The Role of Adolescent Social Inclusion in Educational Attainment Among Vulnerable Youth

Abstract

Background

Completing high school enables access to educational and employment opportunities associated with better physical and mental health, and improved quality of life. Identifying modifiable factors that promote optimal educational trajectories for youth experiencing disadvantage is an important research focus. Social inclusion has been theorised to play a role in promoting better educational outcomes for this priority population, however limited research has examined this relationship.

Method

This study used three waves of data from the state-representative Australian arm of the International Youth Development Study (youngest cohort, N = 733; 54% female, 95% Australian born) to examine the extent to which vulnerability in primary school (Grade 5; M_age = 10.97, SD = 0.38), and social inclusion in mid-adolescence (Year 10; M_age = 15.50, SD = 0.53), were associated with school completion in young adulthood (post-secondary; M_age = 19.02, SD = 0.43).

Results

Regression models identified an interaction between social inclusion and vulnerability (OR = 1.37, 95% CI [1.06, 1.77], p = .016), indicating that the association between vulnerability and school completion varied as a student’s level of social inclusion increased. Higher social inclusion was beneficial for youth with lower levels of vulnerability but did not appear to influence school completion for the most vulnerable students.

Conclusions

For many young people, promoting social inclusion may support engagement in education and play a protective role. However, further research is needed to better understand the role of social inclusion for highly vulnerable youth, particularly the mechanisms via which social inclusion may have differential effects on school completion.

Young people experiencing disadvantage due to structural and economic factors often have limited access to resources and opportunities in life and may struggle to successfully navigate pathways from school to employment and financial stability. In young adulthood this can have adverse impacts on people’s living conditions, health equity, mental health, and quality of life, further maintaining and potentially compounding social inequities into later life (Marmot et al., 2008). Identifying factors that buffer young people from the effects of disadvantage may redress this imbalance and improve developmental trajectories for young people experiencing disadvantage, including their educational and employment outcomes.

Disadvantage arises through social inequity due to a combination of structural, environmental, and historical factors (e.g., colonisation, racism, governmental policy) (Marmot, 2004). Current discourse in the public health sector on disadvantage and vulnerability terminology highlights concerns regarding usage of these terms as they focus on deficits and lack of autonomy (Katz et al., 2020). It is therefore important that systemic factors which contribute to vulnerability are incorporated into conceptualisations of these terms. Structural factors such as the social and economic determinants of health (e.g., age, gender, education, and geography), and an individual’s exposure to risk factors, arise as a result of systems and policies that perpetuate inequities, leading to unequal access to resources and opportunities (Bronfenbrenner & Morris, 2006; Goldfeld et al., 2018). Evidence suggests that disadvantage affecting children occurs on a continuum, whereby more disadvantaged youth often experience poorer physical and mental health outcomes, as well as lower educational attainment (Marmot et al., 2008; Goldfeld et al., 2018; Renner et al., 2023a). As such, to promote optimal educational outcomes in youth experiencing disadvantage, alongside the imperative to redress structural and systemic causes, it is critical to target factors likely to buffer the effects of disadvantage by recognising the agency of individuals and strengthening protective factors.

Social inclusion is one factor that has been theorised to potentially modify the impact of exposure to disadvantage on young people. Social inclusion can be defined as feeling included in society, or a sense of belonging to a group, whilst also having access to resources and opportunities (Abrams, Hogg, & Marques, 2005). Social inclusion encompasses several key domains including: (a) participation, (b) connectedness and a sense of belonging, and (c) citizenship and rights (Gidley et al., 2010; Cordier et al., 2017). Together, these domains reflect an individual’s involvement in activities and society, the benefits of social networks and engagement with others, and the civic experience of an individual’s societal rights and obligations. These domains intersect across multiple bioecological levels within our social environments including individual, peer, family, school, and community levels. The complexity of social inclusion in adolescence is thus best represented through a multidimensional approach to measurement that captures these varying bioecological levels (Magson, Craven, & Bodkin-Andrews, 2014; Moyano et al., 2020).

Despite the availability of two- and three-dimensional scales suitable for assessing social inclusion in adolescence, few studies have employed these scales to examine the relationship of social inclusion with school completion. Notably, a recent Australian study measured two social inclusion domains (citizenship and rights, and connectedness and belonging) across four bioecological levels (individual, family, school, and community), and found that higher levels of social inclusion in mid-adolescence (average age 15 years) were associated with an increased likelihood of completing secondary high school (Renner et al., 2023b). The broader body of evidence also indicates that a range of individual factors closely aligned with social inclusion are positively associated with educational outcomes (Wu, Tsang, & Ming, 2014; Huat See & Gorard, 2015).

One factor that has emerged from the aforementioned evidence as critical to educational outcomes in young people is social capital (Cemalcilar & Gökşen, 2014). Social capital can be defined as access to resources (e.g., through opportunities like mentoring), often through personal relationships, which enable accomplishment and success in life, in areas such as educational attainment and employment (Scales et al., 2022). Parental involvement and access to opportunities through family, school, and community are forms of social capital associated with school retention and educational attainment (Cemalcilar & Gökşen, 2014; Huat See & Gorard, 2015; Taggart, 2018). These factors are also linked to school completion among young people from culturally diverse and migrant backgrounds (Perreira, Harris, & Lee, 2006; Ryabov, 2011; Wu et al., 2014). Moreover, informal mentoring by a teacher, friend, or community member, is associated with increased educational achievement and attainment (Erickson, McDonald, & Elder, 2009).

Importantly, youth experiencing disadvantage commonly have less access to multiple forms of social capital, adversely impacting their educational outcomes. In particular, family-based social capital is likely to influence educational outcomes differentially relative to school poverty level (Fasang, Mangino, & Brückner, 2014). For instance, when parents in high-poverty schools form strong bonds with each other, there is often a lower probability of their children completing school. By contrast, in low-poverty schools, high levels of parental bonding often result in a higher probability of school completion (Fasang et al., 2014). It has been suggested that other factors, such as peer networks and norms, within these social contexts, are likely to determine whether social capital and/or social inclusion have beneficial impacts on educational outcomes. For example, in neighbourhoods where male youth are exposed to violence, peers are likely to be important for physical protection, yet those same peers may influence male youth to become disconnected from school (Rendón, 2014). Overall, the evidence suggests that the impact of social inclusion on educational outcomes may vary depending on the level of disadvantage and social context experienced by a young person.

Taken together, it has been theorised that social inclusion in adolescence is associated with school completion in young adulthood. Yet few studies to date have examined this relationship, nor whether rates of school completion vary as a function of disadvantage. It is also notable that few studies have employed a multidimensional measure of social inclusion, which would enable assessment of the extent to which inclusion across various societal levels is differentially associated with school completion in young people experiencing disadvantage. To address these gaps in the literature, this study aimed to examine the extent to which a two-dimensional measure of adolescent social inclusion, assessing citizenship, and connectedness and belonging, moderated the impact of vulnerability in primary school on Year 12 school completion in a sample of Australian youth. Based on prior literature, it was hypothesised that higher levels of social inclusion in adolescence would be associated with an increased likelihood of Year 12 completion, except among highly vulnerable youth.

Methods

Participants

This study used data from the International Youth Development Study (IYDS), a prospective cohort study that commenced in 2002 in two sites: Victoria, Australia and Washington State, United States of America (McMorris et al., 2007). The IYDS investigated risk and protective factors, as well as prosocial and antisocial outcomes, across multiple bioecological domains, with a focus on substance use and delinquency. A two-stage cluster sampling approach at the suburb and school level was used to recruit a representative sample of students across grades 5, 7, and 9. In Australia, ethics approval was granted for the IYDS by the Royal Children’s Hospital Human Ethics Research Committee; see McMorris et al. (2007) for further detail on the IYDS design and sampling methodology.

The current study used data from Victoria, Australia. This arm of the study involved ten waves of data collection, from 2002 through to 2018/2019. The current study focused on the primary school (youngest) cohort across three waves of data collection: wave 1 (n = 927; average age 11 years; Grade 5), wave 5 (n = 825; average age 16 years; Year 10), and wave 7 (n = 809; average age 19 years; post-secondary). Informed consent was provided by parents and participants in waves 1 and 5, and by participants (young adults) in wave 7 (McMorris et al., 2007; Evans-Whipp et al., 2013). Data was collected via a parent telephone interview in wave 1 and youth self-report surveys in waves 1 and 5, using an Australian adaptation of the Communities That Care youth survey, administered during class time at school. As respondents had finished school at wave 7, surveys were completed via paper or online self-report surveys, or by telephone interview. Only participants who participated in all three waves (1, 5, and 7), and responded to the school completion item, were included in the analysis (n = 733). Table 1 outlines the sample characteristics.

Table 1.

IYDS Sample Size, Distribution, Key Demographics, and Retention Rates for the Primary School Cohort in Victoria, Australia

Sample	Year	n	Age
			M	SD
Wave 1 (baseline)	2002	927	10.98	0.40
Wave 5	2007	825	15.99	0.39
Wave 7	2010	809	19.03	0.44
Final sample	2010	733	19.02	0.43
Excluded a		75
Lost to follow-up between waves 1 and 7		119
Demographics for final sample (n = 733)
Age
Wave 1	2002	733	10.97	0.38
Wave 5	2007	733	15.50	0.53
Wave 7	2010	733	19.02	0.43
Measured at wave 1		n	%
Gender		733
Male		335	45.7
Female		398	54.3
Country of birth		706
Australia		668	94.6
Country other than Australia		38	5.4
Aboriginal or Torres Strait Islander		4	0.6
Household income by quintiles b		568
Less than $30,000 (lowest quintile)		131	23.1
Between $30,001 and $50,000 (second)		126	22.2
Between $50,001 and $70,000 (third)		123	21.7
Between $70,001 and $90,000 (fourth)		105	18.5
Over $90,000 (highest quintile)		83	14.6
Retention in all three waves (final sample) c			79.1

Note.

^aExcluded represents participants not included in the final analysis due to: (a) missing from wave 5 but participated in wave 7 (n =64); and (b) present in each wave but did not respond to the school completion item (n = 11).

^bHousehold income is in Australian currency (AUD).

^cParticipants included in final analysis.

Measures

Disadvantage/Vulnerability

Disadvantage was assessed via a 13-item measure of vulnerability that captured key social inequities relevant to disadvantage (Goldfeld et al., 2018; Renner et al., 2023a). This measure, collected at wave 1 (baseline) in 2002 (Grade 5), was based on a prior latent class analysis in the IYDS by the authors (Renner et al., 2023a). This analysis identified vulnerability as a three-dimensional construct (sociodemographic, geographical, and risk factor domains) differentiated across four categories: 1 = low, equivalent to a low level of exposure to disadvantage; 2 = normative; 3 = welfare; and 4 = high, equivalent to a high level of exposure to disadvantage (see Appendices A and B in supplementary material for item details). The low vulnerability category was characterised by sociodemographic advantage, with participants who were more likely to live in higher SES urban areas, and have higher levels of parental income, education, and employment. The normative category comprised participants with moderate levels of parental income and education. Next, the welfare category comprised participants with low levels of parental income, education, and employment, and high levels of welfare support. Finally, the high vulnerability category was characterised by high levels of familial and community risk (i.e., family antisocial behaviour and conflict, and unsafe neighbourhood environments), low parental education, and moderate welfare involvement. Cronbach’s alpha for the 13-item vulnerability scale was α = .59 (Hair et al., 2014).

Social Inclusion

The 21-item social inclusion measure, drawn from prior factor analyses in the IYDS by the authors (Renner et al., 2023b), was measured at wave 5 in 2007 (Year 10 – final year of compulsory school in Australia). This measure was operationalised as a latent factor with four sub-factors that aligned with four bioecological levels: (1) Citizenship (six items; individual level), (2) Connectedness to Community (five items), (3) Connectedness to Family (four items), and (4) Connectedness to and Participation in School (six items; see Renner et al., 2023b for item details). Factor scores were used as a continuous measure in the current study ranging from low to high social inclusion, where 1 represented low and 4 represented high social inclusion. Cronbach’s alpha for the 4-factor structure indicated the measure demonstrated good reliability across the four factors (α = .80-.87) and overall reliability (Hair et al., 2014) for the social inclusion measure (α = .75).

School Completion

School completion was measured at wave 7 in 2010, one-year post-secondary school. In Australia, school education is compulsory until the age of 16 or 17 years, or completion of Year 10, varying by state and territory, with Year 12 being the final year of secondary education. The highest year level of secondary school completed was dichotomised: (a) did not complete Year 12 (or equivalent), or (b) completed Year 12 (or equivalent).

Covariates

Covariates measured at waves 1 and 5 were included to improve the precision of the analyses. They covered seven main areas: (a) demographics, (b) ethnicity, (c) substance use, (d) antisocial behaviour, (e) childhood emotional and behavioural problems, (f) parenting problems, and (g) mental health problems, which were selected based on prior literature (Kao & Thompson, 2003; Goldfeld et al., 2018; Gubbels, van der Put, & Assink, 2019). See Appendix C in the supplementary material for more information on these items.

Analytic Approach

The analysis involved four steps. First, descriptive statistics were conducted for each of the key measures and covariates (see Table 2). Analyses included data points at each of the three waves: (1) vulnerability at wave 1; (2) social inclusion at wave 5; and (3) Year 12 completion at wave 7. While some aspects of the vulnerability and social inclusion measures overlapped conceptually, they measured unique concepts and were weakly correlated (r =.14, p < .001).

Table 2.

Descriptive Statistics for Final Sample (N = 733)

Variable	M	SD	Value / range	n	%
Vulnerability a	2.68	0.75	1-4	733
High			1	51	7.0
Welfare			2	209	28.5
Normative			3	395	53.9
Low			4	78	10.6
Missing from wave 1 (N = 927)				194
Social inclusion b	2.52	1.11	1-4	733
First quartile (low social inclusion)			1	175	23.9
Second quartile			2	188	25.6
Third quartile			3	185	25.2
Fourth quartile (high social inclusion)			4	185	25.2
Missing from wave 5 (N = 825)				92
Educational outcome c
Completed Year 12 (highest year of secondary school)	0.78	0.41		733
No			0	159	21.7
Yes			1	574	78.3
Missing from wave 7 (N = 809)				76
Covariates
Gender a	1.54	0.50		733
Male			1	335	45.7
Female			2	398	54.3
Parental ethnic background a	0.81	0.39		704
Non-Australian			0	132	18.75
Australian			1	572	81.25
Current alcohol use (past 30 days) b	0.67	0.47		732
No			0	245	33.5
Yes			1	487	66.5
Perceived availability of drugs in the community b	2.16	0.83	1-4	722
Perceived rewards for antisocial behaviour b	2.29	0.91	1-5	727
Parental attitudes favourable to antisocial behaviour b	1.34	0.47	1-4	728
Childhood concentration/attention problems b	2.64	0.80	1-4	726
Sensation seeking b	2.67	1.34	1-6	727
Parental overcontrol b	2.10	0.81	1-4	728
Depression symptomology b	1.60	0.52	1-3	728

Note. Means (M) and standard deviations (SD) are not standardised.

^aMeasured at wave 1.

^bMeasured at wave 5.

^cMeasured at wave 7.

Second, the direction of coding for vulnerability was reversed to improve interpretability of the interaction, that is, from low protection (high vulnerability/low social inclusion) to high protection (low vulnerability/high social inclusion). Due to lower respondent numbers in the low and high vulnerability categories (n = 78 and 51, respectively), both the vulnerability and social inclusion measures were used as continuous variables in this analysis (see Appendix D). Coding for the continuous vulnerability measure therefore ranged from high vulnerability (1) to low vulnerability (4). Following recoding, logistic regression analyses were conducted in StataBE 17.0 (StataCorp, 2021) to investigate the effect of the interaction between vulnerability and social inclusion on Year 12 completion. Clustering effects by school were accounted for using the ‘syvset’ command and ‘svy’ prefix in StataBE 17.0. The analysis adjusted for two covariates measured at wave 1 and eight covariates measured at wave 5.

Third, predictive margins were calculated to examine interaction effects for each category of vulnerability. The ‘marginsplot’ command in StataBE 17.0 was used to plot the vulnerability and social inclusion interaction and explore the association for different vulnerability categories. For each category of vulnerability, an examination of whether the slope of social inclusion on school completion was significantly different from zero was conducted. This was done using the margins ‘dydx’ command in StataBE 17.0 (StataCorp, 2021) to test the statistical significance of the slope. Finally, as sample attrition was higher in the high vulnerability category and the low social inclusion level, sensitivity analyses were conducted to test patterns of missing data via a logistic regression of missingness. A multiple imputation analysis (K=10) was also undertaken to examine whether missing data was systematically impacting the regression results.

Results

Sample Characteristics

The sample consisted of slightly more females than males (only genders measured; 54% female), was predominantly Australian born (95%), and was normally distributed across quintiles of household income (see Table 1). The retention rate in all three waves for the youngest cohort was high (79%), however it was disproportionately lower for Aboriginal and Torres Strait Islander participants (33%), with only four participants in the analysis (0.5% of the final sample). Rates of retention also differed by gender, with a higher percentage of females retained in the final sample (female 83% compared to male 75%).

Overall the sample had moderate levels of vulnerability and social inclusion, and moderately low levels of risk factors (see Table 2). The majority of the sample (78%) completed Year 12, however this varied across the vulnerability categories and by social inclusion level (see Table 3). Specifically, rates of school completion were lower in more vulnerable participants and higher in less vulnerable participants. Conversely, rates of school completion were higher in participants with higher levels of social inclusion and lower in participants with lower levels of social inclusion. Table 3 presents the frequencies and percentages for school completion by vulnerability category and social inclusion level.

Table 3.

Frequencies for Wave 7 Sample (2010; N = 733), Year 12 School Completion by Predicted Class of Vulnerability and by Predicted Level of Social Inclusion

Year 12 school completion (or equivalent)	Total		Vulnerability class								Social inclusion level (quartile)
			Low		Normative		Welfare		High		First (lowest)		Second		Third		Fourth (highest)
	n	%	n	%	n	%	n	%	n	%	n	%	n	%	n	%	n	%
Completed Year 12	574	78.3	75	96.2	328	83.0	141	67.5	30	58.8	117	66.9	143	76.1	153	82.7	161	87.0
Did not complete Year 12	159	21.7	3	3.8	67	17.0	68	32.5	21	41.2	58	33.1	45	23.9	32	17.3	24	13.0
Final sample	733	100	78	10.6	395	53.9	209	28.5	51	7.0	175	23.9	188	25.6	185	25.2	185	25.2
Excluded a	75	8.1
Lost to follow-up b	119	12.8

Note.

^aRefers to participants who were: (a) missing in wave 5 but participated in wave 7 (n = 64), and (b) present in each wave but did not respond to the school completion item (n = 11).

^bRefers to participants who were: (a) missing only from wave 7 (n = 83), and (b) missing from both wave 5 and 7 (n = 36).

Relationship with School Completion

Table 4 shows the non-imputed and imputed results of the logistic regression model for Year 12 school completion. An interaction between social inclusion and vulnerability was observed (OR = 1.37, 95% CI [1.06, 1.77], p = .016), indicating that as inclusion increased by one unit for each unit of vulnerability (which was reverse coded to indicate decreasing vulnerability), the odds of school completion increased by 37%, on average.

Table 4.

Logistic Regression and Interaction of Vulnerability and Social Inclusion on School Completion, Accounting for Clustering at School Level, and Adjusting for Covariates

Variable	Non-imputed (N = 733)			Imputed (N = 825)
	OR	95% CI		OR	95% CI
		LL	UL		LL	LL
Vulnerability	1.57*	1.08	2.30	1.58*	1.10	2.26
Social inclusion	0.86	0.57	1.31	0.96	0.66	1.41
Vulnerability x social inclusion	1.37*	1.06	1.77	1.26*	1.00	1.59
Covariates
Gender – female a	1.42	0.93	2.18	1.18	0.79	1.75
Parental ethnic background – non-Australian a	2.89**	1.52	5.50	2.09**	1.34	3.27
Current alcohol use (past 30 days) b	0.44**	0.27	0.72	0.51**	0.32	0.81
Perceived rewards for antisocial involvement b	1.12	0.88	1.43	1.05	0.83	1.32
Parental attitudes favourable toward antisocial behaviour b	0.88	0.57	1.36	0.83	0.56	1.24
Perceived availability of drugs in community b	1.10	0.83	1.46	1.17	0.90	1.51
Childhood behaviour problems, concentration/attention b	0.77	0.57	1.03	0.80	0.61	1.05
Sensation seeking b	0.97	0.81	1.16	0.96	0.81	1.12
Parental overcontrol b	1.28	0.99	1.67	1.31*	1.03	1.67
Depression symptomology b	1.11	0.26	4.78	1.11	0.74	1.67

^aMeasured at wave 1.

^bMeasured at wave 5.

^*p ≤ .05.

^**p < .01.

Figure 1 provides a visual plot of the predictive margins for the interaction between social inclusion and vulnerability on school completion (see Appendix E). The slope of the line for the association between social inclusion and school completion for both the low vulnerability and normative categories were significantly different from zero (dydx = .06, p < .001; dydx = .06, p = .002, respectively), suggesting there was an associated increase in the likelihood of completing high school. In contrast, the slope for the welfare and high vulnerability categories were not statistically significantly different from zero (dydx = .03, p = .19; dydx = −.03, p = .49, respectively), suggesting social inclusion did not influence school completion for the higher vulnerability categories.

Figure 1.

Diagrammatic Depiction of the Interaction between Vulnerability and Social Inclusion and the Impact on Year 12 Completion (N = 733)

Logistic regression results using imputed values identified a similar yet slightly smaller odds ratio for the vulnerability and inclusion interaction (OR = 1.26, 95% CI [1.00,1.59], p = .05; see Table 4). As the pattern of results was similar in the imputed and non-imputed analyses, the original, non-imputed dataset is presented. Sensitivity analyses also indicated that neither vulnerability nor social inclusion predicted missingness, after adjusting for covariates (OR = 1.04, 95% CI [0.76, 1.43], p = .796 and OR = 1.00, 95% CI [0.78, 1.27], p = .982, respectively; see Appendix F in supplementary material).

Discussion

Prior research suggests that vulnerability and social inclusion are important factors associated with secondary school completion, however research on these factors, and on their potential to break patterns of poor educational outcomes in youth, has been lacking. This study therefore aimed to examine the extent to which social inclusion in adolescence moderated the impact of vulnerability on Year 12 completion, in a state representative sample of Australian youth from primary through to post-secondary school. Results partially supported the hypothesis that higher levels of social inclusion in adolescence would increase the likelihood of Year 12 completion for all youth, except for those experiencing the highest level of vulnerability. That is, social inclusion increased the likelihood of school completion as the level of vulnerability decreased, however was only beneficial for the least vulnerable youth (normative and low vulnerability categories). For youth experiencing sociodemographic disadvantage (welfare category) and youth exposed to high levels of risk (high vulnerability category), social inclusion did not affect the likelihood of completing school; the relationship with social inclusion was not significant for either category. These findings have important implications for the development and implementation of social inclusion intervention efforts, and for prioritising relevant youth populations to improve educational equity.

Study results indicated a small moderating effect of social inclusion on the relationship between vulnerability and school completion, with higher levels of social inclusion associated with an increased likelihood of school completion. Notably, this effect was strongest among youth from more advantaged family backgrounds. Consistent with prior research, higher levels of social inclusion likely support participation and engagement in education and counter exclusionary behaviours that can have negative impacts on academic achievement and subsequent school completion (Emler & Reicher, 2005). Importantly, for youth from culturally diverse backgrounds, social inclusion may promote access to important educational opportunities (Perreira et al., 2006; Ryabov, 2011; Wu et al., 2014), however if sociodemographic vulnerability is present, this may influence the effect of social inclusion. Two of the four categories of vulnerability appeared to benefit from increasing social inclusion levels (low and normative), suggesting that initiatives that improve social inclusion may increase school completion for a large proportion of young people. Importantly, increasing social inclusion among young people with average levels of family income and education, which makes up the majority of participants in the sample (largest category, n = 395), may help to promote educational opportunities. While participants in this category did not experience high levels of sociodemographic disadvantage, nearly one fifth of these students did not complete secondary school. Hence, social inclusion may facilitate educational opportunities and subsequent pathways into adulthood for a large number of Australian youth.

Whilst social inclusion appeared to buffer the impact of vulnerability for a large proportion of young people in our sample (low and normative; 65%) as hypothesised, this association differed for youth experiencing higher levels of vulnerability. For youth experiencing sociodemographic disadvantage (welfare category) the relationship with social inclusion was not significant. Perhaps for youth in this category, who were categorised predominantly by parental receipt of welfare support, social inclusion is insufficient to offset the impact of the structural and systemic factors that contribute to disadvantage. There is some evidence to suggest that the interaction between school-level poverty and parental social connections established within school settings reduces educational attainment in high-poverty schools, with the opposite effect in low-poverty schools (Fasang et al., 2014). This suggests that the impact of social inclusion on school completion may differ according to the nature and extent of resources available within social networks, such that the benefits of social inclusion are contingent on access to high-quality resources.

Similarly, there was not a significant association with social inclusion for those in the high vulnerability category. This category was characterised by high levels of risk factors at both the family and community level. Exposure to community disorganisation, family antisocial behaviour, and family conflict, were key characteristics of this category. In addition, low parental education and high levels of welfare support (nearly 50%) reflected the structural complexities of disadvantage for this category of vulnerable young people. A combination of exposure to familial antisocial behaviour and living in an unsafe neighbourhood with high rates of crime and violence may increase the likelihood of socialising with antisocial adults and peers. For example, in neighbourhoods with high population density, crime, violence, and drug use, youth frequently socialise with peers and adults in the community likely to model antisocial behaviours and early school leaving (Evans-Whipp et al., 2013; Toumbourou et al., 2015). Identifying as part of, or feeling included in, an antisocial peer group, may in turn increase early school leaving through pressure to adhere to social norms that reject school-level authority, and may influence youth to enter the workforce early rather than complete high school (Emler & Reicher, 2005). This has implications for initiatives aimed at increasing social inclusion, and highlights a need for further research to explore the impact of prosocial peers and community members, specifically for youth experiencing high levels of vulnerability.

Strengths and Limitations

This study used data from a state representative cohort study with comprehensive assessment of youth development spanning childhood, adolescence, and young adulthood (McMorris et al., 2007). Whilst this study used rigorous methodology, there are several limitations. First, there may be reporting bias due to the self-report nature of the survey, however given the survey was confidential and reporting was de-identified, and the scales are reliable and validated in adolescent populations, the impact of this bias is likely to be minimal. Second, the reliability of the vulnerability scale was not optimum, indicating possible inconsistencies between items. Third, the vulnerability and social inclusion measures were categorical, but due to small samples in the low and high vulnerability categories, both variables were used as continuous in this study. In particular, examination of the interaction for the high vulnerability category was limited due to low cell counts. As such, results for the higher vulnerability categories in this study are likely to represent conservative estimates. Future research using a larger sample of youth experiencing high vulnerabilty would improve understanding of the interaction between social inclusion and disadvantage, and the subsequent impact on school completion.

Implications

The results of this study are informative for educational psychologists, and school leadership and wellbeing teams, to inform initiatives designed to promote school completion outcomes in young people. Implementing universal programs to increase social inclusion may be a way to improve school completion for many students, however these programs may be most beneficial for the least vulnerable and potentially perpetuate the advantages of those who are more privileged. For youth experiencing high levels of vulnerability, social inclusion does not improve school completion and therefore developing specific targeted interventions that focus on building prosocial connections to school and community may better support this priority group. Further research is needed among highly disadvantaged young people to explore the relationship with social inclusion, to better address social inequities.

Importantly, results indicate that alongside inteventions aimed at promoting better educational outcomes via schools, communities, young people, and their families, it is essential to target structural, economic, and systemic factors that perpetuate exposure to disadvantage (Marmot et al., 2008). Ensuring stakeholders are engaged across contexts, and designing initiatives that extend beyond social inclusion, is critical to breaking the cycle of disadvantage in vulnerable communities.

Boosting social inclusion may also benefit young people beyond their educational outcomes. For youth from culturally diverse or First Nations backgrounds, improving social inclusion may help buffer the effects of discrimination and prejudice and the associated inequities (Perreira et a., 2006; Thomas & Griffin, 2021). Future research examining the effect of social inclusion on mental health and wellbeing, with a particular focus on the benefits of cultural and ethnic capital, among First Nations youth and for youth from culturally diverse backgrounds, is an important next step.

Conclusion

Using multidimensional measures of vulnerability and social inclusion, this study found the relationship between vulnerability and school completion in young people varied relative to the level of social inclusion experienced in adolescence. Specifically, results indicated that among most young people with lower levels of vulnerability, social inclusion in adolescence was associated with a greater likelihood of school completion in young adulthood. In contrast, for youth experiencing higher levels of vulnerability, social inclusion did not appear to influence the likelihood of school completion. While these results suggest promoting social inclusion may be beneficial for many youth, caution should be exercised for youth experiencing high levels of vulnerability, with additional research needed to explore the factors influencing this relationship. These findings suggest adolesence may be a point of intervention to modify life trajectories, however, for youth exposed to structural and enduring disadvantage, additional methods are needed to improve employment pathways and quality of life.

What is known?

• Youth disadvantage has a negative impact on physical and mental health, education, employment, and subsequent life trajectories.

• Social inclusion may moderate the effect of disadvantage by improving school completion, however few longer-term studies have examined this relationship.

What is new?

• Social inclusion in mid-adolescence was associated with subsequent school completion in young people characterised by low to moderate levels of vulnerability.

• This protective effect was not observed among youth characterised by higher levels of vulnerability.

What is significant for clinical practice?

• Increasing and enabling social inclusion may improve school completion in many young people, with the potential for lifelong benefits in employment opportunities and later health and wellbeing.

• Universal implementation of social inclusion programs should be done prudently.

• Further research with larger samples of youth experiencing high levels of vulnerability is needed to understand the impact of social inclusion on educational outcomes in this priority group.