Despite that overparenting is a well-intended practice, is also generally understood that excessive protection of their children and solving their problems for them (APS) have an adverse impact on children’s outcomes (Luebbe et al., 2018; Reed et al., 2016; Segrin et al., 2015). Self-determination theory (Ryan and Deci, 2000) could help explain why these behaviors could be related to maladjustment. According to this theory, contexts that prevent the satisfaction of autonomy, competence and relatedness (three basic psychological needs) negatively influence on psychological and behavioral development. Parents that provide excessive assistance and anticipatory problem solving implicitly convey negative feedback to children, suggesting that they are not competent and unable to tackle problems on their own. Consequently, these implicit messages would undermine children’s sense of competence and control (Hong and Cui, 2020).

From an empirical point of view, much of the overparenting research has been conducted with students in high school or college, showing that this parenting behavior can lead to high levels of narcissism, dependence on others and depression and low levels of self-efficacy in the students (Bradley-Geist and Olson-Buchanan, 2014; Odenweller et al., 2014; Schiffrin et al., 2015; Segrin et al., 2013). Nevertheless, these same parenting behaviors may be developmentally appropriate in other developmental periods. For example, in infants to 3–4 years old, benefits of high levels of parental assistance to their children in anticipating and solving problems include cognitive, physical and socioemotional development (Schiffrin et al., 2015). However, the influence is detrimental to children’s development when parents are not able to reduce these monitoring behaviors as their children grow. Although there is also little research on the impact of overparenting on children in elementary schools, some studies have shown that overprotective parents avoid the experience of failure to their children, complicating the development of their problem-solving and task-management skills on their own (Fearon and Roisman, 2017; Randall et al., 2015) as well as their independence and ability to choose (Liss and Schiffrin, 2014). This may become problematic for children’s emotional development (Hussain et al., 2021; Perry et al., 2018), and inability to socialize and deal with failure, and even developmental delays (Padilla-Walker et al., 2021; Valente et al., 2017). It is not known when anticipatory problem solving could begin to have negative effects on children’s development. Therefore, more research is needed to improve our understanding of the impact of these parenting behaviors on elementary school children (Turner et al., 2022).

On the other hand, parent educational attainment is an important background factor and often linked to parenting practices (Azad et al., 2014; Hoff and Laursen, 2019). Popularly, excessive involvement in helping their children solve their problems tends to be used more frequently by families with a higher socioeconomic level, given the belief that these parents have more time to focus on raising their children and their success is assumed as their own (Segrin et al., 2013; van Ingen et al., 2015). Thus, these overparenting behaviors have been associated with higher parents’ educational level in the media (Gibbs, 2009). However, the empirical evidence available in this regard is scarce; rather, the few studies conducted show that parents with a high level of education were less likely to use overprotective behaviors (e.g., Sleddens et al., 2014).

More research is needed to better understand the associations between sociodemographic characteristics, overparenting and children’s outcomes in elementary school (Arlinghaus et al., 2023). Considering findings of previous research, we hypothesized that Anticipatory Problem Solving by fathers is not only associated with the school problems of their children, but also mediates the relations between fathers’ educational level and children’s school problems. Understanding the role that anticipatory problem solving plays in promoting children’s school success could have relevant implications for parenting and education practices.

The Environmental Sensitivity Theory (Greven et al., 2019; Pluess, 2015) is known as an evolutionary-developmental theory, which explains that humans differ in their sensitivity to contextual factors. Several studies conducted under the umbrella of Environmental Sensitivity Theory have reported gender differences in favor of females in Sensory Processing Sensitivity (Benham, 2006; Chacón et al., 2021; Konrad and Herzberg, 2017). Sensory Processing Sensitivity has been defined as a genetic-based personality trait characterized by an ability to respond to external factors and process sensory information more deeply (Assary et al., 2021; Pluess, 2015). This personality trait involves a higher sensitivity to various stimuli, such as environmental influences from groups, families, schools, etc. (Aron et al., 2012; Aron and Aron, 1997; Jagiellowicz et al., 2011; Pluess, 2015).

Moreover, there are other reasons for testing for differences depending on a child’s gender in the associations of contextual paternal factors and children’s outcomes. Firstly, since the 90s, several studies have shown, in countries committed to equal opportunities, gender differences in favor of girls in diverse measures related to school performance (Epstein et al., 1998; Freudenthaler et al., 2008; Steinmayr and Spinath, 2008; Wong et al., 2002). Secondly, there is also some research exploring the relations between child gender and factors commonly associated with overparenting as parental monitoring or involvement that suggests child gender does influence levels of these parental behaviors (Padilla-Walker and Nelson, 2012) with female children generally reporting greater rates (Bradley-Geist and Olson-Buchanan, 2014; Klevens and Hall, 2014). Thirdly, several studies testing for gender differences in the relation between parenting and child outcomes suggest that girls appear to be more sensitive to parenting practices than boys (e.g., Lewis-Morrarty et al., 2015), hence parenting may be a more intense predictor of child’s outcomes for girls than for boys (Leung, 2020).

There are also some studies, carried out on adolescents or emerging adults, that have analyzed the moderating role of gender in the association of overparenting with child’s outcomes. Although, some of these researches have not found gender differences on this association (e.g., Darlow et al., 2017; Scharf et al., 2017), most of them have suggested that overparenting seems to increase risk of maladjustment concretely for girls (Barton and Kirtley, 2012; Janssens et al., 2009; Kouros et al., 2017; Patock-Peckham and Morgan-Lopez, 2006). In the present study, we analyze the moderating role of the child’s gender in the effects of father’s educational level on APS and school problems of children during childhood. Although we have not found previous studies in this period of age, we hypothesize, based on the results discussed, that these effects will be stronger among girls than among boys.

On the other hand, we also tested surgency as an additional moderator. Surgency (extraversion) is defined as the level of arousal and positive affect regarding an activity or change, as well as a low level of behavioral inhibition or shyness (Rothbart, 2007). Some research has showed that higher levels of surgency were predictive of higher levels of motivation for school (Chen and Zhang, 2011) and higher levels of academic success in middle childhood (Bramlett et al., 2000; Colom et al., 2007; Rudasill et al., 2010). In contrast, shyness (an indicator of low surgency) has been sometimes negatively associated with academic outcome (Hughes and Coplan, 2010; Zhang et al., 2017). Thus, for highly shy children, even going to school can be a challenge, causing them academic difficulties due to their low levels of emotional and behavioral engagement, low levels in language skills, low school liking and less participation in teacher-child interactions (Cameron, 2009; Eggum-Wilkens et al., 2014; Rudasill, 2011; Valiente et al., 2012).

Moreover, research carried out in the framework of Environmental Sensitivity Theory has found that introversion (or low surgency) is related to higher levels of sensitivity to the environment, such as parental influences (Asscher et al., 2016; Hentges et al., 2015; Pluess et al., 2010). For instance, parents’ negative affect predicted childhood anxiety problems for highly shy children (an indicator of low surgency), but not for other children (Lindhout et al., 2009). However, other studies have found that children with high levels of surgency seem to be more sensitive to the effects of parenting; for example, del Puerto-Golzarri et al. (2022) found that boys with high levels of surgency whose fathers used an authoritarian parenting style showed more reactive aggression. We have only found a study that examines whether surgency would moderate the link between APS and child’s behavior problems, such as aggression (Ruiz-Ortiz et al., 2023). This study showed that only among children with high levels of surgency, overprotection reduced levels of child’s aggression.

Lastly, it seems there is a relation between surgency and gender. Else-Quest et al. (2006) found that surgency showed a difference, favoring boys, in particular with boys scoring higher than girls. Other studies have found gender differences in the components or surgency. Thus, boys are higher on activity level and high-intensity pleasure (Gaias et al., 2012; Muris and Meesters, 2009). Moreover, findings about shyness have suggested that this temperament trait may be less socially acceptable for boys than for girls, even parents seem to react more negatively to shy behaviors in boys as compared to girls (e.g., Bosacki, 2008; Rubin and Coplan, 2004). For all these reasons, we are interested in including surgency as an additional moderator in our study and the previous findings lead us to hypothesize that we will find triple interactive effects of paternal factors with child’s gender and surgency on children’s school problems. Given the lack of prior specific research about the possible direction of this moderation, this research aim was exploratory.

The sample was selected by age and sex, according to geographical localization. Five governmental schools in southern Spain (Cádiz) agreed to participate in the study including 14 classrooms with 335 students. A comprehensive verbal and written description of the purpose of the study was given to the children’s parents and teachers. Participation in the study was voluntary and anonymous and that the study adhered to all ethical guidelines. Written consents from 252 families were obtained. Once the families returned the completed questionnaires and the data was coded, a final sample of 203 schoolchildren, 96 boys (47.3%) and 107 girls (52.7%), aged 7–8 years (M = 92.42 months, SD = 3.52) participated in the study. Fathers’ average ages were 37.59 years (SD = 7.97). The 26.6% of fathers completed primary education, 10.8% secondary education, 47.7% high school or higher vocational training studies and 14.8% university studies.

In order to avoid common method biases, sample data was collected at three time points. Data about fathers’ educational level was obtained at T1 (first term or school year) by an ad hoc sociodemographic questionnaire. During the second term of school year (T2), fathers filled the Overparenting Scale, concretely the Anticipatory Problems Solving subscale (Segrin et al., 2012) and fathers and mothers jointly informed about children’s surgency by the Children’s Behavior Questionnaire (CBQ; Rothbart et al., 2001). In the third term of school year (T3), teachers reported information about their students’ school problems completing the Behavior Assessment System for Children (BASC T-2, Reynolds and Kamphaus, 2004). Given common method bias basically occurs when all data are obtained from the same sources or informants, we tried to prevent it by implementing a procedural remedy, as recommended by Podsakoff et al. (2003). Thus, we collected the measures of the variables from key but different informants (APS by father; School problems by teacher; child’s surgency and demographic variables by father and mother jointly in different time). This procedure can prevent the informant’s mentality from biasing the observed relationship between the variables, thus eliminating the effects of implicit theories, social desirability, or evaluator moods (Podsakoff et al., 2003).

Fathers reported their educational level through the ad hoc sociodemographic questionnaire, based on Antolín et al. (2009). Fathers’ education levels (predictor variable) were coded into five categories: (1) primary school, (2) secondary school, (3) high school, (4) higher vocational school, and (5) university degree or higher. Fathers indicated their highest level of education in the five points scale. Fathers’ education levels were coded from 1 to 5. The larger the numbers are, the higher the education level is.

Due to their documented association with school problems, father’s employment status, number of siblings and father’s age were treated as covariates. It seems be clear that parents’ employment status plays a significant role on students’ academic performance, either because of a better employment status provide necessary facilities needed for the enhancement of their children education (Suleman et al., 2012) or because of parental employment affects the time spent with their children (Schildberg-Hörisch, 2016). Furthermore, some studies have pointed out that educational performance is negatively influenced by a high number of siblings (Baert et al., 2022; Biblarz and Raftery, 1999; Björklund and Sundström, 2006; Garasky, 1995; Ghysels and Van Vlasselaer, 2008). On the other hand, parent age and the number of siblings are likely to influence interactively on children’s performance. Hence, we control for each factor separately.

The information about age, fathers’ employment status and number of siblings was reported by fathers through the ad hoc sociodemographic questionnaire. Father’s age and number of siblings were used as continuous variables. Fathers’ occupation was coded into five categories: (1) active, (2) unemployed with allowance, (3) unemployed without allowance, (4) retired and (5) others.

The study used SPSS 21 to carry out descriptive statistics and bivariate correlation analyses in order to test H1 y H2. The other hypotheses were tested using the PROCESS macro for SPSS (Hayes, 2018), using bias-corrected bootstrapping with 5,000 resamples to account for the non-normal distribution in the outcome variables (MacKinnon et al., 2004) and to generate confidence intervals (Preacher and Hayes, 2008). Specifically, in the current study models 4, 15, and 19 were used following Hayes’s recommendations (Hayes, 2018). Concretely, we used Model 4 and calculated 5,000 bootstrapped samples to estimate the 95% bias corrected and accelerated confidence intervals of the indirect effect to study direct and indirect effects of father’s educational level and APS on children’s school problems (H3). A mediation test is significant when the lower and the upper bounds of the bootstrap confidence intervals of the indirect effect between the predictor and the outcome do not include zero (Hayes, 2018). Instead of the traditional Sobel test, PROCESS model bootstrap approach was used, because the bootstrap method has higher statistical power and makes more realistic assumptions about the sampling distribution of the indirect effect (Abu-Bader and Jones, 2021; MacKinnon et al., 2004).

Further to test H4 and H5, we used Model 15 with bias-corrected bootstrap confidence intervals (BC; 95% CI) based on 5,000 bootstrap resamples. Model 15 was used to evaluate the moderation effect of children’s gender in both the direct and indirect effects of fathers’ educational level and APS on children’s school problems (Hayes, 2018). Lastly to test H6 and H7, Model 19 was used with bias-corrected bootstrap confidence intervals (BC; 95% CI) based on 5,000 bootstrap resamples. Model 19 was conducted to test three-way interactions, which is similar to model 15, but with a moderator influencing the indirect path by the mediator that depends on a second moderator (a moderated moderated mediation). This model was used to test if the moderated effect by child’s gender for father’s educational level on child’s school problems via APS would be at the same time moderated by child’s surgency (Hayes, 2018). Significant interactions were probed using a traditional simple slope test, the pick-a-point approach (Aiken et al., 1991), and the Johnson-Neyman technique for regions of significance (Hayes and Matthes, 2009; Preacher et al., 2006), which identifies the regions of significance of an association when the moderator is a continuous variable (Hayes, 2018). Since regression analyses included interaction terms, continuous variables that constitute an interaction product were mean centered to minimize multicollinearity issues (Field, 2013).

In the study, to determine the correlations between the variables, bivariate correlation analysis was applied. Table 1 provides the correlations for all variables. As the result of the analysis, between father’s educational level and APS and also between father’s educational level and child’s school problems, statistically significant negative correlations were found. In addition, a statistically significant positive correlation was detected between APS and child’s school problems. Thus, H1 and H2 were supported. The moderating variables (child’s gender and surgency) did not show a significant relationship with any other variable. Lastly, father’s age showed a statistically significant positive correlation with father’s educational level and negative with the child’s school problems. Father’s employment status was negatively associated with father’s educational level (note that regarding educational level, the larger the numbers are, the higher the education level is, and regarding employment status, the larger the numbers are, the less socially recognized employment status is).

Following the procedures outlined by Preacher et al. (2007), we used PROCESS macro in the SPSS software by selecting Model 4 (Hayes, 2018) to address H3, which stated the effect of father’s educational level on child’s school problems is mediated by APS. The analyses included father’s age and employment status and number of siblings as control variables. The results of the mediation analysis are presented in Table 2. As can be seen in Table 2, firstly the findings revealed that between father’s educational level and child’s school problems, after controlling for father’s age and employment status and number of siblings, there was a tendency toward statistical signification for negative association (b = −0.978, p = 0.060). Secondly, the results also indicated that father’s educational level was negatively associated with APS (b = −0.144, p = 0.006). Thirdly, APS was positively related to child’s school problems (b = −1.793, p = 0.015). Moreover, the residual direct association between father’s educational level and child’s school problems was not significant. Additionally, regarding control variables, father’s age negatively related to child’s school problems (b = −0.365, p = 0.016) and number of siblings had a positive and significant relation with child’s school problems (b = 2.333, p = 0.034), regardless of father’s educational level and APS. The bootstrapping analysis with 5,000 iterations revealed that APS mediated the relation between father’s educational level and child’s school problems. The 95% bias-corrected confidence interval for the indirect effect did not include zero (indirect effect = −0.258, SE = 0.142, 95% CI [−0.583, −0.033]) and the direct effect included zero (direct effect = −0.719, SE = 0.521, 95% CI [−1.748, 0.309]). Thus, H3 is supported. The mediation effect accounts for 26.40% of the total effect of father’s educational level on child’s school problems (Wen and Fan, 2015).

In order to test H4 and H5, a moderated mediation model using PROCESS Model 15 (Hayes, 2018) was conducted to test whether gender had a moderating role in the direct effects of father’s educational level and APS on child’s school problems, as well as, in the indirect effect of father’s educational level on child’s school problems via APS, controlling for father’s age and employment status and number of siblings. The results in Table 3 showed that child’s gender moderated the relationship between APS and child’s school problems (b = 4.271, p = 0.004), whereas having no significant effect on the relation between father’s educational level and child’s school problems. In order to further analyze the APS × child’s gender interaction, a simple slope test was performed using the pick-a-point approach (Aiken et al., 1991). A significant positive relation between APS and child’s school problems was found among girls (b = 3.424, p = 0.000), while a non-significant negative relation was found among boys (b = −0.847, p = 0.450), as shown in Figure 2. Additionally, the index of moderated mediation was significant (index = −0.615, SE = 0.311, 95% CI [−1.333, −0.125], 5,000 bootstraps); only among girls, the indirect effect value of the father’s educational level on child’s school problems by APS was significant (boys, indirect effect = 0.122, SE = 0.182, 95% CI [−0.210, 0.530]; girls, indirect effect = −0.493, SE = 0.232, 95% CI [−1.027, −0.122]). Thus, H4 was only partially supported because child’s gender moderated the direct effect of APS on child’s school problems (only among girls, this direct effect was significant), but did not moderate the direct effect of father’s educational level on child’s school problems. H5 was supported because the indirect effect of fathers’ educational level on children’s school problems via APS was significantly different for boys than for girls.

In order to test H6 and H7, a moderated moderated mediation model using PROCESS Model 19 (Hayes, 2018), with child’s school problems as the outcome variable, father’s educational level as the predictor, APS as the mediating variable and child’s gender and surgency as the moderating variables, controlling for father’s age and employment status and number of siblings.

The model predicting APS revealed 6.1% of explained variance [F_{(4, 189)} = 2.973, p = 0.021]. Father’s educational level was the only significant predictor of APS, whereas any covariates did not significantly predict the criterion (Table 4). The regression model predicting child’s school problems revealed 21% of explained variance [F_{(14, 189)} = 3.239, p = 0.000]. As expected and in line with H6 (see Table 4), the three-way interaction Surgency × Gender × APS was statistically significant (b = −3.966, p = 0.025). However, the three-way interaction Surgency × Gender × father’s educational level was not statistically significant (b = −1.859, p = 0.109), hence H6 was only partially supported. The three-way interaction Surgency × Gender × APS significantly increased the amount of explained variance by 2.3% [F_{(1, 174)} = 5.106, p = 0.025]. The conditional effects of APS × child’s gender on child’s school problems at different levels of surgency revealed that this two-way interaction was a significant positive predictor of school problems when surgency was low (b = 9.853, p = 0.000) and medium (b = 5.844, p = 0.000), but not when surgency was high (b = 1.835, p = 0.360). Thus, as shown in Table 5, the relation between APS and school problems was significant for girls, regardless the surgency level, whereas among boys, only when they had a low level of surgency, this relation was significant. Simple slopes analyses were computed at the mean of surgency and ± 1 SD, producing the plots in Figure 3. Johnson-Neyman technique provided additional information, showing that the value 0.653 of surgency is the one that splits different regions of signification, in such a way that the interactive effect APS × child’s gender on school problems is significant for surgency values below of 0.653 (74.1%) and is non-significant for surgency values above of 0.653 (25.9%).

The overall index of moderated moderated mediation was significant (index = 0.583, SE = 0.337, 95% CI [0.003, 1.311]), supporting H7. More specifically, results revealed that the moderated mediation effect was significant in the conditions of low and medium levels of surgency (index = −1.447, SE = 0.634, 95% CI [−2.835, −0.373] and index = −0.858, SE = 0.375, 95% CI [−1.696, −0.231], respectively), but not in the condition of high level of surgency (index = −0.270, SE = 0.334, 95% CI [−1.023, 0.298]). In the condition of low level of surgency, the effect of father’s educational level on child’s school problems was significant and positively mediated by APS for boys (index = 0.874, SE = 0.420, 95% CI [0.162, 1.809]), but significant and negatively mediated for girls (index = −0.573, SE = 0.348, 95% CI [−1.369, −0.011]). In the condition of medium level of surgency, the effect of father’s educational level on child’s school problems was positive and non-significantly mediated by APS for boys (index = 0.345, SE = 0.219, 95% CI [−0.016, 0.851]), but significant and negatively mediated for girls (index = −0.513, SE = 0.240, 95% CI [−1.041, −0.132]). Lastly, although the index of moderated mediation was not significant in the condition of high level of surgency, it is interesting to note that the effect of father’s educational level on child’s school problems was negative and non-significantly mediated by APS for boys (index = −0.184, SE = 0.215, 95% CI [−0.634, 0.213]), and also negative but significantly mediated for girls (index = −0.454, SE = 0.301, 95% CI [−1.155, −0.008]).

The current study provided important knowledge to advance our understanding of the direct and indirect influence of paternal factors on children’s problems at school. However, our research still has some limitations that point out future research directions. First, we used a community sample of children from Cádiz (southern Spain); thus, our profile of temperamentally shy boys may not capture the extreme temperamental trait and these findings may not be generalizable to other groups and other geographical regions. Future studies are needed to replicate these findings in more diverse samples of children. Second, the current study included different informants (fathers reported their overparenting behaviors, mother and father jointly informed about children’s surgency and teachers reported children’s school problems) and the measuring instruments selected are widely used. However, the use of parent-reported data as our measure of surgency could be restricted to the contexts that parents observe their children, and their responses may have been biased by social desirability. Future work should improve reliability using additional informants for each measure and using complementary other methods as direct observation or physiological measures. Third, the PROCESS macro in SPSS was used to test the proposed mediation–moderation model which was based on previous literature and theory. However, and although data was collected at three time points, the used cross-sectional design does prevents us determining the direction of causality among the variables. Future studies using longitudinal designs are needed to confirm these findings. Fourth, our initial interest focused on the role of overparenting as mediator of the relation between fathers’ educational level and children’s school problems and the moderating role of gender and surgency. Nevertheless, future studies could explore alternative explanations concerning how a low educational level of the father can lead to the school problems of his children.

Self-determination theory (Ryan and Deci, 2000) has been shown as a suitable psychological framework to assess the influence of overparenting behaviors on children’s school problems. Considering that the basic human need for autonomy requires to be satisfied in order to assure an individual’s wellbeing, the theory explicitly supports that overparenting behaviors could threaten their children’s school adjustment preventing the satisfaction of their needs for autonomy. The current research adds that the consequences of this parental style depend on variables such as gender and age, so that boys and girls may differ in their need for personal autonomy based on their rate of development, challenges, and school transitions.

On the other hand, our findings add empirical evidence to the Environmental Sensitivity framework (elaboration on the various theories as environmental susceptibility or diathesis-stress; Ellis et al., 2011; Pluess, 2015), showing that the effects of two environmental factors (paternal educational level and overprotection) on school maladjustment are moderated by individual variables, such as gender and surgency. Moreover, the current study is in line with others that show that environmental sensitivity is better conceptualized as multidimensional and better understood when it is studied employing a person-centered approach (Markovitch et al., 2023; Moran et al., 2017). In this sense, our findings suggest that it is the combination of one’s tendency to shyness and low extraversion with gender that possibly has a unique role in environmental sensitivity rather than each of these individual characteristics separately. Understanding the complex individual differences in environmental sensitivity requires more research by various analytic strategies.

Given the noted relevance of adjusting to elementary school to the development of later socioemotional and academic skills, and the growing trend of parents in western countries to overprotect their children, the present study poses some practical implications. First, for teachers and parents, our findings highlight the importance of an early identification of children who display high levels of shyness, especially among boys, in order to reinforce the support so that they can overcome the challenges and threats at school, gradually promoting autonomy. Moreover, the fact that girls appear to be more sensitive to adverse effects of overprotective behaviors of their fathers, programs guiding parents to handle these behaviors which play a key role in their daughters’ school problems. Further, these programs are especially crucial for those fathers with low educational attainment. These probably well-intended fathers should be aware that helping their daughters too much to solve their problems at school, when they could deal with them, could have harmful consequences for their progress in educational settings.

In this sense, findings raise implications for educational settings since it can support programs aimed at parents with a low educational level and at risk of displaying overparenting behaviors, considering children’s characteristics.