There are two categories defining parental involvement: the static structure view and the dynamic process view.

The static structure view suggested that parental involvement was divided into several specific types. Epstein (1995) presented there are six types of parental involvement including parenting, communication, volunteering, learning at home, decision making, and collaborating with the community.

Similarly, Xu et al. (2024) reviewed about 30 primary studies, and found that parental involvement in the homework field can be divided into five types, including autonomy support, content support, parental control, frequency, and mixed. Autonomy support mean parents attend to students’ ideas and support their homework initiatives. Content support mean parents provide direct support on the content of assignments. Parental control mean when parents function to monitor, control, and interfere with homework assignments. Parental involvement was coded as “frequency” when it focused on the frequency or amount of its involvement. A study was coded as “mixed” when it included more than one dimension of involvement.

The dynamic process view holds that parental involvement as a whole is a dynamic process (Fiskerstrand, 2022). Fiskerstrand (2022) classified parental involvement into 12 categories, including competence, belief, motivation, emotion, presence, framing, parenting, activity, talking, helping, choosing, and outing. From the perspective of being, doing, and thinking, these 12 categories can be grouped into four classes. They are the parental thinking (including competence, belief, motivation, emotion), the parental being (including presence, framing, parenting), and the parental doing (including activity, talking, helping, choosing, outing) (Fiskerstrand, 2022).

Considering the primary studies included, the present project adopted the Xu et al. (2024) framework to classify parental involvement. Then the parental involvement were divided into four subtypes: autonomy support (AS), parental control (PC), behavior control (BC) and content support (CS).

Previous studies mainly focused on parental involvement and academic achievement, schooling and children’s adjustment (including truancy control, STEM learning, reading, computational thinking), student engagement (including home-based digital learning), and resilience.

Some studies have explored the relationship between parental involvement and academic achievement, and found that parental involvement can promote academic achievement. Wilder (2014) synthesized the results of nine meta-analyses and the results indicated that the relationship between parental involvement and academic achievement was positive, regardless of a definition of parental involvement or measure of achievement. Similarly, Kim (2022) has conducted a second-order meta-analysis of parental involvement and achievement research over the past 50 years (including 1,177 primary studies in 23 meta-analyses), the results show that there is a positive association between parental involvement and achievement. For the impact of specific types of parental involvement, Kim (2022) and Wilder (2014) had consistent results, pointing to the strongest effect for parent expectations and aspirations and mixed results for homework help. The meta-analysis results of Jiang et al. (2023) shows an overall positive link between supportive parent homework involvement and students’ mathematics achievement and a negative link between intrusive parent homework involvement and students’ mathematics achievement. Xu et al. (2024) used three-level meta-analysis to explore the relationship between parental homework involvement and students’ achievement. The results revealed an overall weak negative relationship between parental homework involvement and students’ achievement. Within specific categories of parental involvement, students’ achievement was positively related to autonomy support, but largely unrelated to content support, parental control, frequency, and mixed.

Some studies have studied the relationship between parental involvement and schooling and children’s adjustment (including truancy control, STEM learning, reading, computational thinking) and revealed positive association between parents’ involvement in children’s schooling and children’s adjustment. The meta-analysis by Barger et al. (2019) revealed small positive associations (r = 0.13 to 0.23) between parents’ naturally-occurring involvement in children’s schooling and children’s academic adjustment (i.e., achievement, engagement, and motivation) that were maintained over time. Parents’ involvement was also positively related to children’s social (r = 0.12) and emotional adjustment (r = 0.17) and negatively related to their delinquency (r = −0.15), concurrently. Kanungo et al. (2024) conducted a scoping review analysis (including 17 articles) of the forms of parental involvement for truancy control, the findings indicated that volunteering and communication are adequate indicators of truancy control and both indicators can significantly impact truancy prevention programmes. The review of Thomas et al. (2020) about parent involvement and its influence on children’s STEM learning found that parent involvement positively affects children’s quantitative skills and problem-solving skills in the terms of the specific skills in STEM content areas.

In addition, some studies have found that parental involvement also promotes students’ reading (Çaliskan and Ulas, 2022) and computational thinking (Cai and Wong, 2023). Çaliskan and Ulas (2022) employed a pretest-posttest quasi-experimental design with a paired control group, data were collected from a total of 100 fourth graders studying in two different primary schools. The findings showed that there were significant differences between the experimental group and the control group, meaning that the parent-involved reading activities developed by the researchers had a positive effect on the students’ reading comprehension, reading motivation, and attitudes towards reading. Cai and Wong (2023) conducted a systematic review about parental involvement in computational thinking education found that three ways in which parents were involved (including affective, behavioral, and cognitive participation) all play an active role in students’ computational thinking education.

Some studies have tapped the relationship between parental involvement and student engagement and revealed there was a positive correlation between parental involvement and student engagement (home-based digital learning belongs to its subcategory). Erol and Turhan (2018) investigated 1,488 students and used the Parental Involvement Scale and Engagement to School Scale to measure parental involvement and student engagement. The results showed a significant positive correlation between parental involvement and student engagement. Increasing and encouraging parental participation in the educational process can enhance students’ engagement with the school. A review of Yang et al. (2023) revealed that parental involvement was the most crucial aspects of social support for students’ school engagement. The results also found that student engagement was reflected via ABC dimensions (i.e., affective, behavioural, and cognitive). Qualter (2024) conducted a narrative literature review about shaping the role of parental involvement in home-based digital learning found that parental involvement plays the most central role in home-based digital learning. The enhancement of parents’ self-efficacy and motivation in home-based digital learning is conducive to the development and education of children’s digital technologies.

Some studies have discussed the relationship between parental involvement and resilience and revealed parental involvement has a positive impact on student engagement. Kovács et al. (2022) conducted a systematic literature review about resilience of parental involvement found that whether the goal is to build upon resilience as a personality trait or target its development as a consequence, strong collaboration between the parents, teachers and professionals concerned in the process can significantly contribute to the child’s psychological, emotional and academic development. Olaseni (2020) conducted an empirical study with a sample of 347 adolescents in Nigeria and found that parental involvement significantly predicted academic resilience in such a way that high parental involvement is linked with high academic resilience. Cui et al. (2024) studied 105,641 Chinese students using latent profile analyses (LPAs) and multivariate analysis of covariance (MANCOVA) and found that parents’ more frequent involvement in students’ everyday lives, coupled with less frequent involvement in their study matters, may effectively foster academic success and enhance the development of resilient traits.

Previous findings have yielded three perspectives: the facilitating view, hindering view, and non-correlation view. Bronfenbrenner’s early theory (Bronfenbrenner, 1977; Bronfenbrenner and Morris, 2007; Woolfolk, 2013) broadly proposed environmental influences on development without specifying directionality. Yeh (1999, 2004) delineated four positive parental involvement types promoting creativity through “appropriate family climate,” “creative activity experience,” “creative skill guidance” and “appropriate parenting styles.” However, additional involvement styles exist, including psychological control and detachment. When examined holistically using omnibus measurements, the cumulative role of parental involvement should demonstrate a blended influence across facilitating, hindering and null effects. Therefore, we propose the following opening hypothesis:

The relationship between parental involvement and student creativity in the following studies may have been moderated by several variables: parental involvement type, grade, parental gender, region, and publication type.

Parental involvement can be described using different models. From the measure perspective, concerning, parental involvement can be divided into four subtypes: autonomy support (AS), parental control (PC), behavior control (BC) and content support (CS) (Xu et al., 2024).

AS refers to respecting children’s perspectives, encouraging independent problem-solving, providing choices (Grolnick and Ryan, 1989; Ryan et al., 2006). Tang et al. (2022) surveyed on 5,523 students and evidenced AS positively predicted elementary/middle schoolers’ creativity (r = 0.100). Chen et al. (2021) investigated a sample of 258 7th graders and revealed a significant correlation between parental autonomy support and creative thinking (r = 0.200).

Alternatively, PC entails coercive, psychologically intrusive tactics inducing compliance via love withdrawal or guilt, it often divided into psychological control (PyC) and behavioral control (BC). PyC encompasses parental intrusion into the child’s inner world, aiming to align the child’s thoughts, behaviors, and feelings with parental demands through strategies such as withdrawal of love and guilt induction, whereas BC regulates behavior through monitoring and discipline to social norms (Barber, 1996; Barber and Harmon, 2002; Silk et al., 2003). Chen et al. (2021) found that BC dimensions positively (r = 0.270, 0.320 and 0.200), and PC dimensions negatively (r = −0.080, −0.070 and −0.060) correlated with the three dimension of creative thinking (fluency, flexibility, and originality). Ren et al. (2017) found that BC dimensions positively (r = 0.110, 0.110 and 0.100), and PC dimensions negatively (r = −0.100, −0.100 and −0.040) correlated with the three dimension of creative thinking.

Finally, CS represents direct academic involvement like guidance and home-school communication (Cho and Lin, 2011; Oh et al., 2014; Zheng et al., 2020) linked CS to heightened creativity (r = 0.290) on the creative problem solving in math/science scale among 733 elementary/middle schoolers (Cho and Lin, 2011).

These differential effects align with self-determination theory positing autonomy, relatedness, and competence as undergirding motivation and creativity (Ryan and Deci, 2019, 2020). In contrast, parental psychological control, such as “love recycling” and “guilt triggering,” thwarts intrinsic motivation, narrowing exploratory behavior and hampering creativity. Accordingly, we propose the Hypothesis 2.

The parental involvement-creativity link may vary by children’s grade level. As grade level increases, opportunities for involvement in creative development may diminish for several reasons: increasing academic difficulty and specialization exceeding parental expertise, long times of school life and study limiting parent-child engagement, and children’s advancing cognitive skills (Said-Metwaly et al., 2021).

Empirically, moderating effects of grade have been demonstrated. Cho and Lin (2011) surveyed 733 elementary and middle school pupils and found positive family processes more strongly correlated with creative problem-solving in elementary than middle or high schoolers. Liang et al. (2021) also showed grade level differences in the involvement-creativity relationship for both mini-c and little-c creativity among 526 students. Collectively, these results indicate parental involvement exerts greater impact at elementary relative to secondary levels as academic contexts shift. Therefore, we propose Hypothesis 3.

Theoretically, both paternal and maternal involvement could encompass autonomy support, psychological control, behavior control and content support. However, differential links with creativity have been proposed. For instance, among 65 5th graders, Wallinga and Crase (1979) evidenced significant positive father-creativity but not mother-creativity correlations. Yet with 550 Chinese high schoolers, Liu et al. (2013) found maternal involvement more strongly favored creative thinking development. Due to the inconsistency among various research, we propose Hypothesis 4:

The influence of region on the interplay between parental involvement and student creativity warrants examination. Regions are often distinguished as Eastern (e.g., Chinese, Korean, and Japan), associated with psychological control and content support, and Western (e.g., European countries such as England and France and their former colonies) emphasizing autonomy support and behavioral monitoring (Niu and Sternberg, 2003; Ofole and Ezeokoli, 2014; Oh et al., 2014; Man et al., 2015; Fanchini et al., 2018; Zhang, 2023).

Eastern culture may have a dampening influence on creativity. Niu and Sternberg (2001), through the experimental study of students from two different cultures in China and the United States, showed that the artistic creativity of Chinese students was more likely to be reduced as a function of restrictive task constraints or of the absence of explicit instructions to be creative.

Western culture may have a stimulating influence on creativity. Niu and Sternberg (2001) showed that an independent self-oriented culture (western culture) is more encouraging of the development of artistic creativity than is an interdependent self-oriented culture (eastern culture). Shao et al. (2019) found that individuals from different cultures, particularly those from individualist and collectivist cultures, show differences in preferred creative processes and creative processing modes. To be specific, usefulness seems more important than novelty in the East, whereas novelty seems equally important as usefulness, if not more so, in the West when they are engaged in creative endeavors.

These varying socialization patterns cultivate differing levels of creativity across cultures (Niu and Sternberg, 2001). Overarching cultural orientation effects also emerge empirically. For example, in a cross-cultural study across eight countries, Zhang et al. (2021) found the involvement-creativity link varied significantly, with nonsignificant correlations in most Eastern countries (China, Kosovo, Russia, Saudi Arabia) in contrast to a significant negative correlation in the Western country of Chile. Therefore, we hypothesize:

According to previous research (Card, 2012), publication type may also be a potential moderating variable. During journal review processes, core publications preferentially select significant findings over nonsignificant results published in peripheral outlets like dissertations or conference papers (Borenstein et al., 2009; Card, 2012). Our literature review observed such patterns; for example, Yin (2019) found a nonsignificant parental involvement-creativity association in a master thesis. By contrast, among studies published in core educational psychology journals, like Man et al. (2015) in a journal found that parental behavioral control and psychological control and social creativity were significantly correlated with correlation coefficients of 0.090 and −0.210. This expected effect of publication bias leads to our final hypothesis:

All research hypotheses are organized as a diagram as shown in Figure 1.

A systematic search was conducted across major Chinese (Chinese National Knowledge Infrastructure (CNKI), Wanfang Data Knowledge Service Platform, VIP Chinese Journal Service Platform) and English (SAGE, Wiley, Springer, Taylor and Francis, ScienceDirect, ProQuest, JSTOR, Web of Science) databases combining parental related terms (“parental,” “paternal,” “maternal”) with involvement terms (“involvement,” “assistance”) and creativity terms (“creativity,” “creative thinking”) in title, abstract and keyword fields. To minimize publication bias, additional searches utilized public search engines and reference lists, including relevant reviews (e.g., van der Zanden et al., 2020). The final search date was November 6, 2023 resulting in an initial pool of 3,777 records. Figure 2 diagrams the systematic search and selection process.

Studies that met the following criteria were included in the meta-analysis: (1) reported correlation coefficients between the total score or dimensions of parental involvement and student creativity and its sub-dimensions, or other statistics that could be transformed into correlation coefficients, such as F-values or beta values. Multiple regressions were excluded due to difficulty in recovery. (2) The study population consisted of students from kindergarten through undergraduate level. Primary means K-6, middle school and high school mean 7–12. (3) Sample sizes for independent samples are reported. (4) Measurement instruments for and parental involvement and creativity were reported.

Studies that met the following criteria were excluded in the meta-analysis:

More than 40 studies were initially retrieved, and a total of 30 studies met the above criteria (see Table 1), containing 37 independent samples, with a total participation of 20,906, including 7 Chinese-language studies, 23 English-language studies, 24 journal articles, and 6 dissertations. The publication years span from 1991 to 2022.

Each study was coded according to the characteristics (see Table 1). It is important to note that effect sizes were calculated in terms of independent samples and were coded separately if more than one independent sample was reported in a single study at the same time.

The effect size used for this research is the correlation coefficient r (zero-order r) (Borenstein et al., 2009). In the calculation process, each r value was first converted to the corresponding Fisher’s Z score with the formula Z = 0.5 × In (1+r1–r). The Fisher’s Z values were then converted to calculate the zero-order correlation coefficient r with r = e2z+1e2z−1 (Borenstein et al., 2009). Where the variance of Z is V_Z =1n−3, n denotes the sample size, and the standard error of Z is SEz =VZ. Krumm et al. (2015) reported the F-value. Fanchini et al. (2018) reported the β-value. r=FF+dfe (Card, 2012), r = β (simple linear regression) (Kim, 2011) which was transformed into an r-value to be included in the calculation.

The current study employed a three-level random effects meta-analytic model (Assink and Wibbelink, 2016; Harrer et al., 2021) that assumes error at the sample (level 1, participations), outcome (level 2, effect sizes), and study levels (level 3). This approach estimates intra-cluster shared effect sizes through a random effects model, while permitting the same cluster share a same effect.

In this study, I² and Tau² were used to represent the between-study variance. It is suggested that 25, 50, and 75% of I²could be used as thresholds for low, moderate, and high heterogeneity (Borenstein et al., 2009). When a three-level model is used, the Tau² value is equal to the sum of two components, Tau² (level 2) and Tau² (level 3) (Assink and Wibbelink, 2016).

Publication bias refers to the phenomenon that studies with significant results are more likely to be published (Borenstein et al., 2009). Rosenthal’s Fail safe N (Nfs), Funnel plot, trim and fill method, Egger’s regression are commonly used to detect publication bias (Borenstein et al., 2009).

The Basic Quality Assessment of Primary Study, BQAPS (Xu et al., 2024 press) was used as a quality assessment tool to assess the quality of studies included in the meta-analysis. Based on the scoring criteria, score of 0–6 categorized as low-quality, 7–12 categorized as low- medium level, 13–18 were categorized as medium-high level, and 19–24 were categorized as high quality.

All statistical calculations were conducted in R (R Core Team, 2022). The metafor package was used for the calculation of the three-level model (Viechtbauer, 2010; Harrer et al., 2021). The meta package was used for the creation of the funnel plots (Assink and Wibbelink, 2016; Balduzzi et al., 2019; Harrer et al., 2021).

The parental involvement type had a significant moderating effect (F _(3,66) = 17.591, p < 0.001) (see Table 4). Autonomy support was positively associated with student creativity with an r-value of 0.144 [95% CI (0.092, 0.194)]. The content support group had an r-value of 0.131 [95% CI (0.084, 0.179)]. The r-value for the psychological control group was −0.117 [95% CI (−0.189, −0.045)]. The r-value for the behavioral control group was 0.133 [95% CI (0.045, 0.218)].

The moderating effect of grade was not significant (F _(4,60) = 0.900, p = 0.552). The r-values were 0.150 [95% CI (−0.029, 0.319)], 0.119 [95% CI (0.035, 0.203)], 0.081 [95% CI (0.023, 0.137)], 0.087 [95% CI (0.018, 0.156)], and 0.106 [95% CI (−0.056, 0.263)] for pre-school, elementary, middle school, elementary and secondary, and university, respectively. There were no significant differences between the r-values for the subgroups of grade.

The moderating effect of parental gender was not significant (F _(2,67) = 0.300, p = 0.742). The r-values for maternal involvement and parental involvement were 0.096 [95% CI (0.011, 0.180)], and 0.108 [95% CI (0.061, 0.154)], respectively. The r-value for father involvement was 0.065 [95% CI (−0.037, 0.165)]. The r-values for the three subgroups of parental gender were not significantly different.

The moderating effect of region was not significant (F _(1,64) = 0.098, p = 0.756). The r-values were 0.085 [95% CI (0.044, 0.125)] and 0.098 [95% CI (0.021, 0.174)] for the Eastern and Western culture groups, respectively.

The moderating effect of publication type was not significant (F _(1,68) = 0.934, p = 0.337). The r-values was 0.111 [95% CI (0.066, 0.154)] and 0.062 [95% CI (−0.029, 0.151)] for the journal group and master’s thesis group, respectively.