Self-Compassion as a Mediator of the Longitudinal Link Between Parent and Adolescent Depressive Symptoms

Abstract

The literature has consistently shown the link between parental depressive symptoms and adolescent depressive symptoms. Emotion regulation processes may explain this association, with self-compassion acting as a potential mediating mechanism. This study aimed to explore these developmental trajectories using a longitudinal design. The sample comprised dyads of parents and adolescents (N = 311), assessed across two assessment waves (T1 and T2) over 12 months. The adolescents’ mean age was 13.63 years (SD = 1.30; 173 males). Mediation analyses using structural equation modelling supported the role of self-compassion in mediating the relationship between parental and adolescent depressive symptoms over time (53% of outcome variance explained). The results suggest that fostering self-compassion early in adolescence may help prevent the later adverse effects of depressive symptoms.

1. Introduction

Depression is a primary contributor to disease and economic global impact [1], and its prevalence is likely to grow until 2050 [2]. Particularly during adolescence, depression prevalence is notably high, with a lifetime prevalence among individuals aged 13 to 18 of 11% [3,4]. In the Portuguese context, data from the last decade have consistently pointed to concerning levels of psychological distress among adolescents, with frequent reports of emotional symptoms such as sadness, anxiety, or irritability, and an upward trend in internalising problems already noted before the COVID-19 pandemic [5,6]. The causes of depression are multi-determined and complex [3]. Nonetheless, evidence demonstrates an augmented risk from the interaction between genetic and developmental factors, psychosocial adversity, and hormonal changes in adolescence [7].

Adolescence is a developmental phase characterised by multiple biopsychosocial modifications, including hormonal changes, physical maturation, sexual emergence, complex cognitive models about others and self, independent self-identity, autonomy from parents, and concerns about acceptance and fit in the peer group [8,9]. These developmental phenomena substantially increase the odds of depression during adolescence [9], and when it occurs, it is particularly linked to a higher probability of recurrence, chronicity, physical health problems, psychiatric comorbidity, and suicidality [4,7,10,11]. Depression during adolescence has long-lasting effects on social, academic, and emotional functioning [12].

Longitudinal research on modifiable factors associated with the variability of depressive symptoms is critical to the design of preventive strategies in developmental psychopathology [13]. Given that adolescence marks a key phase for depression onset [3,4], it is of utmost importance to examine variables that could relate to the variability of depressive symptoms observed in this developmental period. This type of study may also be relevant for the early identification and efficient early prevention of depression in at-risk groups [14,15,16].

1.1. Parental Depression and Adolescent Depression

Parental depression has been associated with caregiving patterns marked by emotional unavailability, social withdrawal, and diminished responsiveness, which may interfere with adolescents’ emotional development and regulation.

The intergenerational transmission of depressive symptoms can be understood through several psychological theories. Social Cognitive Theory suggests that adolescents may internalize emotional and cognitive patterns observed in their parents, especially when repeatedly exposed to expressions of hopelessness or self-critical behaviours, through observational learning and cognitive processing [17,18]. Sociometer Theory offers a complementary perspective, proposing that perceived parental rejection or neglect—common in the context of depressive caregiving—can erode adolescents’ self-worth and sense of social belonging, thereby contributing to depressive symptoms [19,20].

Empirical studies have consistently shown that adolescents with a family history of depression, particularly with a depressed parent, tend to report earlier onset and greater intensity of depressive symptoms, irrespective of maternal or paternal depression [21,22,23,24]. Furthermore, subthreshold depressive symptoms during adolescence have been linked to more persistent and impairing clinical trajectories over time [25,26,27,28].

Based on these theoretical and empirical foundations, it is hypothesised that parental depressive symptoms will positively predict adolescent depressive symptoms.

1.2. Self-Compassion

Self-compassion is rooted in mammalian motives for caring, altruism, and connection to others and is a source of distress regulation [29,30]. Neff [31] conceptualizes self-compassion as consisting of three core elements: (i) self-kindness, treating oneself with kindness and gentleness instead of harsh self-judgment; (ii) common humanity, recognising that suffering is a universal human experience, not an isolating one; and (iii) mindfulness, adopting a mindful attitude that embraces painful thoughts and feelings with openness and balance, avoiding excessive identification with them. Related to these three components are the opposite attitudes when experiencing suffering, namely self-judgement (i.e., the tendency to be critical and hostile towards themselves when difficult situations occur), isolation (i.e., the tendency to perceive the internal experience as a separate or isolating experience from others), and overidentification (i.e., the tendency to overly identify with internal experiences, such as thoughts, emotions, memories) [31,32]. This definition has been studied and validated in several studies, which confirm the multidimensional system of self-compassion composed of six components and a latent variable of general self-compassion in both adult [33,34] and adolescent samples [35,36].

1.3. Self-Compassion in Adolescence

While the investigation of self-compassion during adolescence is still developing, several studies have demonstrated its importance for this age group. In general, research shows a positive link between self-compassion and secure attachment, satisfaction with life, and overall well-being [35,37,38,39]. Conversely, self-compassion tends to be inversely related to negative emotions, trauma symptoms [40], substance abuse [41], aggression [42], self-harm without suicidal intent [43], and social anxiety [44]. Marsh et al. [45] carried out a meta-analysis of 19 studies with a large adolescent sample (ages 10–19), finding a significant inverse association between self-compassion and psychological distress, as indicated by anxiety, depression, and stress levels (r = −0.55). Likewise, a systematic review of 11 cross-sectional studies consistently found that lower self-compassion is strongly linked to increased depressive symptoms during adolescence [46]. Specifically, self-compassion is proposed as a possible mediator. For example, Neff and McGehee [39] found that self-compassion indirectly influenced the connection between family dynamics and depressive symptoms in adolescents. Positive family bonds and secure attachment were related to higher self-compassion, which then corresponded with lower levels of depressive symptoms [39]. Comparable partial mediation results were observed in adolescents aged 14 to 17, where maternal warmth was found to promote self-compassion, which in turn was linked to reduced depressive symptoms [47]. A longitudinal investigation involving at-risk adolescents exposed to potentially traumatic stress showed that higher initial self-compassion predicted decreased depressive symptoms across three assessment points [48]. Similarly, the positive aspects of self-compassion—self-kindness, common humanity, and mindfulness—demonstrated a predictive role in diminishing depressive symptoms over two time periods [49].

1.4. Self-Compassion, Parental Depression, and Adolescent Depression

Gilbert [50] and other authors [51,52] argue that self-compassion is rooted in an evolved mammalian system that regulates attachment and caregiving behaviours. When kindness and concern are available, individuals tend to feel safe, connected, and soothed. In contrast, self-criticism is linked to threat-focused systems associated with dominance, submission, and social rank. Consequently, individuals raised in secure and supportive environments are more likely to develop self-compassionate attitudes, while those raised in insecure, stressful, or threatening contexts tend to become more self-critical, hypervigilant, avoidant, and defensive in their internal and external environments [53]. Additionally, Neff [33] highlights that self-compassion fosters self-acceptance, reduces harsh self-judgement, enhances connectedness, and mitigates ruminative cycles that may contribute to emotional dysfunction.

Parental depression is often associated with parenting behaviours that reflect emotional unavailability, inconsistency, or threat, which may resemble the insecure or stressful caregiving environments described above. Thus, depressed parents may create family contexts characterised by emotional detachment or psychological threat, rather than warmth and safeness, potentially contributing to the development of depressive symptoms in their offspring. At the same time, this type of parenting environment may undermine the development of self-compassion in adolescents.

Accordingly, it is hypothesised that parental depression negatively predicts adolescent self-compassion, which in turn may increase depressive symptoms. Therefore, self-compassion is expected to mediate the association between parental and adolescent depressive symptoms.

1.5. Gender and Age

Evidence consistently shows that adolescent girls exhibit higher rates of depressive symptoms than boys, with prevalence estimates indicating they may be nearly twice as likely to be affected [4,7,54,55]. Longitudinal investigations further reveal that this disparity tends to widen over the course of adolescence, particularly during its middle and later phases [56,57]. This developmental trend is thought to result from the interplay of multiple factors—biological, emotional, and social.

Several studies found that different factors may influence depressive manifestations to varying degrees, depending on developmental stage. Age, in particular, may act as a potential confounding factor, and longitudinal research also suggests a lack of stability regarding depressive symptoms across time [13,58,59].

Self-compassion has been linked to depressive symptoms in studies using both cross-sectional and longitudinal designs; however, the test of its mediatiating effect is scarce [39,45,46,47,48,49]. Previous evidence indicated that parental depressive symptoms directly impact adolescents’ depressive symptoms [21,22,23,24]. Nevertheless, understanding the underlying mechanisms that cause depressive symptoms in youth is still limited [23,24,60,61]. The present study is intended to slightly contribute to a better understanding of those potential mechanisms, particularly the modifiable ones, and thus enlighten the interventions often aimed at treating symptoms rather than the associated conditions with the goal of diminishing depressive symptoms occurrence and augmenting the well-being of young individuals [62,63]. Additionally, longitudinal studies of the development of self-compassion, how early experiences relate to it, and the role of self-compassion between early experiences and later adjustment are still scarce, too [64]. Specifically, this study will test, using a longitudinal design, the relationships between the selected factors that seem to be associated with adolescent depressive symptoms, and particularly the mediation role of self-compassion in the mentioned intergenerational transmission of depression.

1.6. The Current Study

The potential mediating role of adolescent self-compassion in the association between parental depressive symptoms and adolescent depression remains underexplored. To date, the literature lacks longitudinal studies examining this temporal mediation pathway. While prior research has consistently demonstrated that lower self-compassion is associated with higher depressive symptoms in adolescents, and that parental depressive symptoms increase the risk of depression in offspring, these associations have mostly been studied in isolation. Studies have predominantly focused on partial relationships rather than testing integrated mediation models. Moreover, few studies have explored these dynamics using dyadic data from parents and their adolescent children while controlling for important covariates such as prior adolescent depression, age, and gender. The present study addresses this gap by testing a theoretically grounded longitudinal mediation model over a 12-month period, offering novel insights into the mechanisms underlying the intergenerational transmission of depressive symptoms.

This study, conducted at two time points (Time 1 and Time 2, with a 12-month follow-up), involved adolescent–parent dyads and aimed to (i) examine the relationships among parental depressive symptoms, adolescent self-compassion, and adolescent depressive symptoms; and (ii) longitudinally assess whether adolescent self-compassion mediates the impact of parental depressive symptoms on adolescent depression. Based on prior research highlighting gender and age differences in depression, these demographic factors were incorporated into the mediation model as covariates, alongside adolescents’ baseline depressive symptoms. The proposed theoretical framework is illustrated in Figure 1.

Based on the theoretical background for the proposed mediation model, the hypotheses will be tested as follows:

H1: 

Parental depressive symptoms have a direct effect on adolescents’ depressive symptoms.

H2: 

Parental depressive symptoms negatively predict adolescents’ self-compassion.

H3: 

Diminished adolescent self-compassion impacts adolescents’ depressive symptoms.

H4: 

Adolescents’ self-compassion mediates the association between parental and adolescents’ depressive symptoms, with parental depression negatively predicting self-compassion and thereby increasing adolescents’ depressive symptoms.

Also, in these hypotheses, it is predicted that gender may have a direct impact on adolescents’ depressive symptoms, according to previous literature. Finally, we hypothesized that age may have an impact on the proposed model.

2. Materials and Methods

2.1. Participants

In middle schools, the data were collected in two assessment moments: Time 1 (October 2013), T1, and Time 2 (November 2014), T2 (after 12 months). At Time 1, aligned with the aim of testing the hypotheses in adolescent-parent dyads, the total initial sample comprised 470 adolescents. However, for data analysis purposes (using Structural Equation Modelling), only the complete data of all variables under study (at T1 and T2) were used.

Therefore, the Time 1 sample comprised 311 adolescents and an equal number of their parents. In the sample of adolescents, 199 were female (64%), and 112 were male (36%) between 11 and 17 years old (M = 13.67; SD = 1.38). No age differences emerged between genders, t(309) = −0.103, p = 0.92. Concerning the educational level, ranging from the 7th to 10th grade (M = 8.18; SD = 1.00), females and males did not differ in educational level, t (256.23) = −1.07, p = 0.28. Adolescents were distributed across both urban (n = 71, 22.8%) and rural (n = 240, 77.2%) residential areas. Attrition (n = 159, 33.83%) was mainly due to adolescents moving to other schools. Concerning sociodemographic data for parents, 188 parents answered the gender variable (123 missing participants; 39.5%), of which 142 were female (45.7%) and 46 were male (14.8%); and 135 parents answered the age variable (176 missing participants; 56.6%), ranging from 24 to 75 years old (M = 43.44, SD = 7.45).

2.2. Procedure

This longitudinal study involved two data collection points spaced 12 months apart and was conducted in seven middle schools in central Portugal. Ethical clearance was obtained from the Portuguese Data Protection Authority and the Ministry of Education. School administrations and pedagogical councils approved the research.

Parents and adolescents were fully informed about the study’s purpose and the voluntary nature of participation, and written informed consent was obtained from both. Contact details of the research team were provided for any inquiries or support needs. Only adolescents with consent from themselves and their parents were included.

To ensure confidentiality and data integrity, each participant was assigned a unique code. Adolescents completed the questionnaires in classrooms, supervised by a teacher and a researcher, who offered assistance when requested. Parent questionnaires were distributed through adolescents, completed at home, and returned in sealed envelopes via the adolescents. Adolescents received a small token of appreciation (a bracelet) for participation.

2.3. Instruments

The sociodemographic questionnaires include questions for adolescents and their parents about gender, age, area of residence, school performance, present or past clinical conditions, and present or past treatments.

2.3.1. Brief Symptoms Inventory

The Brief Symptoms Inventory (BSI [65,66]) is a self-administered tool designed to evaluate psychological distress and symptoms of psychopathology. Its brevity and ability to overview various psychological symptoms comprehensively are some of the advantages of this measure. This scale consists of 53 items and nine dimensions about several symptoms of psychopathology (e.g., “your feelings being easily hurt”). Respondents indicate the extent to which they experienced distress from each symptom during the past week using a 5-point Likert scale, ranging from 0 (“not at all”) to 4 (“extremely”). The BSI has demonstrated strong psychometric properties in various studies, and its multifactorial structure has been confirmed in both general and clinical populations internationally [67,68], including Portugal [65]. The Portuguese version also showed internal consistency coefficients (Cronbach’s alpha) for the nine symptom dimensions typically ranging from 0.71 to 0.85, indicating good reliability. In this study, we focused exclusively on the depressive symptoms subscale (6 items; e.g., “feeling lonely”) of the self-administered Portuguese adaptation [65] to assess parental depressive symptoms. Cronbach’s alpha for the used subscale was 0.85 at Time 1 and 0.88 at Time 2.

2.3.2. Self-Compassion Scale

The Self-Compassion Scale (SCS [31,35]) is a self-report questionnaire to assess the attitude of being kind, tolerant, and compassionate towards oneself in the face of life-challenging experiences. It is composed of 26 items and six subscales: Self-kindness (5 items; e.g., “I try to be loving towards myself when I’m feeling emotional pain”), Self-judgment (5 items; e.g., “I’m disapproving and judgmental about my own flaws and inadequacies”), Common humanity (4 items; e.g., “When things are going badly for me, I see the difficulties as part of life that everyone goes through”), Isolation (4 items; e.g., “When I think about my inadequacies it tends to make me feel more separate and cut off from the rest of the world”), Mindfulness (4 items; e.g., “When I’m feeling down I try to approach my feelings with curiosity and openness”), and Over-identification (4 items; e.g., “When something upsets me, I get carried away with my feelings”). Items were rated on a 5-point scale (1 = “almost never” to 5 = “almost always”). In the current study, the total score was used, with higher scores indicating greater self-compassion. Psychometric studies demonstrated that this measure has a higher-order model, and a six-factor correlated model in both adult [31,33] and adolescent samples [35], supporting the use of the general self-compassion score and six subscales. Also, the measure showed good internal consistency for Portuguese language adolescents (Cronbach’s alpha of 0.88 for total score and ranging between 0.70 and 0.79 for subscales) [35]. In the present study, the internal reliability was excellent for the total score (α = 0.92) at Time 1 and good at Time 2 (α = 0.82).

2.3.3. Children’s Depression Inventory

The Children’s Depression Inventory (CDI [69,70]) is a widely used self-report inventory for assessing depressive symptomatology in school-aged children aged 7 to 17. This inventory consists of 27 items, each with three response options in order of increasing severity, from no problem (0) to severe problem (2). The questionnaire covers negative mood, interpersonal problems, ineffectiveness, anhedonia, and negative self-esteem (e.g., “I almost always feel sad”). The total score ranges from 0 to 54 points. Higher scores indicate elevated depressive symptoms. Regarding the factor structure, a first-order factor to assess the general depression factor is considered a robust psychometric structure in diverse studies in clinical and community samples of adolescents [71,72]. Thus, this measure is a valuable screening tool for detecting depressive symptoms. Both original [73] and Portuguese versions [74] showed good psychometric properties, with Cronbach’s alphas (α) ranging from 0.71 to 0.89. In the current study, this measure was used at Time 1 (α = 0.85) and Time 2 (α = 0.92).

2.4. Data Analysis Strategy

Data analyses were performed with PASW (Predictive Analytics Software) version 29 (SPSS, Chicago, IL, USA). Path analyses within Structural Equation Modeling (SEM) were conducted using AMOS version 22 (Amos Development Corporation, Crawfordville, FL, USA).

Before running the structural equation modeling and path analyses, basic descriptive statistics including means and standard deviations were calculated for each variable. To investigate the relationships among variables, Pearson correlation coefficients were computed and interpreted according to established benchmarks [75,76]. These initial correlations provided insight into variable interrelations, serving as a basis for the SEM procedures that followed.

After the initial analyses, path analysis was applied to evaluate the hypothesised relationships between variables outlined in the theoretical model (Figure 1). This SEM technique tests presumed causal connections based on prior theoretical assumptions [77].

The tested path model examined several direct effects: first, from parental depressive symptoms at Time 1 to adolescents’ depressive symptoms at Time 2, addressing Hypothesis 1; second, from parental depressive symptoms at Time 1 to adolescents’ self-compassion at Time 1, corresponding to Hypothesis 2; and third, from adolescents’ self-compassion at Time 1 to their depressive symptoms at Time 2, testing Hypothesis 3. Finally, to evaluate Hypothesis 4, the analysis also tested whether adolescents’ self-compassion at Time 1 mediated the relationship between parental depressive symptoms at Time 1 and adolescents’ depressive symptoms at Time 2. The model controlled for demographic factors, including gender (coded as 0 = male, 1 = female), age, and adolescents’ baseline depressive symptoms (assessed with the CDI), to control for their effects.

The estimation method used was maximum likelihood (ML) to assess the significance of all model path coefficients and to compute the statistics of fit indexes [77]. Model fit was assessed using the Chi-square test (χ²), which quantifies the difference between the observed data and the hypothesised model; lower values reflect a closer fit [77]. However, because this index is susceptible to sample size [74], it was used in conjunction with other global fit indices. To assess overall model adequacy, the following fit indices were examined: Comparative Fit Index (CFI ≥ 0.95 indicating good fit); Tucker–Lewis Index (TLI ≥ 0.95 indicating good fit); Standardized Root Mean Square Residual (SRMR ≤ 0.08 indicating good fit); and Root Mean Square Error of Approximation (RMSEA), with values ≤ 0.05 considered good, ≤0.08 acceptable, and ≥0.10 poor, along with 90% confidence intervals [78]. Significance of direct, indirect, and total effects was tested using the Bootstrap resampling technique. This procedure, using 2000 Bootstrap samples, was employed to create 95% bias-corrected confidence intervals. Effects were deemed statistically significant (p < 0.05) when the 95% bias-corrected confidence interval for the indirect effect did not include zero [75,76,77]. Model evaluation also included examining the explained variance (R²) for each endogenous variable, based on recommendations by Kline [76].

3. Results

3.1. Preliminary Data Analysis

Data were examined for univariate normality, with no significant deviations observed (|Sk| < 3 and |Ku| within 8–10 [77]). Multivariate outliers were assessed using Mahalanobis distance squared (MD²), revealing some extreme values. However, analyses conducted with and without these cases yielded consistent results, so they were retained to preserve sample variability [77]. Independence of residuals was supported by a Durbin–Watson statistic of 2.059. Multicollinearity was assessed via tolerance values (>0.2) and variance inflation factors (VIF < 5), indicating no multicollinearity issues among variables [77].

Table 1. Descriptive Statistics and Correlations for continuous study variables (N = 311).

Variable	M	SD	1	2	3	4	5	6
1. Age (T1)	13.67	1.38	-
2. Adolescent depressive symptoms (T1)	10.32	6.08	0.20 **	-
3. Parental depressive symptoms (T1)	0.80	0.75	0.05	0.13 **	-
4. Self-compassion (T1)	77.35	18.82	−0.23 **	−0.29 **	−0.22 **	-
5. Self-compassion (T2)	72.66	12.66	0.09	0.10	−0.01	0.04	-
6. Adolescent depressive symptoms (T2)	15.62	9.97	0.23 **	0.24 **	0.28 **	−0.73 **	0.15 *	-
7. Parental depressive symptoms (T2)	0.67	0.73	0.08	0.10	0.83 **	−0.14 *	0.05	0.12

Note. * p ≤ 0.05; ** p ≤ 0.01; T1 = Time 1; T2 = Time 2.

presents the means, standard deviations, and Pearson product-moment correlations among the continuous study variables, where applicable, at Time 1 and Time 2.

Results showed that parental depressive symptoms (T1) were negatively and moderately correlated with adolescents’ self-compassion (T1). In addition, parental depressive symptoms were positively and moderately correlated with adolescents’ depressive symptoms (T2). Regarding the mediator variable, adolescents’ self-compassion (T1) was negatively and strongly associated with depressive symptoms (T2). Adolescents’ depressive symptoms (T1) were positively and moderately associated with parental depressive symptoms (T1). A positive and moderate correlation was found between depressive symptoms at T1 and T2. Lastly, depressive symptoms (T1) were negatively and moderately associated with self-compassion (T1). Age was moderately and positively correlated with adolescents’ depressive symptoms at T1 and was also negatively and moderately correlated with self-compassion (T1).

A post hoc power analysis was conducted using G*Power (version 3.1) to assess the achieved statistical power for the multiple regression model predicting adolescent depressive symptoms at Time 2. The analysis was based on the observed effect size (f² = 1.13), a total sample size of 311 adolescent–parent dyads, and five predictors (parental depressive symptoms, adolescent self-compassion, gender, age, and baseline adolescent depressive symptoms). Setting the alpha level at 0.05, the analysis yielded a power estimate of 1.00, indicating adequate statistical power to detect the observed effects.

3.2. Path Analysis

Grounded in prior evidence and proposed hypotheses, the present study tested the direct effect of parental depressive symptoms at Time 1 (T1) on adolescents’ depressive symptoms at Time 2 (T2; Hypothesis 1). The second hypothesis was tested through the direct effect of parental depressive symptoms at Time 1 (T1) on adolescents’ self-compassion (T1). The third hypothesis was tested through the direct effect of adolescents’ self-compassion (T1) on adolescents’ depressive symptoms at Time 2. Finally, it was examined whether the effect of parental depressive symptoms at Time 1 (T1) on adolescents’ depressive symptoms at Time 2 is mediated by adolescents’ self-compassion measured at T1. To minimise potential confounding effects, adolescents’ baseline depressive symptoms (T1), along with age and sex, were statistically controlled by specifying covariances among all exogenous variables. The model was fully saturated, estimating 27 parameters in total. The following paths were not statistically significant: the direct effect of gender on self-compassion at Time 1 (b = 0.34, SE = 2.09, Z = 0.16, p = 0.871, β = 0.009); the direct effect of gender on depressive symptoms at Time 2 (b = 1.16, SE = 0.81, Z = 1.43, p = 0.153, β = 0.06); the direct effect of age on depression symptoms at Time 2 (b = 0.34, SE = 0.29, Z = 1.17, p = 0.242, β = 0.05); and the direct effect of depressive symptoms at Time 1 on depressive symptoms at Time 2 (b = 0.07, SE = 0.07, Z = 1.13, p = 0.258, β = 0.05). Thus, these nonsignificant paths were sequentially removed, and the model, consisting of 23 parameters, was respecified and recalculated (Figure 2). This respecified model revealed an excellent model fit: CMIN/DF = 1.49, p = 0.203; CFI = 0.994, TLI = 0.976, SRMR = 0.02, RMSEA = 0.039, 90% C.I. = 0.000 to 0.101, p = 0.526. In the respecified model, all paths were statistically significant, and the bootstrap resampling method further confirmed the significance of indirect effects. The model explained 15% of adolescents’ self-compassion (T1) and 53% of the variance in adolescents’ depressive symptoms (T2) (Figure 2), which, according to Kline [76], is considered a strong effect.

Direct effects are reported as standardized coefficients (β) with 95% confidence intervals, whereas indirect effects are reported as unstandardized coefficients (b) with 95% confidence intervals. Results showed a direct effect of parental depressive symptoms (T1) on adolescent depressive symptoms at Time 2, β = 0.09, 95% CI = [0.010 to 0.168], p = 0.032. Parental depressive symptoms were also negatively linked to adolescents’ self-compassion at T1 (β = −0.23, 95% CI = [−0.329 to −0.124], p = 0.001. Finally, self-compassion at Time 1 showed a statistically strong association with adolescents’ depressive symptoms at T2, β = −0.70, 95% CI = [−0.753 to −0.638], p = 0.001, suggesting a meaningful contribution to the model’s explanatory power.

Regarding the indirect effects, there was a significant indirect effect of parental depressive symptoms (T1) on adolescents’ depressive symptoms (T2), b = 2.15, 95% CI = [0.087 to 0.229], p = 0.001, through self-compassion (T1), even when other variables were controlled for. This indirect effect indicated that more parental depressive symptoms are longitudinally associated with adolescent depressive symptoms through their effect on lessening adolescents’ self-compassion.

Regarding covariate variables, results demonstrated that age contributed directly and negatively to self-compassion at T1, β = −0.19, 95% CI = [−0.290 to −0.079], p = 0.001. Also, age had a significant indirect effect on depressive symptoms at T2, b = 0.95, 95% CI = [0.055 to 0.208], p = 0.001, through its effect on self-compassion, even when other variables in the model were controlled for. Additionally, adolescents’ depressive symptoms at Time 1 were negatively and directly associated with self-compassion at T1, β = −0.18, 95% CI = [−0.278 to −0.067], p = 0.002. Adolescents’ depressive symptoms at Time 1 were also positively and indirectly associated with depressive symptoms at Time 2, b = 0.20, 95% CI = [0.049 to 0.199], p = 0.002, through self-compassion at Time 1. Overall, these findings highlight self-compassion as a relevant process in explaining later adolescent depressive symptoms.

Detailed direct and indirect effects are reported in

Table 2. Unstandardized regression coefficients, standard errors, t-values, p-values, and 95% confidence intervals for direct and indirect effects in the final mediation model.

Path	b	SE	t	p	95% CI
Direct effects
Parental depressive symptoms (T1) → Adolescent depressive symptoms (T2)	1.24	0.53	2.34	0.019	[0.168, 2.248]
Parental depressive symptoms (T1) → Adolescent self-compassion (T1)	−5.79	1.32	−4.40	<0.001	[−8.314, −3.137]
Adolescent self-compassion (T1) → Adolescent depressive symptoms (T2)	−0.371	0.02	−17.43	<0.001	[−0.413, −0.329]
Age → Adolescent self-compassion (T1)	−2.55	0.74	−3.45	<0.001	[−4.041, −1.115]
Adolescent depressive symptoms (T1) → Adolescent self-compassion (T1)	−0.54	0.17	−3.24	0.001	[−0.872, −0.212]
Indirect effects
Parental depressive symptoms (T1) → Adolescent self-compassion (T1) → Adolescent depressive symptoms (T2)	2.15	-	-	0.001	[0.087, 0.229]
Age → Adolescent self-compassion (T1) → Adolescent depressive symptoms (T2)	0.95	-	-	0.001	[0.055, 0.208]
Adolescent depressive symptoms (T1) → Adolescent self-compassion (T1) → Adolescent depressive symptoms (T2)	0.20	-	-	0.002	[0.049, 0.199]

Note. T1 = Time 1; T2 = Time 2. Effects were estimated using 2000 bootstrap resamples.

.

4. Discussion

4.1. Adolescence, Depression, and Self-Compassion: Background and Aims

Adolescence is a formative stage of life marked by multiple biopsychosocial changes and an increased likelihood of depressive symptoms occurrence [3,79]. Beyond identifying distal and proximal factors of depressive symptoms, it remains particularly relevant to understand the interplay between those factors. The way adolescents manage stressful events, as growing with parents struggling with depressive symptoms, may explain maladaptive or adaptive developmental trajectories. The literature shows that the internal dialogue focused on kindness, understanding, and compassion in facing difficult situations is consistently associated with greater psychological well-being [45] and lower depressive symptoms [46]. The current study aimed to explore the relationships between parental depressive symptoms, adolescent self-compassion, and depressive symptoms over one year using a longitudinal design. The main outcomes shed light on how parental depressive symptoms might shape adolescents’ emotional adjustment, notably through the intermediary role of self-compassion. A key novel contribution of this study lies in its use of a longitudinal and dyadic design, which allowed us to test the temporal mediating role of adolescent self-compassion in the association between parental depressive symptoms and adolescent depression over a 12-month period.

4.2. Main Findings

In line with prior evidence, higher levels of parental depressive symptoms were linked to higher levels of depressive symptomatology in adolescents [21,22,23,24]. Also, adolescents reporting higher levels of depressive symptoms tend to exhibit lower self-compassion, including in longitudinal analyses [39,47]. As hypothesised, adolescents who are more compassionate and gentler to themselves tend to report fewer depressive symptoms. Parental depressive symptoms were significantly associated with self-compassion only at T1, not at T2. Notably, parental depression at T2 was still associated with self-compassion measured at T1, not with any other variables. This pattern suggests that it may be particularly important to foster self-compassion in adolescents whose parents have depressive symptoms as early as possible. It also seems to highlight the value of longitudinal designs over cross-sectional approaches to explore the proposed hypothesis.

The main contribution of the present study lies in analysing the indirect effect of parental depressive symptoms at T1 on adolescent depressive symptoms 12 months later, mediated by adolescents’ self-compassion at T1, while rigorously controlling for gender, age, and baseline adolescent depressive symptoms. The inclusion of these covariates enhances the robustness of the longitudinal mediation model by reducing potential confounding influences and strengthening the validity of causal inferences. This approach addresses an important gap in the literature, as prior studies have predominantly relied on cross-sectional designs or have examined these variables in isolation, rather than testing dynamic mediational pathways over time. This study advances the literature by testing a longitudinal mechanism linking parental depressive symptoms to adolescent depressive symptoms via self-compassion—longitudinally. While previous studies have focused mainly on parenting style, the prospective impact of parental psychopathology, specifically depressive symptoms, on adolescent self-compassion remains largely unexplored. Consistent with our hypotheses, results from the structural equation modelling analyses supported the proposed model. Results suggest higher levels of parental depressive symptoms were associated with lower levels of adolescent self-compassion, which, in turn, predicted higher depressive symptoms 12 months later. This partial mediation aligns with prior research [39,40,41,42,43,44,45,46,47], which suggests that parental depressive symptoms may influence adolescents’ emotion regulation. A substantial body of evidence suggests that early adverse experiences often shape a self-concept marked by insecurity, inadequacy, and shame—core emotional patterns that underpin vulnerability to various forms of psychopathology [51,80]. Parental depressive symptoms also have a significant direct effect on adolescents’ depressive symptoms across time, underscoring the persistent and intergenerational nature of depressive disorders. Parental depression is considered to influence adolescent depression due to both the heritability of depression and the stressor events that it may potentially lead to [81]. Moreover, an internal climate marked by rumination and self-criticism is frequently linked to the persistence of depressive symptoms [82,83,84]. These results reinforce the theoretical framework proposing self-compassion as an adaptive process [29,45]. The longitudinal mediating role of self-compassion between parental depressive symptoms and their adolescent offspring is what provides a novel contribution. In other words, although the mediation is statistically partial, self-compassion has a strong effect in depressive symptoms, suggesting a meaningful contribution and substantial explanatory pathway linking parental depressive symptoms to adolescent outcomes after 12 months. These results underscore the promising role of self-compassion as a key target for preventive interventions aiming to disrupt the intergenerational continuity of depressive symptomatology.

Both age and baseline adolescents’ depressive symptoms were found to exert significant indirect effects on depressive symptoms at T2 only through self-compassion. Younger adolescents are more likely to exhibit elevated self-compassion levels, also suggested by descriptive data, which, in turn, seems to diminish the influence on depressive symptoms over time. This aligns with existing literature that suggests depressive symptoms are not static over time but can be shaped by developmental factors [13,58]. The small but significant indirect contributions of age and baseline depressive symptoms further suggest that developmental and prior symptom factors impact adolescent outcomes through self-compassion. Interestingly, and contrary to the hypothesis based on previous literature, gender did not significantly affect depressive symptoms or self-compassion, which opposes some evidence indicating gender disparities in how depression begins and evolves during adolescence [7,54,55,56,57], and congruent with other findings on self-compassion [39]. These findings suggest that self-compassion may serve as a convergence point between individual and contextual factors, modulating the pattern of depressive symptoms as adolescents grow.

4.3. Strengths and Clinical and Practical Implications

This study presents several strengths, practical implications, and limitations that warrant consideration and offer directions for future research. This study employs a longitudinal design, enabling the analysis of the temporal relationships between variables, and, as mentioned, to test the suggested theoretical hypothesis, this design proved to be the indicated one. Additionally, a strength of this study lies in its dyadic design, incorporating paired data from adolescents and their respective parents. This approach enables a nuanced examination of the intergenerational transmission of depressive symptoms. The results offer meaningful implications for clinical practice, school-based mental health programs, and family-focused interventions. As highlighted elsewhere, modification of causal factors is the aim of intervention, as this may result in a change in the outcome variable [85]. The capacity to treat oneself with kindness and understanding in moments of suffering or perceived failure is considered a flexible psychological process that can be enhanced through structured training and practice [86,87,88,89]. Psychological interventions should aim to promote self-compassionate attitudes. Incorporating self-compassion training into depression disorder onset prevention as early as possible may be particularly beneficial. Such interventions could also be integrated into mental health programs or school-based prevention initiatives. Even in clinical psychology settings, particularly with depressed adolescents, these results highlight the critical need to inquire about parents’ mental illness—even when not spontaneously mentioned. Addressing parental depressive symptoms as a relevant contextual factor may enhance case conceptualisation and treatment planning. A family-centred approach to treating adolescent depression should be considered. Family therapy and psychoeducation for parents may be effective in improving family dynamics and reducing the transmission of depressive symptoms from parents to children [24].

4.4. Limitations and Directions for Future Research

It is important to acknowledge the study’s limitations, which offer valuable guidance for future investigations. Data were collected prior to the COVID-19 pandemic, which may limit comparability with more recent cohorts. However, the psychological mechanisms examined in this study are developmentally grounded and remain relevant across different historical periods. The one-year longitudinal study may limit the efficacy in determining causality from the results. It also implies that extending the findings to other age groups must be conducted with caution. The sample consisted solely of Portuguese adolescents and their parents, which may limit the generalisation to other cultural contexts. The study relied on self-report measures for both adolescents and parents, which can introduce biases, such as social desirability or inaccuracies in reporting depressive symptoms. While the predictors showed meaningful effects, future research should examine a broader range of factors to better understand their relationship with depressive symptoms and clinically diagnosed depressive disorders. Based on these limitations, future longitudinal studies may benefit from replication on other cultures and from including more follow-up assessment moments to cover time, change, or stability of causality models. Subclinical, depressed, and non-depressed adolescent groups should be compared in larger follow-up assessments to differentiate causal factors from correlated symptoms. The significant indirect effect of age suggests that older adolescents tend to exhibit lower levels of self-compassion, highlighting the need for further investigation into how developmental stage interacts with psychological processes such as self-compassion. Although gender did not show significant effects in the present model, some prior evidence indicates differences between genders in depressive symptoms. Subsequent studies may examine whether gender differences emerge in other cultural contexts, age ranges, or different methodological designs. Future studies should investigate, in a longitudinal design, other variables, including parenting practices, attachment styles, and peer support, given that self-compassion is fundamentally linked to secure attachment and early nurturing caregiving experiences. It should also be pondered whether family-based interventions can improve both parental and adolescent mental health outcomes, providing valuable insights for clinical practice and preventive research. Adapting the interventions to the Portuguese cultural and healthcare context will enhance their effectiveness and accessibility. Additionally, cross-cultural comparative studies involving Portugal and other countries would provide valuable insights into universal versus culture-specific mechanisms underlying self-compassion and its impact on adolescent mental health.

5. Conclusions

For several decades, research on factors linked to depression and on depression prevention has been flourishing. Some of those efforts were directed to early intervention. Despite those remarkable efforts, depression manifestations are still a serious public health concern [1,2,90]. This study may offer valuable insights to guide mental health policies and early intervention strategies specifically designed to address the distinct needs of young people. Improved understanding of the intergenerational transmission of depressive disorders, exploring how parental depressive symptoms affect their adolescent offspring’s depressive symptoms, represents a meaningful advancement toward understanding mechanisms that may contribute to the prevention of depressive disorders. The identification of nurturing self-compassion, as it is a relevant mediating process in the association between parental and adolescent depressive symptoms, highlights its promise as a focus for early intervention strategies designed to alleviate individual suffering and promoting emotional well-being across development. The continuous study of how self-compassion may explain the effect of personal and environmental factors on depressive symptoms, as on other mental health outcomes, may be a valuable research avenue for prevention and treatment studies.