Several databases (Scopus, EBSCO, Eric, PsycINFO, and Science Direct) were chosen based on their relevance to the study subject area and accessibility through university subscription. Potential keywords were extracted from the review questions provided. To combine keywords, a comprehensive search string was used, utilizing the Boolean operators OR and/or AND, phrase searching and truncation. Thus, the final search was carried out using the following search terms: (e.g., “school-based mentoring” AND “program effect*” OR “program success*” OR “program evaluation*” OR “program assessment* OR “program outcome*”). The systematic search covered the period from 2010 to January 2025 in English-language peer-reviewed journal articles. The publication time frame was chosen because studies conducted before 2010 were included in Wood and Mayo-Wilson (2012) meta-analysis, indicating that the systematic reviews and meta-analyses were completed in 2010 or earlier. The initial search across the selected databases resulted in 719 articles, of which 437 remained after duplicate articles were removed from the search. Next, the article title and abstract were checked by two authors based on inclusion criteria. The flowchart in Figure 1 illustrates the procedure for selecting the studies for analysis.

Inclusion criteria: (1) The presence of a mentoring program (group or individual) in the school environment. (2) Participants included school-aged youth, defined as those expected to be between 6 and 19 years of age. (3) Empirical studies should focus on the SBM. Two authors assessed all studies based on the inclusion criteria. Through title and abstract screening, 386 publications were excluded due to irrelevance based on the predefined inclusion criteria. This process identified 51 studies that met the initial inclusion criteria. In the next stage of the search, the full texts were retrieved and reviewed for eligibility with the following additional criteria: (1) Studies that focused on group or individual SBM programs were included. (2) Studies that assessed the program’s effects on at least one outcome variable, as well as those that examined correlations or predictions related to positive youth development. (3) Each study was required to include a quantifiable program evaluation achieved either through pre- and post-program assessments or by comparing a control group that did not receive the intervention. After further elimination, 27 studies met the inclusion criteria for this review (Figure 1).

The appraisal process was performed using the Mixed Method Appraisal Tools (MMAT) and its user guide developed by Hong et al. (2018). The MMAT is suitable for systematic reviews with diverse methodologies. The appraisal begins with two screening questions: (1) Are the research questions clear 2) Do the collected data allow the research questions to be answered? All 27 studies met the criteria. We then applied the MMAT to assess the methodological quality of each study. Each study design (qualitative, quantitative, randomized controlled trials, non-randomized studies, and mixed methods studies) was evaluated using five methodological criteria questions, with three possible answers: yes, no, or cannot tell.

For instance, the five criteria for quantitative studies are as follows: Is the sampling strategy relevant to addressing the research question? Was the sample representative of the target population? Are the measurements appropriate Is the risk of non-response bias low? Is the statistical analysis appropriate to answer the research question?). Articles that met at least three of these criteria were included in this study. Studies were assigned an overall quality score ranging from (0/5) to (5/5) based on methodological quality criteria. Overall, most of the studies included were of high quality: 14 studies achieved a score of 5/5, nine scored 4/5, and four scored 3/5. No study was excluded based on the quality score.

Most studies used established measurement scales and statistical analyses to address their research questions. There were just as many or more “Cannot Tell” responses compared to “No” suggesting that in some studies, lower appraisals may reflect issues in reporting rather than actual quality. Two researchers independently conducted the assessments of the included studies. The quality assessment of each study is presented in (Supplementary Table S1).

The SLR was conducted qualitatively, and relevant data were extracted as statements. Qualitative thematic analysis was used to extract data from the reviewed studies, which could then be used to address the research questions. Thematic analysis is a method for identifying, analysing, and reporting patterns (themes) within data. Inductive thematic analysis was considered the most appropriate method for synthesizing a mixed research design (Flemming et al., 2019). The analysis was guided by the six-step thematic analysis method prescribed by Braun and Clarke (2006).

The full texts of the 27 included studies were used to create a database for this review. The authors divided up the articles. The process began with two authors familiarizing themselves with the articles by individually reading the results section of the studies. Subsequently, we identified significant segments of the results that were related to outcome measures. Key information necessary to answer the research questions was then collected. Subsequently, the extracted data (outcome measures) from each study were organized through coding, which involved assigning codes to the outcome measures by interpreting their meanings. The relevant coded data extracts were then grouped into common themes for analysis. At this stage, four main themes and 18 subthemes were established. The two authors maintained frequent communication during the data extraction process.

The next step involved reviewing the themes, whereby the authors checked the relevance of the constructed themes to the extracted data. The following section reports the themes and subthemes, labeled as domains and subdomains, respectively. The extracted bibliographic data from the included studies were also reported (e.g., contextual information, name of the mentoring program, research design, methods of data collection, length of the program, target groups, samples, and estimated effect sizes).

Of the selected articles, 20 studies were conducted in the United States, and the others were conducted in Korea, Sweden, Australia, Portugal, and the UK. In terms of publication years, three were published in 2010, three in 2011, two in 2012, two in 2013, one in 2014, two in 2015, two in 2017, two in 2018, two in 2019, one in 2020, three in 2021, two in 2022, and two in 2024. The studies were conducted across various types of schools: five in elementary schools, seven in middle schools, four in secondary schools, and the rest in a mix of elementary, middle, and secondary schools (See Table 1).

Of the 27 studies included in the review, 23 were quantitative, one was qualitative, and three used a mixed methods design. Most of the studies incorporated a control group, with most experimental designs employing random assignment of participants (Bodin and Leifman, 2011; Herrera et al., 2011; McQuillin et al., 2011; Schwartz et al., 2012; Núñez et al., 2013; Simões and Alarcão, 2014; Bayer et al., 2015; Heppen et al., 2018; Goulet et al., 2018; Lyons et al., 2019; Jablon and Lyons, 2021; Green et al., 2021, McQuillin and McDaniel, 2021; Green, 2022; Martins et al., 2024). When random assignment was not possible, eight studies adopted a quasi-experimental approach (Choi and Lemberger, 2010; Elledge et al., 2010; Nixon and Werner, 2010; Gordon et al., 2013; Wasburn-Moses and Noltemeyer, 2018; Larose et al., 2018; Chan et al., 2020; Alwani et al., 2023). Four studies utilized an exploratory design with pre- and post-surveys (Keller and Pryce, 2012) Crooks et al., 2017; Meltzer et al., 2020; Butler et al., 2024) due to the nature of the study.

In 11 studies, the program duration was less than one year; seven studies covered one school year, and nine studies extended over a period ranging from one and a half to three years (See Table 1).

In line with the study purpose and qualitative approach, the domains are presented in order from the most frequently mentioned in the literature to the least, indicating the measurement priorities of scholars. The review resulted in four main domains and 16 subdomains. The school functioning domain was the most widely assessed in the literature, followed by psychological well-being, social and interpersonal functioning, substance use, and behavioral conduct. A summary of each study’s outcome domains and measures is provided in Table 2.

The majority of quantitative studies reported standardized effect size measures, specifically Cohen’s d, which identified certain d values as indicative of small, medium, and large effects. Several studies have cited partial eta squared (ηp²), and Eta squared (e.g., Simões and Alarcão, 2014; Wasburn-Moses and Noltemeyer, 2018), along with Hedges’ g (e.g., Herrera et al., 2011; Green et al., 2021; Green et al., 2022) as measures of effect size. Cohen’s d and Hedges’ g values are interpreted in a similar manner (e.g., small effect = 0.2, medium effect = 0.5, large effect = 0.8). However, the interpretation of the partial eta squared (η²_p) is different: small effect = 0.01, medium effect = 0.06, large effect = η²_p ≥ 0.14.

Few studies did not include a measure of effect size (e.g., Crooks et al., 2017; Goulet et al., 2018; Larose et al., 2018; Jablon and Lyons, 2021) despite reporting significant positive outcomes from the mentoring program. These studies primarily emphasized beta coefficients (β) and R-squared values derived from multivariate models to interpret the magnitude of these effects within their study contexts. For these studies, we used an Effect size calculator to compute effect sizes based on either the observed squared multiple correlations (R²) or the means and standard deviations of the treatment and control groups (Table 3).

Most studies (n = 23) found that SBM interventions were effective for at least one outcome. Among the 27 studies, one study indicated a large effect size (Elledge et al., 2010). Martins et al. (2024) also reported a large effect size across all dependent variables; however, when considered individually, the effect sizes were medium to small, depending on the dependent variable. Three studies reported large to medium effect sizes for at least one outcome measure (Nixon and Werner, 2010; Gordon et al., 2013; Green et al., 2021), two studies reported medium effect sizes (Simões and Alarcão, 2014; Green et al., 2022), seven studies reported medium to small effect sizes (Keller and Pryce, 2012; Gordon et al., 2013; Núñez et al., 2013; Goulet et al., 2018; Lyons et al., 2019; McQuillin and McDaniel, 2021; Martins et al., 2024), and eight reported small effect sizes (Choi and Lemberger, 2010; Bodin and Leifman, 2011; Herrera et al., 2011; Bayer et al., 2015; Wasburn-Moses and Noltemeyer, 2018; Crooks et al., 2017; Chan et al., 2020; Jablon and Lyons, 2021). Five studies showed no positive changes resulting from mentoring, precluding any conclusions about program effectiveness (McQuillin et al., 2011; Schwartz et al., 2012; Heppen et al. (2018); Alwani et al., 2023; Butler et al., 2024) (Table 3).

This study supports the effectiveness of the SBM program in positively influencing youth’s development. The program has proven effective in fostering positive outcomes across a diverse range of school populations, including elementary, primary, and secondary school students. Most studies highlighted significant gains in one or more outcome domains. It has successfully improved academic achievement (Choi and Lemberger, 2010; Núñez et al., 2013; Wasburn-Moses et al., 2015; Larose et al., 2018; Lyons et al., 2019; Meltzer et al. (2020); McQuillin and McDaniel, 2021; Martins et al., 2024), increased school engagement (Gordon et al., 2013; Goulet et al., 2018; Martins et al., 2024), enhanced school attendance (Gordon et al., 2013; Simões and Alarcão, 2014), and self-regulated learning (Meltzer et al., 2020; Martins et al., 2024), and decreased disciplinary referrals (Gordon et al., 2013; McQuillin et al., 2021; Wasburn-Moses and Noltemeyer, 2018). Evidence of the positive impact of mentoring on various school-related outcomes suggests the potential benefits of mentoring programs in enhancing students’ educational experiences. This could be due to the fact that SBM program are often employed to increase students’ academic performance and educational achievement (DuBois and Karcher, 2014) resulting in stronger impact on school-related outcome domains.

The program also helped students develop a variety of skills, such as interpersonal communication, problem solving, and decision-making (Meltzer et al., 2020; Green et al., 2021, 2022). The results also highlight the effectiveness of the program in positively increasing youth self-esteem and self-efficacy (Núñez et al., 2013), resiliency (Green et al., 2021, 2022), connectedness with peers, teachers, and family (Gordon et al., 2013; Crooks et al., 2017), mental health (Crooks et al., 2017; Meltzer et al. (2020); Green et al.,2021, 2022;), and overall life satisfaction (McQuillin and McDaniel, 2021). Additionally, this review analysis revealed that the program led to reductions in peer victimization (Elledge et al., 2010; Nixon and Werner, 2010), behavioral problems (Gordon et al., 2013; Jablon and Lyons, 2021), aggressive behavior (Nixon and Werner, 2010; Keller and Pryce, 2012; Jablon and Lyons, 2021), delinquency (Lyons et al., 2019), and depression among youth (Keller and Pryce, 2012).

These findings are consistent with previous research and reviews (e.g., McQuillin et al., 2011; McArthur et al., 2017; McCoy, 2017; Karaferye, 2018; Grey, 2019; May et al., 2021; Jefferson, 2022; Fehérvári and Varga, 2023; Richardson, 2024; Turgut and Tasçi, 2024), which demonstrate the positive impact of SBM programs on academic, social, and behavioral outcomes. Especially, SBM programs had a statistically significant positive effect on students’ school attendance, grade point average, social behavior, and discipline referrals (McArthur et al., 2017; McCoy, 2017; May et al., 2021; Richardson, 2024). Our findings also correspond with the literature (Moor, 2018; Raposa et al., 2019; Keller et al., 2020; Kuperminc et al., 2020), indicating that the school functioning, self-esteem, and well-being of the mentored group significantly improved after mentoring.

Despite the positive effects of SBM, the program was not entirely successful in several studies (e.g., Bodin and Leifman, 2011; McQuillin et al., 2011; Herrera et al., 2011; Schwartz et al., 2012; Bayer et al., 2015; Heppen et al. (2018); Alwani et al., 2023; Butler et al., 2024). In these studies, where mentoring failed to produce significant benefits, the main issue was program implementation fidelity (e.g., Bodin and Leifman, 2011). Schwartz et al. (2012) found that 39% of the participants reported that teachers did not consistently permit students to attend the mentoring program during school hours. As a result, the program was ineffective during school hours; however, when meetings occurred after school or during lunch, notable academic benefits were observed. In fact, the impact of SBM on academic outcomes was moderated by the time of day that matches met.

Another factor contributing to the program’s ineffectiveness was mentors’ lack of knowledge and preparation for it, as highlighted by Bayer et al. (2015), Goulet et al. (2018), and Alwani et al. (2023). Alwani et al. (2023) found that the mentoring program was associated with a decrease in students’ life satisfaction, self-esteem, behavioral engagement, and motivation to learn. The authors further elaborate that a mentoring program may fail when mentors discuss goals and academic challenges in ways that suppress the mentee’s voice, potentially making the experience feel discouraging or unpleasant. Goulet et al. (2018) emphasized that lower academic achievement in one elementary school was linked to program implementation issues, specifically in staff selection and training, which may have hindered effective program implementation. The negative outcomes or adverse effects of the program were linked to the program itself (Foulkes and Stringaris, 2023). Schwartz et al. (2012) also reported negative effects on students’ academic outcomes, suggesting that the program may inadvertently harm students. The evidence of potential harm highlights the need for careful implementation. SBM programs, while aimed to support young people, can sometimes lead to iatrogenic effects, causing harm rather than benefit. Therefore, addressing potential harm is crucial for designing and implementing effective mentoring programs in the future. To ensure the mentoring programs are effective, it is crucial to train mentors or hire those with specialized expertise and experience, particularly in facilitating discussions with students aimed at behavior change (Ambrosetti, 2012; Karaferye, 2018; Hickman and Anderson, 2019).

Heppen et al. (2018) observed that the timing of the program and variations in the target population contributed to the ineffectiveness of their programs. They also noted that significant challenges, including high student mobility, large mentor caseloads, and a lack of clear service differentiation between treatment and control groups, affected program quality, and consequently, students generally did not experience the program as intended. Additionally, Herrera et al. (2011) noted that the mentoring program’s limited effectiveness was related to program design and delivery, engagement levels, program evaluation, and methodological constraints in their study. Similarly, Butler et al. (2024) noted methodological limitations in their study, which employed a pre/post design similar to many other assessments of violence prevention initiatives. However, they did not include an appropriate control group, which constrained their ability to establish a direct causal link between the program and any observed changes. The authors emphasized the necessity for further research with a more robust study design.

Another factor highlighted in these studies, where mentoring did not yield significant benefits, was the program’s duration. Two studies found that a short-term SBM program (lasting less than six months) did not offer substantial advantages to participants (e.g., McQuillin et al., 2011; Schwartz et al., 2012). This aligns with the literature suggesting that longer mentoring relationships, particularly those lasting a year or more, are associated with improved outcomes in various domains such as mental health, behavioral development, and academic performance (Damm et al., 2022; DeWit et al., 2016; Johnson et al., 2011).

In contrast, among the studies we reviewed, significant positive outcomes were observed even with short mentoring durations of 3 to 9 months (e.g., Green et al., 2021; Simões and Alarcão, 2014; Meltzer et al., 2020). These results are consistent with previous research (Weiler et al., 2015; McClain et al., 2021). Therefore, while the duration of mentoring relationships appears to be important, it is not the only factor influencing program effectiveness. This implies that prioritizing the quality of mentorship and the support structure of the mentoring program may be more effective in improving outcomes than simply extending the program length. For instance, Heppen et al. (2018) evaluated long-term impacts over three years; however, they did not find significant benefits, which they attributed to the quality of program implementation and fidelity. Additionally, as noted, this finding suggests that, although the duration of the mentoring program may be an important determinant of youth outcomes in the SBM program, it may be a less important factor in overall program effectiveness (Raposa et al., 2019).

Our findings from studies that did not show significant benefits from mentoring align with earlier systematic reviews and meta-analyses by Wood and Mayo-Wilson (2012) and Wheeler et al. (2010). Wood and Mayo-Wilson (2012) attributed the lack of positive outcomes to factors such as inadequate mentor training, ineffective mentor-mentee pairing, poor program implementation, and the short duration of the programs, with the longest lasting only a few months for most young participants. The authors further suggested that larger studies with multiple replications may be necessary. Likewise, Wheeler et al. (2010) highlighted factors such as the absence of supplementary implementation support and a relatively high percentage of youth in the intervention group who were never paired with a mentor, which affected certain outcomes.

Our analysis further indicated that although most studies reported positive and significant changes, the overall effect sizes, where reported, were generally small or small to medium. This aligns with earlier reviews suggesting that the impacts of mentoring are significant but not large (Hall, 2003; McQuillin et al., 2011; Tolan et al., 2014; Raposa et al., 2019). Similarly, Wheeler et al. (2010) reported that SBM can be modestly effective for improving selected outcomes, particularly one year of participation in the program, which generally leads to a moderate impact on a range of school-related outcomes. Nonetheless, it is crucial to highlight that even slight improvements in various areas of youth functioning can significantly affect positive youth development (Tanner-Smith et al., 2018). Our study findings contradict previous systematic analyses and meta-analyses by Wood and Mayo-Wilson (2012), which reported that SBM programs were ineffective. In this review, the mentoring program did not improve any intended outcomes.

This review has important limitations that are worth noting. First, our search strategies may have missed studies that met the inclusion criteria, and we may have missed relevant studies in non-English-speaking contexts, as our review was confined to English publications. Second, this review focused on SBM programs as a common and important setting for implementing mentoring programs to foster positive development (DuBois and Karcher, 2014). This context has the potential to reach a broader population of young people, providing them with opportunities to address their specific needs, which directly impact their success. In addition, by developing relationships within the school environment, mentors can interact with students over long periods as they advance through various grade levels (Jefferson, 2022). However, focusing on SBM programs limits the generalizability of our findings to other youth populations and contexts.

Third, the results should be generalized with caution because of the significant variability identified among the different studies. For instance, there are considerable differences in methodology, sample size, program delivery, and the contexts of the studies, which range from elementary to middle and secondary schools from which participants were drawn. Additionally, most of these studies were conducted in the USA. Another limitation of this study is that it focused only on peer-reviewed articles as a source of data. Future reviews should include other sources, such as books, theses, and dissertations. The implications of this study for research and practice highlight the importance of incorporating stronger theoretical underpinnings and adopting broader outcome measures in youth mentoring programs. The findings suggest several recommendations for future studies. Notably, research on the effectiveness of mentoring programs shows significant room for improvement (Farruggia et al., 2011; Schwartz et al., 2012). Given the critical role of youth mentoring programs, more studies are needed to rigorously evaluate their impacts and effectiveness.

Most articles in this review were quantitative; future studies should consider using more qualitative or mixed-methods designs. A qualitative perspective provides an in-depth analysis and detailed explanation of mentoring effectiveness. Therefore, while some areas can be measured through survey questionnaires, others need to be qualitative to ensure that the assessment meets the specific needs of both mentors and mentees. Geographically, empirical studies on youth mentoring have mostly been conducted in the United States. Although mentoring programs are popular in many countries, few studies have been conducted on mentoring program outcome assessments outside the United States. More research is needed on mentoring program assessments in other countries. Finally, an overall assessment must come from a variety of sources so that the information can be used effectively to make program decisions. Future program assessments must use a multi-informant approach and consider including data from parents, teachers’ reports, and mentors to increase the validity of the employed measurement scales (Marino et al., 2021). Overall, this review calls for future research efforts to be guided by robust theoretical foundations while also encouraging the broadening of the scope of outcome measures that better capture the impact of youth mentoring programs.