Transition matters: pupils’ experiences of the primary–secondary school transition in the West of Scotland and consequences for well‐being and attainment

Abstract

The importance of school transitions for pupil adjustment, particularly their impact on later well‐being and attainment, remains contested. This paper draws on data from a longitudinal, school‐based study of over 2000 Scottish pupils, first surveyed in 135 primary schools (age 11) in 1994, and followed up in 43 secondary schools (age 13 and 15) and again after leaving school (age 18/19) in 2002/3. The length of follow‐up makes this study unique in transition research. After a year in secondary school (age 13), the majority recalled having had difficulties of adjustment to both school and peer social systems at the beginning of secondary education. While the primary (but not secondary) school played a small part in accounting for different transition experiences, controlling for a wide range of sociodemographic and other factors, personal characteristics were much more important. Respondents of lower ability and lower self‐esteem experienced poorer school transitions; those who were anxious, less prepared for secondary school and had experienced victimisation, poorer peer transitions. Further analysis of the impact of school and peer transitions on well‐being and attainment revealed that each had specific independent effects both within and beyond secondary education. At age 15, a poorer school transition predicted higher levels of depression and lower attainment; a poorer peer transition, lower self‐esteem, more depression and lower levels of anti‐social behaviour. Although reduced in size, similar results extended to outcomes at age 18/19. These effects bear comparison with those associated with gender and school disengagement, clearly demonstrating the importance of successful transition for later well‐being and attainment.

Keywords:

• primary–secondary transition
• predictors of transitions
• school effects
• pupil well‐being
• educational attainment
• longitudinal study

Introduction

The transition from primary to secondary school in the UK, and its equivalent elsewhere, has been depicted both as ‘one of the most difficult in pupils’ educational careers’ (Zeedyk et al. 2003, 67) and, more broadly, as a ‘key rite of passage’ (Pratt and George 2005, 16) in young people’s lives. The former reflects the view that the move from the smaller, more personal environment of the primary school to the larger, impersonal world of the secondary school at a minimum involves some problems of adjustment; the latter that, in addition to new educational demands, exposure to many more (older) pupils, and associated peer groups and pressures, poses a more fundamental challenge to identity (Measor and Fleetham 2005; Warin and Muldoon 2008). The key question is how much the transition really does matter for pupils, both in respect of the formal (school) and informal (peer) social systems (Measor and Woods 1984), and specifically the impact of their transition experiences on future learning and well‐being.

Surprisingly perhaps, and despite the importance attributed to a ‘smooth transition’ in UK educational policy (Morris and Pullen 2006; HMIe 2006; Scottish Executive 2006), the question remains controversial, in large part because the evidence base is inconsistent and incomplete. Although there are a number of excellent reviews (Anderson et al. 2000; Galton, Morrison, and Pell 2000; McGee et al. 2003; Boyd et al. 2007), a plethora of local educational authority reports (e.g. Highland Council 2000) and evaluations of ‘transition projects’ (e.g. Shepherd and Roker 2005), with some notable exceptions (Galton, Gray, and Ruddock 1999, 2003; Galton and Morrison 2000) the primary–secondary transition has not featured strongly on the UK academic research agenda (Zeedyk et al. 2003; Boyd 2005). The research has also lacked continuity, the focus over the last two decades switching back and forth between pupils’ experience of transitions (e.g. Measor and Woods 1984; Chedzoy and Burden 2005), the widely documented post‐transfer ‘dip’ in educational attainment (e.g. Nisbet and Entwistle 1969; Galton, Gray, and Ruddock 1999) and curricula and pedagogic continuity (e.g. Galton, Gray, and Ruddock 2003; Morris and Pullen 2006). Much is assumed about the connections between these issues, but there is little direct investigation of the links, particularly between transition experience and educational outcomes, and even less on well‐being.

Notwithstanding this lack of continuity, three main strands of research are identifiable, focusing respectively on pupils’ experience of transition, predictors of poorer transitions, and consequences for educational attainment and pupil well‐being.

Previous research

Pupils’ experience of transition

A major focus of (mainly UK) research into the primary–secondary transition has involved investigation of pupils’ pre‐transfer expectations and post‐transfer experiences. There is near universal agreement that the majority of pupils express some concerns and anxieties prior to transfer about a range of issues associated both with the formal school system (e.g. the size of the school) and informal system of peer relations (e.g. making friends) (Measor and Woods 1984; Fouracre 1993; Anderson et al. 2000; Ward 2000; Graham and Hill 2003; Chedzoy and Burden 2005; Shepherd and Roker 2005). The evidence in relation to post‐transfer experiences is less consistent (Lohaus et al. 2004), though a general consensus exists that the worries expressed by primary pupils dissipate over the first year, or even first term, of secondary school (Galton and Morrison 2000; McGee et al. 2003; Boyd 2005). For example, Graham and Hill (2003) found that, after only a month in secondary school, two‐thirds of their Glasgow sample reported no anxieties, and most said they stopped being concerned almost immediately. By contrast, Zeedyk et al. (2003), in another Scottish study, found that though a majority of pupils reported secondary school to be better than expected, worries about peer relations (especially bullying) and the school system (e.g. getting lost) were more, not less, prevalent than in a comparison group of primary school pupils.

This more negative picture of the transition experience is consistent with several qualitative studies which stress the ‘critical’ nature of change for young people’s identity and well‐being (Measor and Woods 1984; Lahelma and Gordon 1997; Pointon 2000; Jackson and Warin 2000; Tobbell 2003; Pratt and George 2005; Muldoon 2005). These researchers draw attention to aspects of the secondary school system (e.g. lack of personal space) and highlight the importance of the informal school system as it shapes peer relations and friendships. This is consistent with Measor and Wood’s (1984) earlier ethnographic study which found pupils’ concerns fluctuated over the course of the first year in secondary school, some of the most personal ones (particularly about peer relations) only being expressed at the end of that period. However, this contrasts with the view that for the great majority of pupils, transition anxiety is of relatively short duration, and its significance for educational and other outcomes has been over‐exaggerated at the expense of other educational issues (Galton, Gray, and Ruddock 1999, 2003; Galton, Morrison, and Pell, 2000).

Vulnerable groups: characteristics of pupils making poorer transitions

Even on the most favourable assessment of the transition experience, it is widely acknowledged that some pupils are more vulnerable than others (Anderson et al. 2000). This second strand of (mainly US) research has identified a wide range of pupil characteristics which distinguish between those making better and worse transitions.

Among sociodemographic factors, younger age has consistently been identified as a predictor of poorer transition (Nisbet and Entwistle 1969; Galton, Morrison, and Pell 2000). In respect of gender, there is less consistency. US studies suggest girls are more vulnerable, especially with regard to peer relations (Blyth, Simmons, and Carlton‐Ford 1983; Anderson et al. 2000) while evidence from New Zealand suggests the reverse (McGee et al. 2003). Several studies have also examined the impact of family socioeconomic status (SES), the consensus from international research being that lower SES youth experience more transition problems (Anderson et al. 2000: McGee et al. 2003). In the US, the importance of family SES is attested to by a number of studies focusing on transitions experienced by poor black youth (Gillock and Reyes 1996; Reyes et al. 2000; Guttman and Midgley 2000). Surprisingly, in the UK, less attention has been directed to the impact of SES, though its importance is generally assumed (Galton and Morrison 2000). Ethnicity has similarly received little attention, though Graham and Hill (2003) found pupils of South Asian (Pakistani Muslim) origin had more pre‐ and post‐transfer concerns and adjustment problems than their white Scottish counterparts.

The importance of the family is also often acknowledged (Anderson et al. 2000; McGee et al. 2003). However, in respect of family structure, apart from reference to pupils from one‐parent families having more problems adjusting to secondary school (Galton and Morrison 2000), there is little direct evidence. With regard to parenting styles, there is some evidence from US studies that pupils whose parents are both more caring and encouraging of autonomy exhibit better adjustment to junior high (Lord, Eccles, and McCarthy 1994) and high school (Newman et al. 2000). In the latter study, family members (especially mothers) were most often cited as sources of support for academic and peer related challenges. Having older siblings in the same school (Measor and Woods 1984; Ward 2000) and peer support (Hirsch and Dubois 1992) has similarly been found to be associated with better adjustment to secondary school.

Among personal factors, a consistent finding is that pupils of lower ability experience more transitional stress and anxiety than their more able peers (Anderson et al.2000; Chedzoy and Burden 2005), making them a particular target of interventions to promote smoother transitions (e.g. Bryan, Treanor, and Hill 2007). Closely related is low self‐image or self‐esteem which has been linked to low preparedness for transfer and subsequent poorer transitions (McGee et al. 2003). One UK intervention (Shepherd and Roker 2005), focusing on shy and withdrawn children, reported reduced fear about transition following programmes to boost self‐esteem. Such children had often experienced bullying in primary school, itself associated with lower self‐esteem (Hawker and Boulton 2000) and likely to heighten anxiety about future victimisation. Another factor, given high prominence by Anderson et al. (2000) in a US context is problem behaviour, disruptive or aggressive pupils having greater problems adjusting to junior high school (Berndt and Mekos 1995). In the UK, despite the current concern about anti‐social behaviour among young people (Collishaw et al. 2007), and its implications for disengagement from education, it has not featured much in relation to the primary–secondary transition.

Consequences for educational attainment and well‐being

The potential importance of the primary–secondary transition is greatly increased because of the impact it may have on educational attainment. International data are consistent in revealing a ‘dip’ in attainment following transfer to secondary school, the significance of which is increased because it occurs at different ages in different educational systems, thereby making other explanations (e.g. pubertal onset) less likely (McGee et al. 2003). However, while this suggests the importance of transitional experiences, and even who might be most affected, it does not demonstrate a causal link. In a series of studies, Galton et al. (Galton and Willcocks 1983; Galton, Gray, and Ruddock 1999, 2003; Galton, Morrison, and Pell 2000; Galton and Morrison 2000) found the attainment ‘dip’ was less associated with post‐transfer anxiety than with a decrease in enjoyment of school and pupil motivation. These authors have argued that too much attention is given to the transition process itself as a factor in under‐achievement and too little to (dis)continuities in education.

In the US, the fall in attainment following transition is also widely referred to (Eccles et al. 1993; Gillock and Reyes 1996; Reyes et al. 2000; Anderson et al. 2000), but in addition, research has focused on other dimensions of pupil functioning, notably self‐esteem, and less often psychological symptoms. The evidence in respect of self‐esteem is not consistent, several studies reporting a ‘dip’ following transfer to junior high school in both boys and girls (Simmons et al. 1979; Wigfield et al. 1991; Anderson et al. 2000), others that the effect is confined to girls (Blyth, Simmons, and Carlton‐Ford 1983), and still others finding no change (Hirsch and Rapkin 1987; Crockett et al. 1989). In one, focusing on self‐esteem after transfer to junior‐high school (Lord, Eccles, and McCarthy 1994), the main predictors (after prior self‐esteem) were positive self‐concepts (e.g. appearance and peer relations), family environment (e.g. support for autonomy) and school adjustment (liking school). The potential importance of school life has also been noted in other studies of self‐esteem (Wigfield et al. 1991) and in the few that examine other measures of well‐being. Hirsch and Rapkin (1987) found an increase in psychological symptoms (particularly among girls) accompanied by a ‘dramatic’ decline in the quality of school life. Similar changes in psychological symptoms were found in a later study which also demonstrated a protective effect of peer support specific to the period of transition (Hirsch and DuBois 1992).

Summary and aims

The cumulative evidence from the international literature suggests some strongly consistent patterns but also inconsistencies. There is considerable agreement that the majority of pupils experience pre‐transfer anxieties, but much less about both their intensity and duration post‐transition. There is also general agreement about the characteristics of pupils who have more negative experiences. Vulnerable pupils tend to be younger, less able and more disruptive, to have lower self‐esteem, to have parents who are less encouraging of autonomy, and to come from lower SES backgrounds. Finally, there is universal agreement that pupils experience a post‐transfer ‘dip’ in educational attainment, though what part the transition itself plays remains uncertain. By contrast, the evidence relating to well‐being, particularly self‐esteem, is much less consistent, and here too the role of transition is unclear. Simply put, despite some major advances in understanding, we still do not know whether, or how much, transition matters, particularly over the longer term.

In evaluating research into the primary–secondary transition, there are three main problems which contribute to the lack of a solid evidence base. First, the research is characterised by a wide variety of perspectives and methodologies, which may partly account for some of the inconsistencies in findings. In general, qualitative studies have found the transition to be a more profound experience in its consequences for identity and well‐being than studies using quantitative (questionnaire) methods (Measor and Woods 1984; Tobbell 2003; Pratt and George 2005). However, many of these studies are small and unrepresentative, and there are also differences in the length of follow‐up in both qualitative and quantitative investigations. A key recommendation in the literature is for longitudinal studies of greater duration (Galton and Morrison 2000).

Second, in contrast to the extensive focus on individual predictor variables (particularly in US studies), there is a curious lack of attention to differences in respect of the very situation the transition takes place in, schools themselves (Eccles et al. 1993; Lord, Eccles, and McCarthy 1994). In large measure this is because most studies focus on transition to a single secondary school, the only direct evidence coming from a few studies showing variation in high school adjustment by elementary school (Gillock and Reyes 1996). Without sufficient numbers of schools, investigation of ‘school effects’ (Fitz‐Gibbon 1996; Smyth 1999) on transition is virtually precluded, and this in turn inhibits the extent to which the impact of formal school policies (e.g. preparedness for secondary school) and/or characteristics of the peer group (e.g. the extent to which transfer occurs in friendship groups) can be examined. There is some evidence that in an era of parental choice, transition has become more difficult for (placing request) pupils who do not follow the traditional (feeder primary/associated secondary) transfer route, and may do so without the support of friends (Galton, Gray, and Ruddock 1999; Boyd 2005).

Third, and perhaps most important, research on the primary–secondary transition has largely failed to focus on the impact of the transition itself (Tobbell 2003). This is best illustrated by the different foci of UK and US research. The former has been especially strong in describing the experience of transition on the one hand, and changes in outcomes (principally attainment) on the other, but it has largely failed to link the two, relying instead on inferences about linkage from two different sets of evidence. US studies, with a very few exceptions (e.g. Mitman and Packer 1982), have not focused on the transition itself, but have instead placed considerable emphasis on a wide range of individual characteristics (e.g. goals, mastery, self‐esteem), sometimes with additional family and peer group factors, which predict adjustment to junior high school, usually measured by educational grades and self‐esteem. This lack of attention to the experience of transition has been acknowledged as a gap in US research for some time (Eccles et al. 1993; Lord, Eccles, and McCarthy 1994). To bring these two research traditions together, longitudinal studies are required in order to investigate the specific impact of transitional experiences (school and peer) on educational and other outcomes, taking account of sociodemographic, family and personal characteristics together with those of the school.

Against this background, this paper draws on data from a longitudinal study conducted in the 1990s in the West of Scotland, where virtually all pupils transferred, and continue to transfer, to secondary school at age 11–12 after seven years of primary education. Although the study was not specifically designed to investigate the primary–secondary transition, and notably has no data on pre‐transfer concerns, data are available on post‐transfer concerns (reported one year after transfer to secondary) which distinguish between formal (school) and informal (peer) ‘transition experiences’. Data are also available on a wide range of individual characteristics (pre‐ and post‐transition), and on contemporaneous and subsequent well‐being (principally self‐esteem, depression and anti‐social behaviour) and educational attainment. The key strength of the study resides in the length of follow‐up, enabling investigation of the impact of transition experiences on pupil outcomes throughout secondary school (age 13 and 15) and beyond school‐leaving (age 18/19). The study design also allows the issue to be investigated within a (primary and secondary) school context, thus enabling an assessment of the extent to which transition experiences vary by schools. There are six main aims:

1. To describe the extent and nature of school and peer concerns reported by pupils following transfer to secondary schools;

2. To examine how such concerns vary by primary and/or secondary school, and what school‐based factors might account for differences in pupils’ experiences of transition;

3. To identify the social and personal characteristics of pupils who report better and worse transitions;

4. Taking account of such characteristics, to investigate whether transition experiences independently impact on well‐being and academic attainment;

5. To establish whether a ‘transition’ effect is transient (limited to contemporaneous well‐being) or of longer duration (both within and beyond secondary school); and

6. To estimate the relative importance of transition experience for well‐being and attainment relative to other well known predictors; namely, school disengagement and gender.

Methods

Data are derived from the West of Scotland 11 to 16 Study (West and Sweeting 1996). ‘11 to 16’ was a longitudinal study of a cohort of young people in mainstream education who were resident in and around Glasgow in the mid to late 1990s. The cohort was first surveyed aged 11 in their final primary school year in 1994 and followed up in secondary schools on two occasions, aged 13 in the first term of their second year in 1996, and again aged 15 in their fourth year in 1999. Rebranded as the 16+ Study, the cohort was followed up beyond school‐leaving, aged 18/19 in 2002/3 (Sweeting et al. 2005). Data relating to the primary–secondary transition therefore refer to the academic year 1995/6.

The sample was designed to be representative at both primary and secondary levels and, reflecting a major aim of the study to investigate ‘school effects’ on health and health behaviours (West, Sweeting, and Leyland 2004), was clustered within primary and secondary schools. Full details of the sampling strategy are available elsewhere (Ecob et al. 1996), but briefly it involved a reverse sampling procedure comprising: (a) the random selection of 43 secondary schools stratified by educational district (four divisions), religious denomination (Catholic/non‐denominational) and deprivation (proportion receiving clothing grant) together with a separate stratum of independent schools; (b) the random selection of 135 primary schools either associated (and similarly stratified) with the selected secondaries or making large numbers of placing requests to them; and (c) the random selection of classes of 11‐year‐old pupils in each of these primary schools. The inclusion of pupils from placing schools necessitated surveying many more primary pupils than actually transferred to the selected secondaries, the percentage of placing requests being 11% of that total.

Of the eligible sample of 2793, 2586 (93%) participated in the baseline (age 11) survey, 86% of parents also completing a questionnaire. In addition, class teachers filled out a short form on almost all children (n = 2581) which involved ratings of behaviour, temperament and ability. By age 13, sample losses reduced the number of participants to 2371 (85%), and by 15 to 2196 (79%), much bigger losses in the post‐school period reducing the sample size at 18/19 to 1258 (45%). Although the baseline sample was representative of the sex and social class composition of 11 year olds in the study area (Sweeting, Der, and West 2001), thereafter differential attrition made it less so, losses being greater among lower class groups, school truants, and those rated by their primary class‐teachers as being of lower ability and having more emotional and behavioural problems. To compensate for these biases, a weighting scheme was derived, the effect of which is to elevate prevalence estimates of certain behavioural outcomes (West et al. 2003: Sweeting, Der, and West 2001). However, it has a negligible effect on any of the analyses presented here, including those at 18/19, and in consequence unweighted data are used.

In addition to data derived from children, parents and class‐teachers, some data are available on schools themselves. In respect of primary schools, 128/135 head‐teachers completed a questionnaire covering a range of issues including the physical characteristics of the school, pupil intake, overall ethos and selected policies and health education provision. Most unfortunately, no questions were asked about policies and practices directly relating to the primary–secondary transition. In respect of secondary schools, other than that obtained from pupils, the only information available is from official sources, together with independent ratings of ethos made by the research team at the time of fieldwork (West, Sweeting, and Leyland 2004).

Procedures and measures

‘11 to 16/16+’ was approved by Glasgow University’s Ethics Committee. In the school‐based phase, data were collected via self‐complete questionnaires administered in exam‐type conditions, supplemented by mini‐interviews conducted by survey team members (one to five pupils in primary schools, one to eight in secondaries). Participants were informed that all information provided was confidential, the only people seeing their answers being the survey team who checked questionnaires on completion. In the post‐school follow‐up, data were collected via computer‐assisted personal interview together with self‐complete questionnaires in a variety of settings of participants’ own choosing including a central location, previous school and their own home.

The measures of relevance to this paper refer to transition experiences, predictors of transitions, and well‐being and educational outcomes, the study waves at which they were collected being indicated thus: primary (age 11), secondary (13 and 15) and post‐secondary (18/19). Transition experiences (age 13) are conceptualised both as dependent variables, and in relation to outcomes as one of several independent variables. This is represented schematically in Figure 1 which identifies the specific variables involved, first as predictors of transitions (age 11 and 13), and then along with transition experiences as predictors of outcomes within (age 13 and 15) and beyond (18/19) secondary school. In the former, the independent (predictor) variables primarily refer to data collected prior to transition (age 11) and broadly divide into individual (sociodemographic, family and personal attributes, together with pre‐secondary experiences), and contextual influences (primary school and peer group characteristics). In addition, some comparable (post‐transition) data mainly relating to secondary school (age 13) are available. In respect of outcomes, any effect of transition experiences involves both a cross‐sectional (age 13) and longitudinal (15 and 18/19) focus. Details of all the measures are outlined below.

Figure 1 Diagrammatic representation of predictors of (a) transition experiences, and (b) well‐being and attainment according to time (age) of data collection.

Transition experiences (school and peer concerns [age 13])

As indicated, no data on pre‐transfer anxieties or concerns are available. Information on respondents’ post‐transfer experience was elicited in the first term of secondary year 2, and as such their reports are retrospective. This is later than most studies and raises questions both about the definition of ‘transition’ (when it begins and ends) and the extent to which current experiences, and associated feelings (affectivity), shape reports of earlier experiences. However, as suggested by some investigators (Measor and Woods 1984; Gillock and Reyes 1996; Tobbell 2003) accounts obtained later in the transition process may be more indicative of its impact on identity than earlier post‐transfer concerns.

Respondents were asked: ‘Now think back to when you started secondary school. How did you cope with the first few weeks in S1?’, the response categories being ‘very easy’, ‘quite easy’, ‘quite hard’ and ‘very hard’. Reflecting a range of well‐documented school and peer‐related concerns, the items were: the size of the school, the daily routine and timetable, the amount of work (school and homework), having lots of different teachers, meeting and getting to know kids from different areas, bullying or teasing, making new friends, lots of older teenagers, and preparing for each day and travelling to school. Factor analysis (varimax) supported two clear dimensions labelled ‘school concerns’ (e.g. size of school, different teachers, work volume) and ‘peer concerns’ (e.g. different kids, older teenagers, bullying) with Eigenvalues of 1.30 and 3.20 respectively. They reflect the two main dimensions (school and peer) reported in the literature and are treated as separate variables in all analyses, higher factor scores representing poorer transitions.

Predictors of transitions (11 and 13)

(1) Individual characteristics

Age: Measured in months and rescaled as a difference of one year to distinguish between older and younger pupils in the same school year group.

Parental social class: Based on occupational data derived mainly from parents or, in the absence of a parental questionnaire, from reports provided by their 11‐year‐old children, which we have found to be reliable (West, Sweeting, and Speed 2001). All occupations were coded to the Registrar General’s classification (OPCS 1990) from which a measure of the head of household is derived, defined as father’s current or previous occupation if not currently employed or, in the absence of a father figure, the mother’s current or previous occupation. The full six‐fold classification is used, together with a small category of missing data.

Religion: Included because of its continuing importance in the West of Scotland, notably for education (Abbotts et al. 2001), religious affiliation is derived from the parental questionnaire and distinguishes between Protestant, Catholic, other (mainly Islam), none, and a dummy for missing data.

Family structure: From information provided by respondents, this variable distinguishes between those in families with both birth parents, step families and lone parent families.

Parenting: Respondents completed the eight‐item Parental Bonding Instrument (PBI) (Klimidis, Minas, and Alta 1992) which includes two dimensions referring to parental care (e.g. parental love, help, understanding) and (over)control (e.g. treated like a baby). Higher scores represent greater care and greater control (or less autonomy) respectively.

Temperament and ability: ‘The Child at School’ (Kysel et al. 1983) form completed by primary class‐teachers contains a series of nine opposite attributes (e.g. ‘happy and contented’ vs ‘unhappy, anxious or worried’) to which raters indicate how much (‘mostly true’, ‘usually true’, ‘both equally true’, ‘usually true’, ‘mostly true’) it applies to the child. Factor analysis (varimax) of these items produced two dimensions labelled ‘aggression’ (Eigenvalue 4.80) and ‘anxiety’ (Eigenvalue 1.18). Each is entered as an independent variable in all analyses, higher scores indicating greater anxiety and aggression respectively. In addition, teachers rated children’s ability in English and Reading, and Maths on a five‐point scale (‘well above’, slightly above’, ‘about average’, ‘slightly below’ and ‘well below’). As the measures are highly correlated (r = 0.86), only one (maths) is retained as a proxy measure of ability (high scores indicating lower ability).

Self‐esteem: Respondents completed a modified version of the 10‐item Rosenberg (1965) self‐esteem scale, higher scores indicating higher self‐esteem.

Victimisation: Respondents were asked how often (every day, most days, weekly, less often or never) they were (a) ‘bullied’ and (b) ‘teased or called names’. From these two items, an index was constructed to reflect victimisation frequency, higher scores indicating more frequent experience.

Disengagement from school: Assessed via three questions (e.g. ‘I think school is a waste of time’), another index was created from agree/disagree responses, higher scores indicating greater disengagement.

Preparedness for secondary (age 13): Respondents were asked: ‘How well did your primary school prepare you for secondary school? – very well (39%), quite well (52%) or not well (9%)’. This is the only individual predictor collected post‐transition, higher scores representing greater un preparedness.

Pre‐secondary experiences: Comprising three variables which might impact on transitions; whether or not respondents had older siblings (likely to have preceded them into the same secondary school); number of friends already attending secondary school (none, a few, half, most or all); and number of primary schools attended. Because data on older siblings and number of primary schools are derived from parents, cases without a parental questionnaire are dummy coded 0 on each variable.

(2) Primary school and peer characteristics

School type and roll: Whether or not respondents were in associated or placing primary schools and primary school roll (quintiles).

School policies: In the absence of data on transition policies, two indices have been constructed from data derived from primary head‐teachers. The first refers to whether or not the school had a written/unwritten policy on four issues (truancy, bullying, smoking and health‐education), from which a policy index was developed, each written policy (coded 2) being weighted higher than an unwritten (coded 1) policy, with no policy coded 0. The second refers to the extent to which parental involvement in the school was encouraged, and is based on eight items (e.g. PTA, open access to head‐teacher, parents helping in class), from which another index was computed. Both indices are normally distributed (higher scores indicating more policies and greater parental involvement). The assumption behind these two indices is that schools with more progressive policies generally are also likely to be more progressive in preparing pupils for transition.

Peer characteristics: Seven variables were constructed to reflect primary school peer influences, which are defined as the mean scores of primary classes for ability, anxiety, aggression, self‐esteem, victimisation, disengagement and preparedness for secondary school. These variables allow us to distinguish contextual effects from individual level effects (e.g. ability, anxiety etc) described above and are scored in the same direction.

(3) Secondary school factors

These also divide into individual and school characteristics, and refer to data collected at age 13. The former is represented by disengagement from secondary school (similar to the age 11 measure but based on seven items), higher scores indicating greater disengagement. As indicated, very limited information exists on the secondary school context, the only relevant variables being school size (quintiles) and two ethos indicators based respectively on researchers’ ratings of schools at the time of fieldwork, and school means of pupil ratings of school facilities, teacher–pupil relationships and pupil involvement (higher scores, poorer ethos). These variables are fully described elsewhere (West, Sweeting, and Leyland 2004).

Outcome variables ‐ well‐being and attainment (ages 13, 15 and 18/19)

As far as possible, the outcome variables examined are consistent across study waves, so enabling control for identical pre‐transfer (age 11) measures of well‐being. However, while this applies to each outcome across the school‐based phase of the study, and self‐esteem across all waves, comparable measures of depression and anti‐social behaviour were not obtained at age 18/19. Accordingly, depression is replaced by a measure of ‘psychological distress’, assessed via the General Health Questionnaire (GHQ), which captures both anxiety and depression (Goldberg and Williams 1988).

Self‐esteem (13, 15 and 18/19): As at age 11, respondents completed the Rosenberg self‐esteem scale.

Depression (13 and 15): The six‐item version of Kandel and Davies’ (1982) depression scale which asks how often (never, sometimes, most of the time) each of the following occurred in the past month: felt too tired to do things; had trouble going to sleep or staying asleep; felt unhappy, sad or depressed; felt hopeless about the future; felt nervous or tense; and worried too much about things. To reflect a mental health measure of potential clinical relevance, we use a measure of ‘caseness’ defined as the proportion scoring over 19.5, one of several recommended cut‐offs. The percentage of the sample meeting this criterion is 21% at age 13 and 29% at 15.

Anti‐social behaviour (13 and 15): An index based on three indicators at each school wave: truancy from responses (agree/disagree) to the statement ‘If I get the chance to skip school I do’; predicted trouble with the police by age 21 (very unlikely to very likely); and an antisocial identity scale comprising three summed items (‘I get into fights’, ‘I am a rule breaker’ and ‘I take risks’ with four responses, ‘very true’ to ‘very untrue’). Confirmatory factor analysis reveals these form a single dimension with good reliability at each wave (Young, Sweeting, and West 2007), higher scores representing greater anti‐social behaviour.

Psychological distress (18/19): Derived from the shortest (12 item) version of the GHQ (range 0–12), and using a recommended cut‐off (2/3), respondents with a score of 3+ meet the criterion for ‘caseness’ (21%), a score indicative of potential clinical significance.

Educational attainment (18/19): Measured by two commonly used indicators of examination outcomes in Scotland; the number of ‘Standard Grades’ at Credit/General level, and number of ‘Highers’. Standard Grades, which are broadly equivalent to GCSEs in England and Wales, are normally taken at the end of year 4 in secondary school, the ‘Credit/General’ award being widely recognised as a criterion of superior attainment. ‘Highers’ are normally taken in years 5 and 6 in secondary school, and are requirements for entry into higher education. These qualifications were reported by respondents at age 18/19 and, in addition to those achieved in school, could also include those obtained elsewhere (e.g. colleges of further education).

In analyses of outcomes, with two exceptions, all the independent variables described above are retained as controls. The exceptions refer to the replacement of baseline measures of family structure and parenting with identical contemporary measures in analyses of outcomes at age 13 and 15, and the use of contemporary measures of secondary school characteristics in the same analyses at 15. Specific controls for prior well‐being and attainment are as follows: age 11 self‐esteem, depression and anti‐social behaviour for equivalent outcomes at 13, 15 and 18/19; age 11 depression for GHQ caseness, and maths ability for attainment at Standard Grade and Higher levels.

Analyses

As outlined in Figure 1, the principal focus of analyses presented here is on predictors of: (a) transition experiences as represented by school and peer transition concerns; and (b) well‐being within and beyond secondary school, together with educational attainment, including the transition variables as independents. However, the clustered nature of the sample allows us to address a related question of the extent to which there are ‘school effects’ on transitions, i.e. variation in school and peer concerns between primary and/or secondary schools after taking account of individual level characteristics. Furthermore, although data are limited, we can also examine the extent to which any ‘school effects’ are accounted for by school characteristics such as size, policies or preparedness of pupils for transition. To do this, each of the analyses is conducted in three stages: first entering all variables (sociodemographic, family, personal and pre‐secondary) unaffected by secondary school; second, adding primary school and peer context variables, and finally, further adjusting for secondary school factors. The latter also allows us to test the extent to which individual disengagement from secondary school explains any transition effect. While ‘school effects’ on transitions might occur by reference to both primary and secondary schools, in this sample the former is more important. Accordingly, analyses of transitions as dependent variables are presented within a primary school framework to take account of school level variance and estimate its contribution alongside individual predictors. In respect of well‐being and attainment, where transition concerns are independent variables, analyses are presented within a secondary school framework to capture contemporaneous (school and family) influences as well as those prior to entry into secondary school. This analytic framework extends to outcomes at 18/19.

All analyses utilise multi‐level modelling (Rasbash et al. 2000; Leyland and Goldstein 2001) to take account of both school and individual level factors and exclude missing data except for variables where this is represented by a dummy. All continuous (dependent and independent) variables have been transformed into z scores to enable direct comparison of effect sizes (+/− one standard deviation), these analyses utilising OLS linear models. In the case of the two categorical dependent variables (depression and GHQ caseness), logistic regression is used, the results being presented as a series of odds ratios (OR).

Findings

Transition concerns

Figure 2 shows the percentage of respondents who, looking back to the start of secondary school, reported finding it ‘very’ or ‘quite’ hard to cope with a range of issues presented by the new school and new set of pupils. In general, school concerns were more pronounced than peer concerns. Thus, nearly half the sample indicated they had problems coping with the size of the school and a third with the timetable and volume of work. Over two thirds (68%) reported finding one or more of these school issues hard to cope with, though a much smaller number (17%) found them ‘very hard’. Among peer concerns, around a quarter indicated problems coping with ‘older teenagers’, a fifth with bullying or teasing and mixing with pupils from different areas, smaller proportions finding it hard getting organised and travelling to school, and making new friends. Just under half (47%) reported finding one or more of these issues hard to cope with, while 14% found them very hard. Overall, these responses indicate a very varied transition experience, around a quarter reporting no problems at all, another quarter finding transition very hard to cope with.

Figure 2 Transition experiences (school and peer concerns) – % reporting (very/quite) hard to cope at start of secondary school.

Predictors of transition concerns

School variation

Table 1 shows the unadjusted variance in each of the transition variables between the 135 primary and 43 secondary schools respectively. For school concerns, the variance attributable to primary schools (0.021), though small (2% of the total), is significant, whilst that for secondary schools is not. For peer concerns, the primary school variance is larger (4%), and in this case that between secondary schools is also significant. A cross‐classification analysis of both school and peer concerns by primary and secondary schools reveals that the amount of variance attributable to the former is not only greater in these unadjusted models but also in all models which adjust for individual level variables, underpinning our decision to examine predictors of transition within a primary school context.

Table 1. Variation in transition experience (school and peer concerns) between primary and secondary schools – unadjusted school and individual level variance (and standard error) – n = 2034.

	School concerns	Peer concerns
Primary school
School level variance	0.021∗(0.010)	0.040∗(0.013)
Individual level variance	0.971(0.031)	0.960(0.031)
Secondary school
School level variance	0.004(0.005)	0.017∗(0.008)
Individual level variance	0.989(0.031)	0.982(0.031)
Note: ∗5% significance of variance indicated by a value greater than standard error × 1.65 (Snijders & Bosker 1999).

Individual level predictors (univariate)

Taking account of school level variance, Table 2 shows the univariate relationships between each of the individual level variables and both school and peer concerns. There is considerable overlap in the predictors of school and peer concerns, but also some important differences.

Table 2. Individual predictors of (poor) transitions (school and peer concerns) – univariate associations (standardised) – n = 2034.

	School concerns	Peer concerns
Sociodemographic (age 11)
Age (12 months)	−0.11	−0.11
Gender (Male)	−0.03	0.08
Social class (I)
II	0.01	0.02
IIIn	0.04	−0.05
IIIm	0.09	−0.10
IV	0.07	−0.15
V	0.15	−0.19
Missing	0.25	−0.05
Religion (Prot)
Catholic	−0.03	−0.03
Other	0.16	0.17
None	0.03	0.08
Missing	0.00	−0.05
Family (2 birth) (age 11)
Step	0.09	0.08
Lone	0.13∗	0.07
Parenting (age 11)
Control	0.08∗∗	0.09∗∗∗
Care	−0.06∗∗	−0.07∗∗
Personal attributes (age 11)
Maths ability (poor)	0.12∗∗∗	0.06∗∗
Anxiety	0.04∗	0.16∗∗∗
Aggression	0.07∗∗	−0.05∗
Self‐esteem	−0.12∗∗∗	−0.17∗∗∗
Bully victim	0.08∗∗	0.16∗∗∗
Disengagement	0.08∗∗	−0.05∗
(Un)prep secondary (age 13)	0.06∗∗	0.11∗∗∗
Pre sec experience (age 11)
Older sibs	−0.03	−0.02
Pals at sec	0.01	−0.12∗∗∗
Number of primaries	0.03	0.09∗∗∗
Primary school
Placing school	0.11	0.03
School roll	0.05	0.01
School policies	0.02	−0.01
Parental involvement	0.02	0.00
Peer group (class means [age 11])
Maths ability (poor)	−0.02	−0.03
Anxiety	−0.02	0.03
Aggression	−0.06∗	−0.06∗
Self‐esteem	−0.06∗	−0.05
Bully victim	−0.01	0.08∗∗
Disengagement	−0.04	−0.05
(un)prep secondary	0.02	0.08∗∗
Secondary school (age 13) Disengagement (pupil) (Contextual)	0.18∗∗∗	0.10∗∗
School roll	0.01	0.02
Ethos poor (researchers)	−0.03	0.01
Ethos poor (pupils)	0.05	0.03
Notes: All pre‐secondary variables measured at age 11 except preparedness for secondary school (13). Reference categories for categorical variables in parentheses.∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

In general, sociodemographic factors are not strongly related to transition experiences. There is no difference between boys and girls in either school or peer concerns, and though age (younger) and religion (‘other’) suggest poorer transitions, neither is significant. However, while social class is unrelated to school concerns, there is a small (but significant) inverse gradient with peer concerns, indicating progressively fewer concerns with falling social class. Overall, family factors are more important, respondents with (over)controlling parents reporting more school and peer concerns, those with higher parental care the reverse. School concerns are also more marked among those from lone parent families. Personal factors, however, are the most important predictors of both transitions, though not all in the same way. Respondents of lower ability, judged more anxious by class‐teachers, who had been victimised, had lower self‐esteem, and were less well prepared for secondary school, expressed more school and peer concerns (effect sizes being stronger for peer concerns except in the case of ability). Those who were more aggressive and disengaged from primary school had a poorer school, but better peer transition. These different relationships with the two dimensions are also apparent in respect of pre‐school experiences. While there is no relationship with school concerns, respondents with friends already at secondary school experienced a better peer transition as did those attending fewer (normally one) primary schools.

With respect to contextual characteristics, none of the primary school variables is significantly related to transition though respondents from placing schools tended to have more school concerns. However, several peer group characteristics are predictive, respondents in primary classes characterised by higher self‐esteem expressing fewer school concerns, those in classes with more experience of victimisation and less preparation for secondary school more peer concerns. Interestingly, a higher level of class aggression is associated with better school and peer transitions. Finally, excepting pupil disengagement from secondary school, none of the secondary school contextual factors is related to school or peer concerns.

School and individual effects (multivariate)

We next examine school and individual level factors together, employing multi‐level modelling in three stages. In model 1, all variables unassociated with primary school (sociodemographic, family, personal and pre‐secondary) are entered, any remaining school‐level variance indicating ‘school effects’. Model 2 adjusts for primary school and peer factors, any reduction in the school‐level variance suggesting explanations of ‘school effects’. Finally, model 3 adjusts for secondary school factors. The models also allow us to see which of the individual level variables have independent effects. The results are presented in Table 3.

Table 3. School and individual effects on transition experiences (school and peer concerns) – (standardized) – n = 2034.

	School concerns	Peer concerns
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
School level	0.021 ∗(0.010)	0.009(0.008)	0.008(0.008)	0.027 ∗(0.010)	0.019∗(0.009)	0.016∗(0.009)
Sociodemographic (age 11)
Age (12 months)	−0.10	−0.08	−0.10	−0.08	−0.07	−0.07
Gender (Male)	−0.03	−0.02	−0.03	0.01	0.01	0.01
Social class (I)
II	−0.03	−0.04	−0.06	−0.01	0.01	0.00
IIIn	−0.02	−0.01	−0.04	−0.04	−0.02	−0.03
IIIm	0.06	0.08	0.04	−0.08	−0.06	−0.07
IV	0.01	0.04	0.01	−0.14	−0.12	−0.13
V	0.07	0.12	0.07	−0.21	−0.21	−0.22
Missing	0.15	0.22	0.18	−0.14	−0.12	−0.14
Religion (Prot)
Catholic	−0.02	0.02	0.03	0.00	0.02	0.01
Other	0.16	0.13	0.13	0.06	0.06	0.06
None	0.02	0.02	0.00	0.00	0.00	−0.01
Missing	−0.12	−0.09	−0.12	0.03	0.04	0.02
Family (2 birth) (age 11)
Step	0.04	0.04	0.06	0.00	0.00	0.01
Lone	0.07	0.06	0.06	0.00	0.00	0.00
Parenting (age 11)
Control	0.04	0.04	0.04	0.04∗	0.04	0.04
Care	0.00	0.00	0.01	0.00	0.00	0.00
Personal (age 11)
Maths ability (poor)	0.08∗∗	0.08∗∗	0.09∗∗	0.05	0.05	0.05
Anxiety	−0.01	0.00	−0.01	0.12∗∗∗	0.12∗∗∗	0.12∗∗∗
Aggression	0.00	0.02	0.01	−0.06∗	−0.05	−0.06∗
Self‐esteem	−0.08∗∗	−0.07∗∗	−0.06∗	−0.11∗∗∗	−0.11∗∗∗	−0.10∗∗∗
Bully victim	0.04	0.05∗	0.04	0.11∗∗∗	0.09∗∗∗	0.09∗∗∗
Disengagement	0.05∗	0.07∗∗	0.04	−0.06∗∗	−0.05∗	−0.06∗∗
(Un)prep secondary (age 13)	0.04∗	0.04	0.00	0.09∗∗∗	0.08∗∗	0.06∗∗
Pre sec experience (age 11)
Older sibs	−0.06	−0.05	−0.06	−0.04	−0.04	−0.05
Pals at sec	0.01	0.01	0.01	−0.08∗∗	−0.09∗∗	−0.09∗∗
Number of primaries	0.03	0.02	0.02	0.08∗∗	0.07∗∗	0.07∗∗
Primary school
Placing school		0.10	0.14		−0.03	−0.01
School roll		0.03	0.05		−0.05	−0.05
School policies		0.00	−0.01		−0.01	−0.02
Parental involvement		0.03	0.03		0.01	0.01
Peer group (class means [age 11])
School ability (poor)		−0.03	−0.04		−0.04	−0.05
Anxiety		−0.01	0.00		−0.01	−0.01
Aggression		−0.06∗	−0.04		−0.02	−0.01
Self‐esteem		−0.03	−0.03		−0.02	−0.02
Bully victim		−0.03	−0.03		0.05	0.05
Disengagement		−0.05	−0.05∗		−0.05	−0.05
(Un)prep secondary		0.00	−0.01		0.04	0.04
Secondary school (age 13)  Disengagement (pupil)  (contextual)			0.15∗∗∗			0.06∗∗
School roll			−0.02			0.00
Ethos poor (researchers)			0.00			0.04
Ethos poor (pupils)			0.03			0.02
Notes: As Table 2.∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

Focusing first on school concerns, Model 1 shows that after adjusting for sociodemographic, family, personal and pre‐secondary characteristics, the school level variance (0.021) is unchanged (see Table 1), providing evidence of a primary ‘school effect’. The model also shows that the (univariate) effects of the family variables, together with those of anxiety, aggression and victimisation, are rendered non‐significant. Lower ability, disengagement and unpreparedness remain as independent predictors of poorer school transitions, self‐esteem as predictive of a better transition. In Model 2, the primary school and peer group variables are added, the effect of which is to reduce the school level variance to non‐significance, meaning that the ‘school effect’ is explained by these factors. The picture is unaltered in Model 3 when secondary school characteristics are added. At this stage, with all variables in the model, only two pre‐transition characteristics remain significant; irrespective of school, respondents of lower ability, and who had lower self‐esteem, had poorer school transitions. As in the unadjusted model, pupil disengagement is the only secondary school variable related to school concerns.

The results for peer concerns also reveal evidence of a ‘school effect’, though in this case it remains unexplained. While adjustment for individual factors (Model 1) reduces the school level variance (from 0.040 to 0.026), further adjustment for primary school and peer factors (Model 2) has only a small effect, the school level variance remaining significant in the fully adjusted model. The analysis also shows that in contrast to school concerns many more pupil characteristics are independent predictors, including the (inverse) relationship with class which remains significant after controlling for all other variables. Again, however, it is personal attributes that predominate; respondents who were anxious, victimised and unprepared for secondary having a worse transition; those with higher self‐esteem, who were more aggressive and disengaged from primary school having better transitions. Unlike school concerns, adjusting for these variables renders the effect of ability non‐significant. The effects of having a friend at secondary and the number of primary schools attended remain as independent predictors. Finally, none of the secondary school contextual factors is significant, the only association being with pupil disengagement.

Consequences of transitions within and beyond school

We now move on to examine the effects of transition experiences on well‐being within (self‐esteem, depression and anti‐social behaviour) and beyond school (self‐esteem and ‘psychological distress’) together with educational attainment (Standard Grades and Highers). The analyses of well‐being within school refer both to contemporaneous effects (measures of transitions and outcomes at age 13) and longer‐term (age 15) effects (two years later). Each analysis was conducted in three stages as before, the fully adjusted models controlling for all variables included in previous analyses of the transitions themselves. Although modelled within a multi‐level (secondary school) context, there are no ‘school effects’ on self‐esteem, depression or anti‐social behaviour at either 13 or 15, nor on self‐esteem or ‘psychological distress’ at 18/19. In the case of both attainment outcomes, significant school level variance remains after controlling for all variables.

Within school

We focus first on well‐being within school together with Standard Grades, representing outcomes up to the end of statutory education. The results for the fully adjusted models are presented in Table 4, which shows standardised regression coefficients for the continuous variables (self‐esteem, anti‐social behaviour and Standard Grades) and odds ratios for the categorical variable, depression. To simplify the results, only those predictors with significant independent effects on one or more outcomes are shown.

Table 4. Significant predictors of well‐being (ages 13 and 15) and attainment (Standard Grades) (standardised).

	Self‐esteem (β)	Depression (odds)	ASB (β)	Grades (β)
	13(n = 2039)	15(n = 1781)	13(n = 2127)	15(n = 1870)	13(n = 2088)	15(n = 1828)	(n = 1057)
Gender (Male)	−0.46∗∗∗	−0.50∗∗∗	2.13∗∗∗	3.06∗∗∗	−0.19∗∗∗	−0.19∗∗∗	0.12∗∗
Social class (I)
II	0.01	0.15	0.91	0.73	−0.01	0.00	−0.05
IIIn	−0.02	0.13	0.71	0.72	−0.14	−0.04	−0.20∗
IIIm	−0.05	0.13	0.71	0.72	−0.03	0.03	−0.39∗∗∗
IV	−0.01	0.23∗	0.71	0.87	−0.06	−0.06	−0.50∗∗∗
V	−0.06	0.16	1.03	0.65	−0.06	−0.02	−0.54∗∗∗
Missing	0.03	0.21	0.37∗	0.60	−0.08	−0.01	−0.69∗∗∗
Religion (Prot)
Catholic	0.10∗	0.04	0.97	1.37∗	0.05	0.07	0.00
Other	0.10	0.13	1.34	1.38	−0.12	−0.07	−0.04
None	−0.01	0.04	0.87	0.80	−0.02	0.00	0.07
Missing	0.07	−0.06	1.06	1.39	0.19∗	0.17	−0.17
Family (2 birth) (age 13/15)
Step	−0.11	0.11	1.47∗	1.32	0.08	0.10	−0.02
Lone	−0.01	0.01	1.03	1.31	0.04	0.10	−0.10
Parenting (age 13/15)
Control	−0.11∗∗∗	−0.11∗∗∗	1.15∗∗	1.08∗	0.01	0.01	−0.02
Care	0.09∗∗∗	0.10∗∗∗	0.88∗∗	0.82∗∗∗	−0.03	−0.09∗∗∗	−0.04
Personal (age 11)
Maths ability (poor)	0.00	− 0.03	0.93	0.94	0.01	−0.03	−0.41∗∗∗
Aggression	0.01	0.01	1.08	0.97	0.10∗∗∗	0.12∗∗∗	−0.13∗∗∗
PRIOR +	0.33∗∗∗	0.22∗∗∗	2.85∗∗∗	2.05∗∗∗	0.29∗∗∗	0.24∗∗∗
Bully victim	−0.01	−0.01	1.20∗∗	1.21∗∗	−0.03	−0.05	0.01
Disengagement	0.04∗	0.03	0.99	0.79∗∗	−0.06∗∗	−0.03	−0.03
(Un)prep secondary (age 13)	−0.06∗∗	−0.02	1.15	1.11	0.03	0.00	0.04
Pre sec experience (age 13)
Pals at sec	0.00	0.01	1.04	1.09	0.06∗∗	0.07∗∗	−0.02
Number of primaries	0.00	−0.03	0.98	0.89	0.02	0.05∗	−0.01
Primary school
Placing school	0.15∗∗	0.11	0.87	0.75	0.09	0.03	−0.17∗
School roll	−0.01	−0.01	1.03	1.05	−0.02	−0.02	0.06∗
Peer group (means [age 11])
Anxiety	−0.04∗	−0.03	1.00	1.12	−0.02	0.00	0.02
Aggression	−0.04∗	0.02	0.99	0.99	−0.03	−0.02	0.04
Disengagement	−0.01	0.01	1.10	0.89	0.02	0.02	0.06∗
Transition (age 13)
School	−0.06∗∗	−0.02	1.14∗	1.18∗∗	0.06∗∗	0.03	−0.06∗∗
Peers	−0.17∗∗∗	−0.07∗∗	1.35∗∗∗	1.33∗∗∗	−0.09∗∗∗	−0.09∗∗∗	0.03
Secondary school (age 13/15) Disengagement (pupil)	−0.19∗∗∗	−0.20∗∗	1.28∗∗	1.37∗∗∗	0.43∗∗∗	0.43∗∗	−0.16∗∗∗
Ethos poor (researchers)	0.01	0.01	0.98	1.26	−0.05∗∗	−0.01	0.02
Ethos poor (pupils)	0.01	0.01	0.86	0.85	−0.04	−0.02	−0.18∗∗
Note: Adjusted for all variables (as Tables 2 and 3).+ PRIOR = (age 11) self‐esteem, depression, ASB and Maths ability in relevant analyses.p <0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.

To contextualise the importance of transition experiences for well‐being and attainment, some comment is warranted on the effects of other independent variables. The well‐known effects of gender are evident on all outcomes at both timepoints, girls having lower self‐esteem together with higher (odds of) depression, and lower levels of anti‐social behaviour and more Standard Grades. The effect of social class on attainment is also apparent, a pattern contrasting with the lack of association with well‐being. While both religion and family structure have relatively minor and inconsistent effects, lower self‐esteem and higher depression are associated with (over) controlling parents at both timepoints, parental care by contrast having a protective effect which extends to anti‐social behaviour at 15. Among personal attributes, in addition to large effects of prior well‐being and attainment on their respective outcomes, there are persisting effects of aggression on antisocial behaviour and Standard Grades, and of victimisation on depression, together with an effect of disengagement and unpreparedness for secondary school on self‐esteem. Age 11 disengagement is also associated with raised odds of depression at 15. Among pre‐secondary school variables, respondents with more friends in secondary school, and attending more primary schools, had increased levels of anti‐social behaviour. The primary school and peer contextual variables are generally unrelated to well‐being, though placing school is associated with higher self‐esteem, peer anxiety and aggression with reduced self‐esteem, each at age 13 only. By contrast, reduced attainment is associated with placing school, increased attainment with larger primary schools and greater peer disengagement. This last finding is the opposite of a persisting association between individual disengagement with secondary school and reduced well‐being (all measures) and attainment. Among the secondary school variables, there are additional effects of both (researchers’ and pupils’) ethos measures on anti‐social behaviour (age 13) and Standard Grades respectively.

The analyses also reveal that controlling for all these variables, there are independent effects of both transition variables on most outcomes. In respect of self‐esteem, both school and peer concerns are associated with reduced self‐esteem at age 13, the latter being the larger effect. However, by 15, the effect of school concerns is not significant, and that of peer concerns considerably reduced. For depression, the picture is different, both school and peer concerns increasing the odds of ‘caseness’ at both timepoints, and with no reduction in effect size. The picture changes again in respect of anti‐social behaviour. In this case, the effect of school concerns is to increase anti‐social behaviour, the effect of peer concerns to reduce it. These opposite effects persist over time although that of school concerns just falls short of statistical significance at 15. Finally, in respect of attainment, another pattern is revealed. Among respondents who reported more school concerns, there is a reduction in the number of Standard Grades achieved. There is no significant effect of peer concerns. This contrasts with the picture for well‐being where the effect of peer concerns consistently outweighs that of school concerns. These transition effects are only marginally attenuated by the addition of individual disengagement from secondary school in the models (stage 2 results not shown).

A comparison of the effect sizes (β weights and OR) of variables included in the analyses allows us to assess the relative importance of school and peer transitions for well‐being and attainment. Two factors in particular are relevant here, gender and disengagement from secondary school, each consistently associated with the outcomes. In respect of self‐esteem, this shows that at 13 the combined transition effect (β school + β peers = 0.23, a difference of one standard deviation on each variable) is bigger than that of disengagement (β = 0.19) alone, though this is reversed over the next two years. The comparison with gender reveals that a difference of +/− 1 SD in both transition scores is crudely equivalent to the difference between boys and girls. In respect of depression at 13, a comparison of odds ratios reveals that the combined transition effect (1 SD peer + 1 SD school, OR = 1.59) is greater than that of disengagement alone (OR = 1.28), and crudely equivalent to the gender effect, reducing only slightly over time. By comparison, the combined effect of transition on anti‐social behaviour at 13 (β = 0.15) is only about a third that of disengagement (β = 0.43), and even less at 15, and crudely equivalent to that of gender at both timepoints. Finally, the effect of school concerns on standard grades (β = −0.06) is equivalent to about a third that of disengagement (β = −0.16) and about the same size as gender.

Beyond school

The comparable results for age 18/19 outcomes are shown in Table 5. In general, the effects of other independent variables, such as gender on self‐esteem and social class on educational attainment (‘Highers’), persist, though the effect sizes tend to be smaller than observed at 15. Predictors of ‘psychological distress’ (GHQ caseness) are also generally similar to those for depression at 15 with one or two exceptions, notably prior (age 11) depression which is non‐significant in this analysis. Much more important, though, for present purposes are the findings in relation to the transition variables which show a significant effect of either school or peer concerns on each of the outcomes. Thus, controlling for all other variables, respondents who reported more school concerns at 13 had lower self‐esteem six years later, the coefficient for peer concerns just falling short of significance. For ‘psychological distress’, the reverse is the case, peer concerns increasing the odds of ‘caseness’. Finally, echoing the result for Standard Grades, respondents with more school concerns reported fewer ‘Highers’. While generally the effect sizes of the transition variables on these outcomes is reduced compared with earlier on, they still bear comparison with those of other important variables. For example, the combined transition effect (β school + β peers = 0.11) on self‐esteem is similar to that attributable to disengagement at 15 (β = 0.14), and about a third that of gender; that on GHQ caseness (OR = 1.33) is bigger than that of disengagement (OR = 1.02), and again a third that of gender; and the effect on attainment (β = 0.08) though small by comparison with disengagement (β = 0.21) is crudely equivalent to that of gender. In short, the results demonstrate persisting effects of transition variables on well‐being and attainment beyond the end of secondary education.

Table 5. Significant predictors of well‐being (18/19) and attainment (Highers) (standardised).

	Self esteem (β)(n = 1050)	GHQ case (odds)(n = 1063)	Highers (β)(n = 1078)
Gender (Male)	−0.36∗∗∗	1.92∗∗	0.10∗
Social class (I)
II	0.27	0.70	−0.11
IIIn	0.31	1.33	−0.15
IIIm	0.28	0.77	−0.47∗∗∗
IV	0.24	1.04	−0.41∗∗∗
V	0.28	0.73	−0.40∗∗∗
Missing	0.34	0.75	−0.56∗∗∗
Family (2 birth) (age 15)
Step	−0.04	1.12	−0.07
Lone	0.02	1.58∗	−0.15∗
Parenting (age 15)
Control	−0.12∗∗	0.98	0.01
Care	0.02	0.84∗∗	−0.02
Personal attributes (age 11)
Maths ability (poor)	−0.01	0.93	−0.33∗∗∗
Anxiety	0.07∗	0.85	−0.04
Aggression	0.00	0.95	−0.10∗∗
PRIOR+	0.17∗∗∗	1.22
Bully victim	−0.02	1.30∗∗	0.04
(Un)prep secondary (age 13)	−0.06∗	1.27	0.08∗∗
Pre sec experience (age 11)
Number of primaries	0.07∗	1.13	−0.03
Peer group (class means[age 11])
Bully victim	0.02	0.49∗∗	−0.02
Transition (age 13)
School	−0.06∗	1.14	−0.05∗
Peers	−0.05	1.19∗	0.03
Secondary school (age 15) Disengagement (individual) (Contextual)	−0.14∗∗∗	1.02	−0.21∗∗∗
School roll	0.03	0.90	0.14∗∗
Ethos poor (researchers)	0.08∗	0.66∗	0.01
Ethos poor (pupils)	0.05	1.04	−0.15∗∗
Note: Adjusted for all variables (as Tables 2 and 3).+ PRIOR = (age 11) self‐esteem, depression and maths ability in relevant analyses.p <0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.

Discussion

Despite both the importance attributed to the primary–secondary transition in UK educational policy (Morris and Pullen 2006; HMIe 2006; Scottish Executive 2006) and considerable improvements in pastoral care arrangements for pupils in transition (Boyd 2005; Boyd et al. 2007), the evidence base on transition and its consequences remains incomplete. Research in this area is characterised by a number of disparate strands and methodologies, and less consistency in findings than often assumed. The overall aim of this paper has been to provide some connections between these different areas of research, and in particular to place pupils’ experience of transition at the heart of the enterprise and examine whether it really matters for well‐being and attainment.

Data relating to the primary–secondary transition were collected much later than in most studies, and are retrospective rather than current reports of the early post‐transfer period. In the view of some commentators (Measor and Woods 1984; Gillock and Reyes 1996; Tobbell 2003), this is likely to produce a more reflective account of the experience. In the event, we found that, although around a quarter of the sample recalled no problems, the majority had some difficulty in coping at the start of secondary school, a quarter finding the experience very difficult. In contrast to several studies (e.g. Chedzoy and Burden 2005), more concerns were expressed about the formal school system than the informal system of peer relations. Overall, the picture of the transition experience is much closer to that found in a number of qualitative studies (e.g. Pratt and George 2005), which stress its ‘critical’ nature, than in studies using quantitative methods, which generally present a more favourable picture (e.g. Graham and Hill 2003). There is some evidence however, that the earlier period in secondary school is characterised by considerable fluctuation in anxieties, and that the more deep‐seated ones only emerge later (Measor and Woods 1984; Muldoon 2005). This suggests that the concerns expressed by ‘11 to 16’ respondents were indicative of a more profound impact of the transition experience on identity.

The study also found that respondents’ school and peer concerns constitute two separate dimensions, underlining the importance of understanding the primary–secondary transition as navigations through both the formal (school) and informal (peer) social systems (Pratt and George 2005). Pupils can be successful in one arena, but not the other. The distinction is also apparent in respect of the evidence relating to predictors of transitions, one potentially key influence being schools themselves. Surprisingly perhaps, the analysis of ‘school effects’ revealed little or no ‘natural variation’ in transition experiences between secondary schools. By contrast, ‘school effects’ were found for primary schools, but while the effect on school concerns was explained by a cluster of school and peer characteristics, this was not the case for peer concerns. In other words, whatever impact the primary school had was much more apparent in preparing pupils for school issues than coping with peers.

While the primary school played a part in shaping school transitions, individual characteristics were much more important predictors; and among these, personal attributes were more important than sociodemographic, family or other factors. Some of these were common to both school and peer concerns, others underlined their distinctiveness. Among the former, and consistent with the literature (Anderson et al. 2000; Galton, Morrison, and Pell 2000; McGee et al. 2003), respondents with lower ability, lower self‐esteem and unprepared for secondary were particularly vulnerable to poorer school and peer transitions. We did not find gender differences and, though in the expected direction, younger respondents were not significantly disadvantaged in their transition experiences. We also confirmed the importance of family factors (Lord, Eccles, and McCarthy 1994), particularly parenting styles which limit autonomy.

In addition to these common predictors, there were several which were transition specific, notably in respect of peers. Respondents with higher anxiety, who had been victimised and had attended more primary schools were no more likely to report school concerns, but reported more peer concerns. Conversely, those regarded by their primary teachers as more aggressive, who were already disengaged from primary school, who had a friend in secondary, and who came from a lower class background were more likely to experience a positive transition. This last finding in particular runs counter to conventional wisdom (Anderson et al. 2000; McGee et al. 2003), but it vividly underlines the point that the primary–secondary transition is not unidimensional. Paradoxically, for particular groups of pupils conventionally regarded as difficult or disadvantaged, navigating the informal system of peer relations may be easier than for their more advantaged counterparts.

The longitudinal design of the study also allowed us to investigate the effects of respondents’ transition experiences on a range of outcomes, both contemporaneously (age 13) and over the longer term within (15) and beyond (18/19) school. The analyses revealed a complex pattern of relationships which again testify to the distinctiveness of the formal and informal social systems of the secondary school. First, in relation to self‐esteem, we found independent effects of both transitions on the contemporaneous measure (larger for peer than for school). While these weakened over time, peer concerns remained a significant predictor two years later, and school concerns a significant predictor six years later at age 18/19. This suggests, however, that the main impact of transitions on self‐esteem is in the earlier stages of secondary school, a finding consistent with the post‐transfer ‘dip’ found in US studies (Wigfield et al. 1991; Anderson et al. 2000). Second, in relation to depression, the results showed independent effects of both school and peer transitions which persisted over two years in school and did not diminish in magnitude. The comparable analysis of ‘psychological distress’ suggests the effect persists beyond school‐leaving. These findings again bear comparison with some US studies (e.g. Hirsch and Rapkin 1987), which found a different picture for self‐esteem and depression, and demonstrates that poorer school and peer transitions can have long‐term consequences for mental health. Third, the results for anti‐social behaviour also reveal transition effects, but in this case the effect of school concerns appears to increase such behaviour while peer concerns have the opposite effect. Conceivably, each is interpretable as an adaptation, the former as a reaction to school stresses, the latter as a strategy to avoid adverse peer reactions. Finally, in relation to educational attainment, we found evidence of a transition effect specific to school concerns which extended beyond examination results at age 15 to more advanced qualifications. In this case, how well respondents got on with other pupils following transfer to secondary school didn’t seem to matter; how well they coped with school issues, like the workload or different teachers, did.

The cumulative evidence from ‘11 to 16/16+’ strongly suggests that the impact of the primary–secondary transition goes beyond immediate post‐transfer anxieties to have a much more significant, longer‐term effect on pupil well‐being and learning. The experience is diverse, distinguishing pupils who have problems coping with different aspects of the formal and informal social systems, and the impact is diverse in its consequences. Its importance is also underlined by the fact that the impact of transition is comparable to that of two well‐established predictors. This is particularly so in relation to depression, where the combined effect size of school and peer concerns outweighs that of secondary school disengagement and approaches that of the single biggest predictor, gender. Transition experience is less important for anti‐social behaviour and attainment, where other influences predominate. The comparison with school disengagement is also pertinent because it is only a very small part of the explanation of transition effects.

In drawing these conclusions, it is very important to acknowledge several limitations of this study. First, it was not designed as a detailed investigation of the primary–secondary school transition. No data were collected on pre‐transfer concerns and anxieties, and an ideal design would involve the collection of such data over at least the first year of secondary school. Even more regrettably, no data were collected from either primary or secondary schools on transition policies, so allowing a much more robust investigation of school effects. The data derived from primary head‐teachers, though generally suggesting a ‘progressive’ policy effect, are far from ideal. However, the study has many advantages over others in the field, particularly its large sample, extensive range of data on respondent characteristics and the particularly long follow‐up period, which may to some extent compensate for its limitations.

Second, and perhaps most important, the fact that transition data were collected at the start of year 2 in secondary school, one full year after transfer from primary school, and that respondents were asked to recall their experiences of the ‘first few weeks of secondary’, means that their reports are not only retrospective but also constructed in the context of their present situation. As such, it might be argued that this conflates experiences of transition (e.g. victimisation) with current (post‐transition) experiences (e.g. continued victimisation). However, this in turn raises the almost unanswerable question as to whether transition is an episode, with a beginning and end point, or a continuing process. In either view, apart from the victimisation example, it is difficult to see how other features of the concerns reported by respondents, particularly school concerns (e.g. size of school, increased number of teachers), are anything other than characteristics of a new situation. Related to this is a second issue concerning the concurrent reporting of transition data with that on the well‐being outcomes at age 13. This raises the possibility that one explanation of the relationship between transition concerns and self‐esteem, depression and anti‐social behaviour at this timepoint is a reverse one; that is, rather than the experience impacting on pupils’ well‐being, those respondents in poorer mental health simply reported a more negative transition. This possibility is reduced by the number of controls included in the analyses, especially those relating to prior well‐being. It is also made much less likely by the fact that the effects of transition persist over time, both within school and after school‐leaving (six years later) and, perhaps most importantly, because the effects of transitions are specific to particular outcomes. This would not be expected if negative affectivity was a major factor.

Finally, perhaps the major limitation refers to the fact that the findings may only have historical value, the ‘11 to 16’ sample experiencing the primary–secondary transition over a decade ago in 1994/5. On some assessments (e.g. Boyd 2005), changes in pastoral care arrangements have transformed the transition experience for pupils, such that if the study were conducted today we would find a much more favourable picture. However, given the gaps and weaknesses of much research in the area, and in particular the lack of longitudinal studies, such an assessment may be over‐optimistic and certainly exceeds the evidence‐base. There is also an irreconcilable gap between research on contemporary transitions and those focusing on long‐term outcomes which are by definition always subject to the criticism that they are out‐of‐date. Our findings suggest transition matters in a more profound way than is often assumed. At the very least, they strongly endorse existing educational policies concerning the importance of a positive experience for pupils transferring from primary to secondary school. Such policies should recognise the importance of both the formal and informal social systems involved.

Notes on contributors

Professor Patrick West leads the ‘Youth and Health’ research programme in the MRC’s Social & Public Health Sciences (SPHSU) at Glasgow University. His research interests include young people and health (particularly mental health) and health behaviours (particularly smoking), youth lifestyles, school influences, transitions and health inequalities. He is on the editorial boards of Health Education and the Journal of Youth Studies.

Dr Helen Sweeting is a research scientist within the SPHSU ‘Youth and Health’ programme, prior to which she was employed as a clinical psychologist. Her work focuses on the health (currently obesity in particular), and health behaviours of children and young people, and investigates how they are influenced by factors such as gender, social class, family life, lifestyle and the peer group, and how they may have changed over time.

Mr Robert Young is also a researcher within the SPHSU ‘Youth and Health’ programme, prior to which he worked in Academic Psychiatry at Imperial College, Medical School. His research primarily focuses on young people’s mental health and how this varies by ‘traditional’ factors such as social class compared with ‘newer’ influences such as youth culture, the media and position within the peer group. His other major interest is the analysis and design of studies using innovative and advanced research methods.

Acknowledgements

The authors extend their thanks to Marion Henderson, Jacki Horne, Malcolm Hill, Sally Macintyre and Janine Muldoon for very helpful comments on previous drafts of this paper, and to Geoff Der and Alastair Leyland for statistical advice. This work was funded by the UK Medical Research Council as part of the Youth and Health Programme (WBS U.1300.00.007) at the Social and Public Health Sciences Unit.