Longitudinal Relationships Between Parental Psychological Control Dimensions and Adolescent Developmental Outcomes in the Chinese Mainland: Random-Intercept Cross-Lagged Panel Models

Conceptualization of Multidimensional PC and Its Effects on Adolescent Outcomes in Chinese Contexts

Barber’s (1996) seminal work defined parental PC as a type of maladaptive parenting practice that “potentially inhibits or intrudes upon psychological development through manipulation and exploitation of the parent–child bond (e.g., love withdrawal and guilt induction), negative, affect-laden expressions and criticisms (e.g., disappointment and shame), and excessive personal control (e.g., possessiveness, protectiveness)” (p. 3297). Since then, this conceptualization has been widely adopted in PC studies. Despite the emphasis on PC’s complexity and multi-faceted nature that manifests in specific forms of parental controlling behavior, empirical studies typically measured parental PC as a unidimensional construct, using a holistic score derived from global measures, including the most frequently cited 8-item Psychological Control Scale–Youth Self-Report (PCS-YSR; Barber, 1996) and its translated and adapted versions (e.g., Bacikova-Sleskova et al., 2024; Rogers et al., 2020; Zhu & Shek, 2021).

According to self-determination theory (SDT), parental PC is by definition manipulative and coercive and thus violates the child’s three basic psychological needs, namely autonomy (behaving in freely chosen and volitional ways), competence (a sense of efficacy in completing tasks and achieving goals), and relatedness (a feeling of connection with important others), leading to low intrinsic motivation and self-worth, feelings of failure, maladaptive emotional and behavioral responses, and difficulties in adaption and coping (Ryan & Deci, 2017; Soenens & Vansteenkiste, 2010). Across Western cultures, the unidimensional approach links parental PC to disruptions in adolescent functioning, particularly their self-development, emotional states, psychosocial well-being, and behavioral adjustment (Chyung et al., 2022; Pinquart, 2017; Yan et al., 2020).

Unidimensional operations typically employ short scales (e.g., PCS-YSR and other adapted measures) consisting of no more than 10 items, which arguably tap common or even fundamental domains of parental PC. Nevertheless, they only cover a few selected aspects of parental PC (e.g., constraining verbal expressions and personal attacks) rather than its full conceptual range. For example, shaming and guilt induction, which are commonly used by Asian parents (Fung & Lau, 2012), are excluded in PCS-YSR. The narrow scope of items also hinders the effective factorial assessment of possible distinct dimensions of parental PC. Barber et al. (2012) commented that “there is ample need for refinement both in terms of conceptualization and measurement” (p. 274). They considered PC as not only manipulation and coercion, but also intrusion into the personal domain and disrespect, the latter of which was not fully captured by PCS-YSR. They further developed a new subscale, the “Psychological Control–Disrespect Scale,” to assess disrespect for individuality (e.g., unfavorable comparison with others and ignorance), which represents a psychometrically distinct dimension from that measured by PCS-YSR. Progress in refining the understanding and operation of PC also includes separating the dependency-oriented PC (i.e., parental attempts to make children psychologically and emotionally dependent on their parents) and the achievement-oriented PC (i.e., excessively high parental expectations for children’s achievement) as distinct dimensions (Soenens et al., 2010).

The initial unidimensional conception of PC and the aforementioned efforts to define multidimensional PC did not account for culturally distinct meanings and functions of parental PC. Choe et al. (2023) noted that intrusiveness and emotional manipulation are culturally sensitive concepts whose meanings may vary across cultures. For example, in China, where Confucian values (e.g., parental authority and the child’s filial piety) and collectivism (parent–child interdependence) shape parenting and parent–child interactions, parents are normatively more involved in managing their children’s lives to ensure their healthy adjustment and compliance with family obligations and social expectations (Ng & Wang, 2019). Thus, Chinese parenting is often characterized by heightened control and relationship-oriented practices (e.g., shared shame), which are considered coercive PC tactics in Western cultures, but are not necessarily indicative of parental negativity and rejection in China (Chen et al., 2016).

Grounded in the cultural concerns of parental PC, scholars have highlighted its nuanced distinctions in the meanings and functions of individual dimensions in Chinese contexts. Notably, researchers have distinguished Chinese parents’ relationship-oriented PC practices (e.g., guilt induction and shared shame) from harsh forms of PC (e.g., Fung & Lau, 2012; Rudy et al., 2014). On the one hand, harsh PC tactics, such as constraining verbal expression, invalidation, and personal attack that are typically measured by PCS-YSR, represent universally harmful practices that indicate parental hostility and rejection. According to the aforementioned SDT, the harsh PC is typically needs-frustrating, implicitly or explicitly forcing children to change their thoughts, feelings, and behaviors to conform to parental demands, thereby impeding their psychological needs, self-functioning, and healthy adjustment (Barber et al., 2012; Cheah et al., 2019).

On the other hand, Chinese parents’ PC often manifests through relational (e.g., shaming) and guilt-inducing practices (e.g., invoking familial sacrifice)—practices that emphasize parental involvement, emotional interdependence, and parent–child interconnectedness, which may be perceived as normative and care-oriented rather than overtly coercive (Chen-Bouck & Patterson, 2021). Unlike Western contexts, where relational tactics like guilt induction correlate strongly with parental harshness (e.g., parental rejection), Chinese relational induction coexists less consistently with harmful strategies, reflecting its embeddedness in filial piety and intergenerational obligations (Fung & Lau, 2012). From the perspective of SDT, while the relational PC also exerts pressure on children to change to comply with family obligations and social expectations, it does not inherently convey parental rejection or devaluation in collective Chinese cultural contexts (Li et al., 2004; Yu et al., 2015), where relational obligations are socially endorsed, thereby being less harmful to adolescents’ psychological needs and healthy development.

Related work also differentiated two distinct forms of shaming, a typical PC strategy used by Chinese parents: shared shame and social comparison shame (Fang et al., 2022; Helwig et al., 2014). Shared shame emphasizes family honor and collective responsibility, such as when parents say, “Your poor performance would damage the family’s honor,” linking a child’s behavior to the family’s reputation. Rooted in Confucian values, this practice is often seen as a way to teach respect for family duties and societal norms, and it may even foster empathy by raising awareness of how the child’s actions affect others, especially parents. In contrast, social comparison shame uses unfavorable peer comparisons (e.g., “My parents ask me why I cannot be as good as other children”) to criticize children and express parental dissatisfaction. While normalized in China’s culture of “face” (mianzi), this tactic often backfires—children interpret it as personal rejection, leading to resentment, anxiety, or damaged parent-child relationships (Barber et al., 2012; Fang et al., 2022). Repeated comparisons may make a child feel “not good enough,” fueling defiance or low self-appraisals. Using the SDT lens, unlike shared shame, social comparison shame is more likely to undermine children’s psychological needs of competence, autonomy, and relatedness and result in frustration and maladjustment by positioning their self-worth relative to external standards, pressuring self-improvement, and expressing parental dissatisfaction, particularly during adolescence, when sensitivity to social evaluation is heightened (Cao et al., 2025; Fang et al., 2022). The differences between the two forms of shaming highlight that even culturally embedded practices can have different meanings and influences.

Emerging empirical evidence supports the above theoretical differentiation of the individual dimensions of Chinese parents’ PC. For example, Fang et al. (2022) found that relationship-oriented PC (e.g., guilt induction) showed no significant links to adolescent depression, life satisfaction, or academic functioning over time. In contrast, parental social comparison shame and harsh PC consistently predicted worsening depression and reduced life satisfaction among adolescents, aligning with cross-cultural evidence of their harm (Barber et al., 2012). Notably, Zhu et al. (2023) empirically established a three-dimensional framework of Chinese parents’ PC, including harsh PC (e.g., love withdrawal, personal attack, criticisms, and ignorance), relational induction (e.g., shared shame and guilt induction), and social comparison shame (e.g., unfavorable comparison between the child and others), through rigorous procedures of item construction, exploratory factor analysis (EFA), and confirmatory factor analysis (CFA). Their work empirically confirmed social comparison shame as a unique dimension distinct from shared shame, reflecting China’s socio-cultural emphasis on moral guidance and familial obligations (Chao & Tseng, 2002). They also found that relational induction had adaptive or neutral predictive effects on youth psychological well-being, whereas harsh PC robustly predicted adolescent depression. Their validation of parental social comparison shame as a standalone dimension also linked it to lower adolescent life satisfaction, mirroring findings that peer comparisons and/or perceived shame erode youth self-worth (Hamwey & Whiteman, 2021). Nevertheless, other research noted that even culturally adaptive practices may carry risks in modern China, as rapid social development and individualistic values heighten adolescents’ concerns about autonomy, potentially reframing relational induction as coercive (J. Xu, Chen, et al., 2024).

In sum, the three-dimensional framework of PC offers a nuanced conceptual lens for understanding the culturally embedded, differential forms of parental PC practices and their unique associations with adolescent development outcomes in Chinese culture. While a growing body of research has advocated the multidimensional approach in PC studies (e.g., Fang et al., 2022; Xu et al., 2020; Zhang & Wang, 2024; Zhu et al., 2023), they primarily focused on psychological outcomes (e.g., life satisfaction and depression), leaving behavioral and academic effects underexplored or inconsistently reported, with some exceptions (e.g., Xu et al., 2020; Zhang & Wang, 2024). In addition, critical gaps persist in these studies. First, while studies overwhelmingly emphasize parental influences, they largely neglect child effects, although exceptions exist. For example, Xu et al. (2020) demonstrated that adolescents’ adaptive learning beliefs, strategies, and behaviors predicted a significant decrease in parental PC in one year, suggesting bidirectional dynamics. There is also a need to separate within-family (i.e., temporal fluctuations within families; “do increased parental PC predict poorer adolescent outcome later?” or vice versa) from between-family effects (i.e., differences in average levels of adolescent outcomes between families; “do adolescents in families with lower parental PC have better outcomes?”) when examining reciprocal relationships between PC and adolescent outcomes. Second, most studies focused on maternal PC or parental PC in general, with few separating the two parents. Relying on a single parental figure might fail to capture potentially distinct maternal and paternal effects. To advance understanding of parental PC’s role in adolescent development, investigations of bidirectional effects and separate maternal and paternal PC are necessary.

Bidirectional Relationship Between Parental PC and Adolescent Development

While recent meta-analytic evidence generally supports a positive link between parental PC and adolescents’ adverse developmental outcomes, the majority of the studies included are cross-sectional and often adopt a unidirectional framework, emphasizing parental influence on adolescent outcomes while neglecting reciprocity in parent-child dynamics (Bradshaw et al., 2025; Chyung et al., 2022; Salaam & Kyere, 2025). This risks overestimating parental effects and obscuring adolescents’ role in shaping caregiving. Correlational findings cannot confirm whether parental PC causes adolescent maladjustment or whether adolescents’ symptoms elicit stricter parental control. The depression-distortion hypothesis suggests that adolescents with mental health struggles may interpret parental behaviors more negatively, inflating their perceptions of parents’ PC (Laird & Frazer, 2019; Rudolph & Flynn, 2007). Some studies found that adolescents with higher initial internalizing problems and relational aggression later reported increased parental PC (Kaniušonytė & Žukauskienė, 2016; Yu et al., 2021; Zhu & Shek, 2020), suggesting child distress may provoke more perceived intrusive parenting. Another potential mechanism of adolescent effect is that parents may escalate control in response to their child’s observable or externalizing problem behaviors, such as delinquency or academic disengagement (Moilanen et al., 2009; Reitz et al., 2006; Zhu & Shek, 2020). These dynamics highlight a feedback loop: adolescents’ symptoms shape their perceptions of parenting, while parents adjust their strategies based on child behaviors. Without accounting for reciprocity through longitudinal designs, studies risk misattributing adolescent distress solely to parenting.

A handful of studies that have examined the reciprocal relationship between parental PC and different adolescent developmental outcomes show mixed findings, with unidirectional parent effects (Gao et al., 2022), unidirectional child effects (e.g., Otterpohl & Wild, 2015), bidirectional effects (e.g., Janssens et al., 2017; Lin et al., 2020), or insignificant effects (e.g., Otterpohl & Wild, 2015). On the one hand, most studies utilized traditional cross-lagged analyses that conflate within-family processes (temporal fluctuations) with between-family differences (stable traits or contextual factors). Separating these levels is crucial to clarify how changes in parental PC and adolescent outcomes interact over time.

To disentangle these effects, the present study integrated a Random-Intercept Cross-Lagged Panel (RI-CLPM) design. The RI-CLPM distinguishes between stable between-family differences (e.g., socioeconomic factors and broader cultural norms) and dynamic within-family processes (e.g., fluctuations in parenting strategies relative to a family’s baseline; Hamaker et al., 2015; Yang et al., 2025). Between-family differences reflect trait-level characteristics, such as enduring cultural or contextual influences, that shape how parental PC is enacted across households. Within-family effects isolate how deviations from a family’s typical parental PC levels predict subsequent changes in adolescent adjustment, offering stronger causal inference. Prior research conflating these levels obscures whether observed associations stem from cross-family comparisons or intra-family dynamics. For example, daily diary studies that differentiated these effects found that adolescent-driven processes (e.g., symptom-driven increases in parental PC) predominated within-family analyses but not between-family models (van der Kaap-Deeder et al., 2023; Xu & Zheng, 2023). However, their short-term designs (7-30 days) cannot elucidate whether daily interactions accumulate into enduring developmental trajectories. Our study extended prior work by examining bidirectional effects over a longer timescale (6-month intervals) and addressing gaps in understanding long-term parent-adolescent reciprocity.

Adolescent Developmental Outcomes in Different Domains

On the other hand, the aforementioned inconsistent findings regarding the bidirectional relationship between parental PC and adolescent outcomes also suggest that the direction and magnitude of the relationship may depend on the specific outcome measure, including psychological (internalizing/ill-being symptoms and well-being), behavioral (externalizing and prosocial behaviors), and academic measures. Previous studies have found that parental PC is particularly predictive of adolescents’ psychological problems, possibly arising from eroded internal volition and self-worth (Salaam & Kyere, 2025; Yan et al., 2020). Longitudinally, parental PC has been shown to prospectively predict increases in mental health issues (e.g., depression and anxiety) over intervals of six months to one year (Xie et al., 2025; Zhou et al., 2024). In contrast, whether adolescents’ internalizing symptoms predict subsequent increases in parental PC remains unclear, as studies are relatively scarce and the existing results are mixed. Using multi-wave data, Werner et al. (2016) found that adolescents’ depressive symptoms predicted higher levels of maternal PC one year later, and short-term increases in adolescents’ anxiety symptoms were also found to predict stronger parental PC over a two-week interval (Šutić et al., 2025). However, a five-wave longitudinal study among adolescents did not observe significant child-to-parent effects (Zhou et al., 2024). Given the inconclusive picture, more studies on the bidirectional relationship between parental PC and adolescent mental health issues are needed.

According to Bradshaw et al.’s (2025) meta-analysis on the relationship between parental PC and child development, studies on children’s well-being are far fewer than those on ill-being. They also found that while parental PC was robustly correlated with children’s ill-being to a moderate extent (e.g., depression and anxiety; r = 0.26), its association with well-being (e.g., life satisfaction) was smaller yet statistically significant (r = −0.18). Furthermore, the association between parental PC and adolescent well-being, such as life satisfaction, has been mainly examined in cross-sectional studies or as the parent-driven effect in some longitudinal studies, leaving the reciprocal relationship understudied (Bradshaw et al., 2025; Deng et al., 2024). With limited exceptions, Shek (2007) reported a significant bidirectional relationship between parental PC and adolescents’ well-being one year later. However, after controlling for between-person associations, Deng et al. (2024) only identified a stable cross-lagged effect from adolescent well-being to perceived parental PC. Nevertheless, Deng et al. (2024) employed a unidimensional operation of parental PC and did not differentiate between paternal and maternal PC.

Adolescents may display outward-directed behaviors as a result of unmet psychological needs due to parental PC. Previous studies have also suggested that parental PC is reliably associated with adolescents’ behavioral outcomes, such as a broad range of externalizing problems (e.g., antisocial behavior, conduct problems, aggression, and bullying) and prosocial behavior (Fu et al., 2022; Pinquart, 2017). Nevertheless, the lack of longitudinal studies investigating the reciprocity of the association is a limitation commonly noted across studies, including meta-analyses (Yan et al., 2020). Although longitudinal studies have also yielded significant parent-driven or child-driven prospective effects (Agalar et al., 2025; Basili et al., 2021; Gulseven et al., 2025), the limited number of cross-lagged analyses yielded mixed findings. For example, some studies identified significant bidirectional effects between parental PC and adolescents’ aggressive and rule-breaking behavior (Chen et al., 2020; He et al., 2019). However, other studies (e.g., Fu & Zhang, 2020; Otterpohl & Wild, 2015) did not observe significant cross-lagged effects between parental PC and adolescents’ prosocial and problem behaviors. Overall, it is possible that parental PC and adolescents’ behavioral adjustment may mutually reinforce one another over time. Given the insufficient findings, especially those that disentangle between-family and within-family effects, more studies are warranted.

In the academic domain, prior research generally supports a negative association between parental PC and adolescents’ academic functioning (e.g., engagement and motivation) and achievement, especially in the Western context (Fang et al., 2022; Pinquart, 2016). Despite the similar significant association in cross-sectional studies (Lu et al., 2017; Yau et al., 2022), longitudinal findings in Chinese samples are more mixed. Some studies have documented significant negative prospective effects of parental PC on subsequent academic functioning and achievement (Fang et al., 2022; L. Xu, He, et al., 2024), whereas other studies have reported null effects (Fu & Zhang, 2020; Wang et al., 2007). Less attention has been paid to potential child-driven effects in the academic domain and the bidirectional relationship. One exception is Xu et al.’s (2020) study, which found that parental PC significantly predicted an increase in students’ maladaptive academic functioning (but not a decrease in adaptive academic functioning) one year later, and students’ adaptive (but not maladaptive) academic functioning significantly predicted subsequent parental PC.

To sum up, the existing findings regarding the reciprocal relationship between parental PC and adolescent development across various outcome domains in Chinese contexts remain largely inconclusive. In particular, very few studies have considered the multidimensionality of parental PC, multiple outcome measures, and the distinction between within-family and between-family effects.

The Present Study

To address the gaps in understanding the multidimensionality, cultural specificity, and bidirectional dynamics of parental PC, the present study aimed to advance a comprehensive analysis of Chinese parents’ PC by (1) considering three culturally salient PC dimensions (relational induction, social comparison shame, and harsh control); (2) assessing developmental outcomes across psychological, behavioral, and academic domains; (3) evaluating bidirectionality in parent-child effects to disentangle whether parental PC predicts adolescent maladjustment or vice versa; and (4) differentiating within-family effects from between-family effects.

Based on the aforementioned elaborations, we first hypothesized that culturally distinct dimensions of parental PC—harsh control, relational induction, and social comparison shame—would have different effects on adolescent outcomes, with higher levels of harsh control exhibiting the strongest associations with poorer adolescent outcomes, followed by social comparison shame, and relational induction having weak or even nonsignificant associations. Second, such differentiated associations were expected to be observed at both between-family and within-family levels. Third, in addition to parent-driven effects, we expected child-driven effects (i.e., bidirectional dynamics), in which high adolescent maladjustment would prospectively predict increased parental PC across all dimensions. Finally, given the limited evidence directly comparing the links between parental PC and adolescent development across outcome measures in a single study, we sought to explore potential differences in the hypothesized bidirectional relationships across adolescent psychological, behavioral, and academic outcomes without making any specific expectations.

Participants and Procedures

This study was reviewed and approved by the Institutional Review Board at The Hong Kong Polytechnic University (Reference number: HSEARS20220427002 for the baseline survey and HSEARS20230516001 for the follow-ups).

The present study employed a purposive sampling strategy to recruit schools first, then students through the participating schools. We invited five public secondary schools in different cities in the Chinese mainland, where the corresponding author had previously conducted teacher-training workshops and established collaborative relationships. Three of them (two in Jiujiang, Jiangxi Province, East China, and one in Chongqing, Southwest China) provided consent to participate in the present study. With the school principals’ endorsement, a designated teacher at each school received online training on obtaining parental consent and collecting data. Teachers then randomly invited several classes from grades 7, 8, and 10 to join the study. Grades 9, 11, and 12 were excluded due to tighter study schedules. Headteachers of the invited classes also received online training before helping obtain parent consent and administering questionnaires. Parent consent was obtained via a letter (i.e., parental notice) delivered by schools. According to the schools’ feedback, all parents provided consent for their adolescent children’s participation. As such, a total of 1,309 students were invited, and all participated in the present study.

Student questionnaires were administered by headteachers in classrooms. Before each wave, students were well informed about the survey’s purpose, voluntary participation, confidentiality, and their right to withdraw at any time without negative consequences. They were also instructed to respond honestly based on their own perceptions without discussing questions with classmates. After signing the consent form, paper questionnaires were distributed to students, who were given sufficient time to complete them. Data were collected on three occasions (Time 1, Time 2, and Time 3) at six-month intervals. Two attention check questions (e.g., “Please choose ‘strongly agree’ for this question”) were included. Adolescents who failed these checks were excluded as invalid cases, a common method to ensure data quality (Zhou et al., 2024).

Among the 1,309 participating adolescents, some were absent from data collection due to sick leave, personal issues, other school activities, or transfers to another class or school (especially at Time 3). A total of 1,223 (response rate = 93.43%), 1,140 (response rate = 93.35%), and 1,166 adolescents (response rate = 89.08%) responded to the questionnaires at the three occasions, respectively. Among these responses, 1,164 (valid rate = 95.18%), 1,140 (valid rate = 93.29%), and 1,103 (valid rate = 94.60%) were valid (i.e., they answered the two attention check questions correctly).

Missing values across variables and time ranged between 11.23% and 17.80% among the 1,309 participants. Little’s Missing Completely at Random (MCAR) tests showed that missing data were missing completely at random at Time 2, but not at Time 1 and Time 3. To further investigate the impact of missingness, we compared participants who provided valid responses across all three waves (i.e., complete group, n = 895) with those who provided valid responses at only one or two waves (i.e., incomplete group, n = 414). T-tests or chi-square tests revealed no significant differences in age, perceived family socioeconomic status (1 = lowest level, 10 = highest level), and percentage of only-child participants between the complete and incomplete groups. There was a larger proportion of male students in the incomplete group (66.43%) than in the complete group (47.93%, χ² = 38.94, p < .001). Additional multivariate analysis of variance (MANOVA) revealed that the incomplete group reported higher paternal relational induction (F = 5.31, p = .02, η _p ² = .01), paternal harsh control (F = 5.50, p = .02, η _p ² = .01), and conduct problems (F = 5.71, p = .02, η _p ² = .01) compared to the complete group at baseline (Time 1). No significant differences were identified on other key measures at baseline.

Because gender was treated as a covariate and the effect sizes of differences in the three key variables were small, all 1,309 adolescents were included in the final analyses to maximize data use (Muthén & Muthén, 2019). The full information maximum likelihood (FIML) approach was adopted to handle missing values (see the Data Analysis section for more details; Hirose et al., 2016). Among the participants, 53.78% were boys, and 24.83% were only children. At Time 1, 47.14% were in Grade 7, 29.87% in Grade 8, and 22.99% in Grade 10, with a mean age of 13.73 years (SD = 1.17; age range = 11–16 years).

Measures

Measures included adolescent perceptions of parental PC and developmental outcomes, encompassing psychological (mental health issues and life satisfaction), behavioral (prosocial behavior and conduct problems), and academic indicators (school engagement). Results of confirmatory factor analyses (CFA) and related psychometric properties (e.g., average factor loading, composite reliability, average variance extracted) are presented in Table S1 in the Supplemental Materials. Measurement invariance tests across the three time points are outlined in Table S2.

Data Analysis

As all data were collected using self-report questionnaires, which may result in common method bias. Thus, Harman’s single-factor test was performed to assess this issue prior to data analysis. Results showed that a single factor accounted for 29.39%, 29.96%, and 38.40% of the variance at the three waves, respectively, which were below the recommended threshold of 50% (Kock, 2020; Ooi et al., 2023). The results suggested that the common method bias was not a major concern in the present study.

Data analysis was performed using IBM SPSS Statistics Version 29.0 (e.g., missing value analysis, descriptions, and correlations) and Mplus Version 8.5 (e.g., CFA and RI-CLPM). When performing CFA and testing hypothesized longitudinal associations between parental PC and adolescent outcomes, the robust maximum likelihood estimator (i.e., MLR) was employed to mitigate the potential influence of non-normality, and the full information maximum likelihood (FIML) approach was adopted to handle missing values. According to Hirose et al. (2016), the FIML method is reliable for up to 50% missing data, and it does not require missingness to be completely at random. The use of MLR with the FIML method was considered effective for the present study (11.23% to 17.80% missing across waves), as it could improve parameter-estimation robustness while also maintaining optimal statistical power by retaining all available data. This method has been recommended and widely adopted in previous longitudinal research, where data attribution is common (e.g., Newman, 2014; Zhang et al., 2025).

To examine the longitudinal associations between parental PC and adolescent outcomes, we tested RI-CLPMs, which reduce potential bias in traditional cross-lagged panel models (CLPM) by separating between-individual variance from within-individual dynamics. This analysis investigates unique lagged effects between variables while controlling for between-individual correlations, within-individual stabilities, and concurrent correlations. Each adolescent outcome was tested in a separate model. Model fits were evaluated by Chi-square (χ²), comparative fit index (CFI), and root mean square error of approximation (RMSEA). Model fit was acceptable when χ²/df ≤ 5, CFI ≥.90, and RMSEA ≤.08 (Kline, 2015). The covariates mentioned before were controlled in all the tested RI-CLPMs.

To simplify model estimation, constraints were added to RI-CLPMs. First, all cross-lagged paths and within-individual stabilities were freely estimated, and these were considered the basic models (Model 1). Second, constraints were added sequentially: all cross-lagged paths fixed to be time-invariant (Model 2), all stability paths fixed to be time-invariant (Model 3), and both cross-lagged and stability paths fixed to be time-invariant (Model 4). We conducted the Satorra-Bentler chi-square (S-B χ²) difference test to determine whether adding constraints would significantly worsen model fit. However, the S-B χ² difference test can be sensitive to sample size and may be too strict. Thus, we also examined changes in CFI (ΔCFI >.01) and RMSEA (ΔRMSEA >.015) as indicators of worse fit (van Laar & Braeken, 2022). If fixing certain paths to be time-invariant did not worsen the model fit, the constrained model was retained to test our hypotheses. This approach reduces the number of parameters, making the model more parsimonious and easier to interpret (Hihara et al., 2021). It has been adopted in previous research (e.g., Becht et al., 2017).

Between-Family Effects

At the between-family level, maternal and paternal relational induction were both positively associated with adolescents’ mental health issues (Figure 1), negatively with adolescents’ life satisfaction (Figure 2), and positively with adolescents’ conduct problems (Figure 4). No significant between-family correlations were found for parental relational induction with adolescent prosocial behavior (Figure 3) or school engagement (Figure 5).

Both maternal and paternal social comparison shame were positively associated with adolescent mental health issues (Figure 1), negatively with adolescent life satisfaction (Figure 2), and positively with adolescent conduct problems (Figure 4). Maternal social comparison shame was also negatively associated with adolescents’ prosocial behaviors (Figure 3) and school engagement (Figure 5), while paternal associations were non-significant.

Both parents’ harsh control was positively associated with adolescents’ mental health issues (Figure 1) and conduct problems (Figure 4), while negatively associated with adolescents’ life satisfaction (Figure 2). Maternal, but not paternal, harsh control was negatively associated with adolescents’ prosocial behavior (Figure 3) and school engagement (Figure 5).

Within-Family Lagged Effects

Differential Effects of Parental PC Dimensions

Adolescent Effects on Parental PC

The bidirectional associations between adolescent functioning and parental PC dimensions underscore the importance of considering adolescent-driven effects. For example, adolescent-reported mental health problems positively predicted their subsequent perceptions of both maternal and paternal relational induction. This aligns with the depression-distortion hypothesis (Rudolph & Flynn, 2007), wherein adolescents with heightened mental health struggles, such as anxiety or depressive symptoms, may interpret parental communication through a lens of negativity, amplifying their perception of guilt-based tactics. Similarly, lower adolescent-reported life satisfaction was associated with higher perceived maternal and paternal social comparison shame and harsh control over time. This suggests that adolescents who are more dissatisfied with life may be more likely to overinterpret parental comparisons as devaluation or disrespect, even when parents intend them as motivational. They may also consider parents utilizing too many manipulative or coercive behaviors. These patterns highlight how adolescent distress may distort interpretations of parental behavior, inflating perceived PC, consistent with previous findings that internalizing problems predict increased perceived intrusive parenting (Kaniušonytė & Žukauskienė, 2016; Yu et al., 2021; Zhu & Shek, 2020).

The predictive effect of adolescent outcomes on perceived PC may also reflect parental reactivity or adjustment of their control tactics based on adolescents’ observable behaviors. For example, adolescents’ school engagement negatively predicted maternal relational induction over time, suggesting that mothers may escalate guilt-inducing tactics when they notice a decline in their child’s academic effort. Parents appear to intensify PC in response to academic disengagement, reflecting cultural pressures to enforce compliance with educational expectations (Cheung & Pomerantz, 2012) and demonstrating how parents escalate control in reaction to externalizing problems (Moilanen et al., 2009; Reitz et al., 2006; Zhu & Shek, 2020). Such effects may be more accurately captured through the use of multiple reporters. Future research should address this potential perceptual gap by employing multi-informant designs to determine whether adolescent disengagement truly elicits parental reactions or reflects perceptual biases.

Strengths, Limitations, and Future Directions

This study has several key strengths. First, the multidimensional approach to parental PC allowed for nuanced investigation and comparisons of individual PC dimensions. Second, the longitudinal design enabled evaluation of bidirectionality in parent-child effects, disentangling whether parental PC predicts adolescent maladjustment or vice versa, clarifying transactional pathways. Third, the use of RI-CLPM to disentangle within-family fluctuations (e.g., time-specific changes in parenting tactics) from between-family differences (e.g., stable family-level norms) resolved analytical ambiguities. Fourth, the assessment of outcomes across psychological (e.g., mental health problems, life satisfaction), behavioral (e.g., conduct problems, prosocial behaviors), and academic (i.e., school engagement) domains allowed us to determine whether PC dimensions exert universal or domain-specific effects. Together, these strengths provide a nuanced, ecologically valid framework for understanding how psychological control operates across temporal and developmental dimensions.

Despite these strengths, several limitations merit attention. First, the study did not account for broader family dynamics (e.g., marital quality, coparenting interactions) or individual moderators (e.g., adolescent resilience, parental authoritarian traits) that may shape PC’s associations with developmental outcomes. Prior research highlights partner effects in PC: for instance, marital dissatisfaction is directly linked to more negative parenting and a poorer parent-child relationship (Wang et al., 2024). In addition, we did not distinguish between the perceived PC of biological parents and that of stepparents. Previous research also reported that children living with biological parents have different developmental outcomes from those in stepparent families (Nicholson et al., 2002). Future studies should investigate how coparenting dynamics (e.g., alignment or discord in control tactics) and family structure (e.g., biological parents versus stepparents) moderate the effect of PC.

Second, our reliance on adolescents’ self-reports of both parental PC and developmental outcomes is a limitation that may introduce bias. Previous studies generally found that child-perceived parental PC is higher than parents’ self-reported PC, and that the former shows a stronger association with child developmental outcomes than the latter (Hou et al., 2020; Salaam & Kyere, 2025). It can be argued that it is how adolescents perceive and experience parents’ socialization (e.g., PC) that shapes their development, and thus adolescents’ self-reports have been widely used as a valid data collection method in youth research and family studies (Cao et al., 2025; Salaam & Kyere, 2025). Nevertheless, including both adolescents and parents (e.g., mothers, fathers, or both) as informants can incorporate multiple perspectives within the family and offer a more balanced investigation. In addition, solely relying on adolescents’ self-reports may risk overlooking objective behavioral markers (e.g., teacher observations). Nuanced constructs like school engagement require finer-grained measurement to distinguish between behavioral compliance and emotional investment. Thus, future research could employ mixed-method designs (e.g., observational assessments and daily diaries) and include parent-reported measures, thereby providing a more comprehensive understanding of family dynamics and helping address potential perceptual gaps between parents and adolescents.

Third, although we interpreted our findings through culturally unique interpretations of different parental PC dimensions, especially relational induction and social comparison shame, we did not measure adolescents’ actual appraisals of these dimensions and thus could not test our explanations. Previous research reported that adolescents may differ in their ways of interpreting and coping with parental PC, and adolescents’ positive attributions of some parental PC tactics would mitigate their negative effects on adolescent development (Cheah et al., 2019; Smetana et al., 2021). It is possible that adolescents’ interpretations of parents’ intentions (benign and good for children versus negative and disrespectful) behind their social comparison shame contribute to the present complex and context-dependent findings on social comparison shame’s association with adolescent developmental outcomes. Built on our multidimensional operation of parental PC, future research will benefit from directly investigating how adolescents’ appraisals of these dimensions may moderate the link between parental PC dimensions and adolescent outcomes.

Fourth, while some of our findings support the argument that relational PC might be less harmful and even beneficial to children’s development in interdependent cultures, such as in China (Fung & Lau, 2012), other findings suggest that even culturally acceptable PC practices also carry risks to adolescent development in contemporary China (Su & Liu, 2021). One essential factor that may account for such variations in the association between parental PC and adolescent developmental outcome is adolescents’ endorsement of traditional interdependent cultural norms and cultural values. For example, a recent study reported a significant moderation effect of adolescents’ beliefs about reciprocal filial on the relationship between parental social comparison shame and adolescent meaning in life (Cao et al., 2025). Thus, with rapid development in China, a promising future research direction is within-culture heterogeneity in Chinese adolescents’ susceptibility to parental PC, given their differing endorsement of cultural values and family obligations. Investigation of cultural moderators will advance culturally informed models of parental PC and its developmental consequences.

Fifth, the present study involved only Chinese adolescents recruited from three public secondary schools, using a purposive sampling strategy. On the one hand, the present findings on the association between parental PC and adolescent outcomes may not reflect the dynamics among younger children, who lack the same protective factors as adolescents (e.g., peers and activities outside the home) and may thus be more vulnerable to maladaptive parenting practices, such as harsh PC. Meanwhile, a competing hypothesis is that younger children are more likely to depend on their parents and have lower needs for autonomy and independence than adolescents, and thus are less affected by controlling tactics. Given the different possibilities, further research will certainly benefit from involving more diverse samples (e.g., children and adolescents) and from comparing the links between parental PC and developmental outcomes across these samples. On the other hand, a sample from a limited number of schools may have insufficient representativeness, limiting the generalizability of the findings. For example, the present sample of adolescents demonstrated relatively low levels of mental health issues and conduct problems across waves. In addition, missing data across variables due to adolescents’ absences, non-responses, and invalid responses were not missing completely at random at all waves. Therefore, future research should aim to replicate this study with diverse populations (e.g., children, adolescents, and emerging adults across multiple Chinese communities or even in other cultures) and among adolescents who evidence a greater range of adjustment to further verify the present findings.