Interventions to Improve Connectedness, Belonging, and Engagement in Secondary Schools: A Systematic Review and Meta-Analysis

Abstract

School connectedness, belonging, and engagement are linked to improved academic, social, and emotional outcomes, yet no prior systematic review or meta-analysis has synthesised interventions targeting these constructs in secondary schools. This review evaluated randomised controlled trials identified through systematic searches of CINAHL, Medline, ERIC, and PsycINFO (last searched: June 2024). Studies included school-based interventions measuring connectedness, belonging, or engagement with validated tools. Methodological quality was assessed via the Cochrane Risk of Bias Tool, and effectiveness was analysed using meta-analysis (Hedges’ g). Sixteen trials (N = 35,451) met the inclusion criteria, with eleven providing sufficient data for meta-analysis. Overall, interventions significantly improved connectedness, belonging, and engagement (Hedges’ g = 1.056, within-group; 0.642, between-group). Multi-tier programs yielded the largest effects (0.781), with ecological (0.613) and environmental (0.636) approaches; behavioural and intrapersonal strategies were non-significant, possibly due to the small sample sizes. Interventions frequently overlapped in content but differed in theoretical frameworks. Future research should develop more distinctive, multi-tiered interventions and use consistent, theoretically robust measures. The findings underscore the potential of targeting environmental factors and interpersonal techniques to foster connectedness, belonging, and engagement in secondary schools. These results highlight the importance of clarifying the conceptual frameworks and measurement tools to strengthen future interventions’ design and interpretation.

1. Introduction

School connectedness, belonging, and engagement are terms often used interchangeably in the literature, and their importance for adolescents’ academic, social, and emotional outcomes is well established (Arslan, 2019; Corso et al., 2013; Wilkins, 2023). Although often discussed using different terminology, all three have been consistently associated with similar outcomes of promoting academic motivation and success (Korpershoek et al., 2020; Li & Lerner, 2013; Wilson, 2018) and protecting against mental health challenges such as depression and anxiety both during adolescence and in adulthood (K.-A. Allen et al., 2024; Archambault et al., 2019; Raniti et al., 2022). While the existing literature suggests that interventions targeting connectedness, belonging, and engagement are essential for improving student outcomes, few studies have systematically synthesised their effectiveness (K.-A. Allen et al., 2022; Chapman et al., 2013). Further, while authors often use these terms interchangeably, no study has looked at multiple constructs simultaneously, missing key opportunities to further understand how these terms are operationalised and practically and theoretically distinctive within intervention research. Moreover, the lack of meta-analyses conducted within the current literature has made it challenging to identify key intervention characteristics that contribute to success in promoting these constructs in students. This systematic review and meta-analysis aims to address these gaps by evaluating randomised controlled trials (RCTs) that target connectedness, belonging, and engagement by comparing the intervention strategies employed and identifying the characteristics contributing to their effectiveness.

1.1. School Connectedness, Belonging, and Engagement: Theoretical Foundations and Interrelatedness

The concepts of school connectedness, belonging, and engagement are deeply intertwined, stemming from both shared and distinct theoretical foundations that can offer unique insights into the factors influencing students’ academic and socio-emotional experiences. Although these terms are also used outside the school environment, this review focuses exclusively on school-based definitions to inform interventions specific to these constructs.

Belongingness has been defined as “…personal involvement (in a social system) to the extent that the person feels (them)self to be an indispensable and integral part of the system” (Anant, 1967, p. 391). A student who ‘belongs’ to the setting is said to generally experience an emotional connection of safety and inclusion when in the school community, which often involves having positive and welcoming relationships with others (Goodenow, 1993a; Goodenow & Grady, 1993). Belonging is strongly linked to the “belongingness hypothesis”, which posits that a sense of belonging is a fundamental human need (Baumeister & Leary, 1995; Hagerty et al., 1992). Within this conceptualisation, schools can fulfil this need by fostering inclusion in the school community.

Connectedness is closely related to belonging due to its shared emphasis on relationships that foster a strong attachment to the school setting. However, connectedness is often rooted in the “relatedness” construct from self-determination theory, with the two terms frequently used interchangeably. Deci and Ryan (Deci & Ryan, 2012, p. 418) defined relatedness as “The need to be close to, trusting of, caring for, and cared for by others.” This perspective links connectedness to intrinsic motivation, which supports improved academic and socio-emotional outcomes (Ramella et al., 2023).

Distinguishing between belonging and connectedness can be challenging, as both emphasise caring relationships and trust. Many authors reference Goodenow’s work on belonging when discussing connectedness, despite Goodenow’s terminology being specific to belonging (Goodenow, 1993a). Both constructs align with a humanistic approach, as they fit within the “love and belonging” dimension of Maslow’s hierarchy of needs (Maslow, 1954). This alignment likely explains why connectedness is often considered the closest synonym to belonging, compared to terms such as “school bonding” or “school community” (Alink et al., 2023).

In contrast, engagement refers to a student’s active participation and involvement in school-related activities, encompassing behavioural, emotional, and cognitive dimensions (Fredricks et al., 2004). Engagement’s three-dimensional model provides a more consistent definition compared to belonging and connectedness. The behavioural domain of engagement focuses on active student participation in school activities (Appleton et al., 2008), while the cognitive domain emphasises self-regulation skills and psychological investment in learning (Pintrich & De Groot, 1990; Zimmerman, 1990). The emotional dimension aligns closely with belonging, emphasising students’ feelings of being valued by the school community and their affective responses to classroom experiences (Boyle & Allen, 2022). Evidence also highlights strong links between school belonging and behavioural engagement, suggesting that multiple aspects of engagement are intertwined with belonging and connectedness (Korpershoek et al., 2020).

To distinguish between these constructs, belonging can be viewed as addressing students’ emotional needs, connectedness as focusing on relational processes, and engagement as the observable motivation and participation that follows. Engagement appears more behavioural in nature, as students can be engaged in activities without necessarily feeling a sense of belonging or connection to their setting. Conversely, low engagement in school activities may stem from a student’s sense of disconnection from the school community, leading to feelings of exclusion (Boyle & Allen, 2022). This suggests that engagement serves as a visible manifestation of underlying belonging and connectedness. Furthermore, the ‘connecting‘ aspect of connectedness can be characterised as the intrinsic motivational process through which, in the presence of supportive environmental conditions, students develop a sense of belonging within the school environment. This distinction positions connectedness as a fundamental process underlying the concept of belonging while concurrently integrating both terms based on their interrelated characteristics. Clarifying these distinctions emphasises the unique contributions each construct offers in crafting narratives about a student’s subjective experiences of school.

The interconnected nature of these constructs has led many qualitative studies to explore student experiences through two or more of these concepts simultaneously, aiming for a more holistic understanding of school contexts (Armstrong-Carter et al., 2023; Cunningham et al., 2024; Fischl et al., 2017; McCabe et al., 2024; Sulimani-Aidan et al., 2021). However, researchers debate whether one construct of connectedness, belonging, or engagement should be regarded as the most fundamental or influential from a theoretical perspective or if they are best understood as equally interdependent and complementary.

1.2. Convergence of Terms and Operationalisation of Constructs

Despite their theoretical differences, there appears to be a convergence in how the terms are defined and operationalised, particularly in intervention research. Self-determination theory, while more naturally associated with connectedness and belonging, has also been recognised as relevant to conceptualising engagement (Skinner et al., 2008). Similarly, all three constructs have been defined along behavioural, emotional, and cognitive domains, which were initially embedded within the engagement literature (K.-A. Allen et al., 2022; Fredricks et al., 2004; Hodges et al., 2018). This may be due to the practical usefulness of these domains when developing and scoping interventions. Further, all three constructs have been defined within an ecological model (K.-A. Allen et al., 2016; Nguyen et al., 2019; Waters et al., 2009). In this model, there is shared importance of a range of student-level and environmental influences across various domains, including peer relations, teacher support, school policy, extended family, and government legislation (K.-A. Allen et al., 2016; Quin et al., 2018). This reflects a broader trend in school intervention research towards systemic approaches that promote multi-faceted intervention designs (Kim et al., 2023; Navarro & Tudge, 2023). When the ecological and/or three-dimensional models are utilised, it often leads to a broader conceptualisation, further blurring the distinctions and highlighting the similarities between the constructs when used to inform intervention designs.

Given these overlaps and parallels, it is unsurprising that there has been long-standing confusion regarding the use of these terms, often resulting in their interchangeable application (Furlong et al., 2003; Libbey, 2004). In the quantitative literature, reviews frequently use identical measurement tools while adapting terminology to align with their narrative. For example, intervention reviews often include connectedness, belonging, and engagement in search terms (K. Allen et al., 2018; Raniti et al., 2022). One review categorised the measures as connectedness (Hodges et al., 2018), with a subsequent review classifying some of the same outcome evaluation tools as belongingness measures (K.-A. Allen et al., 2022). Within these reviews, some of the tools included also relate clearly to engagement terminology in the measures’ names and conceptualisation. These points highlight how the concepts are interpreted and used interchangeably in the literature. While overlaps are acknowledged, few studies clarify how the different terms affect the practical implications of intervention design. Instead, these conceptual overlaps and inconsistencies are often overlooked, with some authors rebranding studies to align with their chosen narratives. This rebranding exacerbates confusion, making it harder to understand how the differences between constructs might influence intervention development and effectiveness.

The definitional barriers present within the literature often result in intervention studies conceptualising the terms “connectedness”, “belonging”, and “engagement” interchangeably in their narrative rationales (Holt et al., 2008; Walsh et al., 2022), or even neglecting to provide a clear definition altogether (Acosta et al., 2019; Espelage et al., 2015; Orthner et al., 2010; Sawyer et al., 2010). This underscores the frequently ambiguous theoretical foundation of school-based interventions, as the terminology pertaining to the interventions’ rationales, the theoretical relationships between the objectives of the intervention and the mechanisms of change, and the chosen measures for assessing their impact lack consistent alignment. This makes it difficult to draw clear conclusions about why interventions do or do not work, hindering the ability of research to provide evidence-based recommendations for schools.

Although these constructs have shown conceptual overlaps, no study has explicitly examined the differences and similarities in the characteristics of interventions targeting each of these constructs. The absence of meta-analytic studies across all three constructs further hampers the ability to draw firm conclusions about their collective effectiveness, especially as previous reviews have struggled to identify specific intervention characteristics associated with greater efficacy, limiting evaluations of key components (K.-A. Allen et al., 2022; Chapman et al., 2013).

Addressing these gaps provides an opportunity for actionable insights into these interrelated constructs, supporting the development of holistic interventions that simultaneously enhance connectedness, belonging, and engagement. Such efforts could promote student wellbeing, academic motivation, and mental health while offering valuable guidance to educators, researchers, and policymakers seeking evidence-based strategies for fostering inclusive school environments. By synthesising the available evidence, this review aims to unite the bodies of work on connectedness, belonging, and engagement, clarifying variations in intervention characteristics and their effectiveness. This comprehensive approach will facilitate the design of interventions that holistically address these critical constructs, enhancing both student outcomes and broader school practices.

1.3. Aims of the Research

This systematic review and meta-analysis aims to address the lack of clarity between the similarities and differences of connectedness-, belonging-, and engagement-related interventions by comprehensively evaluating school-based interventions. By analysing the effectiveness, characteristics, and quality of RCTs, this review aims to provide actionable guidance for educators, researchers, and policymakers. Specifically, this review has the following objectives:

• Identify intervention studies that measured their effectiveness in improving school connectedness, belonging, and/or engagement in secondary schools;
• Describe the characteristics of interventions and compare within and across the constructs of connectedness, belonging, and/or engagement;
• Conduct a meta-analysis to evaluate the overall effectiveness of current interventions in improving school connectedness, belonging, and/or engagement outcomes;
• Conduct subgroup analyses to identify key intervention characteristics that contribute to their success.

2. Methods

This section outlines the systematic approach and criteria used to ensure a rigorous and transparent assessment of school-based interventions targeting connectedness, belonging, and engagement in secondary school students. The study’s methodology and reporting were guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and checklist (Supplementary Materials S1, S2), which provides a framework for the transparent reporting of systematic reviews and meta-analyses that assess intervention effectiveness (Page et al., 2021).

2.1. Eligibility Criteria

To be included in this systematic review and meta-analysis, studies were required to meet the following criteria:

• Participants were children and young people aged 11–18 years.
• The study described a school-based intervention.
• At least one aspect of school connectedness, belonging, or engagement was assessed as a primary outcome using a validated psychometric tool via student self-report.
• The study employed an RCT design with a comparison group, which could include treatment as usual, an alternative intervention, or a waitlist control.
• The published study was written in English.

Studies were required to use validated instruments that measured at least one of the three constructs (connectedness, belonging, or engagement). Measures were considered conceptually eligible if they were identified by Hodges, Cordier (Hodges et al., 2018), who systematically reviewed and classified measurement tools shared by connectedness, belonging, and engagement narratives. Additionally, measures aligning with the conceptual frameworks of Goodenow (Goodenow, 1993b), who operationalised belonging as a sense of psychological school membership, or Fredricks, Blumenfeld (Fredricks et al., 2004), who defined engagement across behavioural, emotional, and cognitive dimensions, were included. These frameworks were chosen because they are widely cited and provide comprehensive theoretical foundations for understanding these constructs in educational settings. Measures were regarded as validated if they demonstrated evidence of factor analysis for construct validity and internal consistency (Hughes, 2018). Additionally, where available, other forms of validation, such as test–retest reliability, criterion validity (convergent and discriminant), and content validity, were considered to ensure the robustness of the psychometric tools (Mokkink et al., 2010). All included studies relied on student self-reports to assess connectedness, belonging, or engagement outcomes.

No restrictions were placed on the intervention format (e.g., interventionist, intervention type), aside from the requirement that interventions be delivered in a school setting to young people aged 11–18. Hybrid interventions, encompassing components based in schools, families, or communities, were deliberately excluded to ensure that the selected interventions can be implemented solely within school environments. The age range of 11 to 18 years was selected because these students are within compulsory school age and the adolescent developmental phase, which is traditionally the target group for connectedness, belonging, and engagement interventions (Fredricks et al., 2004; Goodenow, 1993a). Only original peer-reviewed journal articles of randomised controlled trials were included, while reviews, cohort studies, editorials, conference abstracts, student dissertations, and case reports were excluded.

2.2. Information Sources and Search Strategy

A comprehensive literature search was conducted across four electronic databases, namely, CINAHL, Medline, ERIC, and PsycINFO, covering all publication dates up to 16 June 2024. To capture the full range of the literature on connectedness, belonging, and engagement interventions, a combination of related terms (free text terms), such as “school bonding”, “school climate”, and “school identification”, was used in combination with subject headings (e.g., MeSH and Thesaurus terms).

Table 1. Search strategies per database.

Database and Search Strategies
CINAHL Ultimate: (((MH “Social Inclusion”) OR (MH “Social Participation”) OR (MH “Social Adjustment”) OR (MH “Social Attitudes”) OR (MH “Membership”) OR (MH “Commitment”) OR (MH “Social Involvement (Iowa NOC)”) OR (MH “Social Inclusion”) OR (MH “Student Experiences”) OR (MH “Social Participation”) OR (MH “Student Attitudes”) OR (MH “Social Adjustment”)) OR (TI (((school) AND (connected* OR connection* OR belonging* OR membership* OR bond* OR attachment* OR engage* OR climate* OR communit* OR affiliat* OR commitment* OR involve* OR disconnect* OR accept* OR experience* OR participat* OR orientat* OR identit* OR relatedness OR quality OR culture OR environment OR inclusion* OR “student attitude*” OR pride* OR value*))) OR AB (((school) AND (connected* OR connection* OR belonging* OR membership* OR bond* OR attachment* OR engage* OR climate* OR communit* OR affiliat* OR commitment* OR involve* OR disconnect* OR accept* OR experience* OR participat* OR orientat* OR identit* OR relatedness OR quality OR culture OR environment OR inclusion* OR “student attitude*” OR pride* OR value*)))) AND (MH “Randomized Controlled Trials”) Limits: English Language; Age: adolescent: 13–18 years
Eric (Proquest): ((DE “Group Membership” OR DE “Group Experience” OR DE “Learner Engagement” OR DE “Educational Environment” OR DE “Classroom Environment” OR DE “School Community Relationship” OR DE “School Involvement” OR DE “Student Participation” OR DE “Peer Acceptance” OR DE “Inclusion” OR DE “Early Experience” OR DE “Educational Experience” OR DE “Group Experience” OR DE “Learning Experience” OR DE “Social Experience” OR DE “Student Experience” OR DE “School Involvement” OR DE “Student Participation” OR DE “Student Attitudes” OR DE “School Attitudes” OR DE “Student Adjustment” OR DE “Student School Relationship”) OR title(((school) AND (connected* OR connection* OR belonging* OR membership* OR bond* OR attachment* OR engage* OR climate* OR communit* OR affiliat* OR commitment* OR involve* OR disconnect* OR accept* OR experience* OR participat* OR orientat* OR identit* OR relatedness OR quality OR culture OR environment OR inclusion* OR “student attitude*” OR pride* OR value*))) OR abstract(((school) AND (connected* OR connection* OR belonging* OR membership* OR bond* OR attachment* OR engage* OR climate* OR communit* OR affiliat* OR commitment* OR involve* OR disconnect* OR accept* OR experience* OR participat* OR orientat* OR identit* OR relatedness OR quality OR culture OR environment OR inclusion* OR “student attitude*” OR pride* OR value*)))) AND (RCT OR (Randomized AND Controlled AND Trial) OR (Randomised AND Controlled AND Trial) OR (Randomized AND Clinical AND Trial) OR (Randomised AND Clinical AND Trial) OR (Controlled AND Clinical AND Trial)) Limits: English, Scholarly journals, Secondary Education OR Middle Schools OR High Schools OR Junior High Schools OR Grade 7 OR Grade 8 OR Grade 9 OR Grade 10 OR Grade 11 OR Grade 12
Medline Ovid: ((school and (connected* or connection* or belonging* or membership* or bond* or attachment* or engage* or climate* or communit* or affiliat* or commitment* or involve* or disconnect* or accept* or experience* or participat* or orientat* or identit* or relatedness or quality or culture or environment or inclusion* or student attitude* or pride* or value*)).ti. or (school and (connected* or connection* or belonging* or membership* or bond* or attachment* or engage* or climate* or communit* or affiliat* or commitment* or involve* or disconnect* or accept* or experience* or participat* or orientat* or identit* or relatedness or quality or culture or environment or inclusion* or student attitude* or pride* or value*)).ab.) AND ((RCT or (Randomized and Controlled and Trial) or (Randomised and Controlled and Trial) or (Randomized and Clinical and Trial) or (Randomised and Clinical and Trial) or (Controlled and Clinical and Trial)).ti. or (RCT or (Randomized and Controlled and Trial) or (Randomised and Controlled and Trial) or (Randomized and Clinical and Trial) or (Randomised and Clinical and Trial) or (Controlled and Clinical and Trial)).ab.) Limits: English language; “adolescent (13 to 18 years)”

presents an overview of the complete database-specific search terms. No restrictions were applied to publication dates, though age limits were imposed to focus on studies involving secondary school-aged students. Additional publications were sourced by screening previous connectedness, belonging, and engagement intervention reviews and reference lists from the included studies up to 24 June 2024.

2.3. Selection Process

The referencing software EndNote version 20 identified duplicate records during the importation and merging process of records from the listed databases. The records obtained from electronic databases were first screened to eliminate duplicates. Next, the abstracts were reviewed against the predefined selection criteria, followed by a thorough assessment of the full texts for studies deemed potentially eligible. To ensure accuracy and consistency, the reviewers conducted all screening stages independently. Any disagreements regarding study eligibility were resolved through discussion with a third reviewer, ensuring a consensus-driven decision-making process.

2.4. Data Collection Process and Data Items

Following the thorough screening and consensus process, a final set of studies meeting all eligibility criteria was identified for inclusion in the review. Data from the selected articles were organised into data extraction tables specifically designed to address the systematic review and meta-analysis aims. These tables ensured the consistent extraction of data characteristics across studies (Islam et al., 2017). One reviewer initially extracted the data, which two other authors then checked for consistency. Data extraction items included the study design, outcome measurement type, intervention characteristics, sample characteristics (i.e., number of schools, sample sizes, student ages), eligibility criteria, and reported outcomes. Intervention characteristics included the intervention aims, approach, components, techniques, delivery type, staff involved in delivery, tier level of support, and the interventions’ frequency, length, and duration.

The study designs, sample characteristics, eligibility criteria, and reported outcomes were directly extracted from the studies. The type of outcome measure was classified according to the terminology used in the questionnaires (i.e., connectedness, belonging, or engagement). Regarding the intervention characteristics, the intervention aims were extracted from each study. The approach was determined based on the theoretical rationale behind the intervention described by the authors (e.g., social learning theory or positive youth development). The intervention components were identified as the specific types of activities within the intervention designs (e.g., curriculum, mentoring, or staff training). The techniques referred to the mechanisms through which the intervention facilitated change (e.g., problem-solving or positive relationships). The delivery type, the staff involved, and the tier level were interpreted from the intervention descriptions. For tier levels, interpretations followed the multi-tiered systems of support model (Stoiber & Gettinger, 2015).

2.5. Risk of Bias Assessment

The methodological quality of the included studies was assessed using the revised Cochrane Risk of Bias Tool for Randomized Trials (RoB-2) (Higgins et al., 2019). The RoB-2 tool evaluates five potential bias domains: randomisation process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of reported results. Each domain was rated as “Low”, “High”, or “Some concerns”, culminating in an overall risk of bias rating for each study.

Two reviewers independently completed the RoB-2 checklist per study, and any discrepancies were resolved through discussion. For example, when differences arose regarding the interpretation of the randomisation process, the reviewers collaborated to align their assessments, and unresolved issues were discussed with a third reviewer to reach consensus. This process ensured consistent and accurate evaluations. Reviewers had no affiliations with study authors, minimising potential conflicts of interest and bias in both data extraction and quality ratings (Parsons et al., 2017).

2.6. Meta-Analysis

Outcome measurement statistics (e.g., pre- and post-mean values, standard deviations, sample sizes) were extracted into a spreadsheet to conduct the meta-analysis. Studies rated with a “High” overall risk of bias were excluded to minimise bias in the results. Data from both experimental and control groups were included to enable between- and within-group analyses.

Data items related to the intervention characteristics facilitated an initial subgroup analysis based on categories such as outcome measure type, RoB-2 categories, study design (e.g., student age, number of schools, intervention duration), and intervention delivery (e.g., delivery method, number of staff groups, number of tiers). These categories formed the basis for the preliminary exploratory analysis, leading to five subgroups being selected for inclusion in the final reported analysis. The decision not to report findings from other subgroups (e.g., student age, number of staff groups, RoB-2 categories) was due to either non-significant results or considerable overlap with the five reported subgroups. The categorisation process for the final reported subgroups is described below.

Authors were contacted if additional data were needed for meta-analysis calculations. In cases where studies had incomplete published data and attempts to contact the authors for the missing information were unsuccessful, these studies were excluded from the analysis. The extracted data were entered into Comprehensive Meta-Analysis, Version 4.0 (Borenstein, 2022). Initially, a within-group analysis evaluated the overall intervention effect on connectedness, belonging, and engagement across all studies. Next, a between-group analysis assessed the intervention-group outcomes compared with the control-group outcomes. Forest plots were generated for the overall within- and between-group results. Finally, studies were grouped for subgroup analysis to examine whether the intervention effects differed based on specific variables between groups. The following variables were used for subgroup analysis: Primary Approach, Primary Component, Primary Technique, Number of Tiers, and Target Group (of students).

A structured classification system was applied to subgroup analyses to address the variability in intervention designs. Using the principles of the Framework Method (Gale et al., 2013), the intervention characteristics were systematically explored and categorised in an iterative process to ensure consistency and accommodate diverse data interpretations. After an initial exploration of the extracted data for intervention ‘components’, it was decided that the Primary Component of each intervention was identified as either curriculum or mentoring, as all interventions included one of these two components. Mentoring was designated as the Primary Component in exceptional cases where both components were present. This classification was iteratively refined through systematic coding and comparison across studies to ensure alignment with the broader intervention aims. Primary Approaches were categorised as Cognitive, Behavioural, Emotional, or Ecological (Walker & Gresham, 2013). Similarly, Primary Techniques were classified based on their underlying mechanisms as Intrapersonal, Interpersonal, or Environmental (Ruiter et al., 2020). A working analytical framework was initially developed using these predefined categories, which were refined iteratively through a best-fit process to ensure they captured the nuances of the intervention descriptions. Tier levels were defined based on intervention descriptions, with studies categorised as “multiple tiers” if they addressed two or more levels (i.e., universal, targeted, or specialist level) or as “one tier” if they focused on a single level (Stoiber & Gettinger, 2015).

Random-effects modelling was applied to calculate effect sizes, accounting for differences in true effects across studies due to variations in intervention characteristics, student demographics, sampling methods, and outcome measures (Riley et al., 2011). The fixed effects model was applied for the subgroup analysis due to the relatively low number of studies included in the meta-analysis (Deeks et al., 2019). Hedges’ g was used to calculate standardised mean differences (SMD) with a 95% confidence interval (CI), with effect sizes interpreted according to Cohen’s d conventions: g ≤ 0.2 (no or negligible effect), 0.2 < g ≤ 0.5 (small effect), 0.5 < g ≤ 0.8 (moderate effect), and g > 0.8 (large effect) (Grove & Cipher, 2020). Prediction intervals assessed heterogeneity in the meta-analysis (Borenstein, 2023).

The Classic Fail-Safe N was calculated to assess publication bias. It indicates the number of additional studies required to nullify the observed effect (Rosenberg, 2005). A large N value relative to the number of included studies suggests a low risk of publication bias.

3. Results

3.1. Study Selection

Overall, 3841 records were retrieved from the four electronic databases searched. After duplicates were removed, 3131 abstracts remained for researcher screening. Abstracts and titles were uploaded on ‘Research Screener’, a semi-automated web application that uses machine learning algorithms to optimise the screening process (Chai et al., 2021). Abstracts were ranked by their relevance to eight seed articles identified through a preliminary literature search (Acosta et al., 2019; Asogwa et al., 2020; Frank et al., 2017; Holt et al., 2008; Lewis et al., 2006; Orthner et al., 2010; Sawyer et al., 2010; Shoshani et al., 2016). The preliminary search involved using a streamlined string of keywords and synonyms in the PsycINFO and PubMed databases, sorted by ‘relevance’ or ‘best match’. Subsequently, references from these seminal articles were meticulously examined to identify further pertinent studies.

The algorithm then produced blocks of 50 articles based on these seed articles. Two independent reviewers completed 11 blocks of abstract screening before reaching the saturation point; three continuous blocks of 50 articles with no abstract flagged as meeting the eligibility criteria indicated saturation (Chai et al., 2021). Of the 3131 records screened, 701 articles were screened at the full-text level. A total of 84 full-text articles were accessed and assessed for eligibility, and the primary reviewer then identified 10 studies that met the eligibility criteria for inclusion in the review. A further 6 additional studies meeting the inclusion criteria were retrieved from manually searching previous connectedness, belonging, and engagement reviews and the reference lists of included articles. A total of 16 studies were included in the review. The PRISMA flowchart (Figure 1) details the search and screening process.

3.2. Description of Studies

3.2.1. Participants

Table 2. Summary of studies.

Reference/ Methodological Quality (RoB2)/ Country	Intervention/ Comparison Condition(s)	Sample Characteristics	Student Eligibility Criteria	School Eligibility Criteria/Sampling Technique	Outcome Measure Type	Outcome Measure Tool	Results: Within Group M (SD)/ Between Group M (SD)/ Effect Size
Reference: (Acosta et al., 2019) RoB2: Some concerns Country: USA	Intervention: Whole-school restorative practice Comparison: Business as usual	Intervention sample: 977 Comparison sample: 1794 Age: 11–12 No. of Schools: 14	All students eligible (All students)	School eligibility: Middle schools in the state of Maine Sampling: Cluster randomised	Connectedness	National Adolescent Health Study (5 Items)	Within group: NR Between group: NR Effect size: Cohen D = 0.64
Reference: (Asogwa et al., 2020) RoB2: Low Country: Nigeria	Intervention: Video-guided educational intervention Comparison: Business as usual	Intervention sample: 25 Comparison sample: 25 Age: 11–15 No. of Schools: 1	Moderate to worse hearing-impaired, low school engagement at pre-test (At-risk students)	School eligibility: Public schools in South-South Nigeria region Sampling: Randomised	Engagement	Student Engagement Scale (19 Items)	Within group: Pre mean: 24.92 (3.53) Post mean: 44.8 (7.8) Between group: Con Post mean: 24.48 (2.9) Effect size: Delta = 2.605
Reference: (Espelage et al., 2015) RoB2: Some concerns Country: USA	Intervention: Second Step Curriculum Comparison: Stories of Us Curriculum	Intervention sample: 47 Comparison sample: 76 Age: 11–12 No. of Schools: 6	Students with any type of diagnosed disability (At-risk students)	School eligibility: Schools across the state of Illinois and Kansas Sampling: Matched pairs cluster randomisation	Belonging	Psychological Sense of School Membership Scale (18 Items)	Within group: Pre mean: 2.17 (0.09) Post mean: 1.59 (0.7) Between group: Con Post mean: 1.51 (0.67) Effect size: Cohen D = 0.12
Reference: (Ferrer-Cascales et al., 2018) RoB2: Low Country: Spain	Intervention: Peer tutoring Comparison: Business as usual	Intervention sample: 987 Comparison sample: 1070 Age: 11–16 No. of Schools: 22	All students eligible (All students)	School eligibility: Public schools in Spain Sampling: Matched pairs cluster randomisation	Belonging	Spanish School Climate Questionnaire—Sense of Belonging subscale (5 Items)	Within group: Pre mean: 6.92 (SD 2.50) Post mean: 11 (SD 2.82) Between group: Con Post mean: 8.94 (2.98) Effect size: Cohen D = 0.06
Reference: (Flannery et al., 2020) RoB2: Some concerns Country: USA	Intervention: Curriculum, leadership team, and peer support Comparison: Business as usual	Intervention sample: 854 Comparison sample: 734 Age: 14–15 No. of Schools: 4	All students eligible (All students)	School eligibility: Schools in one state of the Pacific Northwest region Sampling: Randomised waitlist controlled	Engagement	Motivational and Engagement Scale—High School (44 Items)	Within group: Pre mean: 4.71 (1.04) Post mean: 4.61 (1.13) Between group: Con Post mean: 4.55 (1.17) Effect size: ES = 0.15
Reference: (Frank et al., 2017) RoB2: Low Country: USA	Intervention: Practical-based curriculum on yoga and meditation Comparison: Business as usual	Intervention sample: NR Comparison sample: NR Age: 11–15 No. of Schools: 1	All students eligible (All students)	School eligibility: School in deprived neighbourhood in state of California Sampling: Randomised	Belonging	School Bonding Scale (Items NR)	Within group: Pre mean: 3.37 (0.65) Post mean: 3.35 (0.58) Between group: Con Post mean: 3.13 (0.79) Effect size: Cohen D = 0.45
Reference: (Heppen et al., 2017) RoB2: Low Country: USA	Intervention: Adult–student mentors Comparison: Business as usual	Intervention sample: 276 Comparison sample: NR Age: 13–15 No. of Schools: 10	Students are predicted to be at risk of delayed graduation (At-risk students)	School eligibility: Public schools in the state of California Sampling: Randomised	Engagement	Student Engagement Questionnaire Scale (12 Items)	Within group: Pre mean: NR Post mean: NR Between group: Con Post mean: NR Effect size: NR
Reference: (Holt et al., 2008) RoB2: Low Country: USA	Intervention: Adult–student mentors Comparison: Business as usual	Intervention sample: 18 Comparison sample: 18 Age: 14–15 No. of Schools: 1	Students at risk of academic failure (At-risk students)	School eligibility: Public school in Mid-Atlantic region Sampling: Randomised	Belonging	Psychological Sense of School Membership Scale (12 Items)	Within group: Pre mean: 3.12 (0.51) Post mean: 3.08 (0.57) Between group: Con Post mean: 3.28 (0.41) Effect size: ES = 0.28
Reference: (Kaesornsamut et al., 2012) RoB2: Low Country: Thailand	Intervention: Curriculum-based psycho-education Comparison: Business as usual	Intervention sample: 30 Comparison sample: 30 Age: 16–18 No. of Schools: 1	Students with mild to moderate depressive symptoms, no physical or cognitive disability, no history of mental illness (At-risk students)	School eligibility: Public school in city of Bangkok Sampling: Randomised	Belonging	The Sense of Belonging Instrument (16 Items)	Within group: Pre mean: 46.10 (4.54) Post mean: 47.07 (4.93) Between group: Con Post mean: 42.2 (5.61) Effect size: ES = 0.87
Reference: (Lewis et al., 2006) RoB2: Low Country: USA	Intervention: Practical-based cultural curriculum Comparison: Regular Life Skills course	Intervention sample: 31 Comparison sample: 26 Age: 13–14 No. of Schools: 1	African American Descent (At-risk students)	School eligibility: Middle school in deprived neighbourhood Sampling: Randomised	Belonging	Psychological Sense of School Membership Scale (18 Items)	Within group: NR Between group: NR Effect size: Linear Growth Coefficient = 0.85 Quadratic Acceleration/Deceleration Coefficient = −0.10
Reference: (Orthner et al., 2012) RoB2: Some concerns Country: USA	Intervention: Career-relevant instruction curriculum Comparison: Business as usual	Intervention sample: 1976 Comparison sample: 1655 Age: 11–13 No. of Schools: 14	All students were eligible (All students)	School eligibility: Middle schools in the state of North Carolina Sampling: Matched pairs cluster randomised	Engagement	School Success Profile School Engagement Subscale (3 Items)	Within group: Pre mean: NR Post mean: 4.14 (0.63) Between group: Con Post mean: 4.14 (0.67) Effect size: ES = 1.68
Reference: (Sawyer et al., 2010) RoB2: Some concerns Country: Australia	Intervention: Whole-school cultural change Comparison: Business as usual	Intervention sample: 3037 Comparison sample: 2597 Age: 12–13 No. of Schools: 25	All students were eligible (All students)	School eligibility: Schools in Australia Sampling: Randomised	Belonging	Beyond Blue School Climate Questionnaire—‘Belonging’ Subscale (8 Items)	Within group: Pre mean: 59.12 (9.88) Post mean: 56.28 (10.52) Between group: Con Post mean: 56.01 (10.17) Effect size: Time Coefficient = 0.154
Reference: (Shinde et al., 2020) RoB2: Low Country: India	Intervention: Multiple component intervention (Whole school, curriculum, counselling) SM—Counsellor-led curriculum TSM—Teacher-led curriculum Comparison: Standard government-run life skills adolescence education program	Intervention sample: SM—5084 TSM—4786 Comparison sample: 5362 Age: 13–15 No. of schools: 74	All students were eligible (All students)	School eligibility: Public schools in the state of Bihar Sampling: Matched pairs cluster randomised	Belonging	Beyond Blue School Climate Questionnaire—‘Belonging’ Subscale (8 Items)	Within group: Counsellor Pre mean: 5.01 (1.46) Post mean: 6.79 (1.1) Teacher Pre mean: 5.09 (1.47) Post mean: 4.76 (1.51) Between group: Con Post mean: 5.09 (1.47) Effect size: Counsellor: aMD = 7.33 Teacher: aMD = 0.29
Reference: (Shoshani et al., 2016) RoB2: Some concerns Country: Israel	Intervention: Positive psychology-based curriculum Comparison: Business as usual	Intervention sample: 1262 Comparison sample: 1255 Age: 12–15 No. of schools: 6	All students were eligible (All students)	School eligibility: Middle schools in a district of Israel Sampling: Randomised waitlist controlled	Engagement	School Engagement Survey (51 Items)	Within group: Pre mean: 3.77 (0.87) Post mean: 4.01 (0.78) Between group: Con Post mean: 3.72 (0.76) Effect size: NR
Reference: (Tingey et al., 2020) RoB2: Some concerns Country: USA	Intervention: Cultural- and entrepreneurship-based curriculum Comparison: Recreational sports field days	Intervention sample: 267 Comparison sample: 127 Age: 13–16 No. of schools: 11	Native American ethnicity (At-risk students)	School eligibility: Schools in the state of Arizona Sampling: Randomised	Connectedness	Hemingway Measure of Adolescent Connectedness, and the Awareness of Connectedness Scale (6 Items)	Within group: Pre mean: 3.58 (0.04) Post mean: 3.42 (0.05) Between group: Con Post mean: 3.47 (0.07) Effect size: NR
Reference: (Walsh et al., 2022) RoB2: Some concerns Country: USA	Intervention: Older peer-led activities and discussions Comparison: Business as usual	Intervention sample: 666 Comparison sample: 635 Age: 15–16 No. of schools: 18	All students were eligible (All students)	School eligibility: Schools in the state of North Carolina and New York Sampling: Randomised	Engagement	School Engagement Scale (9 Items)	Within group: NR Between group: NR Effect size: ES = 0.074

provides a detailed description of the studies included in this review. Participants’ ages ranged from 11 to 18 years of age, with studies within the 11–14 (n = 10, 62.50%) and within the 15–18 (n = 6, 37.50%) age bracket. Many studies (n = 9, 56.25%) included all students within the participating schools as eligible for the intervention condition. The remaining studies (n = 7, 43.75%) were reported by the authors to target their intervention at students at risk of low connectedness, belonging, or engagement, grouped by ethnicity (n = 2, 12.50%; n = 1, 6.25% African American descent, n = 1, 6.25% Native American descent), students at risk of academic failure (n = 2, 12.50%), disability (any disability: n = 1, 6.25%; hearing impairment: n = 1, 6.25%), and mild to moderate depressive symptoms (n = 1, 6.25%).

3.2.2. Study Origin, Groups, and Research Designs

The most frequent country of origin for the included studies was the United States (n = 10, 62.50%), followed by one study from Israel, Nigeria, Australia, Spain, Thailand, and India. The range of schools included was between 1 and 74. Most studies involved 10 or more schools in their sample (n = 7, 43.75%), followed by 1 school (n = 5, 31.25%), and 4–6 schools (n = 3, 18.75%). Sample sizes ranged from 36 to 15,232 between studies, with a total sample of 35,451 students broken down by intervention (n = 20,047) or comparison (n = 15,404) groups. This excludes two studies that provided incomplete reporting of their sample size (Frank et al., 2017; Heppen et al., 2017).

All studies randomised participants into groups, with studies using either randomised (n = 9, 56.25%), waitlist controlled (n = 2, 12.50%), cluster (n = 1, 6.25%), or matched pairs cluster (n = 4, 25.00%) sampling techniques. Most studies (n = 15, 93.75%) utilised a single intervention condition compared to a control group, except Shinde et al. (2020), who used two intervention groups and a single control group. Three of these studies used an active control group, with the others (n = 12, 75.00%) receiving treatment as usual. Most studies limited school eligibility geographically (n = 14, 87.50%) by country (n = 2, 12.50%), region (n = 4, 25.00%), state (n = 7, 43.75%), or city (n = 1, 6.25%). Two studies limited school eligibility to the demographics of the school (deprived neighbourhood; n = 2, 12.50%), of which one also limited school eligibility by geographical state (n = 1, 6.25%). Further inclusion criteria for schools include public-only schools (n = 6, 37.50%) or unspecified schools (n = 10, 62.50%).

3.2.3. Outcome Measurement Type and Measurement Tools Used

Most interventions used a measurement tool relating primarily to belonging (n = 8, 50.00%), followed by engagement (n = 6, 37.50%) and connectedness (n = 2, 12.50%). The number of items measuring either connectedness, belonging, or engagement ranged from 3 to 44 items. Frank et al. (2017) did not report the number of items in their measurement tool. Connectedness measurement tools included the National Adolescent Health Study (n = 1, 6.25%) and the Hemingway Measure of Adolescent Connectedness Scale (n = 1, 6.25%). Belonging measurement tools included the Psychological Sense of School Membership Scale (n = 3, 18.75%), Beyond Blue School Climate Questionnaire—Belongingness subscale (n = 2, 12.50%), Spanish School Climate Questionnaire—Sense of Belonging subscale (n = 1, 6.25%), The Sense of Belonging Instrument (n = 1, 6.25%), and the School Bonding Scale (n = 1, 6.25%). Engagement measurement tools included the School Engagement Scale (n = 2, 12.50%), Motivational and Engagement Scale—High School (n = 1, 6.25%), Student Engagement Questionnaire Scale (n = 1, 6.25%), School Success Profile School Engagement Subscale (n = 1, 6.25%), and the School Engagement Survey (n = 1, 6.25%).

3.3. Interventions

A summary of all interventions and their characteristics is provided in Table 2 and

Table 3. Characteristics of interventions.

Reference/ Intervention Name	Intervention Aim(s)	Intervention-Approach/ Components/Techniques	Delivery Type/ Delivered by/ Tier Level	Frequency of Sessions/ Length of Sessions/ Total Sessions/ Duration
Reference: (Acosta et al., 2019) Intervention name: The Restorative Practices Intervention	Increase teacher–student relationships and student skills	Approach: Ecological, Positive youth development, Affect theory (Ecological) Components: Curriculum, Staff training (Curriculum) Techniques: Cultural change, Socio-emotional skills, Positive relationships, Coaching (Environmental)	Delivery type: Individual and group Delivered by: Teachers, Support staff Tier level: Universal, Targeted (Multiple tiers)	Frequency of sessions: Variable Length of sessions: Variable Total sessions: Variable Duration: 2 Years
Reference: (Asogwa et al., 2020) Intervention name: Video-guided educational intervention	Increase school engagement of hearing-impaired students	Approach: Social cognitive learning theory (Cognitive) Components: Curriculum (Curriculum) Techniques: Instructional (Intrapersonal)	Delivery type: Group Delivered by: Specialist staff Tier level: Specialist (One tier)	Frequency of sessions: 1 per week Length of sessions: 45 min Total sessions: 12 Duration: 12 weeks
Reference: (Espelage et al., 2015) Intervention name: Second Step–Student Success Through Prevention	Increase school belonging, prosocial attitudes and behaviours, academic achievement	Approach: Socio-emotional learning (Emotional) Components: Curriculum (Curriculum) Techniques: Socio-emotional skills (Interpersonal)	Delivery type: Group Delivered by: Teachers Tier level: Targeted (One tier)	Frequency of sessions: 1 per week Length of sessions: 50 min Total sessions: 28 Duration: 2 Years
Reference: (Ferrer-Cascales et al., 2018) Intervention name: The TEI Program	Reduce bullying and cyberbullying	Approach: Ecological, Socio-emotional learning, Positive psychology (Ecological) Components: Peer mentoring, Staff training, Community involvement (Mentoring) Techniques: Cultural change, Socio-emotional skills, Positive relationships, Problem-solving (Environmental)	Delivery type: Individual and group Delivered by: Teachers, Support staff, Students Tier level: Universal, Targeted (Multiple tiers)	Frequency of sessions: Variable Length of sessions: Variable Total sessions: Variable Duration: 1 Year
Reference: (Flannery et al., 2020) Intervention name: Freshmen Success	Increase school engagement, attendance, and academic achievement	Approach: Multi-tiered systems of support (Ecological) Components: Peer mentoring, Curriculum, Data-informed decisions (Mentoring) Techniques: Socio-emotional skills, Problem-solving, Positive relationships (Interpersonal)	Delivery type: Group Delivered by: Teachers, Support staff Tier level: Universal (One tier)	Frequency of sessions: 1 per week Length of sessions: 30 min Total sessions: 12 Duration: 1 Year
Reference: (Frank et al., 2017) Intervention name: Transformative Life Skills	Reduce adolescent emotional distress Increase prosocial behaviour, school functioning	Approach: Socio-emotional learning, Stress (Emotional) Components: Curriculum, Yoga (Curriculum) Techniques: Instructional, Mindfulness (Intrapersonal)	Delivery type: Group Delivered by: Specialist staff Tier level: Universal (One tier)	Frequency of sessions: 3–4 per week Length of sessions: 30 min Total sessions: NR Duration: 12–14 weeks
Reference: (Heppen et al., 2017) Intervention name: Check and Connect Mentoring Program	Reduce dropouts Increase student engagement, performance in school, school persistence	Approach: Ecological, Cognitive behavioural theory, Motivational theory, Socio-emotional learning (Ecological) Components: Adult mentoring, Data-informed decisions (Mentoring) Techniques: Socio-emotional skills, Problem-solving, Positive relationships (Interpersonal)	Delivery type: Individual and group Delivered by: Support staff Tier level: Universal, Targeted (Multiple tiers)	Frequency of sessions: Variable Length of sessions: Variable Total sessions: Variable Duration: 1 Year
Reference: (Holt et al., 2008) Intervention name: Achievement Mentoring Program	Increase school-related cognitions and behaviours	Approach: Social cognitive learning theory (Cognitive) Components: Adult mentoring, Staff training (Mentoring) Techniques: Problem-solving, Positive relationships (Interpersonal)	Delivery type: Individual Delivered by: Support staff Tier level: Targeted (One tier)	Frequency of sessions: Variable Length of sessions: Variable Total sessions: Variable Duration: Up to 13 weeks
Reference: (Kaesornsamut et al., 2012) Intervention name: BAND Intervention Program	Reduce negative thinking and depressive symptoms Increase a sense of belonging	Approach: Cognitive behavioural theory, Motivational theory (Emotional) Components: Curriculum (Curriculum) Techniques: Socio-emotional skills (Interpersonal)	Delivery type: Group Delivered by: Specialist staff Tier level: Targeted, Specialist (Multiple tiers)	Frequency of sessions: 2 per week Length of sessions: 60 min Total sessions: 14 Duration: 7 weeks
Reference: (Lewis et al., 2006) Intervention name: The Project EXCEL	Increase communal orientation, school connectedness, motivation to achieve, social change efforts	Approach: Afrocentric approaches (Emotional) Components: Curriculum (Curriculum) Techniques: Instructional, Positive affirmations, Identity development (Intrapersonal)	Delivery type: Group Delivered by: Teachers, Specialist staff Tier level: Universal, Specialist (Multiple tiers)	Frequency of sessions: 3 per week Length of sessions: NR Total sessions: NR Duration: 14–16 weeks (2–16 weeks)
Reference: (Orthner et al., 2012) Intervention name: CareerStart	Increase student engagement, academic performance, and career exploration	Approach: Motivational theory (Behavioural) Components: Curriculum (Curriculum) Techniques: Instructional (Intrapersonal)	Delivery type: Group Delivered by: Teachers Tier level: Universal (One tier)	Frequency of sessions: NR Length of sessions: NR Total sessions: 40 Duration: 3 Years
Reference: (Sawyer et al., 2010) Intervention name: Beyond-Blue intervention	Reduce depressive symptoms Increase individual-level protective factors	Approach: Ecological (Ecological) Components: Curriculum, Community involvement (Curriculum) Techniques: Cultural change, Socio-emotional skills, Parental communication, Access to services, Positive relationships (Environmental)	Delivery type: Individual and group Delivered by: Teachers, Support staff Tier level: Universal (One tier)	Frequency of sessions: 1 per term Length of sessions: 45 min Total sessions: 10 Duration: 3 Years
Reference: (Shinde et al., 2020) Intervention name: School-based interventions for promoting adolescent health	Increase social skills, problem-solving skills, engagement of the school community	Approach: Positive school climate (Ecological) Components: Curriculum, Staff training (Mentoring) Techniques: Cultural change, Socio-emotional skills, Problem-solving, Parental communication, Access to services (Environmental)	Delivery type: Individual and group Delivered by: Teachers or Teachers and Specialist staff Tier level: Universal, Targeted, Specialist (Multiple tiers)	Frequency of sessions: 0.25 per week Length of sessions: 60 min Total sessions: 16 Duration: 17 months
Reference: (Shoshani et al., 2016) Intervention name: The Maytiv school program	Increase positive emotions, engagement, positive relationships, meaning, and achievement of students	Approach: Positive psychology (Cognitive) Components: Curriculum, Staff training (Curriculum) Techniques: Socio-emotional skills, Positive affirmations (Interpersonal)	Delivery type: Group Delivered by: Teachers Tier level: Universal (One tier)	Frequency of sessions: 0.5 per week Length of sessions: 90 min Total sessions: 15 Duration: 2 Years
Reference: (Tingey et al., 2020) Intervention name: Arrowhead Business Group entrepreneurship education program	Increase social wellbeing, economic outcomes	Approach: Ecological, Positive youth development (Ecological) Components: Curriculum (Curriculum) Techniques: Socio-emotional skills, Problem-solving, Identity development (Interpersonal)	Delivery type: Group Delivered by: Specialist staff Tier level: Targeted (One tier)	Frequency of sessions: 0.25 per week Length of sessions: 4–6 h Total sessions: 16 Duration: 1 Year
Reference: (Walsh et al., 2022) Intervention name: Peer Group Connection	Increase school engagement, connectedness among peers, decision-making and goal-setting skills	Approach: Positive youth development (Ecological) Components: Peer mentoring, Staff training, Community involvement (Mentoring) Techniques: Socio-emotional skills, Problem-solving, Positive relationships (Interpersonal)	Delivery type: Group Delivered by: Support staff, Students Tier level: Targeted (One tier)	Frequency of sessions: NR Length of sessions: 45 min Total sessions: NR Duration: 18 months

Notes. NR = Data not reported by authors; BAND = Belonging against Negative Thinking and Depression; EXCEL = Ensuring Excellence through Communalism, African Education, and Leadership.

. Interventions were implemented for either 2–16 weeks (n = 5, 31.25%), 1 year (n = 4, 25.00%), or 18 months to 3 years (n = 7, 43.75%). Intervention approaches were classified as either Ecological (n = 8, 50.00%), Emotional (n = 4, 25.00%), Cognitive (n = 3, 18.75%), or Behavioural (n = 1, 6.25%). Ecological interventions encompassed approaches such as positive youth development programs (n = 3, 18.75%), multi-tiered systems of support (n = 1, 6.25%), school climate (n = 1, 6.25%), or the use of multiple approaches (n = 3, 18.75%). Emotion-based intervention approaches were primarily socio-emotional learning (n = 2, 12.50%) alongside a cognitive behavioural (n = 1, 6.25%) or Afrocentric approach (n = 1, 6.25%). Cognitive intervention approaches involved social cognitive learning theory (n = 2, 12.50%) or Positive psychology (n = 1, 6.25%). The behavioural intervention approach drew from motivational theory (n = 1, 6.25%).

Intervention components were classified primarily as curriculum (n = 11, 68.75%) or mentoring (n = 5, 31.25%) focused. Interventions utilising a curriculum component involved staff training (n = 3, 18.75%), yoga (n = 1, 6.25%), community involvement (n = 1, 6.25%), or solely a curriculum (n = 6, 37.50%). Mentoring-based interventions can be further divided into adult-led (n = 2, 12.50%) or peer-led (n = 3, 18.75%) mentoring.

Intervention techniques were classified as either interpersonal (n = 8, 50.00%), intrapersonal (n = 4, 25.00%) or environmental (n = 4, 25.00%). The most prominent interpersonal technique included using problem-solving (n = 5, 31.25%), many of which simultaneously focused on promoting positive relationships (n = 4, 25.00%). Techniques such as developing socio-emotional skills (n = 1, 6.25%) or developing socio-emotional skills alongside positive affirmations (n = 1, 6.25%) were also utilised as interpersonal intervention techniques. Intrapersonal techniques all included an instructional element (n = 4, 25.00%), while another study also incorporated mindfulness alongside this (n = 1, 6.25%), and another targeted identity development for students (n = 1, 6.25%). Environmental-based techniques all included cultural change (n = 4, 25.00%) alongside multicomponent interpersonal and intrapersonal techniques.

Interventions were delivered either in a group (n = 10, 62.50%), individually (n = 1, 6.25%), or a mixture of settings (n = 5, 31.25%). Interventions were delivered either by one group of staff, such as teaching staff (n = 4, 25.00%), support staff (n = 2, 12.50%), and specialist staff (n = 4, 25.00%), or by multiple staff groups (n = 6, 37.50%). The most common staff collaboration was with teachers and support staff (n = 3, 18.75%), while other studies involved teachers and specialist staff (n = 1, 6.25%), support staff and students (n = 1, 6.25%), or a combination of support staff, teaching staff, and students (n = 1, 6.25%). Interventions chose to focus on either one tier of support (n = 10, 62.5%), including universal (n = 5, 31.25%), targeted (n = 4, 25.00%), and specialist (n = 1, 6.25%), or multiple tiers of support (n = 6, 37.50%). Other studies focused on universal and targeted (n = 3, 18.75%), universal and specialist (n = 1, 6.25%), targeted and specialist (n = 1, 6.25%), or all three levels (n = 1, 6.25%).

Studies differed in the frequency, length, and number of sessions within their intervention. The frequency of intervention sessions was either variable (n = 4, 25.00%), less than once a week (n = 4, 25.00%), once per week (n = 3, 18.75%), more than once a week (n = 3, 18.75%), or not reported (n = 2, 12.50%). The length of sessions was variable (n = 4, 25.00%), less than 60 min (n = 6, 37.50%), 60 min (n = 2, 12.50%), more than 60 min (n = 2, 12.50%), or not reported (n = 2, 12.50%). Total sessions were either variable (n = 4, 25.00%), between 10 and 20 (n = 6, 37.50%), more than 20 (n = 2, 12.50%), or not reported (n = 3, 18.75%).

3.4. Risk of Bias Assessment

The methodology of all included studies was evaluated for bias using the RoB-2 critical appraisal tool (Higgins et al., 2019). Of the 16 studies assessed, 8 were rated as having a low risk of bias overall and across all five domains of risk. The remaining 8 studies were rated as having some concerns. The overall ratings per study are presented in Table 1, and the domain-level ratings for each study are presented in Figure 2.

3.5. Meta-Analysis

Of the 16 studies initially assessed for inclusion in the meta-analysis, 5 were excluded from both the within- and between-group analyses because they did not report the necessary data (Acosta et al., 2019; Frank et al., 2017; Heppen et al., 2017; Lewis et al., 2006; Walsh et al., 2022). Additionally, 1 study was excluded solely from the within-group analysis due to insufficient data (Orthner et al., 2010). Ultimately, 10 studies were included in the within-group analysis and 11 studies in the between-group analysis.

An additional intervention was included in both groups because both intervention conditions from Shinde, Weiss (Shinde et al., 2020) were analysed using the same control group as a comparison. Although the intervention conditions from Shoshani, Steinmetz (Shoshani et al., 2016) were compared against the same control sample, the intervention samples themselves were independent of each other. For Shoshani, Steinmetz (Shoshani et al., 2016), composite scores were calculated from the means of three sub-factors because the authors did not report an overall factor. Additionally, for the Flannery, Kato (Flannery et al., 2020) study, the Global positive engagement subscale from the Motivational and Engagement Scale was used to maintain homogeneity between studies, as no overall variable was reported, and manually computing the variable was not possible.

3.5.1. Overall, Within Groups

The pooled intervention effects for connectedness, belonging, and engagement across all interventions showed a large effect size (z = 4.899, p < 0.001, g = 1.056, 95% CI = 0.634–1.479), with a prediction interval of −0.581 to 2.693. Pre–post-intervention effects varied greatly across the studies, ranging from 0.072 to 3.595. Two interventions produced a negligible effect (g = 0.072–0.092). Four interventions had a small effect size (g = 0.202–0.290), and five interventions had a large effect size (g = 1.153–3.595), two of which had a very high effect size (g = 3.232 and 3.595, respectively). See Figure 3 for full results. The Classic Fail-Safe N for the within-groups analysis was 6487, indicating a low risk of publication bias in the findings. The I-squared was 99.61%, suggesting the random-effects model was appropriate to apply.

3.5.2. Overall, Between Groups

The pooled intervention effects for connectedness, belonging, and engagement across all interventions showed a moderate effect size (z = 3.333, p = 0.001, g = 0.642, 95% CI = 0.265–1.020), with a prediction interval of—0.857 to 2.142. Intervention vs. control effects varied greatly across the studies, ranging from 0.000 to 3.399. Seven interventions showed a significant difference (g = 0.221–3.399), whereas five interventions (g = 0.000–0.394) did not meet the threshold for significance. Figure 4 contains all between-group results. The Classic Fail-Safe N for the within-groups analysis was 3785, indicating a low risk of publication bias in the findings. The I-squared was 99.57%, suggesting the random-effects model was appropriate to apply.

3.5.3. Between-Groups Subgroup Analysis

Findings from the between-group meta-analysis (

Table 4. Between-groups meta-analyses comparing effects for subgroups of included studies.

Subgrouping	n	Hedges’ g	Lower Limit	Upper Limit	Z-Value	p-Value
Primary Approach
Ecological	6	0.613	0.588	0.638	48.162	<0.001 **
Emotional	2	0.372	0.074	0.670	2.446	0.014 *
Cognitive	3	0.403	0.322	0.484	9.788	<0.001 **
Behavioural	1	0.000	−0.065	0.065	0.000	1.000
Primary Component
Curriculum	9	0.529	0.506	0.553	44.100	<0.001 **
Mentoring	3	0.478	0.406	0.549	13.117	<0.001 **
Primary Technique
Interpersonal	6	0.317	0.253	0.381	9.741	<0.001 **
Intrapersonal	2	0.019	−0.046	0.085	0.585	0.559
Environmental	4	0.636	0.610	0.662	48.625	<0.001 **
Number of Tiers
Multiple tiers	4	0.781	0.753	0.810	53.768	<0.001 **
One tier	8	0.113	0.077	0.149	6.134	<0.001 **
Target Group
All students	7	0.521	0.498	0.543	45.337	<0.001 **
At-risk students	5	0.759	0.576	0.942	8.129	<0.001 **

Notes: * = p < 0.05; ** = p < 0.001.

) indicated that all approaches had significant effects except for behaviour-based approaches, which were the least effective intervention overall (p = 1.0, g = 0, 95% CI = −0.065–0.065). Interventions with both curriculum (p = <0.001, g = 0.529, 95% CI = 0.506–0.553) and mentoring (p = <0.001, g = 0.478, 95% CI = 0.406–0.549) components had a similar effect size, suggesting significant positive effects. Interventions using environmental (p = <0.001, g = 0.636, 95% CI = 0.610–0.662) or interpersonal (p = <0.001, g = 0.317, 95% CI = 0.253–0.381) techniques had the largest effect sizes compared to a non-significant effect for intrapersonal (p = 0.559, g = 0.019, 95% CI = −0.046–0.085) techniques. Interventions incorporating multiple tiers had the highest effect sizes overall (p = <0.001, g = 0.781, 95% CI = 0.753–0.810). Interventions that targeted all students had moderate effect sizes (p = <0.001, g = 0.521, 95% CI = 0.498–0.543), while those targeting at-risk students approached large effect sizes (p = <0.001, g = 0.759, 95% CI = 0.576–0.942).

4. Discussion

This study aimed to conduct a systematic review and meta-analysis of interventions designed to improve school connectedness, belonging, and engagement in secondary schools. Sixteen studies were included in the review, with eleven studies contributing to the meta-analysis. The results indicate that the interventions generally had a moderate effect on improving connectedness, belonging, and engagement, with within-group analyses showing large effect sizes and between-group analyses showing moderate effects. Notably, this is the first review to conduct a meta-analysis across connectedness, belonging, or engagement school-based interventions research, addressing a significant gap in the literature. These results suggest that interventions targeting connectedness, belonging, and engagement show considerable promise, especially given their established influence on student outcomes.

4.1. Intervention Approaches

Ecological approaches yielded the largest effect sizes, with cognitive and emotional approaches also leading to significant improvements. In contrast, behavioural interventions yielded non-significant results, though this finding requires cautious interpretation due to the limited number of studies in this category. The significant improvements associated with cognitive approaches align with a previous review (K.-A. Allen et al., 2022) but contrast with prior research suggesting that a combined bio-psycho-social-ecological model is necessary to enhance school belonging for students (K. Allen et al., 2018). These findings suggest that interventions can be successful even when targeting a narrower range of factors.

Connectedness interventions predominantly used ecological approaches and were often supported by a positive youth development framework (Atkiss et al., 2011). Belonging interventions primarily focused on emotions, often adopting an ecological approach involving changes in the school climate or whole-school changes (Salle et al., 2015), or alternatively, approaches highlighting the benefits of developing socio-emotional skills as the foundation behind the intervention. Engagement-focused studies were more likely to take a behavioural or cognitive approach compared with connectedness and belonging, although ecological approaches were also discussed frequently within engagement interventions. This pattern aligns with the conceptualisations that belonging and connectedness emphasise broader contextual factors, as highlighted in positive youth development and school climate models (Atkiss et al., 2011; Salle et al., 2015). Engagement, however, appears more focused on within-student factors, such as behavioural and cognitive mechanisms (Boyle & Allen, 2022). Nevertheless, there are examples of significant overlap between all three constructs, as a previous review found belonging interventions across behavioural, emotional, and cognitive domains, highlighting the importance of behavioural domains in belonging as well as engagement interventions (K.-A. Allen et al., 2022). At the same time, research has conceptualised engagement as closely linked to broader contextual factors, aligning it closely to an ecological approach similar to the belonging and connectedness literature (Nguyen et al., 2019).

The intervention approaches examined in this review are rooted in the theoretical frameworks established by the original authors, providing significant insights into the influence of theory on intervention design. However, these frameworks did not consistently align with the specific components or techniques implemented, highlighting the variability in translating theory into practice. For example, even within the frequently cited positive youth development framework, there were differing interpretations across studies, which is common for this approach (Shek et al., 2019). This variability underscores the absence of robust, conceptually clear frameworks for interventions aimed at fostering connectedness, belonging, and engagement. Such inconsistencies can undermine the theoretical validity and evaluation of interventions. While some studies employed a range of measurement tools to assess impact, they often lacked clarity about the relationships between intervention aims, the processes driving change, and how the measured outcomes influenced each other (Durlak et al., 2011). This highlights the complexity of developing interventions and the crucial need to understand how theoretical foundations underpin practical application.

4.2. Intervention Techniques

Interpersonal and environmental techniques yielded small but significant improvements in connectedness, belonging, and engagement, with environmental factors producing moderate effect sizes. In contrast, intrapersonal techniques, often instructional in nature, had no significant effect. These findings highlight the greater effectiveness of contextual and relational strategies over purely instructional methods (Hofkens & Pianta, 2022; Reinke et al., 2022). This supports a shift toward interventions integrating environmental and interpersonal strategies to foster connectedness, belonging, and engagement (Christenson et al., 2008).

Problem-solving emerged as the most frequently employed subcategory of techniques in the studies reviewed. These techniques often focused on enhancing students’ decision-making skills, either through peer-led or adult-led mentoring programs, while some studies implemented problem-solving in group settings involving school staff. Promoting problem-solving is pivotal for fostering inclusive school environments, as it typically requires creative and responsive solutions tailored to students’ contextual and dynamic needs (Deroncele-Acosta & Ellis, 2024). Problem-solving facilitates students’ successful integration into the school community and strengthens their sense of belonging and connectedness by actively addressing and removing barriers to participation.

The techniques employed across connectedness, belonging, and engagement studies showed considerable overlap, making distinguishing between constructs in practical application difficult. Engagement interventions were more likely to adopt instructional approaches that align with their theoretical focus on behavioural participation (Appleton et al., 2008). However, the practical techniques often blurred the boundaries between constructs. This overlap reinforces the argument that connectedness, belonging, and engagement can be viewed as synonymous within intervention research. For example, while engagement is theoretically linked to student motivation and participation, the techniques used to promote engagement frequently resemble those grounded in interpersonal mechanisms of change.

4.3. Intervention Components

The analysis of intervention components revealed that curriculum-based and mentoring (adult- or peer-led) interventions had small to moderate effects on fostering connectedness, belonging, and engagement. Curriculum-based interventions support positive student outcomes (Durlak et al., 2011; Hattie, 2008) but can be challenging to implement within already crowded curricula (Priestley & Biesta, 2013). Though the results vary, mentoring has shown positive effects on adolescent development (DuBois et al., 2011). Some studies report small or non-significant impacts on academic outcomes and attendance (Wood & Mayo-Wilson, 2012). This may be due to a difference in measurement focus, as mentoring often enhances relationships without directly improving academic metrics (Wood & Mayo-Wilson, 2012). Sustainable impacts may require integrating mentoring into broader school systems (Valdebenito et al., 2018), with flexibility in the session structure to meet individual student needs (Rhodes et al., 2006). Similarly to intervention techniques, no clear patterns emerged in the use of components when comparing connectedness, belonging, and engagement interventions.

4.4. Multi-Tiered and Targeted Interventions

Multi-tiered interventions produced the largest effect sizes of any subgroup, supporting the effectiveness of multi-tiered support systems for interventions that address students’ needs at multiple levels: universal, targeted, and specialised (Stoiber & Gettinger, 2015). Single-tier interventions, while effective, showed smaller effect sizes, reinforcing the importance of addressing students’ needs comprehensively. Multi-tiered approaches were more prevalent in interventions targeting connectedness and belonging, potentially reflecting how their narratives pertain to more multi-faceted interventions (K. Allen et al., 2018; Waters et al., 2009).

Interventions targeting students at risk of low connectedness, belonging, and/or engagement showed larger effect sizes than those aimed at all students. However, interventions aimed at all students were also shown to have significant effects. The studies were roughly split between those targeting at-risk students and those focusing on all students, with a mix of target groups across the three constructs. Previous research suggests that interventions can be particularly effective for vulnerable students, including those with disabilities (K.-A. Allen et al., 2022). These findings reinforce the idea that interventions can be successful at both targeted and universal levels of provision, highlighting how connectedness, belonging, and engagement interventions can benefit all students involved.

4.5. Implications for Practice

This study highlights critical considerations for designing interventions and school environments that enhance students’ connectedness, belonging, and engagement. Ecological approaches integrated with interpersonal and environmental techniques proved to be the most effective.

Effective interpersonal techniques generally include the application of problem-solving strategies, which facilitate the development of more robust relationships among students in the school environment. Increasing students’ problem-solving capacity was often achieved through mentoring, whether through adult-led or peer-led support. This emphasises the critical importance of using problem-solving to foster inclusive environments by addressing barriers with creative and responsive solutions tailored to students’ needs (Deroncele-Acosta & Ellis, 2024). This technique strengthens belonging and connectedness by fostering collaboration among students and educators. Problem-solving should be considered a key technique for school leaders and interventionists seeking to foster connectedness, belonging, and engagement among secondary school students.

Studies within this review also effectively implemented environmental changes. The environmental techniques employed necessitated a whole-school cultural transformation, primarily aimed at fostering relational values such as cooperation and communication among school community members. Additionally, these techniques sought to integrate the principles of restorative practice into the school’s behaviour management policy and to enhance the overall climate of the institution as perceived by students and staff. This underscores the importance of schools’ and interventionists’ consideration of whole-school changes to the educational environment to sustainably foster connectedness, belonging, and engagement among students across the community.

Curriculum-based designs, particularly when integrated into broader school systems, also demonstrate significant benefits. The potential impact of curriculum-based interventions is well documented in the literature (Rhodes et al., 2006). However, practical challenges arise in gaining stakeholder buy-in due to the often crowded curriculum and limited classroom time available for new initiatives (Domitrovich et al., 2010). Despite these challenges, the positive influence of curriculum-based interventions on student outcomes should not be overlooked. Those designing interventions and school environments should consider concise, targeted curricula that promote connectedness, belonging, and engagement.

Overall, a key finding of this review suggests that schools and interventions that combine effective techniques (e.g., mentoring, problem solving, curriculum, and whole-school changes) at both the universal and targeted levels of support are the most successful at improving connectedness, belonging, and engagement for students (Stoiber & Gettinger, 2015). Moreover, it is important to recognise that tailoring the implementation of interventions to specific school contexts is essential, given the need to consider cultural and demographic factors when designing wellbeing-focused interventions (Caldwell et al., 2019; Curran & Wexler, 2017).

4.6. Limitations

Despite its contributions, this review has several limitations. Firstly, the intervention approaches relied on the original authors’ descriptions of their theoretical frameworks, potentially introducing inconsistencies between stated approaches and the actual components or techniques used. However, these discrepancies reflect the current state of the literature and provide valuable insights into the variability in intervention design and reporting.

Significant variability in the measurement tools used was observed, with 13 different tools applied across 16 studies. These inconsistencies align with previous findings (Salmela-Aro et al., 2021; Wong et al., 2024). However, this limitation was mitigated by including only validated measurement tools, ensuring a consistent baseline for comparison (Melendez-Torres et al., 2015). The variability between studies was further addressed through subgroup analyses and the systematic coding of intervention characteristics.

Finally, this review focused solely on RCTs, the gold standard for evaluating intervention efficacy. While this ensures robust findings, it may have excluded effective non-random methods (Reeves et al., 2019) and limited the number of eligible studies, reducing subgroup analysis opportunities. Nonetheless, this focus on high-quality evidence strengthens the reliability of the conclusions drawn.

4.7. Recommendations for Future Research

Future research should focus on clarifying and standardising the definitions and measurement tools for connectedness, belonging, and engagement to enhance consistency and comparability (Flay et al., 2005). Further, to address the inconsistent findings between many of the theoretical approaches cited and the intervention techniques used, future intervention research should develop and apply robust theoretical frameworks to design and evaluate interventions that improve connectedness, belonging, and engagement. The involvement of key school community members, such as students, staff, and caregivers, in refining measurements and intervention frameworks would ensure the comprehensive and practical applicability of future interventions (Ching et al., 2024).

Future research should also investigate the interplay between connectedness, belonging, and engagement to deepen the understanding of their interrelationships and the mechanisms driving effective interventions. Studies should emphasise culturally adaptive and inclusive approaches to enhance relevance across diverse school settings (Caldwell et al., 2019; Curran & Wexler, 2017). Research targeting specific subgroups, such as marginalised or at-risk students, should also be given priority to maximise impact.

Furthermore, longitudinal studies are crucial for evaluating the sustainability of intervention effects and identifying components that contribute to long-term success. Researchers should explore alternative designs beyond RCTs, such as rigorous quasi-experimental or mixed methods approaches, to encompass a broader range of effective interventions.

4.8. Conclusions

This review provides a comprehensive synthesis of interventions designed to enhance school connectedness, belonging, and engagement. It addresses significant gaps in the literature through the analysis of 16 studies and a meta-analysis of 11. The findings demonstrate moderate improvements across these constructs, highlighting the promise of evidence-based interventions in secondary schools.

This review identified a diverse range of interventions, with ecological approaches showing the highest effect sizes, supported by cognitive and emotional approaches. While behavioural interventions were less effective, the findings reinforce the importance of systemic and relational strategies. Intervention approaches often reflected their theoretical frameworks but varied in practical application. Problem-solving emerged as a key technique, fostering inclusion by addressing barriers through collaborative and tailored solutions.

This meta-analysis revealed moderate overall effectiveness, with large within-group and moderate between-group effect sizes. Multi-tiered systems of support, addressing universal, targeted, and specialist levels, were particularly effective, especially for at-risk populations. Tailored approaches considering cultural and demographic diversity further enhanced outcomes. Curriculum-based and mentoring interventions showed small to moderate benefits, with mentoring being particularly impactful when integrated into broader school systems.

The findings of this review highlight several effective techniques that schools and interventionists can use to practically foster connectedness, belonging, and engagement for students. These include implementing mentoring systems, increasing access to problem-solving, considering curriculum designs, and making whole-school changes to the school environment. Using a combination of these techniques across both universal and targeted levels appears to have the most significant practical impact on student outcomes. This review underscores the practical overlap between connectedness, belonging, and engagement, advocating for holistic interventions that incorporate these constructs simultaneously. It also highlights the need for the development of unified theoretical frameworks that integrate all three constructs, guiding the effective incorporation of connectedness, belonging, and engagement in future interventions. Future work can advance academic and socio-emotional outcomes by refining measurement tools, clarifying theoretical frameworks, and developing integrated interventions, ensuring relevance and sustainability across diverse school communities.