How important are school and interpersonal student characteristics in determining later adolescent school connectedness, by school sector?

Waters, Stacey K. ; Cross, Donna ; Shaw, Therese 等

Introduction

Over the past 20 years the focus of Australian schools has shifted from primarily enhancing academic outcomes to recognising the importance of developing the whole student. From the Hobart Declaration on Schooling in 1989 to the Adelaide Declaration 10 years later, the Australian Commonwealth government has implemented several statements and policy directives aimed at providing opportunities for students to gain the knowledge, understanding, skills and values for a full and productive life outside of school.

Coupled with these changes in schools' priorities, there has been a shift toward non-government schooling in Australia. Over the 10-year period between 1997 and 2007, the proportion of students enrolled in government schools increased by less than 2 per cent while the proportion of students attending non-government schools rose sharply by 22 per cent (Australian Bureau of Statistics, 2008).

Characteristics of Australian secondary schools

With the recent increase in non-government school enrolment across Australia, there is a great need to understand potential differences in school characteristics as a function of school sector in order for school policy-makers to make informed decisions about the way in which secondary schools create opportunities for young people to feel connected. Between, as well as within, each of the government, Catholic and Independent school education sectors differing interpersonal and organisational school-level characteristics make each school's environment unique (Waters, Cross & Runions, 2009).These features may also influence the extent to which students feel they belong (Anderman, 2002), their academic achievement (Wingspread, 2004) and their physical, social and emotional health (Carter et al., 2007).

Waters, Cross and Runions (2009) presented an ecological description of a school environment or, more aptly, a school ecology, where a student's interpersonal interactions are surrounded by his or her immediate structural, functional and built environment. The concept of school ecology is based on Bronfenbrenner's (1979) ecological model, which recognises the importance of the environment in shaping human development, and the extrapolation of this work by Moos (1979) to a school setting where the school environment and personal relationships modify academic and health outcomes of young people. In this school ecological model, interpersonal characteristics of schools can be represented by the relationships between students, between staff and between students and staff. Such relationships can range from collegial and respectful to didactic and authoritarian and have been shown to affect the way young people feel about school, achieve academically at school and participate in school activities (Lee & Smith, 1999; Ryan & Patrick, 2001). This component of the ecological framework was shaped largely from the plethora of school climate literature that describes the important role of interpersonal relationships in creating a positive school climate (Brand et al., 2003; Gottfredson et al., 2005; Kuperminc, Leadbeater & Blatt, 2001).

Organisational features of a school can be described in three broad categories: structural, functional and the built environment. They are influenced by well-developed school-based health intervention frameworks such as the Health Promoting Schools (HPS) model (Booth & Samdal, 1997) and the Coordinated School Health Program (CSHP) (McKenzie & Richmond, 1998). The structural features of a school include its size, number of year levels, extent of departmentalisation and leadership (Waters, Cross & Runions, 2009) and differ considerably among Australian schools. Non-government secondary schools are typically larger in size than government schools (Australian Bureau of Statistics, 2009), although this may be confounded by the integration of primary and secondary school students on one campus. Over 77 per cent of employees of government schools were teaching staff compared with 74 per cent of Catholic staff and 70 per cent of Independent school staff (Ministerial Council on Education, Employment, Training and Youth Affairs--MCEETYA, 2005). Many non-government schools incorporate middle-school structures to provide developmentally appropriate learning environments for early adolescents, compared with the government sector in which primary and secondary students are typically separated into different schools.

School functional characteristics are often well considered in models such as the HPS and CSHP, whereby the policies and practices of schools are key to successfully changing behaviour that influences health. In this ecological model, functional characteristics of schools include pastoral care policies and programs, student involvement in decision-making, teaching and learning strategies and an expectation for a high level of learning (Waters, Cross & Runions, 2009). Non-government schools have an additional focus on the development of pastoral care programs and strategies to build the life skills of children and young people compared with government schools (Catholic Education Commission of WA, 2000). Government schools may also provide this support for young people, although it is not a formalised approach.

The third organisational component of a school is the built environment, which extends the HPS framework's focus on the physical environment and borrows from more recent neighbourhood research that suggests that individuals' mental and physical health can be influenced by factors such as the presence of graffiti and access to facilities and services (Weich et al., 2002).Almost 60 per cent of independent schools' income in 2008 was derived from private sources such as parents and other donations (Independent Schools Council of Australia, 2008). In Western Australia in 2004 this represented more than 90 per cent of the budget allocated to buildings and grounds in independent schools (Association of Independent Schools of Western Australia, 2009).

The choice of schooling in Australia

The choice made by parents as to the school sector in which to enrol their child may be influenced by many of these ecological characteristics. In 2008, the Independent Schools Council of Australia (ISCA) (2008) found the most commonly cited reasons for parents choosing an independent school were access to 'good teachers' along with discipline within a supportive and caring environment, good facilities and educational excellence. Similarly, the Australian Council for Educational Research (ACER) (Masters, 2004) reported that parents of students enrolled in all government and non-government schools valued the quality of teachers at a school most highly when choosing a secondary school for their child, followed by a safe and secure school environment in which each student is cared for. Further, the choice of a non-government school was influenced most significantly by the moral and religious values of the school, discipline policies, school traditions and the requirement to wear a school uniform (Masters, 2004).

School ecological features and adolescent outcomes

If, as reported by these recent studies, parents are considering ecological characteristics when choosing secondary schools for their children, more research attention should be focused on student outcomes as a function of school sector. Recent school performance research indicates that students at independent schools are more likely than students at Catholic and government schools (in that order) to gain higher final-year tertiary entrance scores and attend university, even after controlling for other potential confounders of academic achievement and participation in higher education (Lamb et al., 2004; MCEETYA, 2005).While such differences between sectors exist, some government schools still outperform non-government schools, suggesting that there are clearly other factors besides sector which are contributing (Lamb et al., 2004).

Students' interpersonal relationships also vary between sectors. A recent Australian study found government school students more likely to report that they are bullied by others (Cross et al., 2009). Moreover, staff in government schools report seeing or having more bullying reported to them than in non-government schools and spend more time managing such incidents (Cross et al., 2009).

In Australia, students enrolled in independent schools report higher levels of school connectedness than students attending Catholic schools, who are in turn, more highly connected than government school students (Fullarton, 2002). Feeling highly connected to school has been reported as one of the most positive and lasting influences on a student's health and academic achievement (McNeely, Nonnemaker & Blum, 2002; Resnick et al., 1997). Being more highly connected to school has been associated with fewer mental health problems (Shochet, Homel, Cockshaw & Montgomery, 2008;Waters, Cross & Shaw, in press), greater academic achievement (Anderman, 2002; Klem & Connell, 2004) and less involvement in problem behaviours (Resnick et al., 1997). One of the seminal studies reporting the powerful influence of a connection to school came from the National Longitudinal Study of Adolescent Health (Add Health) in America where a cross-sectional sample of more than 12,000 students reported their feelings of connection to school along with other health risk and promoting behaviours (Resnick et al., 1997). From this nationally representative sample, Resnick and colleagues (1997) concluded that school connectedness is indeed one of the most powerful protective factors for young people. Further research from the Add Health study as well as other similar large-scale research studies have been replicated in the USA and Australia, and all report a similar protective role of school connectedness on adolescent health behaviour (Fullarton, 2002; McNeely, Nonnemaker & Blum, 2002).

School connectedness has been defined in many ways, from simply feeling a part of the school through to more complex definitions incorporating the school's support of students' academic interests, personal health and well-being and provision of a safe school environment. Others argue that connectedness can be thought of in three distinct domains of behavioural, cognitive and emotional engagement (Fredricks, Blumenfeld & Paris, 2004). In a school setting, Connell and Wellborn (1991) suggested that ecological characteristics of structure (clarity of expectations); autonomy support (provision of choice); and involvement (interest and support of individuals) help young people to achieve their three basic psychological needs of autonomy, competence and relatedness. These, they argued, create highly connected students who experience greater affect, cognition and behaviour.

This research therefore seeks to answer the following research question: to what extent do school ecological characteristics, represented by school sector, modify the impact of interpersonal relationships and mental health in Year 8 on later school connectedness (Year 9). It is hoped that these findings will provide a starting point for future research to identify the critical features of schools that best enhance adolescent feelings of connectedness to school. Given the plethora of evidence supporting the role of school connectedness in influencing academic, health and well-being outcomes for young people, this research will focus solely on school connectedness as an outcome.

Method

Sample selection and recruitment

Data from two independent studies conducted by the Child Health Promotion Research Centre at Edith Cowan University were combined to investigate the association between school level factors and students' later school connectedness. The Extra-curricular Project (ECP) (student n = 2809) was a naturalistic observational study conducted from 2004 to 2006 and the Supportive Schools Project (SSP) (student n = 3486) was a three-year (2005-2007) cluster randomised trial designed to reduce students' experiences of bullying in secondary school. These two studies involved a total of 39 government and non-government secondary schools.

Both projects received approval from the Edith Cowan University Human Research Ethics Committee and the three Western Australian education sectors. Schools with more than 50 Year 8 students were stratified by school size and socioeconomic status. The SSP recruited schools from the Catholic education sector only. For both studies, within each stratum, schools were randomly selected to participate. Eight schools from each study declined to participate due to competing school priorities and were replaced with the next randomly selected school within their strata. Parents of all Year 8 students were asked to provide consent for their child to participate, using a combination of active and passive consent. Student surveys were completed in class time.

Measures

Individual level

Fourteen Year 8 student-level variables were used as predictors of connectedness, as well as potential interaction terms. School-level variables included school, family and teacher connectedness, mental health, perception of peer support, transition difficulties, academic achievement and classroom management climate. After the factor structures of each composite scale were explored, confirmatory factor analysis (CFA) was conducted in Lisrel. All complete data were used to calculate weighted composites in Lisrel (approximately 6295 students). The CFAs used a weighted least squares estimator to determine the extent to which each of the observed items contributed to the underlying latent construct and to obtain a proportionally weighted continuous score for each, accounting for measurement error in the items.

A model was said to have good fit when the Adjusted [chi square] ([chi square]/df values, representing absolute fit, were less than 5 (Hair et al., 1998), the Root Mean Square Error of Approximation (RMSEA), representing model parsimony, was less than 0.08 (Browne & Cudeck, 1993) and incremental fit, represented by the Adjusted Goodness of Fit Index (AGFI), was greater than 0.95 (Hair et al., 1998).

School connectedness Five statements (I feel part of this school, I feel close to people at this school) on a four-point scale (always-never), were adapted from Sieving and colleagues (2001) to measure the extent to which students felt connected to their school. Adequate model fit of the data was found ([chi square]/df = 8.45; RMSEA = 0.05; AGFI = 0.99), and the measure had good reliability in this sample (a = 0.80) and in other previous studies (a = 0.75-0.82) (Sieving et al., 2001).

Teacher connectedness Students reported on six statements, using a scale of 0-3, how connected they felt to teachers at their school (a teacher at school who cares about me; a teacher at school who tells me when I do a good job). Modified from the California Healthy Kids Survey (WestEd, 2008), this scale had good model fit ([chi square]/df = 5.62; RMSEA = 0.04; AGFI = 0.99) and reliability ([alpha] = 0.83). The shortened version of this scale has in the past reported only modest reliability ([alpha] = 0.63) (McNeely & Falci, 2004).

Family connectedness This scale comprised 13 statements derived partly from Sieving (2001) (for example, a family member home before or after school or both; good relationship with family) and combined with new items developed by the research team (I am an important member of my family; Someone in family listens to my problems). Exploratory factor analyses revealed three factors: students' closeness with their family, parent listens to them and parental presence. Good model fit for the second order model ([chi square]/df = 7.16; RMSEA = 0.05; AGFI = 0.99) was found along with good reliability scores for the 13-item one factor model used in the analyses ([alpha] = 0.88).

Mental health Five sub scales were included comprising five items measuring each of students' conduct problems, peer problems, hyperactivity, emotional symptoms and pro-social behaviours, based on Goodman's (1997) Strengths and Difficulties questionnaire. In earlier reliability testing, the five item scales were found to be reliable, stable over time and highly correlated with other standardised mental health tests (Goodman, 2001). In this sample, satisfactory model fit and reliability were found for each of: conduct problems ([chi square]/df = 5.66; RMSEA = 0.03; AGFI = 0.99; [alpha] = 0.663); peer problems ([chi square]/df = 24.04; RMSEA = 0.07;AGFI = 0.99; [alpha] = 0.565); hyperactivity ([chi square]/df = 9; RMSEA = 0.04; AGFI = 0.99; [alpha] = 0.74); emotional symptoms ([chi square]/df = 13.52; RMSEA = 0.05; AGFI = 0.99; [alpha] = 0.73); and pro-social skills ([chi square]/df = 3.38; RMSEA = 0.02; AGFI = 0.99; [alpha] = 0.69).

Perception of peer support Students' perceived support from peers, adapted from Ladd, Kochenderfer and Coleman (1996) comprised 10 statements measuring peer relationships on a three-point scale (lots of times, sometimes, never) with items such as other students like to work with you; miss you if not at school ([alpha] = 0.85-0.88). Adequate model fit ([chi square]/df = 11.71; RMSEA = 0.04; AGFI = 0.99) and reliability ([alpha] = 0.86) were found in this sample.

Transition to secondary school Students were asked about the extent to which their transition from primary to secondary school was difficult (Akos, 2002). Their responses, originally on a five-point scale, were recoded to binary variable representing those who experienced a difficult transition and those who did not.

Academic achievement Students' self-perceived academic achievement, compared with others in their year group, was measured as either above, average or below average, and later recoded into a binary measure of below average academically and average or above.

Classroom management climate This four-item scale, adapted from McNeely (2002), sought students' perception of the difficulties faced in their classroom, such as trouble getting along with teachers and peers and paying attention in class. Borderline model fit ([chi square]/df = 34.3; RMSEA = 0.07; AGFI = 0.99) and reliability ([alpha] = 0.80) were found.

School level

School-level ecological data were collected in both studies from standard school records and complemented by a survey administered to all project coordinators in each school at the completion of the study.

Built environment A measure of the overall condition of the school's campus (buildings and grounds), the number of buildings with graffiti (every building to no buildings), access to shelter (all had access, most, some, none) and equipment and activities offered at break times (access for all students, rostered, first served, none) for students was sought. Each item was recoded to form two or three categories due to limited response variation.

Functional Measures of the priority schools placed on pastoral care (high or moderate), the type of home room (vertical or horizontal streaming), presence of a house or faction system (yes, no) to which students belong and a pastoral care period in the timetable (yes, no) are used. The government and Catholic education sectors provided the state's benchmark academic testing outcomes, represented as a Year 9 school-level average, for students' reading, writing and numeracy skills. A representative from each school was also asked to report the proportion of teachers at the school who held a master's degree or were in their first two years of teaching, both of which were recoded to form binary variables.

Structural The school's socioeconomic index, as calculated by the Department of Education and Training (DET) in Western Australia and the Commonwealth Department of Education, Employment and Workplace Relations, was used along with the number of students enrolled and the number of year levels at each school (kindergarten to Year 12 against Years 8 to 12). The latter were taken from DET records, along with the sector each school belongs to (government, independent or Catholic).

Approach to data analysis

The basic underlying assumptions of multi-level regression analyses were tested and normality of the dependent variable and linearity were found to hold. While some predictor variables were non-normally distributed, we have followed the recommendations of Altman (1999) who noted that non-normally distributed predictor variables do not influence regression estimation procedures. Random intercept regression analyses with cross-level interactions (between the school-level variable sector and individual-level variables) were performed in STATA (StataCorp, 2007) and used maximum likelihood estimation. A random intercept was included in the model to account for non-independence of students within the same school and the interactions assessed whether the associations between the individual-level variables, such as a student's perceived level of peer support and school connectedness, differed across sectors.

The first stage of the analyses was to build a student-level regression model to explain the individual characteristics of students in Year 8 that predicted their connectedness in Year 9 (with researcher-driven backward stepwise elimination). This was conducted in order to control for these important predictors of adolescent connectedness and further enhances the importance of the study's findings. Next, controlling for the significant student-level characteristics, interactions between individual-level predictors of students' enhanced connectedness and school sector were tested one at a time (assessed by comparing the predicted values from the student level model to one including the interaction term using a likelihood ratio test and presented as a [chi square] value).Thus, a series of models were generated; in each an interaction was added to the student-level regression model generated in the first stage. The final stage of the analyses involved graphing these significant interactions to describe the nature of each interaction. In this article, only the graphs displaying the significant interactions are presented for ease of interpretation of the results. Under each graph, we report the log likelihood ratio chi square test, degrees of freedom and significance of each interaction. Where the p value is below 0.05, a significant interaction is present. All analyses were conducted controlling for the effect of being a participant in the two different studies from which the data were drawn, gender and potential intervention effects of the SSP, as well as the significant interpersonal predictors of Year 9 school connectedness.

Results

Participation

In total, 39 schools were recruited into the two studies: 18 in the ECP and 21 in the SSP. Just over 10 per cent of the SSP student sample and 6.5 per cent of the ECP sample did not have active parental consent to participate so were not asked to complete a survey. While data in the two studies were collected two years apart (2004-2005 for ECP and 2006-2007 for SSP) and then combined, each student cohort was the same age at time of recruitment.

Descriptives

In total, 11 of the 39 schools were government, 24 Catholic and the remaining 4 independent schools. School size ranged from 300 to 1823 (mean 1027) and the majority of schools catered for students in Years 8 to 12. Half of the sample were male (50 per cent) and at data collection were, on average, between 12 and 13 years of age. Students reported high levels of teacher connectedness, few mental health problems and good levels of family connectedness. Most students reported they were average or above academically (see Table 1).

Ecological characteristics

Government Across the 11 government schools in this sample, the average socioeconomic status score was close to the state average, school size was large (on average, 1151), but highly varied. Class sizes averaged 29.2, which was similar to those in the Catholic schools, and slightly higher than those in the independent schools in the sample. School discipline policies were reported to be 'harsh' but no more so than in Catholic and independent schools. Relatively few students were reported to be involved in decision-making. Only 18 per cent of schools reported staff with master's degrees and few teachers were in their first two years of teaching. Almost all reported some graffiti, with one quarter indicating their campus condition was poor. Similarly, few schools reported having a high priority for pastoral care, pastoral care time scheduled in the timetable and house or faction systems in place.

Catholic A broad cross-section (n = 24) of Catholic schools was recruited into this study with the average socioeconomic status close to average for the state. School size was generally smaller than in the government and independent schools sampled, and class sizes were similar to those in the government schools. Catholic schools rated their discipline policies as 'harsh', and only half reported that students had some level of involvement in decision-making. One quarter were kindergarten to Year 12 schools and most reported they had many or some teachers with master's degrees. One third reported some graffiti yet the condition of the campus was reported by most to be good. All reported a high priority for pastoral care, most had a house or faction system and a pastoral care period.

Independent The five independent schools represented in this study were all kindergarten to Year 12 structured schools and 13 points above the state socioeconomic status average. School size was, on average, high, but much less varied than in government and Catholic schools. These schools had slightly smaller class sizes than the government and Catholic schools in the sample. Like the government and Catholic schools, they reported harsh discipline policies, and most indicated that students were involved in decision-making. All schools reported they had some or many teachers at the school with a master's degree and 70 per cent of schools reported they had some or many teachers in their first two years of teaching. All five schools reported no graffiti and a good or very good campus condition. Most had vertical (Year 8-12) home room structures and all had a house or faction system as well as a dedicated pastoral care period and reported pastoral care was a high priority at the school.

Relationship between sector and student-level variables

The relationship between school ecologic characteristics and each of the individual student-level characteristics are shown in Table 1. In this study, there were more girls in government (51 per cent) and Catholic schools (51 per cent) than independent (44 per cent) schools. More students in Catholic schools than government and independent schools found the transition from primary to secondary school difficult and most, regardless of sector, reported they were average or above academically. Students in Catholic schools reported they were more highly connected to school, teachers and family in Year 8 than students enrolled in independent and government schools respectively and reported fewer difficulties in the classroom, but lower levels of perceived support from peers. Students in government schools had, on average, more mental health problems than students enrolled in Catholic or independent schools.

Student-level model

A student-level model was built to determine the significant interpersonal variables in Year 8 that predicted students' enhanced school connectedness in Year 9. Students' connectedness to teachers, emotional symptoms, perception of peer support and academic achievement were not significantly related to school connectedness. The final model included transition experience (p<0.001), connectedness to family (p<0.001), participation in extra-curricular activities (p = 0.025), perception of classroom management climate (p<0.001), pro-social skills (p<0.001), hyperactivity (p = 0.013) and conduct (p<0.001) and peer problems (p<0.001).This was the base model, which was used for each of the subsequent interaction tests.

Cross-level interactions

After testing all potential interactions between sector and each of the individual student-level variables, controlling for the student-level predictors of Year 9 connectedness, five significant interactions were found when comparing the likelihood ratio test (represented as chi-squared) between the models with and without the cross-level (school- and student-level) interaction term (Figures 1-5).

Connectedness to family Figure 1 presents the relationship between family connectedness and later school connectedness modified by school sector after controlling for the Year 8 predictors of school connectedness ([chi square] = 8.40; df = 2; p = 0.015). Students enrolled in Catholic schools who have very low levels of family connectedness also have the lowest levels of school connectedness but, as family connectedness increases, school connectedness also increases to levels higher than in government and independent schools.

Connectedness to teachers In Figure 2, students in any of the three sectors with low teacher connectedness had correspondingly low feelings of school connectedness but for higher levels of teacher connectedness, students in independent and particularly Catholic education schools reported much higher levels of school connectedness ([chi square] = 7.97; df = 2; p = 0.019).

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Hyperactivity Students in Catholic schools with low hyperactivity scores report lower levels of school connectedness than students in independent and government schools respectively, and this difference is greatest for high levels of hyperactivity ([chi square] = 6.87; df = 2; p = 0.032) (see Figure 3).

Emotional problems Figure 4 shows that students with few emotional problems in Catholic schools report higher levels of school connectedness than students in independent or government schools respectively ([chi square] = 7.12; df = 2; p = 0.028). Again, for higher levels of emotional problems, student-reported school connectedness drops to a similar level for students in government and Catholic schools but is slightly higher for students in independent schools.

Perception of peer support In Figure 5, students enrolled at Catholic schools with low levels of peer support experience very high feelings of school connectedness ([chi square] = 28.17; df = 2; p<0.001). Comparatively, students enrolled at government and independent schools with low levels of perceived peer support experience low school connectedness, however as peer support increases, so too does school connectedness. Catholic education students however experience a decline in reported school connectedness where peer support is greater.

Non-significant interactions The relationships of students' school transition experience, perception of their classroom climate, pro-social skills, conduct problems, peer problems and self-reported academic ability with later enhanced school connectedness were not significantly modified by school sector.

Discussion

The aim of this study was to determine the extent to which the relationship between interpersonal characteristics of students and their later school connectedness was modified by school characteristics, represented by the government, Catholic and independent school sectors. Very little research exists around the influence of school characteristics on students' feelings of connectedness to school. In this study, rather than focusing on specific characteristics of schools we have chosen to describe schools by the sector to which they belong. The results therefore are contingent on schools having similar characteristics within and not between sectors.

In this study we have seen that while each school is different, there are substantial between-sector differences in the ecological characteristics of function, structure and built environment, enabling further investigation into how school sector may modify the relationship between students' interpersonal characteristics and their feelings of connectedness to school.

The relationship between five of the 11 student variables and later school connectedness was found to be significantly modified by school sector. These include students' connectedness to family and teachers, hyperactivity, emotional problems and perception of peer support. No relationship was found for students' transition experience, classroom climate, pro-social skills, conduct problems, peer problems and self-reported academic ability.

Higher levels of family and teacher connectedness and perceived support from peers are associated with school connectedness, regardless of school sector. This may be explained in part by the high correlation of school connectedness with family (Shochet, Homel, Cockshaw & Montgomery, 2008) and with teacher connectedness (Anderman, 2003) and perception of peer support (McNeely & Falci, 2004). In this article, students in Catholic schools with low family and teacher connectedness report the lowest school connectedness, yet those with higher levels of family and teacher connectedness report school connectedness higher than that of students in government or independent schools. This finding may be explained by previous research suggesting the important role teachers play in the academic and social and developmental health of young people (McNeely & Falci, 2004).Where there is little or no connection to family or teachers at the school, students' school connectedness is at its lowest. Students with very low family and teacher connectedness have most to gain from targeted school-based strategies to help them engage with their family and other teachers and ultimately protect them from other problem behaviours highly correlated with a lack of school, teacher and family connectedness (Carter et al., 2007; Resnick et al., 1997). That school connectedness in government schools does not increase as rapidly with increasing teacher connectedness compared with non-government schools suggests that non-government schools have developed pastoral strategies that modify the association between teacher and school connectedness more than have government schools.

An unexpected interaction between students' perceived peer support and later school connectedness was found for students enrolled in Catholic schools. In these schools, students with low perceived peer support have high levels of school connectedness one year later but for higher levels of peer support, students' school connectedness is lower than that of government and independent school students. McNeely, Nonnemaker and Blum (2004) and Karcher and Finn (2005) reported similar findings, suggesting that a positive connection to peers may be considered unconventional if peers do not exhibit socially desirable pro-social skills or participate in more risky health behaviours and that these unconventional connections negate any protective effects of being highly connected to school or parents. That this effect may only be present for students in Catholic schools is not explained and requires further investigation.

Students with low levels of emotional problems in Catholic schools have the highest levels of school connectedness one year later, but the data suggest that students with higher mental health risk will continue to feel less connected, irrespective of the school sector they attend. No prior evidence supporting a link between emotional symptoms and school connectedness could be found but low levels of school connectedness in adolescents have been found to predict poorer future general functioning, anxiety and depressive symptoms (Shochet, Dadds, Ham & Montague, 2006) and some empirical support exists for the role of enhanced school connectedness on reduced emotional distress (Resnick et al., 1997).

Limitations

This study has some limitations. Firstly, the student sample was taken from a small number of schools in the Perth metropolitan area only and does not include rural populations. It is anticipated that students in rural locations may be more highly connected to school given the school's central role in rural Western Australia; this requires further investigation. Moreover, the sample of schools, although randomly selected, was potentially biased as eight schools in each of the parent studies that were originally approached to participate declined and within the 39 participating schools, up to 11 per cent of students' parents also declined permission for their child to participate. Furthermore, a self-selection bias exists through students and parents deliberately selecting the school sector into which they are enrolled. Historical effects may be evident between students' responses in both parent studies as they were conducted two years apart. Thirdly, while the ECP cohort was a stratified sample of government, Catholic and independent schools in Western Australia, the SSP cohort was drawn from Catholic schools only. A dummy variable representing the study from which the data originated was included in all analyses to control for systematic differences between the two studies. Fourth, further analyses are needed to investigate interactions by characteristics of schools to determine whether the differences lie between sectors or just depend on individual school characteristics. Finally, some of the data used are self-report and are subject to issues of validity, although extensive pilot testing and test-retest reliability studies have been used in order to minimise this limitation.

Conclusions

There are many practical implications for schools and school staff that should be considered as an outcome of this study. Firstly, while individual schools can do much to develop connected students, it does appear that schools within the independent and Catholic school sectors are better equipped to implement pastoral strategies and policies that keep young people connected to school. Schools with smaller class sizes may also create greater opportunities for young people to become connected to teachers and to the school itself. Second, it appears that students suffering from mental health problems (specifically hyperactivity and emotional problems) benefit from enhanced feelings of school connectedness, regardless of the school sector to which they belong. These findings should encourage all schools to create opportunities for young people to become connected to school, and there was no evidence to suggest that any one school sector currently does this better than another in Western Australian schools. Third, teachers in independent and Catholic education schools appear to have a more central role in enhancing students' feelings of school connectedness than do teachers in government schools. It may be that smaller class sizes, more highly trained teachers and younger teachers and more time for pastoral care in the timetable, all features characteristic of non-government schools in this sample, may lead to greater feelings of connection to teachers which in turn increases later school connectedness.

School connectedness has been linked empirically to providing protection for students from the harms associated with drug and alcohol use, poor physical health and being early school drop-outs (Resnick, Harris & Blum, 1993), yet the effect of Australian school sector on the relationship between school connectedness and interpersonal characteristics has yet to be explored. This study found significant interactions between school sector and the relationship between student characteristics of perceived peer support, connectedness to teachers and family, hyperactivity and emotional symptoms and students' later school connectedness. Therefore, while school characteristics may influence students' school connectedness, students' school connectedness varies based on students' individual characteristics both between and within sectors. Clear implications for future policy and practice are highlighted in this article including the importance of providing dedicated staff and time in the school timetable to facilitate the implementation of pastoral care strategies, a defining feature between government and independent (and to a lesser extent, Catholic) schools in this sample.

Acknowledgements

The authors would like to thank Laura Thomas and Tommy Cordin for their review of earlier drafts of this manuscript as well as the schools and students involved in this research. This work was supported by a Health Promotion Foundation of Western Australia Scholarship.

References

Akos, P. (2002). Student perceptions of the transition from elementary to middle school. Professional School Counseling, 5(5), 339-345.

Altman, D. G. (1999). Practical statistics for medical research. London: Chapman & Hall/ CRC.

Anderman, E. M. (2002). School effects on psychological outcomes during adolescence. Journal of Educational Psychology, 94(4), 795-809.

Anderman, L. H. (2003). Academic and social perceptions as predictors of change in middle school students' sense of school belonging. Journal of Experimental Education, 72(1), 5-22.

Association of Independent Schools of Western Australia. (2009). Funding Independent Schools. Osborne Park: Author. Retrieved 22 July 2009 from http://www.ais.wa. edu.au/funding-indpendent-schools/australian-government-funding/

Australian Bureau of Statistics. (2008). Schools, Australia 2007. Canberra: Author.

Australian Bureau of Statistics. (2009). Schools, Australia 2008. Canberra: Author.

Booth, M. L., & Samdal, O. (1997). Health-promoting schools in Australia: Models and measurement. Australian and New Zealand Journal of Public Health, 21(4), 365-370.

Brand, S., Felner, R., Shim, M., Seitsinger, A., & Dumas, T. (2003). Middle school improvement and reform: Development and validation of a school-level assessment of climate, cultural pluralism, and school safety. Journal of Educational Psychology, 95(3), 570-588.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MS: Harvard University Press.

Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Ed.), Testing structural equation models (pp. 136-162). Newbury Park, CA: Sage.

Carter, M., McGee, R., Taylor, B., & Williams, S. (2007). Health outcomes in adolescence: Associations with family, friends and school engagement. Journal of Adolescence, 30, 51-62.

Catholic Education Commission of WA. (2000). Living the vision. Leederville: Author. Retrieved 11 June 2009, from http://web4.ceo.wa.edu.au/living_vision.asp

Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy and relatedness: A motivational analysis of self-system processes. In M. R. Gunnar & L. A. Sroufe (Eds.), Self Processes and Development: The Minnesota Symposia on Child Development (Vol. 23, pp. 43-77). Hillsdale, NJ: Lawrence Erlbaum.

Cross, D., Shaw, T., Hearn, L., Epstein, M., Monks, H., Lester, L., et al. (2009). Australian covert bullying prevalence study (ACBPS). Perth: Child Health Promotion Research Centre, Edith Cowan University.

Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of the concept, state of the evidence. Review of Educational Research, 74(1), 59-109.

Fullarton, S. (2002). Student engagement with school: Individual and school-level influences. Melbourne: Australian Council for Educational Research.

Goodman, R. (1997). The strengths and difficulties questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581-586.

Goodman, R. (2001). Psychometric properties of the strengths and difficulties questionnaire. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 1337-1345.

Gottfredson, G. D., Gottfredson, D. C., Payne, A. A., & Gottfredson, N. C. (2005). School climate predictors of school disorder: Results from a national study of delinquency prevention in schools. Journal of Research in Crime and Delinquency, 42(4), 412-444.

Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall.

Independent Schools Council of Australia. (2008). Factors affecting school choice. Canberra: Author.

Karcher, M. J., & Finn, L. (2005). How connectedness contributes to experimental smoking among rural youth: Developmental and ecological analyses. Journal of Primary Prevention, 26(1), 25-36.

Klem, A. M., & Connell, J. P. (2004). Relationships matter: Linking teacher support to student engagement and achievement. Journal of School Health, 74(7), 262-273.

Kuperminc, G. P., Leadbeater, B. J., & Blatt, S. J. (2001). School social climate and individual differences in vulnerability to psychopathology among middle school students. Journal of School Psychology, 39(2), 141-159.

Ladd, G. W., Kochenderfer, B. J., & Coleman, C. C. (1996). Friendship quality as a predictor of young children's early school adjustment. Child Development, 67, 1103-1118.

Lamb, S., Rumberger, R., Jesson, D., & Teese, R. (2004). School performance in Australia: Results from analyses of school effectiveness. Melbourne: Centre for Post-compulsory Education and Lifelong Learning, University of Melbourne.

Lee, V. E., & Smith, J. B. (1999). Social support and achievement for young adolescents in Chicago: The role of school academic press. American Educational Research Journal, 36(4), 901-945.

Masters, G. N. (2004). Public or private: Where's the evidence? Melbourne: Australian Council for Educational Research. Retrieved 11 June 2009 from http://www.acer.edu.au/ enews/0408_PublicOrPrivate.html

McKenzie, F. D., & Richmond, J. B. (1998). Linking health and learning: An overview of coordinated school health programs. In E. Marx, S. F. Wooley, & D. Northrop (Eds.), Health is academic: A guide to coordinated school health programs (pp. 1-14). New York, NY: Teachers College Press.

McNeely, C. A., & Falci, C. (2004). School connectedness and the transition into and out of health-risk behaviour among adolescents: A comparison of social belonging and teacher support. Journal of School Health, 74(7), 284-293.

McNeely, C. A., Nonnemaker, J. M., & Blum, R. W. (2002). Promoting school connectedness: Evidence from the National Longitudinal Study of Adolescent Health. Journal of School Health, 72(4), 138-146.

Ministerial Council on Education, Employment, Training and Youth Affairs. (MCEETYA) (2005). National report on schooling in Australia 2005. Canberra: Author.

Moos, R. H. (1979). Evaluating educational environments. San Francisco CA: Jossey-Bass.

Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., Jones, J., et al. (1997). Protecting adolescents from harm: Findings from the National Longitudinal Study on Adolescent Health. Journal of the American Medical Association, 278(10), 823-832.

Resnick, M. D., Harris, L. J., & Blum, R. (1993). The impact of caring and connectedness on adolescent health and well-being. Journal of Paediatric Child Health, 29(Suppl. 1), S3-S9.

Ryan, A. M., & Patrick, H. (2001). The classroom social environment and changes in adolescents' motivation and engagement during middle school. American Educational Research Journal, 38(2), 437-460.

Shochet, I. M., Dadds, M. R., Ham, D., & Montague, R. (2006). School connectedness is an underemphasized parameter in adolescent mental health: Results of a community prediction study. Journal of Clinical Child & Adolescent Psychology, 35(2), 170-179.

Shochet, I. M., Homel, R., Cockshaw, W. D., & Montgomery, D. T. (2008). How do school connectedness and attachment to parents interrelate in predicting adolescent depressive symptoms? Journal of Clinical Child & Adolescent Psychology, 37(3), 676-681.

Sieving, R. E., Beuhring, T., Resnick, M. D., Bearinger, L. H., Shew, M., Ireland, M., et al. (2001). Development of adolescent self-report measures from the National Longitudinal Study of Adolescent Health. Journal of Adolescent Health, 28(1), 73-81.

StataCorp. (2007). STATA/IC 10.0 for Windows. Texas: Author.

Waters, S., Cross, D., & Runions, K. (2009). Social and ecological structures supporting adolescent connectedness to school: A theoretical model. Journal of School Health, 79(11), 516-524.

Waters, S., Cross, D., & Shaw, T. (in press). Does the nature of schools matter? An exploration of selected school ecology factors on adolescent perceptions of school connectedness. British Journal of Educational Psychology.

Weich, S., Blanchard, M., Prince, M., Burton, E., Erens, B., & Sproston, K. (2002). Mental health and the built environment: Cross-sectional survey of individual and contextual risk factors for depression. British Journal of Psychiatry: The Journal of Mental Science, 180, 428-433.

WestEd. (2008). California health kids survey: California school district secondary school survey results fall 07/spring 08. Los Alamitos, CA: Author.

Wingspread. (2004). Wingspread declaration on school connections. Journal of School Health, 74(7), 233-234.

Stacey K. Waters

Donna Cross

Therese Shaw

Edith Cowan University

Stacey Waters is a Lecturer in the Public Health School of Exercise, Biomedical and Health Sciences and a doctoral candidate at the Child Health Promotion Research Centre, at Edith Cowan University. Email: s.waters@ecu.edu.au

Donna Cross is Professor and Director of the Child Health Promotion Research Centre at Edith Cowan University.

Therese Shaw is a Research Fellow at the Child Health Promotion Research Centre at Edith Cowan University.

Table 1 Student-level variables (Year 8) by sector

                                                  SECTOR

                                          Government    Catholic

                                            n     %      n      %

Gender (female)                            904   51.2   1990   0.51
Transition (not difficult)                1339   88.6   2730   76.9
Academic achievement (avg or above)       1411   90.6   3230   92.9

                                          M      SD     M      SD

Connectedness to school (Yr 9) (0-3) *    1.8    0.77   2.2    0.76
Connectedness to family (1-5) *           4.3    0.65   4.6    0.62
Connectedness to teachers (0-3) *         1.7    0.73   2.0    0.79
Perception of peer support (0-2) *        1.5    0.46   1.3    0.35
Pro-social skills (0-2) *                 1.4    0.40   1.6    0.40
Classroom management climate              1.7    0.96   1.4    1.10
  (0-4) ([dagger])
Emotional problems (0-2) ([dagger])       0.6    0.46   0.5    0.47
Conduct problems (0-2) ([dagger])         0.5    0.43   0.3    0.39
Hyperactivity (0-2) ([dagger])            0.8    0.51   0.7    0.52
Peer problems (0-2) ([dagger])            0.4    0.35   0.3    0.33

                                             SECTOR

                                          Independent

                                           n     %

Gender (female)                           245   44.1
Transition (not difficult)                438   88.7
Academic achievement (avg or above)       465   92.6

                                          M     SD

Connectedness to school (Yr 9) (0-3) *    2.0   0.73
Connectedness to family (1-5) *           4.5   0.55
Connectedness to teachers (0-3) *         1.9   0.71
Perception of peer support (0-2) *        1.6   0.43
Pro-social skills (0-2) *                 1.5   0.38
Classroom management climate              1.6   0.82
  (0-4) ([dagger])
Emotional problems (0-2) ([dagger])       0.5   0.43
Conduct problems (0-2) ([dagger])         0.3   0.35
Hyperactivity (0-2) ([dagger])            0.7   0.51
Peer problems (0-2) ([dagger])            0.3   0.32

* Higher score represents more highly connected, supported or skilled

([dagger]) Higher score represents more problems

Table 2 School characteristics by sector

                                           Total          Catholic

                                         M       SD       M      SD
Structural characteristics
School SES                             103.2    7.93    101.5   6.51
School size                            1027.3  402.11   954.9  326.95
Grade 9 reading score                  559.2   59.02    592.8  27.87
  (range: 428-654)
Grade 9 writing score                  547.6   32.94    549.4  24.57
  (range: 437-597)
Grade 9 numeracy score                 507.7   36.03    496.1  18.69
  (range: 445-558)
Harshness of discipline policy          5.6     0.44     5.6    0.42
  (0-6)
Class size                              29.2    2.7l    29.6    2.24

Proportion
Number of Year levels in school         25%              24%
  (1=8-12; 2=K-I2)

Built environment
Condition of campus                     60%     34%      64%    36%
  (2=poor; 3=good; 4=very good)
Presence of graffiti                    47%              37%
  (0=none; 1=some)
Equity of access to equipment           35%     50%      46%    45%
  (1=none; 2=rostered; 3=all access)
Equity of access to shelter             28%     46%      20%    53%
  (1=some; 2=most; 3=shelter for all)
Activities available at break times     36%     32%      45%    32%
  (1=none; 2=rostered; 3=all access)

Functional characteristics
Teachers with master's degree           62%              74%
  (1=none/few; 2=some/many)
Teachers in first two years of          40%              39%
  teaching(1=none/few; 2=some/many)
Priority for pastoral care              95%             100%
  (1=moderate; 2=high)
Homeroom structure                      5l%              52%
  (1=horizontal; 2=vertical)
House/faction system                    80%              93%
 (0=no; 1=yes)
Pastoral care period                    85%              89%
  (0=no; 1=yes)
Student involvement in decision         44%      8%      47%    13%
  making (1=occasionally;
  2=sometimes; 3=regularly)

                                       Government       Independent

                                         M       SD       M       SD
Structural characteristics
School SES                             103.9    7.05    113.5   10.97
School size                            1151.0  546.70   1136.0  88.61
Grade 9 reading score                  481.5   32.95
  (range: 428-654)
Grade 9 writing score                  543.6   46.63
  (range: 437-597)
Grade 9 numeracy score                 534.6   49.5l
  (range: 445-558)
Harshness of discipline policy          5.5     0.34     5.6     0.69
  (0-6)
Class size                              29.2    3.38     26.2    l.35

Proportion
Number of Year levels in school          5%              100%
  (1=8-12; 2=K-I2)

Built environment
Condition of campus                     61%     14%      24%     76%
  (2=poor; 3=good; 4=very good)
Presence of graffiti                    87%               0%
  (0=none; 1=some)
Equity of access to equipment           2l%     50%       0%     76%
  (1=none; 2=rostered; 3=all access)
Equity of access to shelter             38%     31%      59%     4l%
  (1=some; 2=most; 3=shelter for all)
Activities available at break times     13%     44%      48%      0%
  (1=none; 2=rostered; 3=all access)

Functional characteristics
Teachers with master's degree           18%              100%
  (1=none/few; 2=some/many)
Teachers in first two years of          32%              70%
  teaching(1=none/few; 2=some/many)
Priority for pastoral care              79%              100%
  (1=moderate; 2=high)
Homeroom structure                      45%              59%
  (1=horizontal; 2=vertical)
House/faction system                    39%              100%
 (0=no; 1=yes)
Pastoral care period                    69%              100%
  (0=no; 1=yes)
Student involvement in decision         23%      0%     82.5%     0%
  making (1=occasionally;
  2=sometimes; 3=regularly)

Note: Italicised range indicates non-referent category and its
proportion.