Dozens of studies show that good family school relationships can promote students’ academic achievements and positively impact students’ growth (Garcia, 2004; Pomerantz et al., 2007; Zhang, 2011; Kim and Hill, 2015; Herman and Reinke, 2017). According to the Ecological systems theory (Bronfenbrenner, 1979; Bronfenbrenner and Morris, 1998), adolescent development is influenced directly by the interactions that take place within a single microsystem (such as the family, school, or peer group), as well as the similarities and differences in the pattern of the interactions that occur across these systems. Family and school are the two most important environments closely associated with adolescent development (Lohman et al., 2007; Santana et al., 2016).

Research shows that not all family school relationships are good for students (Pomerantz et al., 2007), and it depends on whether parents and teachers have good relations (Nir and Ami, 2005; Emerson et al., 2012; Liang, 2015). Only a relationship based on mutual trust and honesty can rich and frequent interaction be conducive to the development of students (Constantino, 2003; Garcia, 2004). A good relationship between parents and teachers can provide consistent support for students, making students more likely to recognize the benefits of learning (Harris and Goodall, 2009; Emerson et al., 2012). Waller (1932) points out that parents and teachers are natural enemies, not natural collaborators. The complex and crucial connection between families and schools is embodied in the relationships between individual teachers and their students’ families (Driessen et al., 2005; Cole, 2013).

Good parent-teacher relationships are generally effective at maximizing students’ potential, based on the mutually rich and frequent interactions built upon trust and honesty (Constantino, 2003; Garcia, 2004). According to interpersonal communication theory, the interpersonal relationship measurement can be classified into two variables. The first is the dimension of parent-teacher contact, which considers the individual’s judgment of the relationship, such as the frequency of communication, the content, and means of communication, and the efforts the other may make to maintain the relationship. The second is the parent-teacher relationship quality, which mainly refers to the intensity of this judgment and is the perception of the intensity of the former (Dillard et al., 1999). Generally, an individual’s perception of the quality of a relationship comes directly from the state of the communication, such as the frequency of communication and the degree of self-disclosure. Relationships develop over time through a process of self-disclosure (Bylund et al., 2012). Moreover, a relationship comes from the perception of the individual’s influence on others, such as how much influence an individual can have on the decisions or behaviors of others and their evaluation of others as individuals (Zhang, 2016). Such that, there is a difference between parent-teacher contact and the parent-teacher relationship quality. Parent-teacher contact refers to the frequency of interactions between parents and teachers, whereas the parent-teacher relationship quality refers to the quality of their relationship as well as how well their goals are aligned (Serpell and Mashburn, 2012; Stormont et al., 2013; Deng et al., 2018).

In parent-teacher contact, the frequency of ties and activities between families and schools influences the relationship between parents and teachers (Zhao and Hong, 2012). The more connections and activities between families and schools are built and held, the better the relationship between teachers and parents will be (Adams and Christenson, 2000; Minke, 2006). The more frequent the interaction between parents and teachers, the more likely parents are to pass on their educational expectations to their children, thus promoting students’ improvement at the cognitive level and success in academic achievement (Goyette and Xie, 1999; Sebastian and Allensworth, 2012). Parents’ participation in parental meetings, voluntary school activities, and initiatives where they learn from teachers about their children’s learning or behavior can significantly improve their student’s access to school (Perna and Titus, 2005; Pong et al., 2005). Regular parental meetings encourage parents to spend more time helping students and supervising school work, which is beneficial to improving students’ test scores and plays an important role in student attitudes and behaviors (Islam, 2019). Together, parents’ participation in parental meetings, school activities, and teacher contact has a significant positive impact on students with low academic achievements. Furthermore, parental behavior can help students understand the importance that their parents attach to school as well as to their learning behavior (Cheng and Li, 2015; Li, 2017).

In addition, parent-teacher relationship quality variables means the perception of self-influence on others, such as the extent to which one influences other people’s decisions, actions, or evaluations (Zhang, 2016). The subjective relationship quality in the parent-teacher relationship refers to the parents’ and teachers’ emotional perception of the parent-teacher interaction. On the one hand, if parents do not experience a welcoming environment or support from teachers, their participation in constructing the parent-teacher relationship will be reduced. On the other hand, whether teachers receive due respect has a connection to the teachers’ work enthusiasm and students’ academic performance (Chen, 2008). If parents exhibit negative emotions toward the parent-teacher relationship, it will be harder to construct an excellent parent-teacher relationship (Tao, 2016). So, the emotional experience has an important impact on the parent-teacher relationship and students’ academic performance.

Previous studies have shown that contact and communication between teachers and parents are not frequent enough in China (Li, 2013; Zhu and Gao, 2014; Zhang, 2018). It is also possible that schools in China are usually closed, and school-based communication is dominated by schools (Ma, 2010; Li, 2013). Therefore, parents’ participation in school education is limited. At present, the parent-teacher relationship in China is usually manifested by a teacher telling parents about their children’s behavior in school, and then the parents participate in their children’s school education accordingly (Li, 2013; Huang, 2016). From there, most of the links between teachers and parents occur when a student exhibits bad performance, and parents naturally hope that the fewer links there are, the better everyone will be (Zhang, 2003). On the other hand, teachers report that their relationships with parents are a major source of occupational stress (Grolnick and Seal, 2008; Pepe and Addimando, 2013, 2014). Teachers feel more unpleasant experiences are involved in dealing with parent relationships than in other aspects of teaching (Hargreaves and Fullan, 1998).

Although many research studies have shown that the parent-teacher relationship has an important impact on student’s academic achievements (Hill and Tyson, 2009), the parent-teacher relationship in home-school cooperation has escaped the attention of many Chinese researchers, especially in regard to the lack of quantitative research. There are some limitations on the previous study on the topic. Firstly, the existing studies focus on the status of the parent-teacher relationship and the association between the parent-teacher relationship and students’ academic outcome at a certain period (Yuan et al., 2019), that is the use of cross-sectional data. And the long-term effect of the parent-teacher relationship on academic achievement is rare. The unique current social culture and educational environment of mainland China may limit the generalizations of dominant Western research findings on the parent-teacher partnership in China (Deng et al., 2018). Especially the teachers in a public school in China will widely be in charge of the same classes in the middle school; that is, the same teachers teach students in the 3 years of middle school in China, which may lead to the parent-teacher relationship having a carry-over impact. Since teachers will teach the same class for a long time, the original parent-teacher relationship may give parents and teachers a fixed impression, which will have a long-term impact on their interaction, and then have a long-term impact on students’ performance (Deng et al., 2018). Secondly, previous studies take the parent-teacher relationship as a whole variable while didn’t discuss the influence of the parent-teacher relationship at a different level. Assessing different aspects of the parent-teacher relationship may provide more detailed evidence to improve the parent-teacher relationship. Therefore, this paper attempts to address the gaps by aiming to examine the longitudinal association between the parent-teacher relationship and students’ academic achievements in China.

Adolescent development involves various factors, including the social, emotional, and cognitive aspects of the transition from childhood to adulthood (Jaggers et al., 2015). Compared with the performance at the primary school stage, middle school students’ academic performance tends to decline because of the great changes in puberty and their environment (Steinberg and Silk, 2002; Barber and Olsen, 2004; Keating, 2004; Lerner and Steinberg, 2004; Smetana et al., 2004; Hill and Chao, 2009). Additionally, the independence of middle school students in all aspects has gradually increased, resulting in a less close parent-teacher relationship than the one that exists in kindergartens and primary schools, and effective cooperation may be reduced (Seginer, 2006). Therefore, it is worthwhile to use middle school students as research subjects to explore the association between the parent-teacher relationship and middle school students’ developmental outcomes in the context of mainland China as well as the evidence-based potential practical strategies.

This study employs the hierarchical linear model (HLM) to discuss the association between the parent-teacher relationship and middle school students’ academic achievements by analyzing China Education Panel Survey (CEPS) data, which is a large-scale, nationally representative, longitudinal public database. Two questions were addressed in this study to enhance our understanding of the parent-teacher relationship in mainland China:

The data used in this study are from a large-scale longitudinal study [2013–2014 school year (Wave 1)] and [2014–2015 school year (Wave 2)] of the CEPS, a project conducted by the National Survey Research Center (NSRC) at the Renmin University of China. The respondents were middle school students in seventh grade and ninth grade. The sampling method was stratified random sampling based on the average education level and the proportion of migrants in the whole population. In addition, the respondents included parents, teachers, and school administrators. The survey covered 438 classes from 112 schools in 28 county-level administrative areas throughout the country; due to the compulsory education law (Ministry of Education, 2021) in China, the government administrates compulsory education mainly at the county level. The CEPS survey can represent the overall education situation in China. The questionnaire survey procedure starts with the investigator of NSRC contacting the principals at the selected middle schools and then collecting the students’ data by questionnaire.

The students’ ability scores were compiled according to the national students’ cognitive ability on a 20-item test designed by NSRC, which includes the ability of language, graph, calculation, and logic. The score ranges from 0 to 20. It was completed on the spot in class and was strictly time-controlled. The answers were collected and sealed on the spot. The principals then arranged for the related teachers to fill out the questionnaire to collect the teachers’ data. After that, the head teachers of the selected classes asked the parents to fill out the questionnaire. Finally, the questionnaire was mailed to NSRC after it was collected by the head teachers. Many studies have analyzed and discussed the problems of education in China based on these data (e.g., Huang, 2016; Wu, 2016; Zhang, 2016; Li and Zheng, 2017; Pan and Zhang, 2017; Tong, 2017; Wei and Ma, 2017; Zhao et al., 2017; Fang, 2018; Liang et al., 2018). The data are very rich, including basic personal information, students’ situations in school, parent-child interactions, and home-school interactions, all of which can provide information for us to further study the impacts of the parent-teacher relationship on students’ academic achievements.

In this study, we used the data of students in Grade 7 to analyze the carry-over impact of the parent-teacher relationship on students’ achievements for middle school students in China, since the students in seventh grade have the follow-up data 1 year later. After excluding incomplete samples, the sample size of students (parents) was 8,426, and the sample size of classes was 185.

Data used in the current study were collected from three sources, including a child’s responses, responses from one of his/her parents, and responses from his/her classroom teachers. Therefore, we classified variables into two levels, parent-child dyad level and class level.

Due to the data hierarchy, where students were nested in the classroom, the HLM was adopted to account for the dependency among students in the same classroom. Considering that there are only two classes in each grade in each school, it is not suitable to construct a three-level model of student, class, and school. This paper uses a two-level HLM as the analysis model. Level 1 is the parent-child dyad level, and students’ gender, cognitive abilities, parents reported parent-teacher relationship variables, FES used as the predictors. Considering that categorized variables cannot be directly used in the HLM, a dummy coding method was used to transform the FES into two dummy variables. More specially, the first categorical of FES (FES-C1, difficulty level) was set as the reference level, and then two dummy variables, FES.C2 and FES.C3, represented medium and rich levels, respectively, were created.

Level 2 is the class level, and the teachers reported that parent-teacher variables were used as class-level predictors. In addition, corresponding child-level predictors were aggregated at the class level to separate level-1 and level-2 effects (Hoffman, 2015, pp. 344).

We followed the step-up model-building approach as suggested by Raudenbush and Bryk (2002) in the current study.

Step 1: a null model with no predictors were tested to estimate the interclass correlation of the outcome variables.

Model 0:

In this model, j represents different classes; i stands for different students; Y_ij corresponds to the academic achievement of student i in class j; β_0j is the mean of the students’ academic achievements in class j; r_ij is the difference between student i and the class mean; Var(r_ij) reflects the random error at the student/parent level; γ₀₀ is the mean of all students’ academic achievements; u_0j is the difference between the class mean and the all-student mean; andVar(u_0j) reflects the error of the class level.

Step 2: a random intercept model was tested with all level-1 predictors centered at the class mean and corresponding level-2 predictors by aggregating at the class-level to separate level-1 and level-2 effects (Hoffman, 2015, pp. 344).

Where, k represents the k^th level-1 predictors, k ∈ [1,10], meaning no random slopes were specified in this step.

Since SG was a categorical variable and the female as the reference group, the percentage of male students in each classroom was calculated as the level-2 predictor (SGCM). Similarly, the percentage of FES.C2 and FES.C3 at each classroom were computed as the level-2 predictors.

Step 3: a series of random slope models were tested to examine which level-1 coefficients varied significantly across classrooms via conducting a likelihood ratio test to compare models with and without the random slopes.

Where, k represents the k^th level-1 predictors, k ∈ [1,10],μ_kj indicates random slopes were specified in this step. Only the significant random slopes were kept to keep the model as parsimony as possible.

Step 4: class-level predictors were added to models to examine if these class-level predictors could explain the variance of random intercept.

Step 5: class-level predictors were added to models to examine if these class-level predictors could explain the variance of random slopes.

Where, γ₁₁₁ to γ₁₁₅ represented the class-level coefficients for examining the variance in the random slopes. Only the significant coefficients would be kept in the final model.

Table 2 shows the summary of the 8,426 students (parents), including results from both the child and class levels. Table 3 shows the descriptive statistics of the 185 classes.

For the parent-teacher relationship, the degree and form of parent-teacher contact are strongly associated with students’ academic achievement. Students with parents who participated in parental meetings and actively communicated with teachers had higher academic achievement, which a large number of studies have proved (e.g., Goyette and Xie, 1999; Adams and Christenson, 2000; Sebastian and Allensworth, 2012). The frequency of parent-teacher interaction could pass on educational expectations to their children, then contribute to their academic achievements.

The cooperation from parents to supervise their children to complete teachers’ assignments was an important way for parents to engage in their children’s schooling (Cook et al., 2018), which was supported by our findings as well. More specifically, results showed that the less the parents cooperated with teachers’ requirements, the lower students’ academic scores were.

Our results also showed that parental participation in parent-teacher meetings is also positively related to students’ academic achievements.

On the one hand, parental participation in parent-teacher meetings reflects the level of parent involvement in children’s learning to some extent. The positive relationship between parental involvement in their children’s education and students’ success in school is widely documented in the research literature (Fan and Chen, 2001; Barnard, 2004; Todd and Wolpin, 2007; Houtenville and Conway, 2008; Cheung and Pomerantz, 2011). On the other hand, evidence also suggests that some schools failed to fully engage parents and provide them with information about their children’s learning and how they are performing in school (Noel et al., 2013). Therefore, it is critical for parents to actively keep themselves informed about their children’s classroom activities, events, and requirements by communicating with teachers actively.

However, we found that parents reported teachers’ active communication with parents had negative relationships on students’ academic performance, which might be because, in the face-to-face communication between parents and teachers, teachers always have the clear aim of reflecting on the students’ misconduct and requesting that parents cooperate with regulations (Xiong, 2007; Zhang, 2011). If parents are informed of their children’s misconduct in China, they tend to criticize them verbally or even scold them since the children are supposed to obey the rules of parents. So that, telling parents about bad behavior is a punishment in Chinese published schools, which may aggravate students’ negative emotions, further affecting their academic performance. The relationship between teachers and parents in China is protagonist-supporting as a result of the current education system in China, which is oriented toward a screening function rather than a cultivation function (An, 2005), leading to strengthening the importance of exams and the status of teachers (An, 2005; Peterson et al., 2011). Although policymakers, administrators, and teachers have gradually realized the vital role of parental participation, parents’ involvement in children’s education is still in its initial stage in China (Wu et al., 2017).

The results of model 3 showed that teachers’ perception of respect from parents and the extent of teachers’ stress from parents’ requests were significantly associated with academic achievement. The higher levels of stress teachers perceived from parents’ requests, the lower academic achievement; meanwhile, the higher levels of respect teachers perceived from parents, the higher average academic performance. Similar to previous meta-analysis findings, high teachers’ stress level is linked to poor student outcomes, and the parent-teacher relationship is one of the main sources of teacher stress (Cameron and André, 2005). Also, having parents’ respect is a well-known protective factor for teacher retention (Ouyang and Paprock, 2006; Canter, 2009).

In addition, results reveal the random effects of the extent of parents’ afraid of communicating with teachers could be explained by the number of parents that teacher knew; the more parents the teacher knew, the lower level of fear that parents felt about communicating with teachers would become more positive. Together, the results echo that a good parent-teacher relationship link to good academic performance of students. However, the random effects also reveal the non-negligible variation across classrooms. Teachers can teach so many students that they are not familiar with all their children and their parents (Sizer, 1992; Meier, 2005), especially when there are many large classes (more than 56 students in a class) in China (Zhou and Xian, 2012; Zhang, 2013; Fu and Xu, 2018). Hu and Luo (2014) investigated large classes in two cities in Guangxi Province and found that large classes were more abundant in middle and high schools than in primary schools; the number of students in the largest class was 103. Therefore, some classes make it difficult for teachers to be familiar with each student’s parents in China. Yang (2006) found 45.2% of parents contacted teachers when their children fell behind, 19.4% of whom contacted the teacher when their children exhibited problem behavior, while 13.7% took the initiative to communicate with teachers only when the children had special conditions (such as sickness, incomprehension of fees, opinions on teaching methods, and corporal punishment). Thus, the current results also highlight the inequity of education resources, especially the human resources of teachers, in middle schools.