The relationship between parental control types and mental health types in Chinese adolescents

Abstract

This study investigated the relationship between parental control types and mental health categories among Chinese adolescents. About 2240 adolescents (1267 males; M _age = 14.09) were recruited and completed the Parental Control Scale, Patient Health Questionnaire‐9, Life Satisfaction Scale at two time points. Results revealed that adolescents' parental control could be classified into seven types, while mental health could be classified into three types. The key finding demonstrates significant dynamic interactions between these variables: at T1, the “high behavioral control‐low psychological control” parental control type (e.g., behavioural guidance type) significantly promoted adolescents' transition towards more optimal mental health categories; conversely, adolescents classified in the “complete mental health” category at T1 were more likely to have parents exhibiting the “high behavioural control‐low psychological control” positive parenting pattern at T2. This “virtuous cycle” pattern was confirmed, although the “vicious cycle” commonly observed in variable‐centred research between psychological control and poor mental health did not fully emerge in this study. These findings elucidate the complex bidirectional relationships between perceived parental control and mental health development among Chinese adolescents.

BACKGROUND

Adolescent mental health is critical for individual development and societal progress; however, global trends indicate a decline (Yu, 2022). Worldwide, an estimated 10%–20% of children and adolescents experience mental health challenges (Kieling et al., 2011), and one in seven individuals aged 10–19 is diagnosed with a mental disorder (WHO, 2024). The situation in China is particularly concerning. According to the National Mental Health Development Report of China (2019–2020), the depression detection rate among Chinese adolescents is 24.6% (Fu et al., 2021). A meta‐analysis further indicated that between 2010 and 2020, the prevalence of depression and anxiety among Chinese secondary school students ranged from 23.9% to 28.0% (Yu & Huang, 2023). Given these high prevalence rates, identifying antecedent variables that influence adolescent mental health is crucial for both preventing the onset and mitigating the severity of mental health issues.

Bronfenbrenner's (1986) ecological systems theory posits that the family, as the most immediate microsystem, exerts a core influence on individual development. Within this framework, parental control is a key family factor, typically categorized into two dimensions: psychological control and behavioural control (Barber et al., 1994). Psychological control involves intruding into a child's inner world and undermining autonomy, while behavioural control entails setting norms, rules, and monitoring activities. Extensive research demonstrates that parental psychological control is a significant predictor of adolescent mental health problems, such as depression and anxiety (e.g., Deng et al., 2022; Ye et al., 2025). In contrast, behavioural control is generally associated with higher life satisfaction and fewer psychological difficulties (e.g., Bacikova‐Sleskova et al., 2024; Gorostiaga et al., 2019; Leung & Shek, 2020; Shek et al., 2021).

Ecological systems theory further emphasizes bidirectional interactions between individuals and their environments (Bronfenbrenner & Morris, 2006). Adolescents are not passive recipients of parenting; their psychological characteristics can also influence parental behaviours. For example, adolescent depressive symptoms (Ye et al., 2025) and generalized anxiety (Wijsbroek et al., 2011) show positive correlations with subsequent parental psychological control, while externalizing problems negatively predict parental behavioural control (Pinquart, 2017). Empirical studies predominantly support the existence of bidirectional dynamics, characterized by “vicious cycles” for psychological control and “virtuous cycles” for behavioural control (e.g., Pinquart, 2017; Ye et al., 2025; Zhang et al., 2024; Zhou et al., 2024).

However, these variable‐centred approaches have inherent limitations. They examine the net effects of isolated variables, yet parents typically employ combinations of control strategies (Galambos et al., 2003), and adolescent mental health is a multifaceted construct comprising both positive and negative indicators (Suldo & Shaffer, 2008). Analysing these complex constructs as single variables may obscure their true combinatorial patterns and interactive mechanisms.

A person‐centred approach can overcome these limitations by identifying latent classes based on multiple indicators, thereby capturing holistic patterns of parental control and mental health (Molenaar & Campbell, 2009). Some studies have used latent profile analysis and traditional latent transition analysis (LTA) to explore the stability and transitions of these types (e.g., He et al., 2023). However, traditional LTA has a critical methodological flaw: it cannot effectively distinguish stable between‐individual differences from genuine within‐individual change, potentially overestimating type stability (Wen & Zhu, 2021). Such overestimation may distort our understanding of bidirectional dynamics. For instance, an apparently stable “vicious cycle” might reflect prolonged exposure to high‐risk environments (a between‐individual difference) rather than a genuine causal effect of psychological problems on parental control (a within‐individual change).

The emerging method of random intercept latent transition analysis (RI‐LTA) addresses this issue (Muthén & Asparouhov, 2022). RI‐LTA incorporates a random intercept to separate individual traits into a time‐invariant stable component (reflecting between‐individual differences) and a time‐varying latent class component (reflecting within‐individual transitions). This model yields “purified” transition probabilities, enabling a more accurate examination of whether a parental control type genuinely predicts subsequent transitions in mental health types after accounting for stable traits, and vice versa. Thus, RI‐LTA provides an ideal analytical tool for testing the dynamic, within‐individual processes implied by ecological systems theory.

In summary, although previous research provides a foundation, methodological constraints have limited our understanding of the complex, type‐based bidirectional dynamics between parental control and adolescent mental health. Therefore, this study integrates a person‐centred perspective with RI‐LTA methodology to achieve three key objectives within Chinese cultural contexts: (1) identify latent types of adolescent‐perceived parental control and adolescent mental health; (2) use RI‐LTA to characterize transition patterns between these types; and (3) rigorously test for bidirectional predictive relationships between parental control types and mental health types. This approach will provide more precise empirical evidence for understanding the interactive mechanisms between family environment and adolescent development.

Types and transitions of mental health in adolescents

The dual‐factor model (DFM) provides the theoretical foundation for classifying adolescent mental health by integrating indicators from both psychopathology and positive psychology (Greenspoon & Saklofske, 2001). This model posits that mental health status is determined by the independent levels of positive (e.g., well‐being) and negative (e.g., symptoms) indicators, resulting in four primary categories (Suldo & Shaffer, 2008): (1) Complete mental health: characterized by high‐positive and low‐negative indicators, reflecting optimal functioning and resilience; (2) Vulnerable: characterized by low‐positive and low‐negative indicators, indicating a lack of psychological resources and potential risk for future difficulties; (3) Symptomatic but content: characterized by high‐positive and high‐negative indicators, where individuals experience symptoms but maintain a sense of well‐being; and (4) Troubled: characterized by low‐positive and high‐negative indicators, suggesting significant impairment. The DFM thus emphasizes the necessity of assessing both well‐being and psychopathology for a comprehensive understanding of mental health.

Extensive research has applied the DFM to adolescent populations, supporting its cross‐cultural validity (e.g., Miao, 2023; Suldo et al., 2016; Wang, Zheng, et al., 2025; Wang, Zhu, et al., 2025). Although some studies identify classifications that deviate slightly from the four original types, their findings remain broadly consistent with the DFM framework. For instance, some researchers have identified a three‐group solution comprising “Flourishing” (high‐positive/low‐negative), “Troubled” (low‐positive/high‐negative), and “Vulnerable” (moderate‐positive/moderate‐negative) profiles (Shen, 2023; Zhou et al., 2020). Others have reported four types, such as complete mental health, moderately mentally healthy, symptomatic but content, and troubled (Moore et al., 2019), or even five types, including separate categories for youth with emotional symptoms versus conduct problems (Petersen et al., 2022). These variations underscore the model's robustness while highlighting potential nuances in different samples.

Longitudinal studies have further examined the stability of these mental health categories. Most evidence indicates that the complete mental health group is the most stable over time (DiLeo et al., 2022; Miao, 2023; Moore et al., 2019; Wang, Zheng, et al., 2025; Wang, Zhu, et al., 2025; Zhou et al., 2020), whereas the troubled group typically exhibits the lowest stability (Moore et al., 2019; Zhou et al., 2020). However, findings are not entirely consistent; some studies report greater stability for the troubled group compared with the vulnerable group (Miao, 2023), or even the highest stability for the troubled group (Shen, 2023). These discrepancies complicate the practical task of assessment and intervention for educators and clinicians.

A potential source of these inconsistent findings lies in the methodological limitations of traditional LTA, which is commonly used to estimate stability. Traditional LTA cannot distinguish stable between‐individual differences from genuine within‐individual change, which may artificially inflate estimates of category stability (Wen & Zhu, 2021). RI‐LTA addresses this limitation by incorporating a random intercept (f) into the model, effectively separating individual traits into a time‐invariant component (f, reflecting stable between‐individual differences) and a time‐varying latent class component (c, reflecting within‐individual transitions) (Muthén & Asparouhov, 2022). This dual‐layer structure isolates inherent stable traits, yielding “purified” transition probabilities that represent unbiased, within‐individual changes (Muthén & Asparouhov, 2022; Wen & Zhu, 2021). For the present study, RI‐LTA will therefore provide a more precise delineation of between‐individual differences and within‐individual changes in adolescent mental health types, leading to a more accurate understanding of their dynamics.

Types and transitions of parental control in adolescents

Only one study has employed latent profile analysis and traditional LTA to investigate typologies and transitions in adolescent‐perceived parental control (He et al., 2023). This study analysed data from 2110 seventh‐grade students across two time points, 1 year apart. The LPA indicators comprised three subdimensions of psychological control (guilt induction, love withdrawal, authoritarian assertion) and two subdimensions of behavioural control (solicitation, behavioural restriction). The analysis identified three consistent parental control typologies at both time points: a low‐moderate control group (low psychological, moderate behavioural control), a moderate–moderate control group (moderate levels of both), and a high–high control group (high levels of both). According to the traditional LTA results, the low‐moderate control group demonstrated the highest stability, followed by the high–high control group, while the moderate–moderate control group was the least stable.

However, this study has several limitations. First, its exclusive focus on seventh‐grade students restricts the generalizability of its findings across other adolescent developmental stages. Second, the use of traditional LTA may have led to an overestimation of latent class stability, as this method cannot disentangle stable between‐individual differences from genuine within‐individual change. Third, the study did not distinguish between paternal and maternal control behaviours. Family systems theory emphasizes that mothers and fathers often play distinct roles in child development (Bowen, 1966). Empirical evidence supports this distinction; for instance, a meta‐analysis found a stronger association between maternal psychological control and student psychological problems compared with paternal control (Xing et al., 2024). Another study reported that paternal psychological control, but not maternal control, had significant longitudinal associations with adolescent internalizing symptoms and aggression (Yu et al., 2021). Although findings are not entirely consistent, they suggest that paternal and maternal control may have differential impacts on adolescent outcomes.

To address these gaps, the current study will recruit participants from a broader adolescent age range and will analyse paternal and maternal control separately. Using LPA and the more advanced RI‐LTA, this research aims to identify typologies of parental control and investigate their transition patterns among Chinese adolescents.

The relationship between parental control and mental health

Research on parental control and adolescent mental health has delineated three primary explanatory models (Lollis & Kuczynski, 1997). The parent effect model posits a unidirectional influence from parent to child, whereas the child effect model emphasizes how children's characteristics shape parenting behaviours. In contrast, the reciprocal effects model underscores bidirectional, mutual influences between parents and children (Sameroff, 1975). Although evidence exists for the parent and child effect models (Deng et al., 2022; Wijsbroek et al., 2011), the preponderance of empirical support favours the reciprocal effects model (Fu & Zhang, 2020; He et al., 2019; Lin et al., 2020; Pinquart, 2017; Ren et al., 2023; Sun et al., 2018; Ye et al., 2025; Zhang et al., 2024), indicating that parental control and adolescent mental health exert mutual influences over time. However, this literature predominantly relies on a variable‐centred approach, which has a fundamental limitation: it reduces complex, multidimensional constructs into isolated variables. In reality, parents employ combinations of control strategies (Galambos et al., 2003), and adolescent mental health is a multifaceted construct comprising both positive and negative indicators (Suldo & Shaffer, 2008). The variable‐centred perspective may therefore obscure the holistic, configurational nature of these phenomena and their dynamic interactions.

Self‐determination theory (SDT) provides a useful framework for understanding these dynamics, positing that the satisfaction of three basic psychological needs—autonomy, competence, and relatedness—is essential for healthy development (Ryan & Deci, 2000). Autonomy involves a sense of volition and psychological freedom; competence refers to feeling effective in one's interactions; and relatedness entails experiencing warmth and connection with others. Parental psychological control, which intrudes upon a child's inner world, directly thwarts the need for autonomy (Grolnick et al., 1997), impedes opportunities for self‐directed mastery thereby undermining competence (Flamant et al., 2023), and can create relational insecurity that damages relatedness (Xie et al., 2024). Conversely, parental behavioural control, characterized by appropriate rule‐setting and monitoring, can strengthen the parent–child bond (Barber, 1996; Yin, Zhao, et al., 2023; Yin, Zhou, & Zhu, 2023), provide guidance that supports the development of competence and autonomy (Manzeske & Stright, 2009; Wang et al., 2007). Consequently, psychological control is consistently linked to unmet needs and poorer mental health (Deng et al., 2022; He et al., 2019; Pinquart, 2017; Yan et al., 2020; Ye et al., 2025), whereas behavioural control is generally associated with need satisfaction and better adjustment (Bacikova‐Sleskova et al., 2024; Pinquart, 2017; Shek et al., 2021).

The interplay between these control dimensions is crucial. Theoretically, a profile of high behavioural and low psychological control should be most beneficial. In contrast, high psychological control coupled with low behavioural control is likely detrimental. Even when both are present at moderate or high levels, outcomes may be concerning. For instance, He et al. (2023) found that a low‐moderate control profile was associated with higher well‐being, whereas a high–high profile was linked to poorer outcomes. Similarly, Leung and Shek (2020) demonstrated that the benefits of behavioural control were attenuated under conditions of high psychological control. This pattern suggests that the detrimental effects of psychological control may potentially override the beneficial effects of behavioural control.

Adolescents are not passive recipients of parenting but active agents who influence parental behaviours (Sameroff, 1975; Soenens et al., 2019). Parenting stress is a key mediator in this process (Grolnick, 2003). Parents who feel responsible for their child's upbringing (Chao, 1994) may experience heightened stress when adolescents exhibit poor mental health. This stress can make autonomy‐supportive parenting seem untenable, leading parents to resort to more coercive, psychologically controlling tactics. Conversely, adolescents with positive mental health are less likely to evoke such stress, allowing parents to feel more efficacious and employ more constructive, behaviourally controlling strategies. Substantial empirical evidence confirms that adolescent mental health significantly predicts subsequent levels of both parental psychological and behavioural control (He et al., 2019; Ren et al., 2023; Sun et al., 2018; Wijsbroek et al., 2011; Ye et al., 2025; Zhang et al., 2024).

To date, no study has examined how holistic types of adolescent mental health predict parental control typologies. Cross‐lagged research suggests that specific negative indicators (e.g., depression) can predict increased parental psychological control, while positive indicators (e.g., sense of meaning) may show non‐significant predictive effects (Ye et al., 2025). This indicates that negative mental health indicators might interfere with or overshadow the influence of positive indicators. Integrating these findings with the theoretical frameworks discussed, we hypothesize that adolescents classified under the complete mental health type will perceive a parental control profile characterized by higher behavioural than psychological control. Conversely, adolescents belonging to the vulnerable, symptomatic but content, or troubled types will perceive a profile where parental psychological control is either higher than or equal to behavioural control.

The current study

Adolescence is characterized by the development of self‐awareness and a drive for autonomy. Universalistic perspectives posit that fulfilling basic psychological needs is crucial for adolescents across cultures, and that parental psychological control typically has negative effects while behavioural control has positive effects (Soenens et al., 2012). This pattern has been confirmed in both Western and Chinese contexts (Wang et al., 2007; Wijsbroek et al., 2011; Yan et al., 2020). However, traditional Chinese cultural values, which emphasize hierarchical parent–child relationships and involve high parental expectations (Pomerantz & Wang, 2009), are associated with more prevalent controlling behaviours compared with Western parenting styles (Ng et al., 2014; Wang et al., 2007). Therefore, the specific nature of the relationship between perceived parental control and mental health among Chinese adolescents warrants detailed investigation.

Although variable‐centred studies have demonstrated interactive relationships between parental control and adolescent mental health (Fu & Zhang, 2020; He et al., 2019; Lin et al., 2020; Ren et al., 2023; Ye et al., 2025), the bidirectional dynamics between holistic types of parental control and mental health categories remain unexplored. This study will employ latent profile analysis and RI‐LTA to identify typologies of parental control and mental health, examine their transition patterns, and investigate their interrelationships. Given the scarcity of prior evidence on these configurational dynamics, we will not propose definitive directional hypotheses regarding their bidirectional relationships.

Some researchers have critiqued the composite indicators often used in the DFM, noting limitations such as insufficient assessment sensitivity and constrained practical feasibility (Hai et al., 2015). These scholars recommend reconstructing the model with simplified, robust indicators (Doll, 2008; Wang & Zhang, 2011). Previous research has successfully used depression and life satisfaction as proxies for negative and positive mental health, respectively, yielding satisfactory results (Hai et al., 2015; Wang, Zheng, et al., 2025; Wang, Zhu, et al., 2025). Consequently, this study will adopt depression and life satisfaction as primary mental health indicators. Depression is a highly prevalent disorder (Word Health Organization, 2023) and a leading contributor to the global disease burden (Vollset et al., 2024). Life satisfaction, as a stable cognitive component of subjective well‐being, serves as a widely used indicator of positive mental health (Antaramian, 2017). Furthermore, as RI‐LTA is a novel methodological advancement, its performance requires empirical validation (Wen & Zhu, 2021). This study will compare the model fit indices of traditional LTA and RI‐LTA to assess the accuracy and reliability of the RI‐LTA approach.

METHODS

Participants

This study employed a longitudinal design, collecting data from the same group of adolescent participants at two time points with a 6‐month interval. The first wave of data collection occurred in December 2023, yielding 2240 valid responses. Among these, 1267 were male (56.6%), 924 were female (41.3%), and 49 did not report their gender. Participants ranged in age from 12 to 18 years (M = 14.09, SD = 1.61). The second wave of data collection was completed in May 2024 with no participant attrition. Missing values across items ranged from 0.1% to 2.7%.

Results from independent samples t‐tests indicated no statistically significant differences between attrited and retained participants on the following measures: T1 depression (t = −0.28, p > .05), T1 life satisfaction (t = −0.02, p > .05), T1 father behavioural control (t = 1.89, p > .05), T2 depression (t = −0.35, p > .05), T2 life satisfaction (t = 0.38, p > .05), T2 father behavioural control (t = 0.86, p > .05), and T2 mother behavioural control (t = 1.15, p > .05). However, significant differences were found for T1 father psychological control (t = 3.44, p < .001), T1 mother psychological control (t = 3.55, p < .001), T1 mother behavioural control (t = 2.12, p < .05), T2 father psychological control (t = 2.23, p < .05), and T2 mother psychological control (t = 1.15, p < .01). Little's MCAR test indicated that the missing data pattern was not completely random (χ ² = 70,233.276, df = 61,287, p < .001). These results demonstrate that the missing data mechanism in this study does not meet the criteria for missing completely at random. Consequently, the study employed full information maximum likelihood (FIML) for subsequent data processing.

Prior to the survey, informed consent was obtained from teachers, parents, and students. The administration procedure involved homeroom teachers distributing paper questionnaires to students in their classrooms, with instructions to respond carefully to all items. Completed questionnaires were collected immediately after administration. All students were supervised by their homeroom teachers throughout the questionnaire completion process.

Measures

Parental control scale

Wang et al. (2007) adapted the parental control scale for the Chinese cultural context, dividing it into two subscales: parental behavioural control and parental psychological control. The scale uses a 5‐point Likert format, ranging from “completely disagree” to “completely agree.” The psychological control subscale consists of 18 items (e.g., “When I fail to meet my parents' expectations, they tell me I should feel guilty.”), while the behavioural control subscale includes 16 items (e.g., “My parents require me to…get their permission before going out again after returning home from school.”). To separately assess adolescents' perceptions of father and mother control levels, the authors modified all items by replacing “parents” with “father/mother” and “they” with “he/she.” The total mean score was used as the final measure, with higher scores indicating higher levels of perceived father/mother psychological or behavioural control. In this study, Cronbach's alpha for the subscales across the two time points ranged from .893 to .942.

Patient health Questionnaire‐9

This 9‐item self‐report tool was developed based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM‐IV) criteria for major depressive disorder (e.g., “Little interest or pleasure in doing things.”). In China, the PHQ‐9 has been validated as a reliable and effective screening tool for depression (Hu et al., 2014). The scale uses a 4‐point scoring system (0 = not at all, 3 = nearly every day), with the total mean score serving as the final measure. Higher scores indicate more severe depressive symptoms in adolescents. In this study, Cronbach's alpha for the PHQ‐9 at the two time points was .895 and .892, respectively.

Life satisfaction scale

Developed by Diener et al. (1985), this scale has been widely used among Chinese adolescents (Yin, Zhao, et al., 2023; Yin, Zhou, & Zhu, 2023; Zhou et al., 2023). It consists of five items (e.g., “In most ways, my life is close to my ideal.”) rated on a 7‐point Likert scale (1 = completely disagree, 7 = completely agree). The total mean score was used as the final measure, with higher scores indicating greater life satisfaction. In this study, Cronbach's alpha for the scale at the two time points was .862 and .860, respectively.

Statistical analyses

First, reliability tests, correlation analyses, descriptive statistics, and longitudinal measurement invariance tests were conducted using SPSS 25.0 and Mplus 8.11.

Second, LPA was performed separately for parental control and mental health at two time points using Mplus 8.11. Common indicators for LPA include: log likelihood (LL), Akaike information criterion (AIC), Bayesian information criterion (BIC), adjusted BIC (aBIC), entropy, Lo–Mendell–Rubin likelihood ratio test (LMR‐LRT), and bootstrap likelihood ratio test (BLRT). A better model fit is indicated by smaller absolute values of LL, AIC, BIC, and aBIC, higher entropy values, and LMR‐LRT and BLRT p‐values below .05 (Wang & Bi, 2018). However, additional criteria must be met: the estimated probability of latent class membership should not be below 3% (Ryoo et al., 2018), and the interpretability, theoretical relevance, and practical significance of the latent classes should be considered (Wen et al., 2023). Based on these criteria, the most appropriate latent class solution was selected.

Third, RI‐LTA was conducted separately for parental control and mental health without covariates. RI‐LTA examines within‐individual changes (T1 → T2) across time points at the measurement level, while between‐individual differences (i.e., random intercept f) are analysed at the individual level (Muthén & Asparouhov, 2022; Wen & Zhu, 2021). LTA was also performed to validate the RI‐LTA results.

Fourth, parental control types at T1 were dummy‐coded as predictors to examine their effects on transition probabilities of adolescent mental health from T1 to T2. The same approach was then applied to investigate the influence of T1 mental health types on transition probabilities of parental control types from T1 to T2.

Fifth, loglinear modelling with latent variables (Hagenaars, 1993) was used to examine (1) the association between T1 parental control types and T2 mental health types, and (2) the association between T1 mental health types and T2 parental control types. To reduce or eliminate bias from other sources when analysing latent categorical variables (Bakk & Vermunt, 2016), Bolck‐Croon‐Hagenaars (BCH) weights were applied. Additionally, pairwise comparisons of conditional probabilities across different predictor profiles were conducted.

RESULTS

Descriptive statistics

Descriptive statistics and Pearson correlation analysis results are presented in Table 1.

TABLE 1.

Descriptive statistics and correlation analysis (n = 2240).

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13
T1 Age	1
T1 Dep	.42**	1
T1 Ls	−.39**	−.58**	1
T1 Fa_Psy	.05*	.32**	−.28**	1
T1 Fa_Beh	−.17**	−.14**	.28**	.20**	1
T1 Mo_Psy	.07**	.35**	−.31**	.82**	.13**	1
T1 Mo_Beh	−.12**	−.06**	.19**	.19**	.73**	.25**	1
T2 Dep	.41**	.69**	−.52**	.27**	−.17**	.31**	−.09**	1
T2 Ls	−.35**	−.51**	.68**	−.28**	.21**	−.31**	.15**	−.58**	1
T2 Fa_Psy	−.01	.27**	−.24**	.64**	.13**	.57**	.12**	.30**	−.30**	1
T2 Fa_Beh	−.24**	−.20**	.30**	.04	.61**	−.02	.43**	−.23**	.30**	.17**	1
T2 Mo_Psy	−.003	.25**	−.25**	.56**	.06**	.65**	.14**	.31**	−.30**	.81**	.09**	1
T2 Mo_Beh	−.23**	−.16**	.27**	.04	.46**	.07**	.57**	−.17**	.28**	.13**	.75**	.19**	1
M	14.09	0.69	4.35	2.36	3.11	2.43	3.42	0.7	4.3	2.42	3.16	2.54	3.47
SD	1.61	0.6	1.37	0.88	0.89	0.92	0.83	0.59	1.32	0.92	0.88	0.95	0.81

Abbreviations: Dep, depression; Fa_Beh, Father‐Behavioural Control; Fa_Psy, Father‐Psychological Control; Ls, Life‐Satisfaction; Mo_Beh, Mother‐Behavioural Control; Mo_Psy, Mother‐Psychological Control; T1, Time 1; T2, Time 2.

*p < .05; **p < .01.

Longitudinal measurement invariance test

Table S2 presents the results of longitudinal measurement invariance tests for depression, life satisfaction, father psychological control, father behavioural control, mother psychological control, and mother behavioural control.

Types and transitions of parental control

Prior to conducting latent profile analysis, confirmatory factor analysis was performed on the four variables: father psychological control, father behavioural control, mother psychological control, and mother behavioural control. The results demonstrated good model fit indices at both time points (T1: χ ² = 4100.828, df = 276, CFI = 0.928, TLI = 0.915, SRMR = 0.054, RMSEA = 0.079 [0.077, 0.081]; T2: χ ² = 3786.100, df = 276, CFI = 0.938, TLI = 0.927, SRMR = 0.051, RMSEA = 0.075 [0.073, 0.077]). These satisfactory fit indices supported the subsequent implementation of latent profile analysis.

Starting with a two‐class model and adding one latent class at a time, we obtained six models (Table 2). Across both time points (T1/T2), absolute values of LL, AIC, BIC, and aBIC decreased with more classes. Both LMR‐LRT (except 6‐class at T2) and BLRT showed statistically significant differences (p < .05). At T1, the seven‐class model revealed a new profile (<3% probability), and showed the second‐highest entropy after the three‐class model. Considering statistical indicators and theoretical interpretability (Wang & Bi, 2018), we selected the seven‐class model as optimal. Similarly, at T2, the seven‐class model had the second‐highest entropy and performed best on all other indices.

TABLE 2.

Fit indices for LPA of parental control.

Classes	LL	AIC	BIC	aBIC	Entropy	LMR‐LRT(p)	BLRT(p)	The smallest group
Time 1
2	−10,642.56	21,311.12	21,385.40	21,344.10	0.77	<.001	<.001	39.90%
3	−10,228.17	20,492.35	20,595.20	20,538.02	0.82	<.001	<.001	13.90%
4	−9903.63	19,853.25	19,984.68	19,911.61	0.78	<.001	<.001	12.50%
5	−9771.69	19,599.38	19,759.38	19,670.42	0.76	<.05	<.001	11.10%
6	−9612.43	19,290.85	19,479.42	19,374.57	0.77	<.001	<.001	8.60%
7	−9508.37	19,092.73	19,309.87	19,189.14	0.78	<.01	<.001	2.70%
Time 2
2	−10,747.49	21,520.97	21,595.26	21,553.96	0.77	<.001	<.001	38.80%
3	−10,346.77	20,729.53	20,832.39	20,775.20	0.82	<.001	<.001	14.40%
4	−10,082.00	20,210.00	20,341.43	20,268.35	0.77	<.001	<.001	13.70%
5	−9898.59	19,853.19	20,013.18	19,924.22	0.80	<.01	<.001	5.20%
6	−9735.40	19,536.80	19,725.37	19,620.52	0.80	>.05	<.001	3.90%
7	−9602.22	19,280.43	19,497.57	19,376.84	0.80	<.001	<.001	3.40%

Note: Bold values indicate the selection of these models as optimal across all compared models.

The seven profiles at both time points differed only in parental control scores while maintaining consistent characteristics (Figure 1, Table S3). They were named: C1: “Behavioral Guidance Type” (moderate behavioural control, low psychological control); C2: “Structural Management Type” (high behavioural control, moderate psychological control); C3: “Balanced Supervision Type” (moderate both); C4: “Behavioural Manipulation Type” (high behavioural control, low psychological control); C5: “Permissive Type” (low both); C6: “Authoritarian Type” (high both); C7: “Emotional Manipulation Type” (low behavioural control, high psychological control).

FIGURE 1.

Latent profiles of parental control. Fa_Beh, Father‐Behavioural Control; Fa_Psy, Father‐Psychological Control; Mo_Beh, Mother‐Behavioural Control; Mo_Psy, Mother‐Psychological Control; T1, Time 1; T2, Time 2.

Subsequently, we examined the transition patterns of parental control types from T1 to T2 using both traditional LTA and RI‐LTA without covariates. Based on model fit indices (LTA: number of parameters = 84, Log Likelihood = −18,140.60, BIC = 36,929.19; RI‐LTA: number of parameters = 88, Log Likelihood = −16,961.85, BIC = 34,602.55), the results revealed significant differences between the two models (Δχ ² = 2357.5, Δdf = 4, p < .001). The smaller BIC value in RI‐LTA indicates its superior performance compared with traditional LTA.

The RI‐LTA results are detailed in Table 3. The diagonal elements of the transition matrix represent the probabilities of participants remaining in their original groups. Adolescents in the behavioural manipulation type (73%) and authoritarian type (69.9%) showed the highest stability, followed by structural management type (60.5%) and balanced supervision type (59.1%), with behavioural guidance type (52.9%) demonstrating moderate stability. The emotional manipulation type (46.6%) and permissive type (45%) exhibited the lowest stability.

TABLE 3.

Transition probabilities of parental control types.

Profiles		T2
		PC1	PC2	PC3	PC4	PC5	PC6	PC7
T1	PC1	.529	0	.17	.108	.09	.065	.037
	PC2	0	.605	.212	.094	.051	.015	.023
	PC3	.007	.202	.591	.007	.025	.155	.013
	PC4	0	.169	.067	.73	.013	.017	.003
	PC5	0	.27	.062	.183	.45	.035	0
	PC6	.011	.014	.2	0	.057	.699	.02
	PC7	.013	.139	.162	.067	0	.153	.466

Note: Bold values indicate transition probabilities between latent profiles.

Abbreviations: PC1, Behavioural Guidance Type; PC2, Structural Management Type; PC3, Balanced Supervision Type; PC4, Behavioural Manipulation Type; PC5, Permissive Type; PC6, Authoritarian Type; PC7, Emotional Manipulation Type; T1, Time 1; T2, Time 2. The same below.

Types and transitions of mental health

Prior to conducting latent profile analysis, confirmatory factor analyses were performed for depression and life satisfaction measures. The results indicated excellent model fit indices at both time points (T1: χ ² = 689.988, df = 76, CFI = 0.960, TLI = 0.952, SRMR = 0.031, RMSEA = 0.060 [0.056, 0.064]; T2: χ ² = 628.160, df = 76, CFI = 0.964, TLI = 0.956, SRMR = 0.030, RMSEA = 0.057 [0.053, 0.061]). These robust psychometric properties supported the appropriateness of proceeding with subsequent latent profile analysis.

We constructed latent class models starting from a two‐class baseline and sequentially adding classes, resulting in three models (Table 4). At both time points, LL, AIC, BIC, and aBIC values decreased with additional classes. The four‐class models at T1 and T2 had LMR‐LRT values exceeding 0.05, leading to the selection of the three‐class model as optimal for both periods. Using the same approach as parental control typology, we named mental health profiles based on characteristics (Figure 2 and Table S4): C1 (“Complete Mental Health”) featured low depression and high life satisfaction; C2 (“Moderate Mental Health”) showed moderate levels in both; C3 (‘Troubled’) demonstrated high depression and low life satisfaction.

TABLE 4.

Fit indices for LPA of mental health.

Classes	LL	AIC	BIC	aBIC	Entropy	LMR‐LRT(p)	BLRT(p)	The smallest group
Time 1
2	−41,402.27	82,890.55	83,136.26	82,999.64	0.90	<.001	<.001	46.30%
3	−39,687.68	79,491.37	79,822.79	79,638.52	0.90	<.001	<.001	14.80%
4	−39,181.28	78,508.56	78,925.7	78,693.77	0.87	>.05	<.001	6.90%
Time 2
2	−40,991.48	82,068.97	82,314.68	82,178.06	0.88	<.001	<.001	41.10%
3	−39,219.94	78,555.89	78,887.31	78,703.04	0.90	<.001	<.001	13.70%
4	−38,689.64	77,525.28	77,942.42	77,710.49	0.87	>.05	<.001	7.30%

Note: Bold values indicate the selection of these models as optimal across all compared models.

FIGURE 2.

Latent profiles of mental health. Dep, Depression; Ls, Life‐Satisfaction; T1, Time 1; T2, Time 2.

Similarly, we examined the transitions of mental health categories from T1 to T2 using both traditional LTA and RI‐LTA without covariates. Based on model fit indices (LTA: number of parameters = 78, Log Likelihood = −78,193.065, BIC = 156,987.84; RI‐LTA: number of parameters = 92, Log Likelihood = −75,441.38, BIC = 151,592.47), the results revealed significant differences between the two models (Δχ ² = 5503.37, Δdf = 14, p < .001). The RI‐LTA demonstrated a smaller BIC value, indicating its superiority over traditional LTA.

As detailed in Table 5, the RI‐LTA results showed that adolescents with complete mental health exhibited the highest stability (91.8%), followed by those with moderate mental health (68.2%), while the troubled group demonstrated the lowest stability (39.9%).

TABLE 5.

Transition probabilities of mental health types.

Profiles		T2
		MH1	MH2	MH3
T1	MH1	.918	.071	.011
	MH2	.229	.682	.09
	MH3	.189	.412	.399

Note: Bold values indicate transition probabilities between latent profiles.

Abbreviations: MH1, Complete Mental Health; MH2, Moderate Mental Health; MH3, Troubled; T1, Time 1; T2, Time 2. The same below.

The relationship between parental control types and mental health types

First, T1 adolescent parental control types (behavioural guidance type, structural management type, balanced supervision type, behavioural manipulation type, permissive type, authoritarian type, emotional manipulation type) were used as predictors; gender and age were included as covariates. These were decomposed into six dummy variables (M1–M6) with authoritarian type as the reference group. Three models were tested: Model 1 assessed direct effects on random intercept f; Model 2 examined direct effects on both random intercept f and mental health types at T1/T2; Model 3 extended Model 2 by adding predictor × T1 mental health type interactions for T2 mental health types.

Results indicated Model 2 was optimal: A significant difference emerged between Model 1 (parameters = 100, LL = −73,399.13, BIC = 147,567.47) and Model 2 (parameters = 132, LL = −73,264.60, BIC = 147,544.56; Δχ ² = 269.06, Δdf = 32, p < .001), while Model 3 (parameters = 180, LL = −73,246.63, BIC = 147,877.84) showed no significant improvement over Model 2 (Δχ ² = 35.94, Δdf = 48, p > .05). Model 2 demonstrated the lowest BIC value, confirming its direct effects on both random intercept f and mental health transitions.

In Model 2, Table 6 details parental control type effects on mental health transitions versus authoritarian type. Behavioural guidance type increased transitions from T1 troubled to T2 complete/moderate mental health but decreased transitions from T1 complete/moderate to T2 troubled. Structural management type, and behavioural manipulation type all showed increased transitions from T1 troubled to T2 complete mental health and decreased shifts from T1 complete to T2 troubled.

TABLE 6.

Effects of T1 parental control types on mental health type transition probabilities.

Predictors	T2
		MH1	MH2	MH3
T1 behavioural guidance type
T1	MH1	1.000 [1.000, 1.000]	0.746 [0.409, 1.360]	0.220 [0.088, 0.547]
	MH2	1.340 [0.735, 2.442]	1.000 [1.000, 1.000]	0.294 [0.119, 0.730]
	MH3	4.555 [1.830, 11.340]	3.399 [1.370, 8.435]	1.000 [1.000, 1.000]
T1 structural management type
T1	MH1	1.000 [1.000, 1.000]	0.731 [0.394, 1.354]	0.422 [0.182, 0.977]
	MH2	1.368 [0.738, 2.536]	1.000 [1.000, 1.000]	0.578 [0.267, 1.251]
	MH3	2.369 [1.023, 5.483]	1.731 [0.799, 3.749]	1.000 [1.000, 1.000]
T1 balanced supervision type
T1	MH1	1.000 [1.000, 1.000]	0.664 [0.372, 1.187]	0.484 [0.230, 1.022]
	MH2	1.505 [0.842, 2.691]	1.000 [1.000, 1.000]	0.729 [0.368, 1.446]
	MH3	2.064 [0.979, 4.353]	1.371 [0.692, 2.719]	1.000 [1.000, 1.000]
T1 behavioural manipulation type
T1	MH1	1.000 [1.000, 1.000]	0.483 [0.222, 1.051]	0.233 [0.066, 0.828]
	MH2	2.071 [0.952, 4.504]	1.000 [1.000, 1.000]	0.483 [0.137, 1.708]
	MH3	4.287 [1.207, 15.220]	2.070 [0.585, 7.322]	1.000 [1.000, 1.000]
T1 permissive type
T1	MH1	1.000 [1.000, 1.000]	0.798 [0.344, 1.849]	0.566 [0.202, 1.586]
	MH2	1.253 [0.541, 2.905]	1.000 [1.000, 1.000]	0.710 [0.274, 1.837]
	MH3	1.766 [0.630, 4.946]	1.409 [0.544, 3.646]	1.000 [1.000, 1.000]
T1 emotional manipulation type
T1	MH1	1.000 [1.000, 1.000]	1.883 [0.652, 5.442]	2.283 [0.709, 7.350]
	MH2	0.531 [0.184, 1.535]	1.000 [1.000, 1.000]	1.213 [0.478, 3.080]
	MH3	0.438 [0.136, 1.410]	0.825 [0.325, 2.094]	1.000 [1.000, 1.000]

Note: Bold numbers indicate significant p‐values, with 95% CI in parentheses.

Table 7 presents the conditional probabilities of adolescents in each T1 parental control type transitioning to T2 mental health categories, where each row reflects the probabilities for a specific control type. For example, adolescents with PC1 (behavioural guidance type) had a 42.2% probability for MH1, 52.6% for MH2, and 5.3% for MH3 at T2.

TABLE 7.

Probabilities from T1 parental control profiles to T2 mental health profiles.

T1 parental control	T2 mental health
	MH1	MH2	MH3
PC1	.422	.526	.053
PC2	.405	.499	.095
PC3	.181	.615	.204
PC4	.856	.111	.033
PC5	.479	.384	.138
PC6	.222	.495	.282
PC7	.043	.444	.514
Difference Test	PC4 > PC1, PC2, PC5 > PC3, PC6 > PC7	PC3 > PC1, PC2 > PC5 > PC4 PC3 > PC6, PC7 > PC4	PC7 > PC3, PC6 > PC1, PC2 PC7 > PC6 > PC5 > PC1, PC4 PC3 > PC2 > PC4

In summary, the results demonstrate that: (1) Adolescents with T1 parental control classified as PC1 (behavioural guidance type) were most likely to exhibit MH1 (complete mental health, 42.2%) or MH2 (moderate mental health, 52.6%) at T2; (2) Those with T1 parental control as PC2 (structural management type) predominantly showed MH1 (complete mental health, 40.5%) or MH2 (moderate mental health, 49.9%) at T2; (3) Adolescents with T1 parental control as PC3 (balanced supervision type) were most frequently categorized as MH2 (moderate mental health, 61.5%) at T2; (4) The PC4 (behavioural manipulation type) group at T1 was most likely to demonstrate MH1 (complete mental health, 85.6%) at T2; (5) For the PC5 (permissive type) at T1, the predominant outcomes were MH1 (complete mental health, 47.9%) and MH2 (moderate mental health, 38.4%) at T2; (6) Adolescents with T1 parental control as PC6 (authoritarian type) were most commonly classified as MH2 (moderate mental health, 49.5%) at T2; (7) The PC7 (emotional manipulation type) group at T1 was most likely to exhibit MH2 (moderate mental health, 44.4%) or MH3 (troubled, 51.4%) at T2.

The study utilized T1 adolescent mental health categories (complete, moderate, troubled) to create dummy variables M1 (complete mental health = 1) and M2 (moderate mental health = 1) with troubled as the reference group. Similarly, gender and age were included as covariates. Three models were compared: Model 1 (parameters = 92, Log Likelihood = −16,467.77, BIC = 33,643.21) showed significantly worse fit than Model 2 (parameters = 140, Log Likelihood = −16,247.75, BIC = 33,572.39) based on chi‐square difference (Δχ ² = 440.04, Δdf = 48, p < .001), but Model 2 and Model 3 (parameters = 308, Log Likelihood = −16,160.41, BIC = 34,689.98) showed no significant difference (Δχ ² = 174.68, Δdf = 168, p > .05). Model 2 was selected as optimal due to its lowest BIC, indicating that predictors directly influenced both the random intercept and parental control type transitions between T1 and T2.

In Model 2, Table 8 presents the effects of complete mental health and moderate mental health on transition probabilities of adolescent parental control types. Compared with adolescents classified as troubled, those with complete mental health showed: (1) Decreased probabilities of transitioning from T1 balanced supervision type to T2 behavioural guidance type; (2) Increased probabilities of transitioning from T1 behavioural guidance type to T2 balanced supervision type; (3) Decreased probabilities of transitioning from T1 balanced supervision type, permissive type, and emotional manipulation type to T2 authoritarian type; (4) Increased probabilities of transitioning from T1 authoritarian type to T2 balanced supervision type, permissive type, and emotional manipulation type.

TABLE 8.

Effects of T1 mental health types on parental control type transition probabilities.

Predictors		T2
		PC1	PC2	PC3	PC4	PC5	PC6	PC7
T1 complete mental health
T1	PC1	1.0 [1.0, 1.0]	0.6 [0.1, 3.8]	2.9 [1.2, 6.7]	1.5 [0.4, 5.8]	2.7 [0.7, 9.9]	0.5 [0.3, 1.0]	1.8 [1.0, 3.2]
	PC2	1.7 [0.3, 10.7]	1.0 [1.0, 1.0]	4.8 [0.6, 35.6]	2.6 [0.3, 23.5]	4.5 [0.6, 35.2]	0.9 [0.1, 5.8]	3.1 [0.4, 20.8]
	PC3	0.4 [0.2, 0.8]	0.2 [0.0, 1.6]	1.0 [1.0, 1.0]	0.5 [0.1, 2.2]	0.9 [0.2, 4.7]	0.2 [0.1, 0.5]	0.6 [0.3, 1.4]
	PC4	0.7 [0.2, 2.5]	0.4 [0.0, 3.6]	1.9 [0.4, 7.7]	1.0 [1.0, 1.0]	1.8 [0.3, 10.3]	0.3 [0.1, 1.4]	1.2 [0.3, 4.3]
	PC5	0.4 [0.1, 1.4]	0.2 [0.0, 1.7]	1.1 [0.2, 5.3]	0.6 [0.1, 3.4]	1.0 [1.0, 1.0]	0.2 [0.1, 0.8]	0.7 [0.2, 2.7]
	PC6	1.9 [1.0, 3.5]	1.1 [0.2, 7.5]	5.4 [2.1, 13.8]	2.9 [0.7, 11.6]	5.1 [1.3, 19.6]	1.0 [1.0, 1.0]	3.5 [1.7, 7.0]
	PC7	0.5 [0.3, 1.0]	0.3 [0.0, 2.2]	1.6 [0.7, 3.5]	0.8 [0.2, 3.1]	1.5 [0.4, 5.8]	0.3 [0.1, 0.6]	1.0 [1.0, 1.0]
T1 moderate mental health
T1	PC1	1.0 [1.0, 1.0]	0.2 [0.0, 1.3]	1.1 [0.5, 2.4]	0.9 [0.3, 2.6]	1.3 [0.5, 3.6]	0.4 [0.3, 0.7]	1.4 [0.9, 2.3]
	PC2	5.7 [0.8, 42.2]	1.0 [1.0, 1.0]	6.4 [0.8, 54.0]	5.0 [0.6, 43.9]	7.5 [0.9, 59.8]	2.4 [0.3, 17.9]	8.1 [1.0, 62.1]
	PC3	0.9 [0.4, 1.9]	0.2 [0.0, 1.3]	1.0 [1.0, 1.0]	0.8 [0.2, 2.6]	1.2 [0.3, 4.5]	0.4 [0.2, 0.9]	1.3 [0.6, 2.6]
	PC4	1.1 [0.4, 3.3]	0.2 [0.0, 1.7]	1.3 [0.4, 4.2]	1.0 [1.0, 1.0]	1.5 [0.4, 6.0]	0.5 [0.2, 1.4]	1.6 [0.6, 4.7]
	PC5	0.8 [0.3, 2.1]	0.1 [0.0, 1.1]	0.9 [0.2, 3.2]	0.7 [0.2, 2.7]	1.0 [1.0, 1.0]	0.3 [0.1, 0.9]	1.1 [0.4, 3.2]
	PC6	2.4 [1.4, 3.9]	0.4 [0.1, 3.1]	2.6 [1.2, 6.1]	2.1 [0.7, 6.0]	3.1 [1.1, 8.6]	1.0 [1.0, 1.0]	3.4 [1.9, 5.9]
	PC7	0.7 [0.4, 1.1]	0.1 [0.0, 1.0]	0.8 [0.4, 1.6]	0.6 [0.2, 1.8]	0.9 [0.3, 2.7]	0.3 [0.2, 0.5]	1.0 [1.0, 1.0]

Note: Bold numbers indicate significant p‐values, with 95% CI in parentheses.

Compared with adolescents belonging to the troubled type, those with moderate mental health showed the following patterns in parental control type transitions: (1) Increased probabilities of shifting from the authoritarian type at T1 to the behavioural guidance type, balanced supervision type, permissive type, and emotional manipulation type at T2; (2) Decreased probabilities of transitioning from the behavioural guidance type, balanced supervision type, permissive type, and emotional manipulation type at T1 to the authoritarian type at T2.

To examine the relationship between adolescents' T1 mental health categories and T2 parental control types, Table 9 presents the conditional probabilities of adolescents with each T1 mental health category belonging to each T2 parental control type.

TABLE 9.

Probabilities from T1 mental health profiles to T2 parental control profiles.

T1 mental health	T2 parental control
	PC1	PC2	PC3	PC4	PC5	PC6	PC7
MH1	.294	.18	.14	.23	.065	.077	.013
MH2	.3	.114	.367	.02	.106	.07	.024
MH3	.095	.075	.404	.018	.108	.181	.119
Difference test	MH1, MH2 > MH3	MH1 > MH2, MH3	MH2, MH3 > MH1	MH1 > MH2, MH3	MH2 > MH1	MH3 > MH1, MH2	MH3 > MH1, MH2

The results indicate: (1) Adolescents with T1 MH1 (complete mental health) were more likely to have T2 parental control types PC1 (behavioural guidance type, 29.4%) and PC4 (behavioural manipulation type, 23%); (2) Adolescents with T1 MH2 (moderate mental health) were more likely to have T2 parental control types PC3 (balanced supervision type, 36.7%) and PC1 (behavioural guidance type, 30%); (3) Adolescents with T1 MH3 (troubled) were most likely to have T2 parental control type PC3 (balanced supervision type, 40.4%).

Supplementary analyses

To address potential data artefacts, supplementary analyses employed traditional cross‐lagged models (as random intercept cross‐lagged panel models require at least three time points). Controlling for T1 gender, age, and educational stage, eight cross‐lagged models showed good fit. Autoregressive paths were significant across all models.

Cross‐lagged path coefficients revealed positive reciprocal predictions between father psychological control and depression, but negative reciprocal predictions with life satisfaction. Father behavioural control exhibited negative reciprocal predictions with depression, while T1 life satisfaction positively predicted T2 father behavioural control. T1 mother psychological control positively predicted T2 depression, with mother psychological control and life satisfaction showing negative reciprocal predictions. Mother behavioural control had negative reciprocal predictions with depression, and T1 life satisfaction positively predicted T2 mother behavioural control (see Table S5 and Table 10).

TABLE 10.

Presents cross‐lagged path coefficients for all models.

Models	β	SE	95% CI
M1
Father‐Psychological Control → Depression	0.07***	0.02	[0.038, 0.101]
Depression → Father‐Psychological Control	0.07***	0.02	[0.034, 0.102]
M2
Father‐Psychological Control → Life‐Satisfaction	−0.11***	0.02	[−0.137, −0.075]
Life‐Satisfaction → Father‐Psychological Control	−0.07***	0.02	[−0.102, −0.036]
M3
Father‐Behavioural Control → Depression	−0.06***	0.02	[−0.091, −0.03]
Depression → Father‐Behavioural Control	−0.07***	0.02	[−0.103, −0.03]
M4
Father‐Behavioural Control → Life‐Satisfaction	0.02	0.02	[−0.012, 0.052]
Life‐Satisfaction → Father‐Behavioural Control	0.1***	0.02	[0.059, 0.133]
M5
Mother‐Psychological Control → Depression	0.1***	0.02	[0.045, 0.086]
Depression → Mother‐Psychological Control	0.03	0.02	[−0.009, 0.099]
M6
Mother‐Psychological Control → Life‐Satisfaction	−0.12***	0.02	[−0.147, −0.084]
Life‐Satisfaction → Mother‐Psychological Control	−0.06**	0.02	[−0.089, −0.023]
M7
Mother‐Behavioural Control → Depression	−0.04**	0.02	[−0.073, −0.013]
Depression → Mother‐Behavioural Control	−0.07**	0.02	[−0.105, −0.028]
M8
Mother‐Behavioural Control → Life‐Satisfaction	0.02	0.02	[−0.012, 0.05]
Life‐Satisfaction → Mother‐Behavioural Control	0.12***	0.02	[0.077, 0.153]

**p < .01; ***p < .001.

DISCUSSION

Existing variable‐centred research has established bidirectional relationships between parental control dimensions—psychological and behavioural—and specific indicators of adolescent mental health (Fu & Zhang, 2020; He et al., 2019; Lin et al., 2020; Pinquart, 2017; Ren et al., 2023; Sun et al., 2018; Ye et al., 2025; Zhang et al., 2024). These dynamics are often characterized as a “vicious cycle” involving high psychological control and poorer mental health, and a “virtuous cycle” involving high behavioural control and more positive outcomes. However, because parents typically employ combinations of control strategies rather than relying on a single type (Galambos et al., 2003), a variable‐centred approach may obscure the holistic nature of these interactions. To address this limitation, the present study adopts a person‐centred approach. Using two‐wave longitudinal data and RI‐LTA, we will examine the dynamic relationships between typologies of perceived parental control and classifications of mental health among Chinese adolescents.

Types and transitions of adolescent parental control

This study provides the first person‐centred analysis to integrate both paternal and maternal psychological and behavioural control. The analysis identified seven distinct typologies of parental control as perceived by Chinese adolescents: Behavioural Guidance, Structural Management, Balanced Supervision, Behavioural Manipulation, Permissive, Authoritarian, and Emotional Manipulation.

Several key patterns emerged. First, adolescents reported comparable levels of psychological and behavioural control from fathers and mothers. This finding aligns with previous research (Tavassolie et al., 2016), suggesting that adolescents tend to perceive their parents' controlling styles consistently. Second, an examination of type prevalence showed that the Behavioural Guidance and Balanced Supervision types were most common, whereas the Emotional Manipulation type—a category not previously identified (He et al., 2023)—was the least prevalent. This distribution indicates that parents seldom rely exclusively on high psychological control; instead, they typically combine it with behavioural control. This supports the view that parenting practices constitute composite profiles rather than singular dimensions (Galambos et al., 2003).

Finally, the stability of these typologies can be interpreted through their functional characteristics. High behavioural control establishes clear, external rules and predictable environments, creating a sustainable framework for management. In contrast, while psychological control also regulates behaviour, adolescents' growing need for autonomy may lead them to perceive high psychological control as hostile (Barber & Harmon, 2002), potentially provoking resistance and prompting parental adjustment. This explains why the three most stable typologies were characterized by high behavioural control, with the greatest stability observed when psychological control was minimal. Conversely, the Emotional Manipulation and Permissive types demonstrated the lowest stability, likely due to their minimal behavioural control, which can undermine disciplinary effectiveness and fail to provide behavioural structure. When combined with low psychological control (as in the Permissive type), this absence of both control dimensions reflects inadequate parental monitoring (Logan‐Greene & Semanchin Jones, 2015), a situation that may exacerbate adolescent maladjustment and subsequently trigger changes in parenting strategies.

Types and transitions of adolescent mental health

The analysis identified three distinct mental health profiles among Chinese adolescents: complete mental health, moderate mental health, and troubled. This three‐profile solution partially deviates from the standard four‐profile structure of the DFM, but such deviations are consistent with prior research. For example, Moore et al. (2019) also identified four profiles, including a “moderately mentally healthy” category instead of a “vulnerable” one, and Petersen et al. (2022) derived five profiles by subdividing the “symptomatic but content” category. The current results, which identified a moderate mental health profile but lacked distinct vulnerable and symptomatic but content profiles, align with findings from other studies of Chinese adolescents (Shen, 2023; Zhou et al., 2020). Shen (2023) attributed their similar three‐profile result to their sample of left‐behind children, whereas the present study and Zhou et al. (2020) involved general adolescent populations. Zhou et al. (2020) proposed that cultural differences or the omission of specific behavioural indicators might explain the divergence from the Western model. However, it is noteworthy that Miao (2023), also studying Chinese adolescents with the same indicators (depression and life satisfaction), successfully replicated the standard four‐profile model.

After considering potential confounds related to sample characteristics and culture, the current findings suggest that the symptomatic but content and vulnerable profiles may represent transient or unstable states within the mental health trajectories of Chinese adolescents. The symptomatic but content profile is defined by co‐occurring high symptoms and high‐positive indicators (Suldo & Shaffer, 2008). This combination may be inherently transitional because: (1) significant psychological symptoms often prompt intervention from parents or teachers, altering the adolescent's situation; and (2) compared with the troubled profile, these individuals possess positive psychological resources that may facilitate recovery or change. Consequently, maintaining a stable equilibrium of high symptoms and high well‐being over extended periods may be difficult. Similarly, the vulnerable profile—characterized by low symptoms and low‐positive indicators—may also be transient. Although currently asymptomatic, the lack of protective psychological resources makes these adolescents highly susceptible to stressors, increasing their likelihood of transitioning to other profiles.

RI‐LTA indicated that adolescents in the complete mental health profile exhibited the highest stability (91.8%), whereas those in the troubled profile showed the lowest stability (39.9%). These results are consistent with most previous studies using traditional LTA (DiLeo et al., 2022; Miao, 2023; Moore et al., 2019; Wang, Zheng, et al., 2025; Wang, Zhu, et al., 2025; Zhou et al., 2020). Adolescents with complete mental health benefit from an absence of significant symptoms and strong positive resources that aid in self‐regulation. In contrast, troubled adolescents experience pronounced psychological distress and a deficit of positive resources, a conspicuous state of maladjustment that often elicits external support, facilitating change. The moderate mental health profile demonstrated intermediate stability, possibly because their mixed signals—moderate symptoms alongside moderate‐positive states – may result in less consistent or targeted intervention from their environment.

The impact of parental control types on mental health types

The analysis indicated that relative to the authoritarian parenting type, most other parental control types – with the exception of balanced supervision, permissive, and emotional manipulation—significantly increased the probability of adolescents transitioning from poor to good mental health while decreasing the likelihood of transition from good to poor. A key commonality among these beneficial parenting styles was that both parents exhibited higher levels of behavioural control than psychological control. This finding reinforces the established protective role of behavioural control in fostering positive adolescent mental health development (Shek et al., 2021) and suggests that parents should prioritize behavioural control strategies, such as clear rule‐setting and monitoring, in their child‐rearing practices.

Specific predictive relationships were observed between T1 parental control types and T2 adolescent mental health profiles. Adolescents in the complete mental health group at T2 were more likely to have parents characterized by the behavioural manipulation type at T1. Those in the moderate mental health group at T2 tended to have parents with a balanced supervision style at T1, whereas adolescents in the troubled group at T2 were more likely to have parents exhibiting the emotional manipulation type at T1. These results align with previous research (He et al., 2023) and consolidate evidence on the positive impact of behavioural control and the detrimental effects of psychological control (Bacikova‐Sleskova et al., 2024; Deng et al., 2022; He et al., 2019; Pinquart, 2017; Shek et al., 2021; Yan et al., 2020; Ye et al., 2025).

When the level of psychological control was held constant, increasing behavioural control (e.g., moving from the permissive type to the behavioural guidance or behavioural manipulation types, or from balanced supervision to structural management) progressively reduced the likelihood of adolescents belonging to the troubled group. However, a significant increase in the probability of complete mental health was only observed at the highest levels of behavioural control. This suggests that while behavioural control helps reduce psychological problems, only robust behavioural control effectively promotes optimal mental health. Conversely, when behavioural control was held constant, increasing psychological control (e.g., from behavioural manipulation to structural management to authoritarian, or from behavioural guidance to balanced supervision) gradually decreased the chances of complete mental health and increased the probability of a troubled classification. Unlike behavioural control, even moderate levels of psychological control exerted a negative influence on positive mental health development.

Comparing profiles with equivalent levels of both controls revealed distinct patterns. The permissive type (low behavioural and low psychological control) resulted in a small proportion of adolescents in the troubled group, potentially due to inadequate parental monitoring (Logan‐Greene & Semanchin Jones, 2015), though the moderate mental health profile was most common. In contrast, simultaneously high levels of both controls (authoritarian type) were associated with worse mental health outcomes, underscoring the particularly harmful effect of psychological control. Some scholars suggest that high levels of both controls may induce feelings of suffocation (Caron et al., 2006), often overshadowing any protective role of behavioural control. Our findings indicate that under the authoritarian type, compared with the balanced supervision type, probabilities decreased for moderate mental health while increasing for both the moderate and troubled groups. This pattern partially supports our hypothesis that only high behavioural control can mitigate the negative effects of psychological control, as this buffering effect was not observed when comparing the permissive and balanced supervision types. The concurrent increase in probabilities for both moderate and troubled mental health may reflect individual differences in how adolescents respond to specific combinations of parental control.

The impact of mental health types on parental control types

The analysis revealed that adolescents with complete or moderate mental health were more likely than those in the troubled group to influence a shift in their parents from high‐level control (authoritarian type) towards lower‐level control. Conversely, the probability of parents transitioning from low‐level to higher‐level control was significantly lower for adolescents in the complete or moderate mental health groups. This pattern supports the premise that adolescent mental health reciprocally influences parenting styles (Sameroff, 1975; Soenens et al., 2019). Adolescents with more favourable mental health likely possess stronger psychological resources and may engage in more effective communication with their parents, either addressing concerns about current parenting practices or expressing preferences for alternative approaches.

Specific longitudinal predictions were observed. Adolescents in the complete mental health group at T1 were more likely to have parents exhibiting the behavioural guidance or behavioural manipulation types at T2. Those in the moderate mental health group at T1 tended to have parents who demonstrated the behavioural guidance or balanced supervision types at T2. In contrast, adolescents in the troubled group at T1 were most associated with parents displaying the balanced supervision type at T2. Favourable adolescent mental health may reinforce parental trust in their child's autonomy through positive feedback, thereby encouraging parenting styles characterized by behavioural guidance rather than psychological intrusion. Conversely, adolescent psychological difficulties can increase parenting stress (Grolnick, 2003), potentially leading to heightened psychological control. However, parental uncertainty in this situation may result in a default “moderate control” approach. This explains why troubled adolescents at T1 had a lower probability of having parents in the authoritarian or emotional manipulation types at T2 compared with the balanced supervision type, yet still had a higher probability than adolescents in the complete or moderate mental health groups. Adolescents with moderate mental health represent an intermediate case; their mild distress might elicit some psychological control, while their comparatively lower positive attributes (relative to the complete mental health group) may lead parents to perceive them as less capable of handling intensive behavioural supervision.

Results from the traditional cross‐lagged panel model revealed distinct dynamics for paternal and maternal influences. Maternal psychological control significantly predicted subsequent adolescent depression, but adolescent depression did not predict maternal psychological control, supporting a parent effect model. Neither paternal nor maternal behavioural control significantly predicted adolescent life satisfaction; however, adolescent life satisfaction positively predicted both parents' subsequent behavioural control, consistent with a child effect model. Bidirectional relationships, aligning with a reciprocal effects model, were found between (1) paternal/maternal behavioural control and adolescent depression, and (2) paternal/maternal psychological control and adolescent life satisfaction. These findings may reflect the analytical precision gained by distinguishing between paternal and maternal influences, revealing both differential and convergent effects on adolescent mental health. Alternatively, the variable‐centred methodology might mask population heterogeneity in family interaction patterns, potentially creating discrepancies between model outcomes and actual family dynamics. These results caution against overgeneralizing conclusions from variable‐centred models and underscore the necessity of accounting for heterogeneity within family systems.

Limitations and further research

This study has several limitations. First, all data were collected via adolescent self‐reports, which may create discrepancies between perceived and actual parental behaviors. A meta‐analysis indicates that the association between psychological control and adolescent emotion regulation is stronger when assessed through adolescent or observer reports compared with parent reports (Beliveau et al., 2023). Future studies should incorporate parental self‐reports to validate the findings and examine informant discrepancies. Second, the study included only two waves of data collection over a six‐month period, which may be insufficient to capture meaningful long‐term changes in parenting or mental health. Longitudinal designs with more assessment points and a longer timeframe are needed to trace developmental trajectories. Third, the exclusive focus on Chinese adolescents limits cross‐cultural generalizability. Although some parenting effects may be universal (Soenens et al., 2012), Chinese parents often display more controlling behaviors due to cultural norms (Ng et al., 2014; Wang et al., 2007). Consequently, the findings may be most applicable to similar cultural contexts, and cross‐cultural replications are warranted. Fourth, although depression and life satisfaction are key mental health indicators, their exclusive use omits other critical dimensions such as anxiety and externalizing problems. This narrow focus may have constrained the comprehensiveness of the identified mental health profiles. Future research should include a broader range of indicators to construct more robust and inclusive typologies.

CONCLUSION

Adopting a person‐centred approach, this study examined the reciprocal relationships between typologies of perceived parental control and mental health categories among Chinese adolescents, thereby addressing limitations of previous variable‐centred research. The findings align with existing variable‐centred studies (Pinquart, 2017; Ren et al., 2023; Sun et al., 2018; Ye et al., 2025; Zhang et al., 2024; Zhou et al., 2024) in supporting the reciprocal effects model (Sameroff, 1975), demonstrating dynamic interactions between configurations of parental control and adolescent mental health profiles.

The results corroborate the established negative effects of psychological control and positive impacts of behavioural control (Shek et al., 2021). Adolescents who perceive high behavioural control coupled with low psychological control experience parenting characterized by discipline, rule‐setting, and supervision – behaviours often interpreted as care (Yin, Zhao, et al., 2023; Yin, Zhou, & Zhu, 2023) that promote better mental health. These adolescents tend to exhibit greater autonomy, emotional stability, and social adaptability. Their positive adjustment signals effective parenting to parents, thereby reinforcing the use of behavioural control strategies (e.g., monitoring) and reducing reliance on psychological control due to the perceived competence of the child.

Conversely, adolescents who perceive high parental psychological control and low behavioural control experience frustration of their basic psychological needs, which contributes to poorer mental health outcomes (Nigela et al., 2015). Although adolescent maladjustment can increase parenting stress and potentially elicit more psychological control (Grolnick, 2003), parents ultimately aim to support their children's development. Upon recognizing that excessive psychological control is ineffective or harmful, parents may adjust their strategies (Deng et al., 2022). Thus, while variable‐centred approaches often describe a “vicious cycle” between psychological control and poor mental health, this person‐centred analysis suggests a more nuanced reality. Parental adjustment may occur because persistent psychological control stems from uncertainty about effective alternatives, which is consistent with findings showing no significant prediction from adolescent mental health to subsequent parental psychological control (Deng et al., 2022).

In summary, parenting styles emphasizing high behavioural control exhibit positive bidirectional relationships with favourable adolescent mental health, constituting a “virtuous cycle.” Although styles dominated by high psychological control are associated with poorer mental health, adolescent distress in this context did not predict increased psychological control. Instead, it predicted a shift towards moderate levels of both control dimensions. These findings indicate that the classic “vicious cycle” identified in variable‐centred analyses was not observed here. However, it is important to note that the two‐wave, 6‐month longitudinal design may have been insufficient to capture the full cyclical nature of these processes and could potentially lead to attenuated effects.

AUTHOR CONTRIBUTIONS

Xu Wang: Conceptualization; methodology; formal analysis; visualization; writing – original draft; writing – review and editing. Ni Zhu: Data curation; investigation; project administration. Xiao Yu: Data curation; project administration; investigation. Mingchen Wei: Formal analysis; writing – review and editing. Shuai Chen: Formal analysis; writing – review and editing. Weijun Liu: Formal analysis; writing – review and editing. Yanling Liu: Conceptualization; investigation; funding acquisition; supervision; resources; writing – review and editing.

FUNDING INFORMATION

This work was supported by the Chongqing Municipal Education Science Planning Project under Grant [number K23YB2020021].

CONFLICT OF INTEREST STATEMENT

The authors declared no potential conflicts of interest.

ETHICAL APPROVAL

All procedures performed in studies involving human participants followed the institutional and/or national research committee's ethical standards and the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the Ethics Committee of Faculty of Psychology, Southwest University [IRB NO. H24178].

PATIENT CONSENT STATEMENT

Before the survey began, informed consent was obtained from teachers, parents, and students.

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.