Authors: Meng Bai, Xiaoqiong Li, Xueqi Yang
Categories: Research, Harsh parenting, Parental alienation, Core self-evaluation, Non-suicidal self-injury, Longitudinal study, Latent profile analysis
Source: BMC Psychology
Authors: Meng Bai, Xiaoqiong Li, Xueqi Yang
This study explored how distinct profiles of harsh parenting are associated with non-suicidal self-injury and whether parental alienation and core self-evaluation statistically explain these associations in a serial pathway. A total of 5,742 college students were recruited through convenience sampling at three waves with three-month intervals. To capture heterogeneity and probe potential mechanisms, we applied latent profile analysis and profile-specific serial mediation analysis. Three parenting profiles were consistently high harsh parenting (Class 1), medium harsh parenting (Class 2), and low harsh parenting (Class 3). Across all profiles, harsh parenting showed indirect associations with non-suicidal self-injury via parental alienation and core self-evaluation, though the magnitude and specific indirect pathways differed by profile. These findings may inform tailored intervention efforts by identifying profile-specific risk patterns linked to non-suicidal self-injury among young adults.
The online version contains supplementary material available at 10.1186/s40359-026-04155-8.
Non-suicidal self-injury (NSSI) refers to the deliberate infliction of harm on one’s own body without suicidal intent, primarily as a means to temporarily relieve psychological distress or stress. The most common form of NSSI is cutting, and other forms include burning, biting, and hitting oneself [1]. In recent years, the prevalence of NSSI has been rising, with approximately 22% of adolescents worldwide reporting engaging in such behaviors [2]. In China, the reported prevalence is particularly high (25%), significantly exceeding the rates found in the United States, Germany, and Scotland [3, 4]. Moreover, NSSI has become increasingly common among university students [5], with a second peak in onset occurring between the ages of 20 and 24 [6]. Beyond causing physical harm, NSSI is strongly associated with various mental health problems, such as bipolar disorder [7] and depression [8], and, more critically, it substantially increases the risk of suicide [9]. In the context of Chinese university students, NSSI is regarded as one of the strongest clinical predictors of future suicide attempts [10]. Given its high prevalence and severe consequences, it is essential to investigate the contributing factors to NSSI in this population to develop effective prevention and intervention strategies.
Harsh parenting refers to a range of coercive and punitive behaviors, including physical aggression (e.g., slapping or hitting) and verbal aggression (e.g., shouting or insulting), often used by caregivers in response to perceived child misbehavior [11, 12]. In China, two studies revealed that 78.1% of parents had applied psychological pressure to their children, and nearly 50% of parents used multiple forms of harsh parenting [13, 14]. In Chinese culture, there is a deeply rooted belief, reflected in sayings such as “spare the rod, spoil the child” and “the finest diamond must be cut”, that verbal abuse and physical punishment are sometimes considered necessary to ensure a child’s success and may be interpreted as expressions of care and love [15]. These beliefs may help explain why harsh parenting remains prevalent in China. However, most existing research focuses on adolescents, with limited studies directly examining the relationship between harsh parenting and NSSI behaviors among young adults, particularly within the Chinese cultural setting. As a prevalent risk factor in family interactions, harsh parenting may be associated with NSSI through multiple psychological mechanisms, potentially by influencing the quality of the parent-child relationship and individual self-cognition [16]. Based on the above, we propose that harsh parenting is associated with NSSI (Hypothesis 1).
Parental alienation refers to a decrease in a child’s attachment to their parents, accompanied by increased feelings of anger, mistrust, and social withdrawal [17]. Several studies suggest that parental alienation may play a mediating role in the association between harsh parenting and NSSI. For example, in a cross-sectional sample of 1,638 Chinese adolescents, harsh parenting was associated with parental alienation, and parental alienation was associated with poorer psychosocial outcomes, including a higher likelihood of NSSI [18]. Similarly, a longitudinal study of 786 Chinese adolescents suggested that harsh parenting may be associated with decreases in emotional belongingness, which may in turn relate to later NSSI [19]. These findings align with the cognitive-emotional model of NSSI, which conceptualizes NSSI as a multifaceted behavior frequently used to regulate intense or distressing emotions [16]. In the context of harsh parenting, caregivers may be less responsive to children’s emotional needs and may constrain the development of autonomy [20, 21]. Under such conditions, some children may resort to NSSI not only to downregulate overwhelming affect but also to communicate distress and elicit attention or responses from others, given the high salience and “cost” of the behavior [20]. Accordingly, during periods of parental alienation, NSSI may function as a maladaptive strategy through which individuals attempt to obtain parental attention or support. Based on the above, we propose that parental alienation mediates the relationship between harsh parenting and NSSI (Hypothesis 2).
Core self-evaluation refers to an individual’s fundamental assessment of their abilities and value, encompassing traits such as self-esteem, general self-efficacy, locus of control, and emotional stability [21, 22]. The negative consequences of harsh parenting, such as social anxiety [23], emotional dysregulation [24], depression [25], and low self-esteem [26], may be associated with NSSI as an unhealthy self-regulation mechanism. Two studies focused on Chinese middle school students and adults aged 18–23, respectively, have reported that harsh parenting was significantly and negatively associated with core self-evaluation [27, 28]. According to the cognitive-emotional model of NSSI, cognitive factors, particularly maladaptive self-schemas (e.g., negative self-beliefs and self-criticism), may constitute an important vulnerability associated with NSSI [16]. In addition to affect regulation, NSSI may serve self-focused intrapersonal functions by redirecting attention inward, facilitating the management of distressing emotions, and, in some cases, enacting self-punishment [29]. More recently, a cross-lagged study focusing on adolescents found that both previous and current core self-evaluation were significantly and negatively associated with NSSI [30]. Based on these findings, this study hypothesizes that core self-evaluation mediates the relationship between harsh parenting and NSSI (Hypothesis 3).
Parental alienation may be associated with an individual’s core self-evaluation. Family interactions play a crucial role in shaping self-perception, as individuals often use how they are treated by family members as a benchmark for self-assessment [31]. Positive parent-child attachment has been linked to stronger self-awareness and higher core self-evaluation [32], whereas parental alienation may be linked to negative emotional experiences within the family, which may hinder the development of a clear self-concept and relate to lower core self-evaluation [33]. The cognitive-emotional model of NSSI highlights that cognitive vulnerabilities, particularly self-related cognitions, are central to the onset and maintenance of NSSI [16]. Negative core self-evaluation may be associated with stronger self-criticism and negative self-views, reduced perceived capacity to cope with distress, and greater reliance on maladaptive cognitive strategies such as rumination [34]. In emotionally volatile family contexts, these cognitive tendencies may be further intensified, making expectancies about NSSI outcomes more salient (e.g., believing NSSI will effectively reduce distress, function as self-punishment, or help communicate unmet needs). Based on these insights, we propose the fourth hypothesis of this parental alienation and core self-evaluation may serially mediate the relationship between harsh parenting and NSSI (Hypothesis 4).
Latent Profile Analysis (LPA) is an exploratory statistical technique designed to uncover unobserved subgroups within a dataset, thereby capturing heterogeneity in individual’s characteristics across multiple dimensions [35]. Applied in the context of family research, LPA enables scholars to classify individuals according to their experiences of harsh parenting and to identify distinct parenting patterns [36]. These profiles can then be examined in relation to NSSI. Empirical evidence indicates that individuals exposed to more severe forms of harsh parenting have an increased likelihood of engaging in NSSI, whereas those experiencing relatively less severe forms report fewer instances of such behaviors [37]. As a person-centered approach, LPA not only advances our understanding of theoretical mechanisms by highlighting heterogeneity in the effects of harsh parenting but also facilitates the examination of subgroup-specific mediating processes, offering deeper insights into how harsh parenting shapes emotional regulation and self-perception, which may contributing to the development of NSSI. Importantly, the findings derived from this approach may help clinicians with insights that may inform more targeted and effective interventions for youth affected by harsh parenting and NSSI.
Despite China’s cultural context of harsh parenting and its relatively high incidence of NSSI, most research on the relationship between harsh parenting and NSSI has been conducted in Western contexts [38, 39]. There is still limited research exploring the connection and underlying mechanisms between harsh parenting and NSSI in the Chinese context. Based on the cognitive-emotional model of NSSI [16], the present longitudinal study examines the relationship between harsh parenting and NSSI among college students and investigates the mediating roles of parental alienation and core self-evaluation. Additionally, LPA will be employed to identify distinct subgroups within the sample based individuals’ experiences of harsh parenting and their respective NSSI behaviors. This approach will allow us to explore the heterogeneity within the population and identify specific profiles that may be more vulnerable to developing NSSI. Figure 1 illustrates the proposed model for this research.
Fig. 1Diagram of the hypothesized chain mediation mode
Data were collected through convenience sampling at a university in Zhengzhou, Henan Province, with three waves of data collection, spaced three months apart. A total of 12,643 participants completed the first survey, 12,471 participated in the second, and 9,473 took part in the third survey. During data cleaning, responses with short completion times, inconsistent patterns, or problematic demographic information were excluded [40, 41]. After cleaning, 9,471 valid responses were obtained in the T1 phase, 9,473 in the T2 phase, and 7,674 in the T3 phase. A total of 5,742 participants completed all three measurements, including 1,467 males and 4,275 females, with an average age of 20.26 (SD = 1.19). This study was approved by the Human Research Ethics Committee of [Anonymized] and the relevant school boards, principals, and teachers (Approval No. [Anonymized]). Written informed consent was obtained from the participants during the online survey. All participants were fully informed of their rights, including voluntary participation and the option to withdraw from the study at any time. Strict confidentiality protocols were implemented to protect the participants’ privacy.
Harsh parenting was assessed by the harsh parenting scale [11, 42]. It comprises two subscales, maternal harsh parenting, and paternal harsh parenting. Each subscale consists of 4 items (e.g., My mother/father got mad at me, and even yelled at me), rated on a five-point Likert-type scale ranging from 1 (never) to 5 (always). Mean scores were computed, and higher scores reflected higher levels of perceived harsh parenting. A Confirmatory Factor Analysis (CFA) supported the one-factor structure of the scale, with the model showing adequate model fit within the study (χ²(24) = 6.128, p <.001, RMSEA = 0.029, CFI = 0.999, TLI = 0.996, and SRMR = 0.006). In this study, Cronbach’ s α of teh total scare was 0.93.
Building on the original Inventory of Parent and Peer Attachment (IPPA) by Armsden and Greenberg [17, 43], a shortened version comprising 10 items was developed, with three assessing trust, three assessing communication, and four assessing alienation. In this study, the alienation subscale was used to measure participants’ perceived emotional distance from their parents. Responses were recorded on a five-point Likert scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”), with higher scores indicating greater levels of perceived parental alienation. A Confirmatory Factor Analysis (CFA) supported the one-factor structure of the scale, with the model showing adequate model fit in the study (χ²(df) = 22.45, p <.001, RMSEA = 0.049, CFI = 0.996, TLI = 0.977, and SRMR = 0.011). In this study, Cronbach’ s α was 0.91.
Core self-evaluation was assessed using the core self-evaluation scale [44], which consisted of 10 items (e.g., “I believe I can succeed in life”). Responses were provided using a five-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). Mean scores were calculated, with higher scores indicating more positive evaluations of one’s own abilities and worth. In this study, Cronbach’ s α was 0.82.
Non-suicidal self-injury was assessed using the Chinese version of the Adolescent Non-suicidal Self-Injury Assessment Questionnaire [45]. The ANSAQ comprises behavioral and functional sections. This research employed the behavioral section, containing 12 items across two dimensions, utilizing a Likert scale ranging from 1 (never) to 5 (always). Higher scores suggest greater severity of non-suicidal self-injury behaviors. Cronbach’ s α in the present sample was at 0.98.
To reduce potential confounding, this study included a range of demographic and contextual variables as controls in the serial mediation model. These variables comprised gender, subjective economic status, and parental education level. Prior research has shown that variables such as parental education level and economic status are closely linked to harsh parenting and NSSI [46, 47].
This study employs a person-centered Latent Profile Analysis (LPA) to identify heterogeneity in harsh parenting among different groups [48]. Parameter estimates were based on the maximum likelihood method, using the Expectation-Maximization (EM) algorithm. Model fit was assessed using AIC, BIC, and aBIC, with smaller values indicating better fit. Classification accuracy was measured by the entropy index, and differences in model fit across categories were compared using the Lo-Mendell-Rubin test (LMR) and the Bootstrap Likelihood Ratio Test (BLRT). A p-value < 0.05 indicates that the K-class model fits better than the K-1 class model [49]. Subsequently, the serial mediation effect of parental alienation and core self-evaluation in the relationship between harsh parenting and non-suicidal self-injury was tested using Model 6 of the PROCESS macro. Indirect effects were evaluated using a nonparametric bootstrap procedure with 2,000 resamples, and bias-corrected 95% bootstrap confidence intervals were derived. An indirect effect was considered statistically significant when its confidence interval excluded zero. LPA was conducted using Mplus 8.0, and the comparison tests were performed using SPSS 26.0, with statistical significance set at p < 0.05 and 95% confidence intervals calculated.
Among the 5,741 participants, 74.45% were female. The mean age was 20.26 years (SD = 1.19). The majority of participants reported a middle to high family monthly income (57.50%). Paternal education levels were as 63.19% completed Junior high school or below, 22.57% High school or equivalent, and 13.53% College or above. Maternal education levels were as 67.58% completed Junior high school or below, 20.61% High school or equivalent, and 11.83% College or above. Regarding residential area, 51.37% of participants lived in rural areas, 24.28% in towns, and 23.99% in cities.
Common method bias was examined using Harman’s single-factor approach. The analysis identified 7 components with eigenvalues above 1, and the first component accounted for 26.12% of the variance. Because the variance explained by a single factor did not reach the commonly used heuristic threshold of 40%, the results did not indicate substantial common method bias, suggesting that common method bias was unlikely to be a major concern in this study.
Based on interpretability, class size, and classification quality, the 3-class solution was selected (Table 1). Although the information criteria continued to improve with additional classes (with the 5-class model showing the lowest AIC/BIC/aBIC), it yielded a very small class (4.95%, below the commonly used 5% threshold). The 4-class model avoided an extremely small class but showed lower entropy than the 3-class model, suggesting poorer class separation. The 3-class model achieved the highest entropy, indicating the clearest classification, and represented a substantial improvement over the 2-class model in terms of fit indices. Consistently, the LMR (p < 0.001) and BLRT (p = 0.010) supported the 3-class model over the 2-class model. Moreover, class proportions in the 3-class model (13.51%, 15.41%, 71.07%) were well balanced and all exceeded the 5% threshold. Overall, the 3-class model offered the most parsimonious and interpretable solution while maintaining acceptable fit and robust class sizes. As shown in Table 2, the average posterior probability for the most likely class membership in each latent class exceeded 0.90, indicating high classification accuracy and supporting the stability of the three-class solution.Table 1Model fit information for latent profile analysesClassAICBICaBICEntropyLMR(p)BLRT(p)Class proportions (%)19685896964969141 27666876835767550.966<0.0010.010 27.20/72.80 36828968515684070.978<0.0010.010 13.51/71.07/15.4146687767164670270.874<0.0010.010 17.50/54.32/13.44/14.7356296463310631450.926<0.0010.010 22.20/50.12/13.38/9.35/4.95Table 2Average posterior probabilities for the 3-class solutionClassNumber of classesPercentage(%)Average Posterior ProbabilityC1C2C3C177613.520.9960.0000.004C288515.410.0020.9790.019C3408171.070.0000.0070.993
In the analysis using a three-class latent profile model, we first applied the expectation–maximization algorithm to estimate the class-specific item means for each item within the identified classes, and then visualized the class profiles (Fig. 2). Examination of the score distributions and the behavioral patterns of harsh parenting led to the identification of three distinct profiles. Class 1, labeled high harsh parenting (n = 775, 13.51%), was characterized by elevated scores across all measured items. Class 2, labeled medium harsh parenting (n = 884, 15.41%), showed a pronounced peak in the first two items, followed by a steep decline in subsequent items, representing a front-high, back-low pattern of harsh parenting. Class 3, labeled low harsh parenting (n = 4,081, 71.07%), presented moderate ratings on the initial items and persistently low ratings thereafter. Table 3 presents the demographic characteristics of the three latent classes. Significant group differences were found in gender (χ² = 30.79, p < 0.001), paternal education level (χ² = 13.28, p = 0.02), and maternal education level (χ² = 11.48, p = 0.03).Table 3Descriptive sample characteristicsCharacteristicsclass1class2class3FCramér’s Vp(N = 776)(N = 885)(N = 4 081)Age, mean (SD)^a^20.26 (1.19)20.30 (1.31)20.25 (1.15)Gender, No. (%) Males250 (32.22)286 (32.32)931 (22.81)30.790.00< 0.001 Females526 (67.78)599 (67.68)3150 (77.19)Subjective SES, No. (%) Lower class353 (45.49)389 (43.95)1699 (41.63)8.840.150.06 Middle class358 (46.13)434 (49.04)2030 (49.74) Upper class65 (8.38)62 (7.01)352 (8.63)Paternal education level, No. (%) Junior high school or below487 (62.76)590 (66.67)2592 (63.51)13.280.300.01 High school or equivalent176 (22.68)178 (20.11)942 (23.08) College or above113 (14.56)117 (13.22)547 (13.40)Maternal education level, No. (%) Junior high school or below534 (68.81)623 (70.40)2723 (66.72)11.480.050.02 High school or equivalent146 (18.81)155 (17.51)882 (21.61) College or above96 (12.37)107 (12.09)476 (11.66)Residential area. (%) Rural420 (54.12)447 (50.51)2231 (54.67)8.280.260.08 Town175 (22.55)221 (24.97)935 (22.91) City181 (23.32)217 (24.52)915 (22.42)
Fig. 2The three harsh parenting profiles identified by latent profile analysis
We first conducted a serial mediation analysis in the full sample, as shown in Fig. 3. Harsh parenting was positively associated with parental alienation (β = 0.35, p < 0.001), which in turn was positively associated with NSSI (β = 0.05, p < 0.001). Harsh parenting was also negatively associated with core self-evaluation (β = −0.10, p < 0.001), which was negatively related to NSSI (β = −0.10, p < 0.001). Parental alienation was negatively associated with core self-evaluation (β = −0.47, p < 0.001), supporting a serial mediation pathway. Bootstrapping (2,000 resamples) confirmed three significant indirect via parental alienation (effect = 0.02, 95% CI [0.00, 0.03], via core self-evaluation (effect = 0.01, 95% CI [0.01, 0.01]), and via both sequentially (effect = 0.02, 95% CI [0.01, 0.02]). The total indirect effect was 0.04 (95% CI [0.03, 0.06]), and the direct effect remained significant (effect = 0.04, 95% CI [0.01, 0.07]), suggesting partial mediation. Taken together, these findings provide empirical support for Hypotheses 1–4 (for detailed information, see Supplementary Tables S1-S2).Fig. 3The serial mediation model of parental alienation and core self-evaluation for all. Note: ^^p < 0.05, ^**^p < 0.001; Values on paths are standardized regression coefficients (β). Solid lines indicate significant paths and dashed lines indicate non-significant paths
Next, based on the 3-class solution from the LPA conducted on harsh parenting, we divided participants into three groups and constructed separate serial mediation models for each group, as shown in Fig. 4. In the high harsh parenting group, harsh parenting was positively associated with NSSI. Additionally, harsh parenting showed an indirect association with NSSI via core self-evaluation. Parental alienation was negatively associated with core self-evaluation, supporting a serial mediation pathway. Bootstrap results indicated a total indirect effect of 0.02 (95% CI [0.00, 0.05]), accounting for 16.67% of the total effect. The indirect effect via core self-evaluation was 0.01, and the serial indirect effect through parental alienation and core self-evaluation was 0.02. However, the indirect mediation effect via parental alienation alone was not significant (effect = −0.02, 95% CI [−0.04, 0.01]), as the confidence interval included 0 (for detailed information, see Supplementary Tables S3-S4). In the medium harsh parenting group, both parental alienation and core self-evaluation were significantly related to NSSI, consistent wit a serial indirect pathway. The total indirect effect was 0.05 (95% CI [0.02, 0.08]), with the indirect effect via parental alienation being 0.03, via core self-evaluation being 0.01, and the serial indirect effect being 0.01. The direct effect was not significant (effect = 0.01, 95% CI [−0.06, 0.07]), suggesting that the association between harsh parenting and NSSI in this group was primarily explained by the indirect pathways through the mediators (for detailed information, see Supplementary Tables S5-S6). In the low harsh parenting group, both parental alienation and core self-evaluation were significantly related to NSSI, forming a serial mediation pathway. The total indirect effect was 0.03 (95% CI [0.02, 0.05]), with the indirect effect via parental alienation being 0.01, via core self-evaluation being 0.01, and the serial indirect effect being 0.01. The direct effect was not significant (effect = 0.00, 95% CI [−0.04, 0.02]) (For detailed information, see Supplementary materials Tables S7-S8).
Fig. 4Group-specific serial mediation models across harsh parenting profiles. Note: Panels (a-c) correspond to Class 1 (high harsh parenting, n = 776), Class 2 (medium harsh parenting, n = 885), and Class 3 (low harsh parenting, n = 4081), respectively; ^^p < 0.05, ^^p < 0.01, ^^p < 0.001; Values on paths are standardized regression coefficients (β). Solid lines indicate significant paths and dashed lines indicate non-significant paths
To assess whether classification uncertainty materially influenced the subgroup-specific mediation results, we conducted a sensitivity analysis by restricting the sample to participants whose posterior probabilities for their assigned (most likely) class memberships was ≥ 0.90, retaining n = 5,587 (97.31% of the original sample). We then re-estimated the chain mediation models within each class. The effect estimates and significance patterns were largely comparable to those from the full-sample analyses, suggesting that the main conclusions were not materially affected by potential misclassification.
Grounded in the cognitive-emotional model of NSSI [16], the present study developed a serial mediation model to examine how parental alienation and core self-evaluation shape the link between harsh parenting and NSSI in young adults, and further employed a longitudinal design to assess the stability of these associations across time. LPA of harsh parenting identified three distinct patterns. Notably, the magnitude and form of the serial mediation effects differed across these profiles, indicating that the pathways from harsh parenting to NSSI are contingent on broader parenting patterns rather than uniform across all young adults. Overall, these results clarify the mechanisms through which harsh parenting contributes to NSSI, underscoring the roles of parental alienation and core self-evaluation.
Unlike Western educational traditions that often emphasize autonomy, emotional expression, and positive reinforcement, Chinese traditional culture has long regarded strict and authoritarian parenting as a means to foster discipline, resilience, and academic success in children [15, 50]. Our findings indicate that experiences of harsh parenting remain associated with adverse outcomes among college students. Systematic reviews and meta-analyses have shown out that harsh parenting, including both physical and verbal aggression, has been consistently linked to the development of negative emotional states such as depression and anxiety [51, 52]. In addition, a longitudinal study involving 4,231 participants across four waves found that harsh parenting was associated with increased emotional vulnerability, often resulting in greater sensitivity to stress and difficulties in emotion regulation later in life [50]. According to the cognitive-emotional model of NSSI [16], individuals who face intense negative emotions but lack healthy coping mechanisms may turn to NSSI for short-term relief. In this context, NSSI acts as a maladaptive way to regulate emotions, helping individuals manage distressing feelings such as sadness, anger, or anxiety, which may be rooted in poor parenting [53, 54].
Based on the cognitive-emotional model of NSSI [16], this study identified three distinct mediating pathways linking harsh parenting to NSSI. First, consistent with previous research [52], parental alienation emerged as a key mediator. Harsh parenting often erodes trust and warmth in the parent-child relationship, fostering feelings of emotional isolation and helplessness [55]. Within the cognitive-emotional framework, such relational disconnection heightens negative affect and reduces perceived interpersonal support. As a result, individuals may be more likely to adopt NSSI as a maladaptive coping strategy to regulate distress and to signal a need for care and attention from significant others [29, 56]. Second, core self-evaluation also mediated the association between harsh parenting and NSSI. Harsh parenting may foster a negative cognitive schema (e.g., “I am worthless” or “I deserve blame”), which undermines self-worth and elicits self-directed negative emotions such as guilt, shame, and self-criticism. In line with the cognitive-emotional model, these maladaptive self-appraisals may intensify emotional pain and motivate NSSI as a means of self-punishment and affect regulation. A meta-analysis suggested that individuals with low self-evaluations may engage in NSSI as a means of self-punishment to alleviate internalized moral pain [57, 58]. Third, our findings indicate a serial mediating pathway in which harsh parenting is associated with parental alienation, which is in turn linked to lower core self-evaluation and a higher risk of NSSI.This pattern closely accords with the cognitive-emotional model of NSSI, which emphasizes that individuals’ cognitive appraisals of interpersonal contexts shape downstream emotional responses and behavioral coping. Specifically, harsh parenting may be construed as rejection or abandonment, consolidating negative beliefs such as “I am not cared for”, which then amplifies emotional distress and increases reliance on NSSI as an emotion-regulation strategy. When parenting involves verbal aggression, corporal punishment, or neglect, it not only disrupts relational security but also impedes the formation of a stable sense of self [56]. The resulting deterioration in core self-evaluation further heightens vulnerability to NSSI [16, 59].
LPA reveals heterogeneous mechanisms linking harsh parenting to young adults’ NSSI, with variations in both the strength of direct/indirect effects and the distinct roles of interpersonal and self-focused pathways across levels of harsh parenting. Within the cognitive-emotional model of NSSI [16], LPA revealed heterogeneity in the mechanisms underlying young people’s NSSI risk across families characterized by different levels of harsh parenting. In families characterized by chronically high levels of harsh parenting, NSSI risk appears to be dominated by the direct effect. Prolonged exposure to a punitive environment may foster stable and generalized negative cognitive schemas and is often accompanied by persistently elevated emotional arousal. Within the cognitive-emotional model of NSSI, this coupling of maladaptive appraisals and sustained affective activation can substantially deplete psychological resources and undermine individuals’ capacity to engage in adaptive self-regulation [60]. As a result, NSSI may become a more automatic and frequent strategy for emotion regulation and self-punishment [61]. From an intervention perspective, priority should be given to modifying caregivers’ parenting practices, thereby altering the adverse cognitive-emotional context at its source [62].
For families with fluctuating harsh parenting, indirect effects prevail, with parental alienation as the strongest mediator. This inconsistent parenting style may lead individuals to form the core belief that relationships are unstable and unpredictable. Individuals repeatedly interpret parent-child interactions negatively (e.g., perceiving rejection or abandonment), triggering intense emotional distress (such as anxiety and feelings of abandonment) [59]. In this process, non-suicidal self-injury (NSSI) primarily serves as a coping mechanism to regulate the emotional distress triggered by these interpersonal cognitions, or to influence the unstable emotions of the caregiver [63]. Therefore, interventions should focus on repairing secure attachment and enhancing emotional regulation, helping individuals rebuild stable and positive interpersonal cognitive patterns [64]. In families characterized by low levels of harsh parenting, the direct effect was not significant, whereas all indirect pathways, including parental alienation, core self-evaluation, and the sequential mediation pathway, remained significant. This pattern suggests that even within a relatively benign external environment, individuals’ pre-existing vulnerable cognitive structures, such as negative self-schemas and heightened sensitivity to interpersonal cues, may still be activated by subtle stressors. Such activation can elicit emotional distress and, in turn, increase the likelihood of relying on NSSI for self-regulation or self-punishment. These findings highlight the independent predictive value of internal cognitive–emotional processes for NSSI risk [29]. Accordingly, prevention efforts should prioritize cognitive restructuring to stabilize self-concept and the cultivation of alternative secure bonds to compensate for or revise latent interpersonal insecurity [65].
These effects, although modest in magnitude, are noteworthy because they emerged over a three-month interval after adjusting for demographic covariates. Small longitudinal associations observed over relatively short time spans may accumulate over time in adulthood and translate into meaningful differences in NSSI severity and related mental health risk [66]. In interpreting practical significance, it is often more informative to consider the total indirect effect of the serial pathway than to focus on any single path coefficient. The combined product of the sequential links captures the strength of the hypothesized cascade from harsh parenting to NSSI severity through parental alienation and adults’ core self-evaluation. Thus, even when each step is small, the overall indirect effect can reflect a coherent and theoretically meaningful mechanistic process [65]. From an applied perspective, these findings point to modifiable targets for prevention, particularly parental alienation and adults’ core self-evaluation. Nevertheless, the modest effect sizes warrant cautious interpretation, and replication with longer follow-ups and stronger designs is needed to establish the robustness and clinical relevance of these pathways.
While this study offers valuable theoretical contributions and practical implications, several limitations should be addressed in future research. First, incorporating a broader range of indicators may enable a more precise classification of harsh parenting patterns, and it is also important to distinguish the developmental consequences of paternal versus maternal harsh parenting. Future researches should therefore develop more comprehensive and nuanced measurement systems for parenting practices. Second, the sample showed a gender imbalance, with most participants being female,. Which may introduce gender-related biases and warrants caution in generalizing the findings. Replication with a more balanced gender distribution is needed to evaluate whether the observed associations hold across gender groups. Third, harsh parenting was assessed using self-report measures, which may be subject to recall bias regarding participants’childhood family experiences. Moreover, parental alienation and core self-evaluation were measured within the same wave, limiting causal inference regarding their temporal ordering. Future research could mitigate these concerns by employing more objective, technology-assisted assessments (e.g., mobile- or wearable-based methods) and collecting multi-wave longitudinal data to more rigorously test temporal sequencing and potential causal pathways. Finally, to control questionnaire length and minimize participant burden, this study considered only family factors in relation to young adults’ non-suicidal self-injury. Future research could adopt an ecological systems perspective by integrating multiple risk factors across individual, family, peer, and school levels, thereby advancing a multidimensional risk accumulation model.
This study provides novel evidence on the long-term impact of harsh parenting on NSSI by employing a longitudinal design combined with a person-centered profiling approach. The results suggest that the relationships between harsh parenting and NSSI vary across distinct parenting profiles. Specifically, the high harsh parenting group appears to show a predominantly direct effects, while the medium harsh parenting group is associated with NSSI mainly through unmet emotional needs and relational alienation. In the low harsh parenting group, subtle dynamics in parent-child relationships and self-evaluation remain influential. These findings underscore the importance of considering parenting patterns in understanding NSSI and suggest the need for profile-specific prevention and intervention strategies.
Supplementary Material 1.