Introduction

Following the transition to formal schooling, families and schools become two of the most important environments responsible for shaping children’s development (Bronfenbrenner & Morris, 2006). Therefore, the three main stakeholders in children’s development (i.e., child, parents, teachers, and schools) tend to form relationships with, and exert influence on one another. To date, most scholarly attention has focused on the unidirectional aspects of the relationship between a child’s home and their school by investigating either pedagogical and instructional practices teachers can apply in the classroom, or certain types of activities parents can engage in at home (e.g., reading to children, assisting with homework) to improve children’s learning, wellbeing, and development (Smith et al., 2020). Many models have been proposed to explain parent factors (e.g., Epstein, 1992; Hoover-Dempsey et al., 1995, 1997). Increasing evidence shows the importance of considering not only the child-caregiver interactions (e.g., teachers’ and parents’ influences on children and the reciprocal children’s influences on their parents and teachers; Albanese et al., 2019; Fang et al., 2021) but also the relationships, communications, and partnerships parents have with primary school teachers and schools (Leenders et al., 2019; Sheridan et al., 2019; Smith et al., 2020).

Represented by Bronfenbrenner’s mesosystems (Bronfenbrenner & Morris, 2006), the terms home-school partnership or family-school partnership have been defined and operationalized in several different ways across studies and programs. This study adopted Bierman and Sheridan’s definition (2022), where the home-school partnership is defined as a multi-dimensional construct that includes contributions made by parents, teachers, and schools, as well as the dynamic quality of parent-teacher communication and relationships to promote the learning and wellbeing of children. Effective home-school partnerships often involve processes occurring at home and at school, roles played by both parents and teachers, behavioral supports, positive attitude and expectations, as well as the parents’ interactions with teachers and schools (Bierman & Sheridan, 2022; Kim & Sheridan, 2015; Smith et al., 2020).

Parents are an integral part of the home-school partnership and their contributions are central to its effectiveness. Past research on parents’ contributions to the home-school partnership has mainly focused on the multi-dimensional construct of parental involvement (Bierman & Sheridan, 2022), which includes home-based involvement (e.g., helping with homework, talking to children about their experiences and activities at school), school-based involvement (e.g., volunteering at school, participating in activities organized by the school), parental expectations and aspirations, and parent-teacher communication (Kim & Sheridan, 2015; Pomerantz et al., 2007; Wilder, 2014). Traditionally, parental involvement had been viewed as a unidirectional construct where teachers and schools have limited influence (Epstein, 1992; Grolnick & Slowiaczek, 1994), whereas increasing evidence suggests that the construct is bidirectional where parents’ expectations, participation in activities, and their proactive communication can be both facilitated or undermined by certain school and teacher practices. General school climate, parents’ relationships with school personnel, and teachers’ attitudes and behaviors can both facilitate or represent a barrier to parents’ involvement (Green et al., 2007; Sheridan et al., 2019; Smith et al., 2020). The parent-teacher communication dimension has also been expanded from simply the quantity of unidirectional communication (e.g., attending parent-teacher conferences) to the quality of bi-directional communication and relationships (e.g., warmth, collaboration; Bierman & Sheridan, 2022). To reflect the above-mentioned increasing recognition of teachers’ and schools’ roles in parents’ participation in this important partnership, while Goodall and Montgomery (2014) differentiate parental involvement and parental engagement, as points on a continuum representing the level of parents’ participation in their child’s education, the term parental engagement has been increasingly used to replace the term parental involvement (Bierman & Sheridan, 2022). From this point forward, this study will use the term parents’ contributions to the home-school partnership to account for the varied definitions of terms.

The Importance of Parents’ Contributions to the Home-School Partnership

Decades of extensive research has demonstrated that parents’ contributions to high-quality home-school partnerships have significant, positive impacts on primary-school-aged children’s learning, with many theories and models about the potential mechanisms involved being proposed and tested (Barger et al., 2019; Bierman & Sheridan, 2022; Smith et al., 2020). Wilder (2014) conducted an umbrella meta-analysis of nine meta-analyses examining the association between parents’ contributions to the home-school partnership, specifically, parental involvement in their child’s learning and their academic achievement. A positive relationship was found, regardless of the definition of parental involvement (e.g., parent–child communication about school, homework assistance, attendance and participation in school activities, education expectations and aspirations; communication with school) or the measure of academic achievement (e.g., standardized test scores, GPA, teacher’s ratings). Another qualitative systematic review involving 75 studies identified four specific examples of parental involvement frequently identified in previous literature to have positive associations with children’s academic achievements: reading at home, holding high expectations and aspirations, communicating with children about school, and encouraging and supporting children’s learning (e.g., praising their children’s performance and efforts; Boonk et al., 2018).

Recent studies have provided a more thorough examination of this relationship. For example, Barger and colleagues (2019) conducted a comprehensive meta-analysis of 480,830 families from 448 independent studies and found that parents’ contributions to the home-school partnership, through their involvement in home- and school-based activities (except homework assistance), had small but meaningful and positive associations with children’s academic achievement (r = 0.13), academic engagement (r = 0.15), and motivation (r = 0.23). Moreover, Kim (2022) conducted a recent umbrella meta-analysis of 23 meta-analyses covering 1177 independent studies published in the last five decades and found similar outcomes to previous studies (Barger et al., 2019; Wilder, 2014). Kim (2022) found a positive association between all types of parent participation except homework assistance and student achievements (r = 0.18). Beyond academic achievements, engagement, and motivation, parents’ contributions toward effective home-school partnerships have been found to be associated with better social, emotional, and behavioral wellbeing of children. In Barger et al.’s (2019) comprehensive meta-analysis, parent participation at home and at school was positively associated with children’s social adjustment (r = 0.12) and emotional adjustment (r = 0.17), and negatively associated with children’s delinquency (r =  − 0.15). Meta-analyses tend to suggest that the strength of these associations is generally consistent across children’s age, gender, ethnicity, and family socio-economic status (Barger et al., 2019; Wilder, 2014). It should be noted that non-significant moderation effects in meta-analyses are often due to lack of statistical power (Barger et al., 2019), and should not be interpreted as evidence for the absence of moderation effects as these factors are often identified as significant moderators in individual studies (particularly for ethnicity and socio-economic status; for details, please see the systematic review conducted by Boonk et al., 2018).

The findings of home-school partnership intervention studies, where the levels of parental involvement were experimentally manipulated, strengthen the findings of the positive outcomes associated with high levels of parents’ contribution to the home-school partnership. A meta-analysis of 77 evaluation studies of home-school partnership interventions (e.g., organizing structured parental involvement activities, facilitating parent-teacher communications through note exchanges) found that the interventions resulted in improved academic outcomes (g = 0.25), greater levels of child participation in learning (g = 0.25), better children’s social-behavioral competence (g = 0.32), and improved mental health (g = 0.34). Additionally, a more recent meta-analysis of 39 studies that focused on children in lower primary school identified comparable effect sizes regarding the influence of home-school partnership interventions on reducing behavioral problems (g = 0.38) and promoting social-emotional skills (g = 0.33; Cosso et al., 2022). A program component analysis suggested that effective home-school partnership programs targeting both the relational aspect (e.g., parent-teacher communication, collaboration, and relationship) and the structural aspect (e.g., both home- and school-based participation) were effective in affecting positive changes in children’s socio-emotional wellbeing (Sheridan et al., 2019). These interventions often involve increases in parents’ contributions to these relational and structural aspects of the home-school partnership. The aforementioned interventions were shown to be more beneficial for children from racial minorities (e.g., African and Latino American) and children living in rural settings. This is likely due to the limitations in traditional communication and engagement practices resulting in a lower quality of pre-intervention home-school partnerships (Sheridan et al., 2019).

Factors Associated with Parents’ Contributions to the Home-School Partnership

Parents’ contributions to the effective home-school partnership are influenced by various factors with several explanatory models proposed (Yamauchi et al., 2017). The two most frequently applied theoretical frameworks of the determinants of parental involvement are Epstein’s types of family involvement and the Hoover-Dempsey model of the parental involvement process (Yamauchi et al., 2017). Epstein (1992, 1995) identified six types of family involvement practices schools can engage in, with each likely to lead to different outcomes in children and their families. These types include parenting (establishing positive home learning environment), communicating (exchanging information about school programs and learning progress between parents and teachers), volunteering (parents volunteering in school-based activities), learning at home (parents supporting children’s homework and other curriculum-related activities at home), decision-making (including parents in school decisions), and collaborating with community (utilizing community resources to optimize children’s learning; Epstein, 1992, 1995).

The Hoover-Dempsey model of parental involvement describes the antecedents of parental involvement, consequences on child adjustments, and mediation pathways (Hoover-Dempsey et al., 1995, 1997; Hoover‐Dempsey et al. 2005). It highlights three key family level factors as central antecedents: 1) parental motivational beliefs, which consist of an active role construction for participation where parents hold strong beliefs about how they should be involved, as well as a strong sense of self-efficacy that they can make positive impacts on children’s learning; 2) perceptions and beliefs about whether the school, the teacher, or the child want them to be involved; and 3) certain contextual factors in the family life that make parental participation easier or harder. Additionally, both models identify a list of factors associated with parents’ contributions to the home-school partnership, such as child age, socio-economic status of the family, school and the community, parents’ own experiences with schooling, culture, ethnicity, parents’ knowledge, skills, time and energy, child learning and behavioral problems, family stressors (e.g., single parents), school climate, as well as teachers’ and schools’ confidence and ability to engage with families (Epstein, 1995; Hoover-Dempsey et al., 1997). Moreover, the important role of parenting practices and self-efficacy has been emphasized in both models and many independent studies (e.g., Jeynes, 2018; Ogg & Anthony, 2019; Wilder, 2014).

The Associations Between Parental Self-Efficacy and the Home-School Partnership

An important parent level factor related to parents’ contributions to effective home-school partnerships is parenting self-efficacy, which is adapted from Bandura’s (1997) theory of self-efficacy and defined as parents’ beliefs in their capabilities to influence their child and the environment, to promote child development and other desirable outcomes (Albanese et al., 2019; Bandura, 1997; Sanders & Mazzucchelli, 2013). As one of the most frequently studied parent-level factors, extensive evidence has established the critical importance of parenting self-efficacy to be positively associated with a wide range of parenting practices and parent outcomes (e.g., effective and positive parenting practices, emotion regulation, parental adjustment, satisfaction with parenting), child outcomes (e.g., child self-regulation, child development, social, emotional, and behavioral wellbeing, learning), and relational outcomes (e.g., parent–child relationship, couple relationship, relationships with other carers such as teachers; Albanese et al., 2019; Fang et al., 2021). Given its importance, significant efforts have been made to promote parenting self-efficacy through the delivery of evidence-based parenting support programs (Sanders et al., 2019). As a key stakeholder in children’s learning and wellbeing, schools are one of the most important settings for providing parenting support and subsequently to develop relationships with parents (Sanders et al., 2021).

Parenting self-efficacy can be either general or task specific (e.g., education focused). Regarding task-specific parenting self-efficacy, many studies investigating the Hoover-Dempsey model of parental involvement have found that parenting self-efficacy in relation to helping their children succeed in school was positively associated with home-based involvement, school-based involvement, and parent-teacher relationship and communication (e.g., Ferretti et al., 2019; Ice & Hoover-Dempsey, 2010; Kwan & Wong, 2016; Newland et al., 2013; Reininger & Santana López, 2017; Semke et al., 2010). These findings have been replicated extensively and the aforementioned relationships were not only identified in studies conducted in the USA but also studies conducted in other countries in Asia and South America (Ferretti et al., 2019; Kwan & Wong, 2016; Newland et al., 2013).

Similarly, an increasing body of work has shown that parents’ self-efficacy related to their general parenting role (not limited just to helping in relation to schooling) has also been shown to be associated with the parents’ contributions toward building strong home-school partnership. A French study of 203 parents of first- and-second-grade children found that general parental self-efficacy was strongly associated with parents’ self-efficacy in supporting children’s schooling (r = 0.51–0.62) and positively associated with a range of parent behaviors (Tazouti & Jarlégan, 2019). Another cross-sectional study of 309 Japanese and 372 Korean mothers of primary school children found a positive association between parenting self-efficacy and parental involvement in their children’s education at home in both samples (Holloway et al., 2016). Similar findings were found in a Chinese cross-sectional study of 237 parents of children with autism spectrum disorders transitioning to school (Hou et al., 2023). Furthermore, a Korean longitudinal study of 1353 parents of primary-school aged children evaluated a three-wave cross-lagged panel model (Choe, 2022). Positive cross-lag regressions between parental self-efficacy and parental involvement in child’s education were found across the three waves.

A study of 192 Israeli parents found that highly efficacious parents tended to be more proactive and more likely to communicate with teachers when dealing with child socio-emotional problems. These parents were also found to be more independent and exhibit less avoidance when dealing with learning difficulties (Harpaz & Grinshtain, 2020). In a follow-up study, Grinshtain and Harpaz (2023) found that although the patterns of association are different across cultures (Jewish versus Arabic), the overall trend was that lower levels of parental self-efficacy were associated with less effective parent-teacher communication practices. In a recent cross-sequential study with a national representative sample of 8152 Australian primary-school-aged children, Ma and colleagues (2024) found that parenting self-efficacy was positively associated with the perceived quality of parent-teacher communication at two years follow-up, and this association was over and above both the dynamic influences of child behavior at school and the teacher–child relationship. Limited influence of child gender, parent gender, or cohort differences was found.

Recent progress on parenting self-efficacy research has moved toward a focus on the broader construct of parental self-regulation. This extends Bandura’s (1997) work on self-efficacy and personal agency, and has been viewed as a more comprehensive explanatory framework for parenting self-efficacy. The parental self-regulation framework views parenting self-efficacy as a core element that is strongly associated with parents’ capacity to self-monitor; to regulate their own attributions, emotions, and behaviors; as well as to problem solve independently to achieve their parenting goals (other core elements are self-sufficiency, self-management, personal agency, problem-solving; Morawska et al., 2019; Sanders & Mazzucchelli, 2013; Sanders et al., 2019). As an emerging area of research, the available evidence on parental self-regulation primarily focused on parenting self-efficacy (Sanders et al., 2019), which has been found to be a strong indicator of parents’ overall self-regulatory capacity (r =  ~ 0.70, factor loading = 0.94; Hamilton et al., 2015; Tellegen et al., 2022). Measures of parental self-regulation have only been recently developed (e.g., Hamilton et al., 2015; Tellegen et al., 2022). However, to date, very limited research has investigated the associations between parental self-regulation and the parents’ contributions to the home-school partnership. Previous studies have mainly focused on the parenting self-efficacy element of parental self-regulation (as indicated by the studies included in this introduction).

In summary, for many decades, many individual research projects have demonstrated the link between parenting self-efficacy and parents’ involvement in children’s education, with a range of positive associations found. To provide further clarity regarding the strength of these associations across different conceptualizations of both parenting self-efficacy and parents’ contributions to the home-school partnership, a meta-analysis is required to synthesize the existing research findings.

The Present Study

A significant number of studies have identified a positive relationship between levels of parenting self-efficacy and levels of parents’ contribution toward effective home-school partnerships with considerable variations found in strengths of these associations. There has yet to be a study synthesizing the existing evidence. The present study aims to fill this gap by examining the associations between parenting self-efficacy and various behavioral, relational, and cognitive aspects of parents’ contributions to the home-school partnership, namely home-based participation (behavioral), school-based participation (behavioral), home-school communication and relationships (relational), as well as parental expectations and aspirations (cognitive), among parents of primary school students. Meta-analytic findings have the capacity to inform future practices for promoting the home-school partnership. An additional exploratory aim of the present study was to identify studies that measured the multi-dimensional construct of parental self-regulation (e.g., self-sufficiency, self-efficacy, self-management, personal agency, problem solving; Morawska et al., 2019; Sanders et al., 2019). A multi-level meta-analytic approach was used as it is one of the most robust existing methods currently available (Cheung, 2019; Van den Noortgate et al., 2015). To our knowledge, this meta-analysis is the first effort to synthesize all the available literature for parents of primary school students. In addition, given that demographic moderators were frequently identified in individual studies (e.g., Grinshtain & Harpaz, 2023; Newland et al., 2013; Sheridan et al., 2019), we also aimed to explore the potential moderation effects of a range of demographic factors (e.g., child gender, child age, parent gender, race/ethnicity, social-economic status). We limited the child age range to primary school age because research suggests that the home-school partnership is very different in day-care settings and that parents tend to participate less in children’s education in secondary education (Barger et al., 2019; Pomerantz et al., 2007, 2012). We hypothesized that self-efficacy (and/or self-regulation) to be positively associated with the home-school partnership in general, and also with the four aspects included in the present study: home-based participation, school-based participation, home-school communication and relationships, and parental expectations and aspirations. We also expected positive associations with both general and education-focused parenting self-efficacy. An exploratory research question was related to the strength of the associations between the two types of parenting self-efficacy (general and education focused). Moderator analyses are exploratory in nature without a priori hypotheses.

Method

This meta-analysis was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA; Liberati et al., 2009) guidelines.

Search Strategy

The search strategy was developed in close collaboration with experienced university librarians to capture both the parenting self-efficacy and the parental self-regulation constructs together with the home-school partnership construct. To identify relevant literature, systematic literature search was conducted with the following databases: PsycINFO (EBSCO), Web of Science (Clarivate), ERIC (ProQuest), Education Database (ProQuest), Social Science Database (ProQuest), and Sociological Abstracts (ProQuest). These databases were chosen to cover a wide range of literature in psychology, education, and sociology. The literature search was limited to articles published in English; however, no date restrictions were implemented in the search. The combination of the following terms was searched in titles, abstracts, and keywords with the proximity operators:

(“efficacy” OR “self-regulation” OR “confidence” OR “self-control” OR “personal agency” OR “self-sufficiency” OR “self-management” OR “problem solving” OR “emotion* regulation” OR “executive function” OR “motivational belief*”). AND [((“school” OR “teacher*” OR “educator*”) NEAR/2 (“home” OR “parent*” OR “famil*”) NEAR/2 (“communication” OR “collaboration*” OR “partnership*” OR “relationship*”)) OR (((“parent*” OR “famil*” OR “father*” OR “paternal” OR “mother*” OR “maternal” OR “home”) NEAR/3 (“involvement*” OR “engagement”)) AND (“school*” OR “education*” OR “student*”))].

The search terms were tailored to each database, including controlled vocabulary, and all database searches were completed in late October 2023. The search strategies applied in each database are provided in Supplementary Material. Search for unpublished doctoral theses was conducted in the PsycINFO (EBSCO). The reference lists of the included studies were examined to identify any further relevant articles. Removal of duplicates and study screening was conducted in the web-based management software Covidence (Veritas Health Innovation 2024).

Study Selection

Title and abstract screening based on the inclusion and exclusion criteria were led by first author with a 10% cross-check by the second author, which is an acceptable method when the inter-rater agreement is high. Agreement on inclusion/exclusion decisions was reached for the majority of articles (the inter-rater agreement rate was 98.4%) with any discrepancies discussed and resolved by consensus between the authors. Full-text screening was completed independently by the first author and a graduate research assistant who was familiar with the topic. At all screening stages, all retrieved sources were screened for eligibility with the following criteria:

  • Only studies with quantitative data were included. Qualitative studies, reviews, commentaries, and conceptual works were excluded. Also, the study needed to be in English language, and with full text available.

  • Only studies focused on outcomes related to home-school partnership as per our broad definition were included (i.e., a shared responsibility of, and the collaboration between, family and school that benefits children’s learning at schools; Bierman & Sheridan, 2022; Goodall & Montgomery, 2014; Smith et al., 2020). This could be behavioral (e.g., assisting with homework, participating in school-based activities), relational (e.g., good parent-teacher relationship and communication), and cognitive (e.g., expectations and aspirations for the child to do well academically). The cognitive aspect was considered meeting our broad definition because a partnership between home and school requires parents to share expectations for children’s learning, as meeting or exceeding the expectations set by schools and teachers. For the present study, the activities that parents participated in need to directly link to the children’s formal schooling. Studies focused on parents’ investment in extra-curricular activities, health education, and sexual education were excluded. Studies examining parents’ participation in non-education-related aspects of child rearing and parenting, and in other settings such as in healthcare settings were excluded.

  • The study needed to focus on the home-school partnership in primary school. Studies focused on the home-school partnership in the childcare or early childhood education settings (include Head Start) as well as secondary schools were excluded. The study was considered eligible if only a small subset of participants (< 20%) were from early childhood education or secondary schools. This threshold was selected for the following reasons: 1) there are natural age range differences between samples due to differences in the education systems across the world (e.g., grade 7 is considered primary school in certain countries); 2) a 20% threshold is unlikely to result in biased estimation; and 3) having < 20% of the sample recruited from early childhood education or secondary schools is unlikely to be the result of targeted recruitment.

  • Only papers examining at least one aspect of parental self-regulation as defined by Sanders et al. (2019) were included (including parenting self-efficacy). The parenting self-efficacy/parental self-regulation must have been in relation to parenting or helping children do well in school. Studies that explored parenting self-efficacy/parental self-regulation in parents’ own abilities related to academic subject matter (e.g., homework, Math) were excluded. Only the self-efficacy component of motivational beliefs in the Hoover-Dempsey model (Hoover-Dempsey et al., 1995, 1997) was considered and subsequently included. Parenting self-efficacy/parental self-regulation had to be parent reported; studies relying on child perception were excluded.

  • Included studies needed to provide sufficient information on the effect sizes of the relationship between parental self-efficacy and the home-school partnership (i.e., Pearson’s r correlations). If bivariate correlation was not reported, studies were only included if there was sufficient information to allow the estimation of the effect size (e.g., t, d, χ2, or ρ). Studies were excluded if they only present results from multivariate analyses (e.g., multiple regression analyses, structural equation modeling, and multilevel modeling) due to the inability to obtain independent effect sizes between parental self-efficacy and the home-school partnership.

  • If the same dataset was analyzed in multiple studies, the studies were examined to select the most representative study using the dataset to avoid redundancy (i.e., the study with the largest sample size; and when the samples were the same, the first published study). For longitudinal studies (including intervention studies), only the correlations reported at the earliest time point were included.

Study Extraction

Data charting and extraction templates were created in Excel. Data were extracted and charted on study characteristics (citation, publication status), sample demographics (region, sample size, sample characteristics, child and parent gender ratios, child and parent age, socio-economic status), as well as data extracted in relation to our research questions (parental self-regulation variable, home-school partnership variable, the internal consistency of measures, effect sizes). During extraction, the signs of a number of original effect sizes were changed to ensure data unity in the analysis. Positive signs in the extraction sheet indicate positive associations between the level of parental self-regulation and desirable home-school partnership outcomes, or negative associations between levels of parental self-regulation and undesirable home-school partnership outcomes. Data extraction was completed concurrently with full-text screening by the first author and double checked by a graduate research assistant.

Coding of Variables

The coding categories were developed in discussions of the study authors. Variables were independently coded by the first author and the third author, with any discrepancies discussed and resolved by consensus between the authors. Given that all included studies only measured the parenting self-efficacy component of parental self-regulation, we named this variable to be parenting self-efficacy (instead of parental self-regulation). Two categories were created for the parenting self-efficacy variable, namely general parenting self-efficacy (e.g., Parenting Sense of Competence Scale; Ohan et al., 2000) and education-focused parenting self-efficacy (e.g., Parent Efficacy for Helping Children Succeed in School Scale; Hoover-Dempsey et al., 2005). Decisions were made based on the descriptions and the wording of the measures.

The home-school partnership construct was coded into four categories, namely relational, behavioral (home-based participation), behavioral (school-based participation), and cognitive. The relational category focused on the home-school relationship and parent-teacher communication (e.g., quality of the relationship or communication, sharing of information about school, parents’ communication style, parents’ perception of teachers’ or schools’ invitation to involve). The cognitive category includes parental expectations and aspirations on children’s learning. The behavioral category includes actual behaviors the parents were doing to help promote children’s learning and wellbeing, either at home or at school. We further broke the behavioral category down to home-based activities and school-based activities. Home-based activities include activities parents do outside of school or without the participation of school personnel (e.g., homework assistance, involvement in child academic life such as through discussion and encouragement). Activities with/at the school include the involvement of school (e.g., attending parent-teacher conference, participation in school activities, volunteering at the school). When the original measure included items capturing multiple aspects of the home-school partnership, it was coded as non-applicable, thus excluded from analyses that involved breaking down the home-school partnership variable.

Statistical Procedures

All analyses were conducted in R Studio (RStudio Team, 2020) using the Metafor Package (Viechtbauer, 2010). Effect sizes based on Pearson’s r correlation coefficients were converted into Fisher’s Z correlation coefficients to adjust for skewed standard errors. Other effect sizes were converted from their original form (e.g., t, d, χ2, or ρ) into Fisher’s z correlations using standard conversion formulas (Hunter & Schmidt, 2004; Lipsey & Wilson, 2001). The aggregated effect sizes were transformed back to Pearson’s r correlations to aid interpretation and reporting using the formulas suggested by Borenstein et al. (2009). Given that the effect sizes were expected to vary across studies, a random effects model was selected (Cheung, 2019; Lopez-Lopez et al., 2018; Van den Noortgate et al., 2015).

Model Estimation

To accommodate the nested structure of the data, where studies often have multiple effect sizes reported for each association between parenting self-efficacy and home-school partnership (e.g., multiple measures were used to capture similar construct), a three-level random effects model approach was utilized (Cheung, 2014; Lopez-Lopez et al., 2018; Van den Noortgate et al., 2015). The multi-level method has significant advantages over traditional methods accounting for the dependent effect sizes, such as ignoring dependence, aggregating effect sizes within studies, selecting one effect size per study, and shifting the unit of analysis, by producing more accurate estimation of standard errors and attribution of variance (Cheung, 2014; Lopez-Lopez et al., 2018; Van den Noortgate et al., 2015). We treated multiple studies using separate cohorts or samples in a single study as independent studies in our analyses. Additionally, when a very small number of studies (in the present study, two studies) reported effect sizes separately from mothers and fathers of the same child, we applied the aggregation method prior to the formal analysis (Cheung, 2014; Lopez-Lopez et al., 2018; Van den Noortgate et al., 2015).

In the three-level model, level one refers to the sampling level (effect sizes), level two refers to the outcome level (associations between parenting self-efficacy variable and home-school partnership variable), and level three refers to the study level. Level one sampling variances were often omitted in the reporting of original studies, where close approximations were obtained from the study sample sizes using the standard formula (Borenstein et al., 2009). Level two random effect (τ22) represents within-study variability with variances component denoted by στ2, and level three 3 random effect (τ32) represents between-study variability of effect sizes, with an expected mean of zero and variance indicated as στ3. We computed Q and I2 statistics to use as indicators of heterogeneity and the proportion of explained variance at each level, respectively (Cheung, 2014, 2019).

We first estimated a three-level random effect model (M1) to estimate the overall correlation between parenting self-efficacy and home-school partnership. Random effect and its variance at each level will also be estimated. Next, we will estimate two three-level mixed-effect models to test the influences of different types of parenting self-efficacy (M2) and different types of home-school partnership variables (M3). In the two mixed-effect models, differences between parenting self-efficacy types and between home-school partnership types were treated as fixed effects in the models. The different types of variables were dummy coded in the analyses. We used restricted maximum likelihood (“REML”) estimator in all analyses.

Correction for Attenuation

The effect sizes reported in each study might be lower than the “actual” effect sizes in the population due to measurement errors, so we adjusted for this potential attenuation by applying the formula proposed by Card (2015). Adjusted correlation coefficients (ρ) were produced by dividing the extracted Pearson’s r values by the root square of measures’ reliability coefficients (Cronbach’s α) multiplied by each other (i.e., \(\rho =r / \sqrt{\alpha 1\times \alpha 2}\)). If Cronbach’s α was not reported in the original study or not applicable for certain measures, we set it to be at 1. These correlations (ρ) were further converted into Fisher’s Z. We re-ran all above-mentioned analyses to provide adjusted pooled effect sizes, which were presented in relevant tables along with the pooled effect sizes calculated by using the original r. Results were converted back to r after the analyses while displayed under the column heading ρ.

Moderation Analyses

Moderation analyses were performed on M1 to examine whether the overall associations between parenting self-efficacy and the home-school partnership were influenced by child gender, parent gender, race/ethnicity, sample type, and socio-economic status. Due to limitations in the available data, child age was not tested as a potential moderator. Studies which did not provide sufficient information on certain moderator variables were excluded from those analyses. Child gender was recorded as the ratio of males in the total sample, while parent gender was coded as the ratio of females. Ethnicity was recorded as the percentage of White/Caucasian participants in the total sample. Meta-regression analyses were conducted with these continuous moderators. Depending on each study’s recruitment strategy and inclusion criteria, we coded the sample type as either general (universal sample, no or limited restrictions applied) or targeted (special inclusion criteria applied, e.g., neuro diverse children, children with behavioral problems). Different studies conducted in different countries and years tend to have different standards in their reporting of family income; for the ease of analysis and interpretation, we treated socio-economic status as a dichotomous variable where each study can be either low socio-economic status (indicated in the study as low socio-economic status or socially disadvantaged) or middle and high socio-economic status (if not indicated in the study as low). Subgroup analyses were conducted with these categorical variables. Although subgroup analysis is the most commonly used and appropriate technique for moderator analysis with categorical variables (Borenstein et al., 2009; Hedges & Pigott, 2001), it often lacks the statistical power necessary to detect significant between-group differences when between study heterogeneity is high (Hedges & Pigott, 2001). Thus, we also reported the results of the power analysis to aid interpretation.

Evaluating Publication Bias

Three tests were conducted to investigate potential publication bias. First, the publication status of each study was examined as a potential categorical moderator using subgroup analysis. Next, we conducted a classical funnel plots analysis and Egger’s regression test was used to evaluate the symmetry of the funnel plots (Egger et al., 1997). To adapt the tests to a three-level model, we followed the guidelines of Rodgers and Pustejovsky (2020). In addition, we performed classic fail-safe N (Orwin, 1983; Rosenthal, 1979). Fail-safe N analysis calculates the number of unpublished studies with null results (i.e., Zr = 0) which would be required to reduce the synthesized effect size between parenting self-efficacy and home-school partnership to a non-statistically significant value (p > .001). Given the unit of analysis in the present study was set at effect size level, the results from the fail-safe N analysis should be interpreted as the number of null effect sizes needed rather than studies needed. Given that conducting publication bias tests with multilevel meta-analysis is still an emerging area of research with no software adaptation available for the fail-safe N analysis (Rodgers & Pustejovsky, 2020), the issue of data dependency was ignored in the fail-safe N analysis.

Results

Study Selection

A total of 3913 papers were identified in database searches. After removal of duplicates, 2357 studies were screened for title and abstracts and the full text of 125 studies were assessed for eligibility (see Fig. 1). We included 43 studies in the meta-analysis with 7 studies having more than one independent cohort of students, thus bringing the total number of unique samples included in subsequent analyses to 50 (hereafter referred to as studies).

Fig. 1
figure 1

PRISMA flow diagram of identification and selection of included studies

Full size image

Study Characteristics

Detailed information about each eligible study including the raw effect sizes (e.g., Pearson’s r) is displayed in Supplementary Material. The 50 eligible studies were published between 1992 and 2024, with 72% published in the last 15 years. Thirty studies were published journal articles while another 20 were doctoral dissertations. A total of 20,043 children were included with the sample sizes ranging from 35 to 8152. The majority of studies were conducted in the USA (31/50) and used cross-sectional methodology (44/50). Most studies recruited general samples (41/50), while the remaining recruited children have medical conditions, disabilities, neurodiversity, or behavioral difficulties. Eighteen studies indicated that the sample was financially disadvantaged. All studies included in this meta-analysis only measured the parenting self-efficacy component of parental self-regulation.

Three-Level Model for Overall Correlation

A three-level random effect model (M1) was estimated to calculate the average correlation between parenting self-efficacy and home-school partnership. A small to medium Pearson’s r correlation at .189 was found (Zr = .192 [95% CI: .110, .273]). Model statistics are reported in Table 1. Significant heterogeneity was found in the correlations, Q(df=184) = 1262.50, p < .001. Both level 3 and level 2 variances were significant, indicating significant between- and within-study variations. These were confirmed by estimating three follow-up models with variance constrained at certain levels, which resulted in significant worse model fit (see Table 2). The correlation between parenting self-efficacy and home-school partnership outcomes not only varied across studies but also varied across different self-efficacy and home-school partnership dimensions. According to I2 (Cheung, 2014), about 51.0% of the total variation attributed to level 3, about 40.0% of the total variation attributed to level 2, and the remaining 8.9% of the total variation attributed to level 1. A forest plot of effect sizes is provided in Supplementary Material.

Table 1 Associations between parenting self-efficacy and home-school partnership outcomes (i = 185; k = 50)
Full size table
Table 2 Model fit comparison
Full size table

Mixed-Effects Model for Parenting Self-Efficacy Categories

To explore whether the strength of the associations between parenting self-efficacy and the home-school partnership differed across the two parenting self-efficacy categories, namely education focused self-efficacy and general parenting self-efficacy, we estimated a three-level mixed effect model (M2; model details are shown in Table 1). A statistically significant between group difference was identified, F(2) = 23.54, p < .001. Education-focused self-efficacy had a statistically significantly stronger association with the home-school partnership outcomes than general parenting self-efficacy (r = .183 versus .114 respectively), t(183) = 1.97, p = .050. Significant heterogeneity in the effect sizes was observed, Q(df=183) = 1261.82, p < .001. Model fit comparison results are reported in Table 2. Between- and within-study variations were found as the level 3 variance components and the level 2 variance components accounted for 50.3% and 41.4% of the total variation, respectively.

Mixed-Effects Model for Home-School Partnership Categories

We estimated another three-level mixed effect model (M3; see Table 1 for details) to investigate whether the effect sizes vary across the four different home-school partnership outcomes. Statistically significant between outcomes variation was found, F(4) = 215.41, p < .001. Parenting self-efficacy was most strongly associated with home-based involvement (r = .248) and parental expectations and aspirations (r = .248). The strength of these two associations did not differ statistically from one another. The association between parenting self-efficacy with home-based participation was statistically significantly stronger than the associations with school-based participation, r = .124, t(176) = 7.70, p < .001, and with parent-teacher communication and relationships, r = .090, t(176) = 9.36, p < .001. Parental expectations and aspirations had statistically significantly stronger associations with parenting self-efficacy than school-based participation, t(176) = 4.34, p < .001, and parent-teacher communication and relationships, t(176) = 5.04, p < .001. Parenting self-efficacy had a statistically significantly stronger association with school-based participation than with parent-teacher communication and relationships, t(176) = 2.06, p = .041. The effect sizes also had significant heterogeneity (see Table 2 for model fit comparisons results), Q(df=174) = 2056.05, p < .001. Based on I2 (Cheung, 2014), within-study variations accounted for 53.1% of the total variation, between-study variations accounted for 38.7% of the total variation, and sampling variations only accounted for 8.1% of the total variation.

Moderator Analyses

Moderation analyses were performed to examine whether the overall associations between parenting self-efficacy and the home-school partnership were influenced by child gender, parent gender, race/ethnicity, sample type, and socio-economic status. This approach was chosen to allow sufficient sample size to be maintained for the moderator analysis. Each moderator was tested in separate models with either subgroup analysis or meta-regression. Studies which did not provide sufficient information on certain moderator variables were excluded from the analysis for that moderator. Detailed results are reported in Table 3.

Table 3 Moderator analyses
Full size table

Child Gender

The percentage of males in the sample was entered as a moderator, which was not statistically significant in the model. The results suggested that child gender (% of male) was not associated with the strength of the overall association, β = .086, F(1) = 1.77, p = .183. The between-study heterogeneity contribution increased to about 60.33% of the total variation while the within-study variation contribution reduced to 32.02%.

Parent Gender

Percentage of female parents who completed the questionnaire was tested as a moderator in meta-regression analysis. Non-statistically significant results suggested that the strength of the overall association was not associated with the gender of the parent (% of female), β =  − .209, t(164) =  − 1.93, p = .053. Variance distribution is consistent with M1.

Ethnicity

Percentage of White/Caucasian participants in the sample was entered in the model. The results suggested that the strength of the association was not statistically significantly related to the ethnicity of the participants, β =  − .022, F (1) = 0.13, p = .719. The sample-level variation attribution to the total variance increased to 18.0%, the within-study variation attribution reduced to 3.6%, and the between-study variation attribution increased to 78.3%.

Sample

Type of sample was tested as a moderator. The association appeared to be stronger in the targeted samples (Fisher’s Z = .288) than in the general sample (Fisher’s Z = .172). However, the difference was not statistically significant, t(183) = 1.86, p = .064. Due to high between-study variations (50.4%), the power to detect statistically significant between-group differences was only at 0.31. Variance distribution is similar to M1.

Socio-Economic Status

Socio-economic status was examined via subgroup analysis. The associations in low socio-economic status samples (Fisher’s Z = .149) seemed to be weaker than the associations in the general sample (Fisher’s Z = .217); however, the difference was not statistically significant, t(184) = 1.41, p = .160. With high levels of between-study variation (50.8%), the power to detect between-group differences was low (0.17). Variance distribution is consistent with the distribution in M1.

Publication Bias

As outlined in the “Method” section, we applied three tests to detect the publication bias, namely testing the publication status as a moderator, examining funnel plots and Egger’s test, and fail-safe N analysis. Publication status was first tested as a potential moderator (see Table 3). Results showed that although published studies had larger effect sizes (Fisher’s Z = .222) than unpublished studies (Fisher’s Z = .149), the difference (∆ Fisher’s Z = .073) was not statistically significant, F(1) = 2.37, p = .124. However, the statistical power to detect statistically significant differences was low (0.19). Funnel plot (see Fig. 2) and Egger’s test of heterogeneity suggested that the distribution of effect sizes was largely symmetrical, thus showing limited evidence of publication bias, t(18) = 0.111, p = .913. Classic fail-safe N analysis suggested that at least 2797 null effect sizes are required to reduce the aggregated effect sizes of the associations between parenting self-efficacy and home-school partnership outcomes to be non-statistically significant (p > .001). Overall, the three analyses suggested the findings of the present study to be robust to the file-drawer problem.

Fig. 2
figure 2

Contour-enhanced funnel plot for potential publication bias. Note: The black dots represent the individual Fisher’s Z. The black-dotted line indicates the region under which we would expect estimated effect sizes to fall if the true effect size was approximately 0.20

Full size image

Discussion

To our knowledge, this is the first meta-analysis examining the associations between parenting self-efficacy and multiple domains of parents’ contributions toward effective home-school partnership. Despite the aim of the search strategy to include key terms that captured the multi-dimensional parental self-regulation construct, due to the recent use of this construct, all included studies measured only its core element, the unidimensional construct of parenting self-efficacy. Overall, parenting self-efficacy had a small to medium correlation with the parents’ contributions to the home-school partnership. The associations were stronger when the parenting self-efficacy had a specific focus on children’s education compared to parenting in general. Additionally, parenting self-efficacy tended to have stronger associations with home-based participation and parental expectations and aspirations than parent-teacher communication/relationship and school-based participation. We identified limited influences of the moderators and limited evidence of publication bias.

The overall positive association between parenting self-efficacy and home-school partnership outcomes was consistent with our prediction. The findings supported the pathways identified in home-school partnership models and frameworks (Yamauchi et al., 2017), such as the frequently applied Hoover-Dempsey model (Hoover-Dempsey et al., 1995, 1997), where the parenting self-efficacy was identified as a key antecedent of parents’ contributions to the home-school partnership (e.g., Ferretti et al., 2019; Ice & Hoover-Dempsey, 2010; Kwan & Wong, 2016; Newland et al., 2013; Reininger & Santana López, 2017; Semke et al., 2010). Findings of robust evidence for the concurrent associations between parenting self-efficacy and the home-school partnership outcomes identified in previous cross sectional studies (e.g., Holloway et al., 2016; Hou et al., 2023; Tazouti & Jarlégan, 2019) strengthened the findings from previous longitudinal studies that found baseline parenting self-efficacy was associated with home-school partnerships outcomes at one year (Choe, 2022) or two years later (Ma et al., 2024).

As noted in the Hoover-Dempsey et al. (2005) frameworks of the home-school partnership and theory of parenting self-efficacy (Bandura, 1997), the associations were likely to be bi-directional where experiences in educational settings can also influence parents’ levels of self-efficacy. For example, parents’ positive experiences of participation in school-based activities and parent-teacher communications can result in increased levels of parents’ self-efficacy, whereas certain teacher or school practices (e.g., not valuing the input from parents, not respecting different cultural and religious backgrounds) can undermine parents’ participation and subsequent self-efficacy levels. This is consistent with a previous study where not only the levels of parenting self-efficacy at baseline were found to be associated with the quality of parent-teacher communication at two years later but also the quality of the parent-teacher communication was longitudinally associated with parents’ levels of self-efficacy in one of the two study waves (Ma et al., 2024).

Parents with high levels of parenting self-efficacy tend to possess the knowledge and skills required to promote children’s learning, to hold strong beliefs in their own ability to have a positive influence, to apply their knowledge and skills in a wide-range of home-based and school-based activities, to attribute positive changes to their own actions or behaviors, as well as to resolve problems that emerge along the way (Morawska et al., 2019; Sanders et al., 2019). Additionally, highly self-efficacious parents have the capacity to work effectively with other caregivers of their children (Albanese et al., 2019). This collaboration in child rearing requires parents to regulate their emotions, cognitions, and behaviors to communicate and collaborate effectively with others, such as teachers and other school staff. The development of parenting self-efficacy in communicating and collaborating with significant others begins early with grandparents and other carers and continues through early childhood education settings. By the time children are transitioning to formal schooling, individual differences in the confidence and capability levels for partnering with others can influence parents’ styles of communicating and relating to others involved in caring for and educating their children.

The present study found stronger associations between home-school partnership outcomes and education-focused parenting self-efficacy than general parenting self-efficacy. This difference is in line with the findings of a previous French study, which was the only other study that administered both measures of parenting self-efficacy in the general parenting role and parenting self-efficacy related to schooling (Tazouti & Jarlégan, 2019). Schooling-related parenting self-efficacy tended to have stronger correlations with a range of cognitive and behavioral home-school partnership outcomes in both mothers and fathers. Although the overall level of parenting self-efficacy is closely related to task-specific parenting self-efficacy (e.g., in the above-mentioned study, the correlations were ranged between 0.51 and 0.62), education-focused parenting self-efficacy is more proximal to the home-school partnership while general parenting self-efficacy is relatively distal. This finding highlighted the differences between, and the importance of, both the general and task-specific parenting self-efficacy. Evidence suggested that compared to general self-efficacy, task-specific parenting self-efficacy is more likely to change as a result of successes and failures, while improving the general parenting self-efficacy tends to improve task-specific parenting self-efficacy at the same time (Sanders et al., 2019).

Similarly, we found that parenting self-efficacy had stronger associations with home-based participation and parental expectations and aspirations than with school-based participation and parent-teacher communication/relationships. These differences were consistent with the patterns identified in individual studies that included measures of different aspects of the home-school partnership in the same study (e.g., Ferretti et al., 2019; Kwan & Wong, 2016; Newland et al., 2013). These findings suggested that parenting self-efficacy tends to have different levels of associations with various home-school partnership outcomes. For example, parents tend to have greater control over their own behaviors and thoughts than they do when in environments that involve another party. Parents’ participation in school-based activities and the relationship they have with the teachers can be considerably influenced by the school and its teaching staff, as well as logistical factors such as parents’ time and energy (Hoover-Dempsey et al., 2005).

Finally, none of the tested moderators (i.e., parent gender, child gender, sample type, ethnicity, socio-economic status) were statistically significant in the models. This is consistent with the findings of previous meta-analyses on similar topics, that most potential moderators tested, that have been frequently identified as having significant influence in individual studies, were not statistically significant in the models (e.g., Barger et al., 2019; Smith et al., 2020; Wilder, 2014). However, as was the case in the present study, subgroup analysis often lacks the statistical power necessary to detect statistically significant between-group differences when between study heterogeneity is high (Hedges & Pigott, 2001). Therefore, failing to find evidence suggesting statistically significant moderation effects should not be interpreted as evidence of the absence of moderation effects. In light of this limitation, even though certain moderators (e.g., child gender and ethnicity) have been found to be associated with different levels of parenting self-efficacy (Fang et al., 2021) and parents’ contributions to the home-school partnership (Bierman & Sheridan, 2022), it is possible that a subset of the moderator variables we tested have limited moderating effects on the strengths of the associations between them.

It is worth noting, that although not statistically significant, the associations between parenting self-efficacy and parents’ contributions to the home-school partnership seemed to be stronger in samples of children with special needs (p = 0.064). Given that parents of children with special needs, often through necessity, have a long history of working with and receiving support from others which may undermine their parenting self-efficacy, it is evident that parents’ own confidence levels play an important role in making decisions about being involved in children’s education and making contact with teachers and schools. Another close-to-statistically-significant moderator identified was the gender of the parent (p = 0.053). It is possible that mothers, often identified in research as the primary caregivers, are more likely to participate in children’s education and build partnerships with schools regardless of their levels of parenting self-efficacy, which has subsequently influenced the fluctuations around this mandatory level of involvement. Fathers’ decisions regarding their participation in the home-school partnership were comparatively more likely to be significantly influenced by their confidence in their parenting role. This is perhaps due to the traditional gender roles that do not mandate them to be involved.

Implications

Policy and Practice

Despite extensive efforts to strengthen the home-school partnership and increase parents’ participation in their child’s education, most have focused on preparing schools to engage parents in school-based activities, with less attention paid to preparing parents to support their children’s learning and few programs focused on building parents’ capacity (Leenders et al., 2019; Sheridan et al., 2019; Smith et al., 2020). Among all the factors influencing the effectiveness of home-school partnership, some are very difficult to change (e.g., socio-economic status), while others are modifiable such as parenting self-efficacy and its broader construct parental self-regulation. One important implication of the present study is that it is important for school leaders and teaching staff to recognize the importance of parenting self-efficacy and parental self-regulation, and how the school can facilitate the development of these constructs. In addition to its associations with a wide range of education-related child outcomes such as their social, emotional, and behavioral wellbeing, self-esteem, school engagement, and achievement (Albanese et al., 2019; Fang et al., 2021; Morawska et al., 2019), parenting self-efficacy also plays an important role in parents’ participation in children’s education as well as the quality of the communication and collaboration between a child’s family and their school. Parenting self-efficacy can be either global or specific to a task such as communicating with teachers or participating in certain educational activities. Different types of parenting self-efficacy can be promoted through a combination of broad-focused and more topic-specific interventions and training programs (e.g., specific training programs aim to promote parents skills and self-efficacy in communicating with teachers; Sanders et al., 2021).

Parents are likely to exhibit varying levels of parenting self-efficacy and indeed, parental self-regulation when their children enter formal schooling (Sanders et al., 2019). These differences are determined by a range of proximal and historical factors, such as parents’ own prior experience of schooling (e.g., non-attendance, school failure), parental mental health conditions, and certain conditions that their child is living with (e.g., neuro-diversity, early onset conduct problems, ADHD, learning difficulties, physical disabilities, medical conditions). These parents often have a history of working with professionals, which can undermine or support their parenting self-efficacy and self-regulation, particularly when they want the school to make special provisions for their child.

One major way to enhance the parenting self-efficacy is through the delivery of evidence-based parenting programs delivered at schools (Moller et al., 2024; Sanders et al., 2021). Extensive evidence suggests that participating in high-quality, culturally informed, evidence-based parenting programs can have positive and lasting effects on parenting self-efficacy and a range of child development, learning, and wellbeing outcomes (Morawska et al., 2019; Sanders et al., 2021; Sanders et al. 2014). Schools are uniquely positioned to provide parenting support services by involving, communicating, and supporting parents as part of schools’ core business (Sanders et al., 2021). An additional incentive for schools to offer evidence-based parenting programs are the findings that indicate that through participating in such programs, parents’ self-efficacy is improved, which in turn, better equips them to contribute to their children’s education and to build stronger home-school partnerships.

Future Research

Research included in the present study only focused on the parenting self-efficacy as one of the key elements of the boarder parental self-regulation construct. Lack of reliable measures of the broader parental self-regulation construct partially contributed to this issue. Future research on this topic should utilize measures of parental self-regulation that have been recently developed and showed promising psychometric properties such as the Me as a Parent measure (MaaP; Hamilton et al., 2015) and the Parenting Self-regulation Scale (PSRS; Tellegen et al., 2022). Research should also consider investigating how other key elements of the parental self-regulation (i.e., self-sufficiency, self-management, personal agency, problem-solving) can influence the strength of the home-school partnership. In contrast to the parenting self-efficacy construct, where similar measures have been used across studies, measures of the home-school partnership vary considerably. This could be partially due to inconsistencies in the definition of the home-school partnership and the dynamic changes in researchers’ understanding of the construct (e.g., from mainly focusing on involvement in structural activities toward understanding the home-school partnership as a bi-directional relationship that takes place in various contexts). Many measures also tend to focus on the quantity of the parents’ participation rather than the quality, and consequently fail to assess the context and the reasons for participation (e.g., larger amount of time contacting teacher could be resulted from either parent proactively checking children’s progress or teacher reaching out to a parent due to high levels of concerns).

In addition, there is increasing recognition of the home-school partnership as a bi-directional relationship, where parents and teachers tend to have influence over each other’s contributions (Bierman & Sheridan, 2022). However, the bidirectional nature is often not captured by current unidirectional measures (Leenders et al., 2019; Smith et al., 2020). Future studies need to access brief, multi-factor, change sensitive measures that better reflect the bi-directional nature of the home-school partnership and capture the quality and complexity of this partnership rather than simply the quantity. Moreover, the current evidence base relies heavily on findings from cross-sectional studies and as is the case for most studies included in this meta-analysis. Additional studies with robust designs are needed to strengthen the evidence base (e.g., longitudinal and experimental design). In particular, school-based randomized controlled clinical trials of parenting programs, where parents’ level of parenting self-efficacy is experimentally manipulated through randomization, provide an opportunity to study the causal relationship between parenting self-efficacy and the home-school partnership outcomes (e.g., Boyle et al., 2023).

Another promising area for future research is applying person-centered approaches to study the complex interplay between parenting self-efficacy and parents’ contributions to the home-school partnership in different domains. The majority of existing quantitative studies adopted variable-centred approaches (e.g., ANOVA, multiple regression, structural equation modeling, meta-analysis), which utilized sets of “averaged” parameters in analyses under the assumption that all participants were similar and obtained from a single population. Person-centered approaches (e.g., latent class analysis, latent profile analysis, latent transition analysis) take into account potential individual differences by allowing participants to be drawn from different subpopulations. For example, longitudinal studies applying the latent transition analysis method would enable researchers to examine the proportion of parents reported the quality of home-school partnership in different domains remained the same, improved, or decreased following key school transition events (e.g., early childhood education to primary school, primary school to middle and high schools). This would also enable the investigation of how parenting self-efficacy contributed to these changes.

Limitations

Our results should be interpreted in light of certain limitations. Firstly, meta-analyses of the associations between two constructs rely heavily on cross-sectional data with correlational analyses. Hence, the present study has limited power to draw definitive conclusions about causal relationships. It is highly likely that the relationships between parenting self-efficacy and the home-school partnership are bi-directional in nature. As mentioned above, more longitudinal and experimental studies are needed. Secondly, a large number of studies only reported results from multivariate analyses and were therefore excluded from the current meta-analysis. This reduced the statistical power to detect moderators and could introduce potential biases in the results (e.g., researchers with different reporting styles, certain journals have strict page limits). Lastly, although the results revealed a positive association between parenting self-efficacy and home-school partnership outcomes, there was high level of between-study heterogeneity. This could partially be due to the differences in the definition and measurement of the home-school partnership construct. Although we attempted to break down the home-school partnership outcomes into four categories and had two independent coders, significant heterogeneity still exists within each category (e.g., talking to children about school can be very different to assisting children with homework). This high level of between-study heterogeneity also limited the statistical power in detecting moderators of the association (Hedges & Pigott, 2001). It is worth noting that statistical insignificance should not be interpreted as evidence for the absence of the moderation effects.

Conclusion

This meta-analysis investigated the associations between parenting self-efficacy and home-school partnership outcomes. Parenting self-efficacy demonstrated positive associations with home-school partnership outcomes. The associations were stronger in education-focused parenting self-efficacy than in general parenting self-efficacy, and were stronger with home-based participation and in parental expectations and aspirations than in school-based participation and parent-teacher communication/relationship. No statistically significant moderating effects of child gender, parent gender, ethnicity, sample type, and socio-economic status were detected. We also detected limited publication bias. In addition, no study that explicitly measured the multi-dimensional construct of parental self-regulation was identified. The findings of the present study strongly suggest that the school leaders and teachers recognize the critical role of parenting self-efficacy and how this construct can facilitate the development and growth of the home-school partnership. The promotion of parenting self-efficacy may result in not only improvements in children’s learning and wellbeing but also with stronger partnerships between a child’s family and their school.