Exposure to caregiver conflict is widely characterized as stressful and associated with poor adolescent mental health (Harold & Sellers, 2018). Like much of the psychology literature (Muthukrishna et al., 2020), the bulk of the evidence for this negative association has been generated by studies with young people from Western, educated, industrialized, rich, and democratic (i.e., WEIRD) countries, especially the USA (Harold & Sellers, 2018; van Eldik et al., 2020). The resilience literature (i.e., accounts of what protects adolescents who are exposed to significant stress from negative outcomes, like mental illness) is similarly grounded in studies with adolescents from WEIRD countries (Masten, 2014). Growing up in a country where western norms and values are not dominant and the majority of people are challenged by resource constraints, poor technological progress, restricted education opportunities, and disabling politics—hereafter referred to as a less-WEIRD country—is likely to impact adolescent resilience. Further, because the resilience literature is characterized by only nascent attention to the impacts of situational or cultural context (Ungar & Theron, 2020), there is poor understanding of the differential resilience-enabling value that specific protective factors might hold for specific adolescent populations (Ungar, 2017). This article, which draws on cross-sectional data generated by a sample of adolescents who were part of the Resilient Youth in Stressed Environments (RYSE) study in Canada (a WEIRD country) and South Africa (a less-WEIRD country), addresses these gaps. It interrogates similarities and differences in the resilience of these adolescents, all of whom reported exposure to caregiver conflict. In so doing, it encourages mental health practitioner attention to the variability of resilience processes across contexts (Ungar, 2019a) and amplifies calls to support healthy development in differentiated ways (Shonkoff, 2020).

While there is no absolute definition for resilience, many studies explain resilience as the capacity of a system, like an adolescent or community, to function normatively despite exposure to significant stressors (Masten, 2001, 2014; Masten & Motti-Stefanidi, 2020; Rutter, 2013). This capacity is facilitated by several resilience-enabling processes, including attachment, problem-solving, self-regulation, and meaning making (Masten & Wright, 2010). Resilience-enabling processes draw on personal resources (e.g., comforting beliefs, meaning making), as well as social-ecological ones (e.g., home routines, caring mentors, quality education, or mental health services), and their interactions (Masten & Cicchetti, 2016; Ungar, 2011). The more resources that a young person has access to, the better their chances should be of adjusting positively to risk exposure (Bondi, Pepler, Motz, & Andrews, 2020; Hobfoll, 2001; Ungar, 2019a). However, the protective value of these interacting resources is not necessarily fixed. The type of risk exposure and contextual and cultural dynamics influence which resources are more, or less, facilitative of positive outcomes (such as mental health) in the face of significant stress (Panter-Brick, 2015; Ungar, 2017).

Caregiver Conflict and Adolescent Mental Health Resilience

Be it overt (e.g., physical hostility) or covert (e.g., strained silences), caregiver conflict potentiates destructive spillover effects for family functioning and parenting (Bradford, Vaughn, & Barber, 2008). For instance, a study with 601 American families showed that reports of parental conflict were followed by subsequent reports of father—adolescent and mother—adolescent conflict (Smith, Nelson, & Adelson, 2019). Typically, distressed caregiver relationships and their spillover effects undermine adolescents’ sense of emotional security and, in so doing, precipitate internalizing and externalizing mental health difficulties (Cummings & Davies, 2002; Davies & Cummings, 1994). There is some evidence that girls are more vulnerable than boys to caregiver conflict and its spillover effects (e.g., Chan, Kelly, & Toumbourou, 2013).

The negative relationship between caregiver conflict and adolescent mental health has been reported for young people from less-WEIRD countries too (Harold & Sellers, 2018). For instance, Bradford et al. (2004) reported spillover effects of parental conflict for parent and adolescent wellbeing in samples of adolescents from Bangladesh, Bosnia, China, Colombia, India, Palestine, and South Africa. Although there is robust evidence for the negative mental health effects of caregiver discord (van Eldik et al., 2020), it is less clear what enables the resilience of adolescents who are exposed to caregiver discord.

Various protective factors have been found to support adolescent resilience to the negative mental health effects of caregiver conflict (Formoso, Gonzales, & Aiken, 2000; Van Voorhees et al., 2008). Consistent with a systemic approach to resilience, adolescent resilience to caregiver discord has been attributed to personal strengths (e.g., the ability to regulate emotion and behavior, positive temperament, comforting beliefs or positive meaning-making; Harold & Sellers, 2018; Kwok, Gu, Synchaisuksawat, & Wong, 2020) and social-ecological resources (e.g., household routine, being emotionally supported by an adult outside of the home, positive connections to siblings or peers; Gershoff, Aber, Raver, & Lennon, 2007; Harold & Sellers, 2018; van Dijk, van der Valk, Deković, & Branje, 2020). Resilience is typically facilitated by an interplay of these resources (Harold & Sellers, 2018). Two of these resources, i.e., comforting beliefs and household routine, are of interest to the current study. Even though these two resources are prominent in accounts of resilience (Masten & Motti-Stefanidi, 2020; Ungar, 2019a), neither resource has been emphasized in studies of adolescent resilience to caregiver conflict.

Comforting Beliefs

Making positive meaning of the threat that caregiver conflict poses to the family and to them personally (e.g., interpreting that threat as minimal or manageable) is strongly associated with an adolescent’s ability to adjust well to caregiver discord (Cummings & Davies, 2002; El-Sheikh & Harger, 2001; Harold & Sellers, 2018; Lucas-Thompson, Lunkenheimer, & Granger, 2017; Simpson, 2019). As explained by Masten (2014), belief systems that encourage comforting viewpoints support the capacity to appraise situations beyond one’s personal control as minimally threatening or manageable. A systematic review of the 2009–2017 South African studies of child and youth resilience found that comforting beliefs—mostly linked to religious teachings or spirituality—facilitated young people’s adjustment to various adversities, including marital discord and family disruptions (Van Breda & Theron, 2018). For instance, the belief that hardship (such as the loss of a parent) was part of a higher being’s plan that would ultimately yield something positive was associated with the resilience of South African families with varied religious affiliations (Greeff & Joubert, 2007). A study with culturally diverse adolescents who were living in Melbourne, and whose origins were WEIRD (e.g., Anglo-Celtic) or less-WEIRD (e.g., Croatian or Indian), reported similar links between spiritual beliefs, optimistic meaning making, and a sense of life ultimately having purpose irrespective of current hardship (Raftopoulos & Bates, 2011). A more recent study with Czech adolescents from nonreligious families also reported that spiritual beliefs, such as the belief in a benevolent higher power, brought comfort, and reassurance to adolescents (Klůzová Kráčmarová, Dutková, & Tavel, 2019). Interestingly, a study that included 21,173 Canadian adolescents confirmed the resilience-enabling effects of spirituality but showed that the strongest effects related to adolescents’ belief that life has purpose and is enjoyable rather than to beliefs in spiritual beings (Michaelson et al., 2019).

Household Routine

Families, particularly parents, enable child and adolescent resilience when they facilitate routine (Masten, 2014; Masten & Motti-Stefanidi, 2020). Domestic routines, such as regularly eating meals together, advance healthy development when they offer opportunities for young people to experience belonging, acquire values and norms, improve self-regulation, and perceive their world as predictable (Barton et al., 2019; Fiese, Foley, & Spagnola, 2006; Thakur & Cohen, 2020). This protective value is heightened when household routines are dependable, fit family resources, align meaningfully to family values and goals, and do not stimulate conflict (Weisner, 2010). In less-WEIRD contexts, like South Africa, the consistency and value of routines can be obstructed by economic challenges that impede resources (e.g., availability of food), space (e.g., a family sharing a one or two-room dwelling), and caregiver energy to interact positively with their offspring (De Goede, 2018). Even so, there is evidence that when routines can be maintained, adolescents who face resource constraints are better supported to adjust to risk-filled lives (Ferguson, Cassells, MacAllister, & Evans, 2013). In WEIRD contexts, household routines have also been shown to be protective for adolescents. For example, a study with 200 mother-child dyads who lived in impoverished circumstances in the USA found protective effects of household routines on adolescents’ problem behavior as well as school engagement and achievement (Taylor & Lopez, 2005). Similarly, a study with 943 American adolescents showed that family routines protected young people with maltreatment experiences against post-traumatic stress and depression (Thakur & Cohen, 2020).

The Present Study

Using a moderated moderation model, the aim of this study was to explore the potential differential interplay of household routines and comforting beliefs on the depression of adolescents with experiences of severe caregiver conflict in a WEIRD and less-WEIRD context. This approach makes it possible to study the effect of four different resource combinations on the relationship between context and depression: having both comforting beliefs and household routines, having either comforting beliefs or household routines, and having neither of these resources. Hence, a moderated moderation model gives insight into which of these resource combinations might have greater or lesser protective effects for depression among adolescents with experiences of severe caregiver conflict from South Africa vs. Canada.

The following hypotheses informed the analysis:

  1. 1.

    Context-unspecific: Since the literature suggests that being well-resourced should generally lead to positive outcomes in stressful environments (Hobfoll, 2001; Ungar, 2019a), it was hypothesized that the level of depression is dependent on the number of available resources in both contexts. That is, both samples will report the lowest level of depression when comforting beliefs and daily home routines are present, a medium level of depression when either comforting belief or household routine is present, and the highest level of depression when both resources are absent. This hypothesis presumes that the two resources have an equally protective impact in both contexts.

  2. 2.

    Context-specific: Given that contextual and cultural factors can influence which resources are more, or less, facilitative of positive outcomes (Panter-Brick, 2015; Ungar, 2017), it was hypothesized that comforting beliefs and daily home routines would hold variable protective effects for adolescents in the WEIRD (Canada) and less-WEIRD (South Africa) context. In keeping with exploratory research, there were no a priori assumptions about these variable effects.

Method

Procedure

The data originate from the Resilient Youth in Stressed Environments (RYSE) project conducted in 2018 with adolescents (age range 14–24 years) from Canada (CA) and South Africa (SA). This age range fits with more recent understandings that adolescence needs to be understood as a developmental period that is longer than originally suggested by developmental theorists (i.e., continues until the age of 24; Sawyer, Azzopardi, Wickremarathne, & Patton, 2018). The respective Institutional Review Boards of the universities where the principal investigators are affiliated in Canada (Health Sciences Research Ethics Board, Dalhousie University, #2017-4321) and South Africa (Faculty of Health Sciences Research Ethics Committee, University of Pretoria, #UP17/05/01) approved the study. In both countries, the sites were communities that are dependent on the oil and gas industry for their livelihood. At the time of the study, both sites were challenged by the economic downturn in the oil and gas industry and associated risks of unemployment and family strain. Additionally, as in much of SA (Canham, 2018; Kaminer, Eagle, & Crawford-Browne, 2018), the SA community was characterized by structural disadvantage and social disorder (e.g., violent protests relating to poor service delivery and local government corruption). In keeping with the structural disadvantage, most of the housing in the community is low-cost and cramped.

At both sites, community-based advisory panels (CAP) consisting of local adolescents and adults were established. They supported the research teams to recruit eligible participants. In SA, the CAP used flyers to advertise the study in the community. In CA, researchers advertised the study by word-of-mouth and used social media. Participants had to be residents of the respective research communities, between 14 and 24 years old, and proficient in English. Prior to participation, participants signed an informed consent form. Canadian adolescents received $25 cash for their participation and South African adolescents a ZAR150 (i.e., about $15) shopping voucher. The incentive amounts were advised by the CAPs.

Data were collected via paper-pencil surveys. In SA, the survey was administered by trained research assistants to small groups of participants as recommended by the South African CAP and previous research (Van Rensburg, Theron, & Ungar, 2019). Items were read aloud by a research assistant. Canadian participants completed the survey independently.

Participants

Canada

Of the Canadian sample (N = 495), n = 152 reported that they experienced a home with fights and severe relationship problems between their parents/caregivers and provided complete data. This sub-sample was characterized by 65.8% young women, an age range of 14–24 years (mean age = 18.81, SD = 2.95), 58.6% school attending, and 82.9% White, 11.8% Indigenous, and 5.3% other race/ethnicities. Further, 42% lived without parents/caregivers, 26% with one parent/caregiver, and 32% with both parents/caregivers. Also, 30.9% reported no home routines and 27.6% reported no comforting beliefs.

South Africa

Of the South African sample (N = 599), n = 150 reported that they experienced a home with fights and severe relationship problems between their parents/caregivers and provided complete data. This sub-sample was characterized by 65.3% young women, an age range of 14–24 years (mean age = 18.27, SD = 2.88), 77.3% school attending, and 92% Black, 6% White, and 2% other race/ethnicities. Further, 21% lived without parents/caregivers, 44% with one parent/caregiver, and 29 with both parents/caregivers. Also, 22% reported no home routines and 11.3% reported no comforting beliefs.

Measures

Caregiver Conflict

Participants reported on a dichotomous item (0 = No, 1 = Yes) if they have lived in a home with fights or severe relationship problems between their parents/parent-figures/caregivers. This variable was not part of the main analysis but served to identify the participants for the main analysis and was used for descriptive purposes.

Context

Participants were grouped according to their study site (1 = Canada, 2 = South Africa).

Depression

The Beck Depression Inventory-II (Beck, Steer, & Brown, 1996) was used to indicate participants’ level of depression. Twenty-one items, each consisting of four statements that indicate an increasing severity of a symptom, ask about the experience of symptoms of depression over the last 2 weeks. Sample item: “0 = I do not feel sad; 1 = I feel sad much of the time, 2 = I am sad all the time; 3 = I am so sad or unhappy that I can’t stand it.” A higher sum score indicates a higher level of depression (range = 0–63). Reliability for the Canadian sample was Ω = 0.96 [0.96, 0.97], and for the SA sample, Ω = 0.92 [0.90, 0.94].

Regular Daily Routine

Participants reported on a dichotomous item (0 = No, 1 = Yes) if they had predictable home routines like regular meals and a regular bedtime during the first 18 years of their life.

Comforting Beliefs

Participants reported on a dichotomous item (0 = No, 1 = Yes) if they had beliefs that gave them comfort during the first 18 years of their life.

Covariates

Sex (female, male) and age (in years) were included. The analysis further controlled for the potential experience of divorce or permanent separation of the parents/parent-figures/caregivers (0 = No, 1 = Yes) which could have ended the risk. Similarly, it controlled for participants living alone, with one of the parents, or with both parents. This was dummy coded: living alone vs. living with both parents (parent home: no vs. both) and living with one parent vs. living with both parents (parent home: one vs. both). The analysis also controlled for the following additional factors that influence level of depression. The 28-item Child and Youth Resilience Measure (CYRM-28; Ungar & Liebenberg, 2011) was included to control for the level of currently available, cross-culturally important resources associated with the resilience of adolescents (e.g., personal skills, peer support, psychological caregiving, community resources). The CYRM-28 uses a five-point Likert scale (1 = Not at all, 5 = A lot) and higher sum scores indicate more available resources. Reliability for the Canadian sample was Ω = 0.94 [0.93, 0.96], and for the SA sample, Ω = 0.89 [0.87, 0.92]. The Child Post-Traumatic Stress—Reaction Index (CPTS-RI; Pynoos et al., 1987) was included to control for the self-reported degree of posttraumatic stress over the last month. The CPTS-RI uses 20 items and a five-point Likert scale (1 = Never, 5 = Most of the time) so that a higher sum score indicates higher posttraumatic stress. Sample item: “Do you wish you had no feelings?”. Reliability for the Canadian sample was Ω = 0.92 [0.90, 0.94], and for the SA sample, Ω = 0.88 [0.85, 0.91]. Finally, a short form of the Highly Sensitive Child scale (Pluess et al., 2018; Pluess, personal communication) was included to control for how strongly the participants respond to negative and positive experiences. The applied short form consists of six statements (e.g., “I notice when small things have changed in my environment”) and participants are asked to rate their agreement with each statement on a seven-point Likert scale (1 = Not at all, 7 = Extremely). Higher sum scores indicate a higher sensitivity. Reliability for the Canadian sample was Ω = 0.70 [0.62, 0.77], and for the SA sample, Ω = 0.52 [0.40, 0.64].

Statistical Analysis

Model Estimation

A robust moderated moderation model (see Fig. 1 for the conceptual model) using the PROCESS Macro v3.4 (Hayes, 2017) for IBM SPSS v25 (IBM Corp, 2017) was estimated exclusively for adolescents with experiences of severe caregiver conflict from both countries. A moderated moderation model is a regression which can be used to test if the effect of a focal antecedent (i.e., predictor) on an outcome is dependent on a moderator and if the effect of the moderator itself is dependent on a second moderator (also known as three-way interaction; Hayes, 2017). In the present model, the predictor was binary and comprised the following two groups: (1) Canadian adolescents with experiences of fights or severe relationship problems at home and (2) South African adolescents with experiences of fights or severe relationship problems at home. The two moderators, i.e., regular daily routine and comforting beliefs, were binary. The continuous outcome was depression. The identified covariates were also included into the model. To account for potential issues with sample size, outliers, normality, and homoscedasticity, a robust regression using bootstrapping (95% confidence intervals with 50,000 bootstrap samples) was applied (Field, 2018). Using bootstrapped confidence intervals (bCI), a significant effect is estimated when the interval does not contain zero.

Fig. 1
figure 1

Conceptual model of the study. PTS, posttraumatic stress. Resources: as assessed by the Child-Youth Resilience Measure. Sensitivity: as assessed by the Highly Sensitive Child scale

Full size image

Model Inference

Mathematically, a moderated moderation model consists of main effects, three two-way interactions, and one three-way interaction. Except for the three-way interaction, these have to be interpreted as conditional effects (for a detailed explanation see Hayes, 2017). Hence, only the three-way interaction can be interpreted in a way that shows how the effect of a focal predictor on an outcome is dependent on different levels of the moderators. PROCESS provides simple slope tests that indicate if the two risk groups differ significantly on a specific combination of the moderators (e.g., the presence of routine and comforting beliefs). Furthermore, contrasts were used to indicate if potential differences between the two groups, i.e., the simple slopes, were significantly different across the different moderator combinations (e.g., presence versus absence of routine and comforting beliefs).

Results

Descriptive Analysis

The overall studied CA RYSE sample reported significantly more experience of caregiver conflict (41%) than the overall SA sample (31%) (χ2 = 12.00, p < .01). An ANOVA was used to test if adolescents with and without experiences of caregiver conflict from the studied CA and SA communities differ in their self-reported symptoms of depression. A significant result (F = 24.95; df = 3, p < .01) indicated significant differences. A post hoc test (Tukey HSD) revealed that CA and SA adolescents with experiences of caregiver conflict did not differ significantly in their reported symptoms of depression (Mcaregiver conflict CA = 21.30, SD = 14.17; Mcaregiver conflict SA = 19.13, SD = 11.18; p = .21), but they had significantly higher depression scores than CA and SA adolescents without experiences of caregiver conflict (Mno caregiver conflict CA = 11.49, SD = 11.22; Mno caregiver conflict SA = 14.02, SD = 8.92; p = .11).

As per the focus of the current article, Table 1 shows the correlations between the study variables for those participants who reported severe caregiver conflicts. While there was a significant negative correlation between regular daily routines and depression and a non-significant correlation between comforting beliefs and depression in the CA sample, the SA sample showed the opposite pattern. Further, age was negatively associated with depression only in SA and sensitivity was positively associated with depression only in CA. Both countries showed that higher symptoms of depression are associated with female sex, higher posttraumatic stress symptoms, and fewer resilience-enabling resources. Also, while the samples did not differ in their level of depression (see ANOVA results above), posttraumatic symptoms (t = 1.05, p = .30), and sensitivity (t = 1.63, p = .10), they showed a significant difference in their reporting of available resilience-enabling resources (t = − 3.92, p < .01) with the CA sample (M = 101.95, SD = 18.56) reporting fewer resources than the SA sample (M = 109.65, SD = 15.41).

Table 1 Intercorrelations for study variables disaggregated by country
Full size table

The Interplay of Context, Home Routine, and Comforting Beliefs on Depression in Adolescents with Experiences of Severe Caregiver Conflict

The full robust moderated moderation model explained 58.3% of the variance in depression (F(15, 276) = 25.75, p < .01). As can be seen in Table 2, the three-way interaction testing the joint interplay of context, home routine, and comforting beliefs on depression was significant (ΔR2 = .01, p = .02). The post hoc analysis of the power of the three-way interaction showed a sufficient power of .81 (at an effect size of f = .25, α = .05, N = 302, number of conditions = 8, number of covariates = 6; Faul, Erdfelder, Lang, & Buchner, 2007). Simple slope tests revealed that the two samples showed significant differences in their depression when a home routine was present and comforting beliefs were absent (effect = .58, t = 2.29, p = .02). As can be seen in Fig. 2, the SA sample showed a significantly higher level of depression (Mdepression SA = 26.01) than the CA sample (Mdepression CA = 18.27). All other moderator combinations (present home routine and comforting beliefs [Mdepression CA = 20.50, Mdepression SA = 19.79, effect = − .05, t = − 0.45, p = .66]; absent home routine and comforting beliefs [Mdepression CA = 24.68, Mdepression SA = 19.66, effect = − .43, t = − 1.32, p = .19]; absent home routine and present comforting beliefs [Mdepression CA = 18.90, Mdepression SA = 19.64, effect = .02, t = 0.10, p = .92]) did not show significant differences between the samples. Even though not significant, Fig. 2 shows a trend that the absence of both resources is worse for the depression of the CA sample than for the SA sample.

Table 2 Results of the robust moderated moderation regression
Full size table
Fig. 2
figure 2

Three-way interaction graphs: effects of all combinations of the studied resources on adolescent depression for each studied context. +, present home routine/comforting beliefs; -, absent home routine/comforting beliefs

Full size image

Table 3 shows the results of the contrasts. Most notably, the non-significant contrast A shows that both samples report similar depression scores when comforting beliefs were present regardless of the absence or presence of a home routine. Hence, both samples do not differ if comforting beliefs are present. In comparison, the significant contrast B shows that the samples report significantly different depression scores when home routines were present and the presence of comforting beliefs varied. Although both samples show no significant differences when both resources were present, adolescents from SA report significantly higher depression scores when only home routines were present. Also, significantly different patterns were found when comforting beliefs were absent and home routines varied (contrast D). When only home routines were present, then the depression of the SA sample was higher than that of the CA sample. No significant differences were found between the samples when both resources were absent. The other contrasts were not significantly different.

Table 3 Contrasts based on standardized model indicators
Full size table

The simple slopes were also investigated to explore sample-specific effects of the moderators. For CA, the following pattern emerged: both resources being absent were always significantly worse for depression (Mdepression = 24.68) compared with when at least a daily home routine (Mdepression = 18.27, effect = − .53, t = − 2.37, p = .02) or comforting beliefs (Mdepression = 18.90, effect = − .46, t = − 2.14, p = .03) were present. Also, there was no significant difference for the CA sample when both resources were present (Mdepression = 20.49), or just daily home routine (Mdepression = 18.27, effect = .17, t = 1.09, p = .28), or comforting beliefs (Mdepression = 18.90, effect = .09, t = 0.62, p = .54).

For the SA sample, Fig. 2 indicates that having a daily home routine and no comforting beliefs is worse for depression than any other combination of the resources. While having a daily home routine and no comforting beliefs (Mdepression = 26.02) showed a significantly higher depression compared with having both resources (Mdepression = 19.79, effect = − .46, t = − 2.06, p = .04), depression scores were not significantly different when both resources were absent (Mdepression = 19.66, effect = .48, t = 1.39, p = .17). Also, no significant differences were found between having only comforting beliefs (Mdepression = 19.64) and having neither of the two resources (Mdepression = 19.66, effect = − .01, t = − 0.02, p = .98) or having both (Mdepression = 19.79, effect = .03, t = 0.17, p = .86).

Discussion

The purpose of this study was to explore the potential differential interplay of household routines and comforting beliefs on the depression of adolescents with experiences of severe caregiver conflict in a WEIRD (Canada) and less-WEIRD (South Africa) context. To do so, a moderated moderation model was used. The results showed that when both resources and only comforting beliefs were present, self-reported symptoms of depression were similarly low for adolescent participants in Canada and South Africa. However, relative to the country context, having neither resource nor only a home routine, protected adolescent participants differently. Given this, the assumption that household routines and comforting beliefs (both of which are widely reported as resilience-enabling; Masten, 2014; Masten & Motti-Stefanidi, 2020; Ungar, 2019a) have an equally protective effect for adolescents challenged by caregiver conflict in WEIRD and less-WEIRD contexts needs to be questioned. This novel finding fits with calls for greater attention to the variability of factors supporting positive outcomes and to designing mental health interventions that are responsive to such variability (Shonkoff, 2020; Ungar, 2019b). In other words, even though the experiences of severe caregiver conflict have similar effects on depression levels for adolescents in the South African and Canadian context, optimal interventions would respond to context-specific protective effects, especially when choosing which resilience resources to prioritize.

Similarities across Contexts

The descriptive analyses mirrored traditional empirical findings on how depressive symptoms are related to gender, comorbid disorders, and level of available resilience-enabling resources in both contexts: self-reported symptoms of depression were higher for young women (e.g., Piccinelli & Wilkinson, 2000), positively associated with posttraumatic stress symptoms (Kilpatrick et al., 2003), and negatively associated with the level of available resilience-enabling resources (Dumont & Provost, 1999).

In both contexts, self-reported depression scores were similar when comforting beliefs were present regardless of the absence/presence of a home routine. To be clearer, having both resources seemed to be as protective as having only comforting beliefs. This suggests that so long as adolescents challenged by caregiver conflict have comforting beliefs, home routines do not seem to matter for depression outcomes. The protective value of comforting beliefs fits with the well-documented understanding that adolescent capacity to make positive meaning of the threat that caregiver conflict poses is strongly protective (Cummings & Davies, 2002; El-Sheikh & Harger, 2001; Harold & Sellers, 2018; Lucas-Thompson et al., 2017; Simpson, 2019). This result has significant implications for intervention design, more particularly when practitioners have limited resources to facilitate intervention planning and operationalization. Since focusing only on comforting beliefs will likely be as effective as interventions that target home routines and comforting beliefs together, intervention requirements (including finances and time) will be reduced, thereby potentially advancing intervention efficiency. Follow-up qualitative work would be helpful to better understand the content of such comforting beliefs, and how similar or different the content might be for adolescents in WEIRD versus less-WEIRD contexts. Better understanding of the content matters for intervention design. For example, should the comforting beliefs be spiritual in nature (as is widely reported in the South African literature; Van Breda & Theron, 2018), then mental health practitioners might be more inclined to explore spiritual beliefs with clients (Whitley & Jarvis, 2015). Similarly, although there is robust evidence that indigenous Canadian adolescents have comforting beliefs that reflect indigenous teachings (Rowhani & Hatala, 2017), it is less clear what informs the comforting beliefs of other Canadian adolescents (Russell, Liebenberg, & Ungar, 2015).

Differences Across Contexts

While the descriptive analyses would suggest independent, context-specific effects of the studied resources (regular home routines were negatively related to adolescent depression only in Canada, and comforting beliefs were negatively related only in South Africa), the moderation analysis showed contextualized interactions between them. For Canadian participants, self-reported depression symptoms were higher than those of the South African sample when both resources were absent and overall highest compared with any other combination of the resources. The results, therefore, seem to indicate that it is worse for Canadian adolescents to have neither home routine nor comforting beliefs than having either one. This is partly in line with literature suggesting that having some resources is generally associated with more positive outcomes in stressful environments but that having more resources is better (Bondi et al., 2020; Hobfoll, 2001; Ungar, 2019a). Further, the fact that Canadian participants’ self-reported depression levels were similar for having either resource or both raises questions about cumulative protective effects for the studied resources. Put differently, even though comforting beliefs or home routines seem to have similar protective mental health effects for Canadian participants confronted with severe caregiver conflict, there appears to be no additive value when both are present. Future studies should investigate what determines whether the protective effects of resources do, or do not, accumulate. Such insights can increase the efficiency of resilience interventions, particularly if this allows for choice between equally protective resources that do not show cumulative protective effects.

The self-reported depression symptoms of the South African participants were significantly higher than those of the Canadian participants and overall highest when they reported a home routine without comforting beliefs. Put differently, it was least protective for South African adolescents to have a home routine in the absence of comforting beliefs than any other combination of these two resources. A possible reason for this could be the disconnect between a South African household that observes meal or bedtime routines and an immediate and extended community that is repeatedly disrupted by service delivery protests and other forms of chaos, as was typical in the South African RYSE community and others (Canham, 2018; Kaminer et al., 2018). In contrast, comforting beliefs would at least support adolescents to make positive meaning of their caregivers’ conflict and of community upheaval. This theorizing fits with South African resilience studies’ attention to the protective value of comforting beliefs for adolescents from disadvantaged and violent contexts, but silence about household routines (Van Breda & Theron, 2018). It is also possible that the cramped housing that was typical of the SA RYSE site (as in other less-WEIRD contexts) meant that household routines were less conducive to positive interaction and so less protective (De Goede, 2018; Weisner, 2010). Possibly, household routines in cramped housing circumstances translated into (greater) adolescent exposure to caregiver conflict. Further, African adolescents are traditionally raised to defer to their elders (Mhlongo, 2019), and so it is possible that if caregiver conflict was not modulated during household routines (such as mealtimes), adolescents would have little recourse but to tolerate stressful caregiver interactions. In households with more individualist or other Western values (e.g., Canada), it is possible that adolescents could voice their discomfort. Lastly, because caregiver conflict is associated with spillover effects such as poorer parenting and adolescent-caregiver conflict (Bradford et al., 2008; Smith et al., 2019), it is plausible that home routines exposed adolescents to poor parenting practices or conflict with their caregiver. If this were the case, then interventions for South African adolescents exposed to caregiver conflict might need to include a parenting component that supports adolescents’ parents/caregivers to be enabling parents. Taking care to not only intervene at the level of the adolescent, fits with systemic understandings of resilience that encourage mental health and other practitioners to enable the resilience of caregivers and other systemic stakeholders (Masten & Motti-Stefanidi, 2020). Again, it would be important to follow up qualitatively to better understand what it was about home routines in the South African context that jeopardized adolescent mental health in the absence of comforting beliefs.

Overall, the above suggests that in the South African context, the facilitation of comforting beliefs is cardinal to interventions seeking to support adolescent mental health in the face of caregiver conflict. Further, it could be counter-productive to support the mental health of African adolescents challenged by caregiver conflict and living in a disadvantaged South African community by facilitating household routines unless comforting beliefs were simultaneously encouraged. While this suggestion fits with the result that having both comforting beliefs and routine were associated with similarly low levels of depression in the South African and Canadian sample, it is possible that an intervention targeting both resources might not be the best way forward in the South African context. It might be that comforting beliefs protect adolescents not only from the negative effects of severe caregiver conflicts but also from the maladaptive effects of home routines in a context where homes are cramped and the wider society is chaotic. If that were the case, an intervention that prioritized comforting beliefs would be a contextually responsive one. This line of thinking gives further evidence for the need to closely consider the context and meaning of different resources in a given context before advocating a set of protective resources (Panter-Brick, 2015; Ungar, 2017). Further, the fact that a combination of absent comforting beliefs and present home routines can be protective for Canadian adolescents but detrimental for South African adolescents points to complex interactions between the individual, their close social network, and the socio-cultural environment. That complexity highlights the criticality of understanding which factors best protect which adolescents for what outcomes and in what contexts (Ungar, 2019b).

The results, therefore, speak more in favor of the context-specific hypothesis than context-unspecific hypothesis, since only the resource combination of having comforting beliefs and no home routine can be interpreted as being similarly protective in both contexts.

Limitations and Future Directions

The methodological advantages of the current study are the similar sample characteristics and sizes, the use of a robust regression by applying bootstrapping, and the sufficient power of the three-way interaction. However, cross-sectional data significantly limits the cause-effect interpretations of the studied variables. While depression was chosen as the outcome, it could also influence comforting beliefs, posttraumatic stress symptoms, social resources, or sensitivity over time. Hence, longitudinal studies are needed that take potential feedback loops between the studied variables into account by applying a systems approach. Further, recent empirical studies of systemic resilience indicate that resilience is best conceptualized by a network of multiple interacting resources (Höltge et al., 2020). This points to the need for future studies of adolescent resilience to caregiver conflict to investigate higher-order interactions that include more interacting resources and methodological advances to interpret models with four-way interactions and higher. Ideally, such future studies also need to consider cumulative protective effects.

The current study did not investigate parenting practices or adolescent-caregiver relationships. Both are reported to degenerate in the face of severe and/or chronic caregiver conflict (Bradford et al., 2008; Smith et al., 2019) and, as discussed, could therefore have played a role in the protective effects of family routine. Similarly, other factors (e.g., household stability versus household chaos) could have influenced how adolescents experienced household routine (Barton et al., 2019). Understandings of how family routines enable or constrain adolescent resilience to caregiver conflict would be improved by the inclusion of these factors in future studies.

As noted in the discussion of the results, their interpretation was limited by the absence of adolescents’ explanations of these results. To better understand why and how home routines and comforting beliefs support the resilience of adolescents in WEIRD and less-WEIRD contexts to caregiver conflict will require the inclusion of qualitative methods in subsequent studies. In particular, the qualitative methods need to give voice, as it were, to adolescents themselves as this has potential to support optimal mental health interventions (Wilson, 2020).

Finally, since this study gives further evidence for the crucial role an individual’s household and community context play in the protective effects of resources on the mental health of adolescents exposed to severe caregiver conflict, future studies are warranted that study the same stressor and resources in other WEIRD and less-WEIRD countries.

Conclusion

Caregiver conflict is associated with poor mental health outcomes for adolescents in WEIRD and less-WEIRD contexts. The results reported in this article are novel in that they urge attention to the resilience-enabling value of comforting beliefs and household routine for adolescents challenged by caregiver conflict in WEIRD and less-WEIRD contexts. Most importantly, the results discourage decontextualized facilitation of either or both resources. They show that comforting beliefs and household routine, which are widely considered as protective (Masten & Motti-Stefanidi, 2020; Ungar, 2019a), cannot be conceptualized as equally protective for adolescents in WEIRD and less-WEIRD contexts. Further, depending on contextual specifics, it is possible that home routine could make it harder for adolescents to adjust well to caregiver conflict. Such harmful effects could be masked by other protective resources (i.e., the presence of comforting beliefs). The identified differential interplay of the studied resources, therefore, urges resilience researchers and mental health practitioners to consider contextual dynamics, and the associated differential impact of resilience resources (Ungar, 2017), more closely. Doing so will facilitate interventions that are optimally enabling.