. However, as noted in the methods section, analyses of the attrition inform us in what way the sample is biased.
First should be noted that at an age of 23 to 24 years, most of these former special needs students seem to be established in the adult world. They do find full time work, they do live independently from their parents, and they do establish romantic relationships. Approximately a quarter of the women and a tenth of the men also have children. This gender difference is close to what was observed in Norway 30 to 40 years ago. The female fertilty is similar to what was found among women with low education level in Norway in the 1990s (cf. Skrede 2002: 268–69).
In this way these young people deviate from the reproductive pattern in Norway. Today the mean age for first birth is about 28 for women and about 31 for men (Statistics Norway 2011a). For many early childbearing is a somewhat ambiguous indicator of adult success, since a third of them do not live with the other parent. Ambiguity may also apply to romantic relationships. Some relationships may be detrimental, whereas other may be favourable. According to Crosnoe and Johnson (2011: 445) scholars increasingly have realised that romance among young people may be ‘developmental positive or negative depending on the characteristics of the partners, the quality of the relationship, and the context in which it occurs.’ However, in this quantitative longitudinal study it has not been possible to collect information of this type.
However, a considerable number of young people −35% of the men and 55% of the women–have no full time job. Furthermore: One-fifth of these young people have never lived outside the parental home, and almost an equal number of them have never had any romantic relationship. In sum, this depicts a darker picture of transition to adulthood than the transition for adolescents without special educational needs.
As indicated by the logistic regression analyses, these transitions are influenced by distal factors, like gender and functional level, and proximal factor factors such as health behaviour and present health condition. On the other hand, the direct effects of school factors seem to be rather modest. However, these covariates–type of class, progress and marks–are important because they decisively influence competence attainment (cf. Myklebust 2007: 226). In turn, competence clearly contributes to successful adult adaptation, especially in the labour market, but also in domains such as independent living and romantic involvement. Possessing a driving licence also seems to be a resource that is connected to the transition into adulthood, especially among men. Being in good health is also, as expected, favourable, especially in the romantic sphere and in the labour market.
Finally, the central role that smoking and drinking seem to play in the transition to adulthood should be emphasised. Those using alcohol establish themselves in various domains of adult life to a greater extent than those not drinking. And this tendency is even more distinct concerning smoking. Tobacco use among these young people evidently has a strong impact on transitions to many domains of adult life (work is, however, an exception). The extreme effect of smoking is interesting and deserves to be analysed more closely later. Here suffice it to say that smoking seems to be a very important marker of adult status among former students with special educational needs. Attention should also be drawn to an article with the somewhat provocative title, Smoking: A special need? (Fidler et al. 1992), where adolescents with emotional and behavioural disorders are identified as heavy smokers.
Problematic transitions are fairly common phenomena that affect far more people than students with specially organised teaching programmes. This is apparent from the fact established in the introduction that a third of the complete admissions cohorts in Norway do not complete upper secondary school after five years.
Transitions are demanding, but they do offer new opportunities. In a context where a social passage is expected, no transition at all will be considered a failure. However, transitions are in themselves risky projects, and these changes can, in particular, be critical for young people with limited school resources. This group is often regarded as vulnerable (cf. Coles 1997: 78–80). Students with special educational needs are, therefore, in a danger zone, not only because they have disabilities of various kinds, but also because their transitions often come at times outside the normal schedule, a delay that is often associated with these same disabilities. The situation can become serious for those who experience protracted and disjointed transitions (cf. Coles 1995: 10), which is typical for many special needs students. They are among the disadvantaged groups that are especially vulnerable in the transition to adulthood.
The crucial question is whether or not the transitions made by this specific sample may be generalised to other vulnerable people, e.g. members of younger cohorts. Do new cohorts repeat the adaptations of their predecessors, or do they adapt in new ways? Such questions are central in cohort analysis, in which the main approach is to compare cohorts that encounter different, as well as similar, conditions in their trajectories in various domains of life (see, for example, Ryder 1965; Riley 1988). Economic conditions are, of course, crucial for young people seeking work. In fact, when these cohorts of vulnerable youth were in their early twenties, the unemployment in Norway was low, about 3%. And in spite of the worldwide financial crisis in recent years, the unemployment in the autumn of 2011 was only 3.2% (Statistics Norway, 2011b). When these facts are taken into consideration, Norwegian youth appear to be rather privileged. However, in addition to economic conditions, demographic circumstances may also impinge on the life course, an idea emphasised by Ryder (1965, 845): ‘Any extraordinary size deviation is likely to leave an imprint on the cohort as well as on the society.’ This theory of cohort size and life course outcomes has been further developed by Easterlin (1987). The fact that later entrants to upper secondary education in Norway belong to larger birth cohorts than those analysed in this article, indicates that vulnerable individuals in younger cohorts may face greater difficulties in their early twenties when they try to find their feet in society.