This study was made possible by access to data from the Three City Study, a longitudinal research program at the department of Law, Psychology and Social work at Örebro
University, Sweden. The research program was supported by a grant from Formas. Exploring the Role of Adolescents’ Co-Rumination With Mothers and Friends:
Longitudinal Associations With Depressive Symptoms
Sanna Lindsten and Charlotte Berg
Department of Law, Psychology and Social Work, Örebro University PS3111: Psychology Master’s Thesis
Supervisor: Lauree Tilton-Weaver, Ph.D. Spring 2021
Abstract
Co-rumination is a communication process that has been linked to depressive symptoms among adolescents. The first aim of this study was to examine the directionality of the associations, regarding rumination with peers and mothers, as well as comparing co-rumination about the mothers’ and the adolescents’ problems. The second aim was to investigate if gender moderates the associations between co-rumination with peers and
depressive symptoms. We used self-report data from two measurement points with a one year interval from the 5-year longitudinal “Three City Study”. Participants were pupils aged 15-17 years at the first measurement point (n = 2,914) and mothers to a subset of the adolescents (n = 44). The results showed that co-rumination with peers was associated with increases in depressive symptoms. Depressive symptoms were associated with decreases in co-rumination with mothers about the mothers’ problems. Preliminary analyses showed that co-rumination with mothers about the mothers’ problems was correlated with co-ruminating about
adolescents’ problems. The findings support the notion that co-rumination, especially with friends, is related to depressive symptoms among youth. This, as well as the ways in which mothers communicate with their adolescents, might be important to consider when discussing interventions for adolescents’ depressive symptoms.
Keywords: co-rumination, depressive symptoms, gender, adolescents, peers, mothers, longitudinal study, regression analysis
Acknowledgements
To our supervisor Lauree Tilton-Weaver – thank you for your guidance and support. You made this an inspiring experience for us, and we could not have done it without you!
Contents
Abstract ... 2
Contents ... 4
Exploring the Role of Adolescents’ Co-Rumination With Mothers and Friends: Longitudinal Associations to Depressive Symptoms ... 6
Associations Between Co-Rumination and Depressive Symptoms... 6
Temporal Ordering of the Associations Between Co-Rumination and Depressive Symptoms ... 8
Co-Rumination With Peers and Parents ... 9
Co-Rumination About Adolescents’ or Parents’ Problems ... 11
The Role of Gender ... 12
The Current Study ... 14
Method ... 15
Participants ... 15
Procedure ... 17
Missing Data and Attrition Analysis... 19
Measures ... 21
Adolescent Reports ... 21
Mother Reported Measures ... 22
Plan of Data Analysis... 23
Results ... 24
Co-Rumination With Mothers About Adolescents’ Problems Predicting Depressive
Symptoms ... 24
Co-Rumination With Mothers About Mothers’ Problems Predicting Depressive Symptoms ... 27
Depressive Symptoms Predicting Co-Rumination With Peers ... 27
Depressive Symptoms Predicting Co-Rumination With Mothers About Adolescents’ Problems... 27
Depressive Symptoms Predicting Co-Rumination With Mothers About Mothers’ Problems ... 29 Gender Differences ... 29 Moderation Analysis ... 30 Summary ... 30 Discussion ... 32 Strengths ... 36
Limitations and Future Directions ... 36
Conclusions ... 38
Exploring the Role of Adolescents’ Co-Rumination With Mothers and Friends: Longitudinal Associations to Depressive Symptoms
Depression is a construct that is well researched within the field of psychology. The last two decades attention has been directed towards co-rumination as a potential mechanism contributing to the development of depressive symptoms. Co-rumination is an interpersonal process where two persons excessively discuss personal problems. This is done in a repetitive manner with focus on the negative aspects and emotions, and the negative problem focus is mutually encouraged (Rose, 2002). Co-rumination is different from, but shares significant overlap with both rumination and self-disclosure, and is associated with both negative and positive aspects, such as internalizing problems and relationship quality (Rose, 2002).
Although co-rumination has consistently been associated with depressive symptoms, there are some issues that have not been fully addressed, including the direction of the association and whether the association is similar across different co-rumination partners. These issues are the focus of our study.
In our study, we focused on adolescents, as it is a developmental period during which depressive symptoms and co-rumination both rise (Costello et al., 2006; Hankin et al., 2010; Rose, 2002; Stone et al., 2011). Moreover, Schwartz-Mette and Rose (2012) found that co-rumination mediated depression contagion only for adolescent friend-dyads, not for children friend-dyads, indicating that adolescence is the period in which the covariation between co-rumination and depressive symptoms emerges. Consequently, adolescence can be considered an important developmental period to focus on in relation to depressive symptoms and co-rumination.
Associations Between Co-Rumination and Depressive Symptoms
Co-rumination was originally proposed to predict internalizing problems, including anxiety and depression, based on its overlap with rumination (Rose, 2002). Rumination, or
ruminative response, involves lingering focus on one's own negative affect and depressive symptoms as well as the causes, meanings and consequences of these symptoms (Rose, 2002). A ruminative response style is considered a predictor of depression, and has been found to be related to increases in duration and severity of depressed mood (Nolen-Hoeksema et al., 1993; Nolen-Hoeksema et al., 1994). Early support can be found in a study showing that co-rumination was associated with internalizing problems, and that rumination was responsible for the association between these two, in that the association was no longer significant when controlling for rumination (Rose, 2002).
Today there is substantial evidence for a positive cross-sectional association between co-rumination and depressive symptoms (Spendelow et al. 2017). Co-rumination has also been associated with a lifetime history of clinical depression (Stone et al., 2010), and to increases in immediate negative affect in an experimental study compared to a problem solving or distraction condition (Zelic et al., 2017). Studies have also suggested that co-rumination mediates depression contagion among friendship dyads (Schwartz-Mette & Rose, 2012; Schwartz-Mette & Smith, 2018) and that it mediates the association between
interpersonal stressors and depressive symptoms (Fan et al., 2016). Also, a moderating effect has been found, where the combination of high levels of co-rumination and stressors
prospectively predicted depressive mood in university students (White & Shih, 2012) and depressive symptoms in adolescent girls (Bastin et al., 2015).
Researchers have also identified and studied different aspects of the co-rumination process. The results of this indicate that dwelling and lingering on negative affect
(co-brooding), rather than speculating and talking extensively about the problems (co-reflecting), might be responsible for the association between co-rumination and internalizing problems (Rose et al., 2014; Bastin et al. 2018).
Despite clear associations, most researchers have assumed that co-rumination causes depressive symptoms. This assumption has, to a large degree, gone unchallenged because most studies have used cross-sectional designs and not considered the potential for the other direction or bidirectionality.
Temporal Ordering of the Associations Between Co-Rumination and Depressive Symptoms
When considering co-rumination as a risk-factor for depressive symptoms, the
temporal ordering of these constructs needs to be established. If co-rumination is the cause, it should only precede depressive symptoms. This direction seems to be supported, as co-rumination predicts prospective increases in adolescents’ depressive symptoms (Rose et al., 2007; Schwartz-Mette & Rose, 2012; Hankin et al., 2010). Co-rumination has also been shown to predict shorter time to onset of clinically significant depressive episodes, as well as severity and duration of these episodes (Stone et. al., 2011). Co-brooding, the aspect of co-rumination referring to dwelling on negative affect, predicted increased depressive symptoms over two months (Bastin et al., 2018). Co-rumination has also been shown to predict
depressive symptoms indirectly over time through rumination (Stone & Gibb, 2015). These results give reason to view co-rumination as a risk factor for depressive symptoms.
However, there is reason to expect another direction, since experiencing depressive symptoms can make negative experiences more salient (Mathews et al., 1996; Neshat-Doost et al., 1998; Segal & Gemar, 1997). As co-rumination is a form of coping with negative experiences, it is reasonable to expect that depressive symptoms could lead to more co-rumination. Indeed, when examined, internalizing symptoms also predict prospective increases in co-rumination. This suggests that the connections between co-rumination and depressive symptoms are bidirectional (Rose et al., 2007; Hankin et al., 2010).
However, bidirectionality has only been examined in the context of adolescents’ friendships. Adolescents co-ruminate with friends, but they also co-ruminate with others, namely parents, and their co-rumination with parents is also tied to depressive symptoms (Waller & Rose, 2010).
Co-Rumination With Peers and Parents
Most research in this area is about co-rumination between peers, especially between same-gender close friends. The results indicate that co-rumination within these relationships is associated with relationship quality as well as depressive symptoms (Rose, 2002;
Spendelow et al., 2017). Thus, friendships are an important context for examining co-rumination and depressive symptoms.
Some interest has also been directed to co-rumination between parents and their children. Since parental communication has been linked to adolescents' adjustment (van Dijk et al., 2014; Yu et al., 2006), researchers suggest that co-rumination with parents might be linked to adolescents’ adjustment in a similar way.
This expectation has not been entirely supported by research, because the findings have been inconsistent. Some studies have found a positive association between mother-child co-rumination and the adolescents’ internalizing problems (Waller & Rose, 2010, 2013), and one study found that adolescents with major depressive disorder co-ruminated with their parents more than twice as much as adolescents without a history of Axis I disorders (Waller et al., 2014). In addition to this, one study found a link between father-child co-rumination and depressive symptoms while controlling for positive communication (Ioffe et al., 2020). However, the same study also found a negative association, with mother-child co-rumination being linked to less depressive symptoms, with and without control for positive
Still, compared to the relationship between co-rumination with peers and depressive symptoms, the relationship between parent-child co-rumination and depressive symptoms has quite limited support. Waller and Rose (2013) found that co-rumination with mothers
predicted internalizing problems in the adolescents, but that co-rumination with friends mediated this link. Based on their analysis, they suggest that co-rumination with peers might be more detrimental than co-rumination with mothers. This remains to be tested.
In summary, both co-rumination with friends and with parents have been linked to depressive symptoms among adolescents. However, there is reason to believe that there may be differences in these associations. Adolescence is also a time-period where interactions between parents and children become more complex and stressful (Hankin & Abrahamson, 2001). It is also a developmental phase where friends become increasingly important, and peer influence might begin to exceed the influence of parents (Buhrmester & Furman, 1987). If this is the case, the link between parent-child co-rumination and youths’ emotional
adjustment could be expected to be weaker than the link between co-rumination with peers and youths’ emotional adjustment.
At this point, most researchers have only separately examined adolescents’ co-rumination between friends and co-co-rumination with mothers. Only one study has examined the associations with depressive symptoms looking at co-rumination with mothers and peers in the same sample, however using cross-sectional data (Waller & Rose, 2013). No studies have examined longitudinally whether the strength of the associations depend on the co-rumination partner. Our study aims to address this gap.
Another aspect to consider is whether the associations differ depending on if parents and adolescents discuss the parents’ or the adolescents’ problems when co-ruminating.
Co-Rumination About Adolescents’ or Parents’ Problems
Very few studies have examined whether the associations between parent-child co-rumination and depressive symptoms differ depending on whose problem is in focus (parents’ problems or adolescents’ problems). Co-ruminating with a parent about the parent’s problem may be a form of parentification. Parentification is a construct describing when a child or adolescent is prematurely expected to take on a parental role in the family. In this case, co-rumination would be considered as emotional parentification (Champion et al., 2009), where the child or adolescent provides emotional support, taking care of the emotional needs of the parents or other family members. These tasks and responsibilities are more often associated with the parental role and can be inappropriate for the child’s age (Champion et al., 2009) and in the worst cases has been compared to childhood neglect (Hooper, 2007). Parentification seems to be related to depressive symptoms for the care-providing youth (Burton et al., 2018; Champion et al., 2009; Cohen et al., 2012; Schier et al., 2015), as well as adult
psychopathology in later life (Hooper et al., 2011). Thus, we argue that when adolescents co-ruminate with their parents about the parents’ problem, it can be considered a form of
emotional parentification. Looking at the associations between parentification and depressive symptoms, co-rumination could therefore be thought to be especially problematic when focusing on the parents’ problems.
A study by Waller and Rose (2010) supports our suggestion, finding that
pre-adolescents’ and pre-adolescents’ co-rumination with mothers was associated with internalizing problems, and that the association was stronger for co-rumination focused on the mothers' problems, compared to focusing on the adolescents' problems. In a later study, co-ruminating about the mothers’ problems, but not co-ruminating about the adolescents’ problems, was associated with adolescents' internalizing problems (Waller & Rose, 2013).
In addition, previous research also suggests that gender might play an important role when it comes to the associations between co-rumination and depressive symptoms. While many studies have looked into this, there are still some aspects in need of further
clarification.
The Role of Gender
Substantial attention has been directed to developmental differences between girls and boys. Entering adolescence, gender differences start to appear and increase, with girls
exhibiting more depressive symptoms than do boys (Angold et al., 2002; Twenge & Nolen-Hoeksema, 2002). This has been called paradoxical, considering girls’ tendency to have closer friendships, which is a protective factor against internalizing problems (Rose, 2002). However, girls also seem to co-ruminate more than boys do with both friends and mothers (Rose, 2002; Waller & Rose, 2010). This greater prevalence of co-rumination has been thought to be the source of the gender differences in internalizing problems, a hypothesis that has gained support from multiple studies (Calmes & Roberts, 2008; Felton et al., 2019; Rose, 2002; Stone et al., 2011; Tompkins et al., 2011).
Gender has also been suggested as a moderator of the association between
co-rumination and depressive symptoms, with some arguing that co-co-rumination is more adverse for girls than boys, and that it might even be protective for boys under certain circumstances. This has been discussed in light of girls being more interpersonally oriented and having closer friendships characterized by self disclosure and exchanges of emotional support (Rose & Rudolph, 2006). As well as providing an arena for co-rumination to take place, the greater emphasis on intimate friendships for girls, compared to boys, might leave girls more
vulnerable to the strains in these relationships, thereby making co-rumination a greater risk factor. Supporting this theoretical premise, girls have been found to be more reactive to interpersonal stressors than boys (Flook, 2011; Hankin et al., 2007), and to report more
empathetic distress in response to friends’ problems than boys (Smith, 2015; Smith & Rose, 2011).
Co-rumination also includes self-disclosure, contributing to closer friendships and social support. Boys, relative to girls, tend to engage in less self-disclosure and emotional conversations (Rose & Rudolph, 2006). Given that, the benefits of the social support that co-rumination provides might be more salient for boys, perhaps protecting them from the
negative aspects associated with internalizing problems. Also, low emotional competence has been linked to increased co-rumination for girls but not for boys (Borowski & Zeman, 2018). In the same study, girls engaged in co-rumination more than boys, except for at especially high levels of emotional competence, where boys actually engaged in co-rumination more than girls did. If emotional competence is linked to co-rumination differently in girls and boys, that could contribute to different emotional outcomes, such as more depressive symptoms for girls.
Supporting the notion of gender as a moderator, co-rumination and co-brooding between same-gender friends have been found to prospectively predict depressive symptoms in adolescent girls, but not boys (Bastin et al., 2018; Rose et al., 2007). Bastin et al. (2015) also found that low levels of co-rumination, as compared to high levels for girls, interacted with interpersonal stress in predicting depressive symptoms for boys.
Despite this, these results seem to be limited to a few studies. In fact, most studies, including a meta review (Spendelow et al., 2017), have failed to find a moderating effect of gender on the association between co-rumination and depressive symptoms (Hankin et al., 2010; Rose et al., 2014; Schwartz-Mette & Rose, 2012; Schwartz-Mette & Smith, 2018; Stone et al., 2010, 2011; Stone & Gibb, 2015). Gender also did not moderate the association with depressed mood in a daily diary study (White & Shih, 2012). These findings indicate that co-rumination might have a similar association to depressive symptoms in girls and boys,
and that the gender differences in depressive symptoms during adolescence should be attributed in part to the higher prevalence of co-rumination in girls, rather than a gender specific effect of co-rumination.
In short, the research on whether gender moderates the associations between co-rumination and depressive symptoms is mixed. So, it is important then to continue to consider adolescents’ gender as a potential moderator.
The Current Study
The first aim of our study was to further examine the temporal ordering of the associations between rumination and depressive symptoms. That is, we compared co-rumination predicting change in depressive symptoms to depressive symptoms predicting change in rumination. Another aim of our study was to compare adolescents’ co-rumination with peers to co-co-rumination with mothers, to find out if the associations are different across relationships. As part of this aim, we also compared co-ruminating with mothers about the adolescents’ problems to co-ruminating about the mothers’ problems. Our first hypothesis was that co-rumination and depressive symptoms would be associated longitudinally (Hypothesis 1), with co-rumination with peers (Hypothesis 1a), with mothers about the adolescents’ problems (Hypothesis 1b) and with mothers about the
mothers’ problems (Hypothesis 1c), predicting increases in depressive symptoms as well as depressive symptoms predicting increases in co-rumination with peers (Hypothesis 1d), with mothers about the adolescents’ problems (Hypothesis 1e) and with mothers about the
mothers’ problems (Hypothesis 1f).
Further, we expected that co-rumination with peers would be a stronger predictor of depressive symptoms than co-rumination with mothers, whether they are co-ruminating about the adolescents’ problems or mothers’ problems (Hypothesis 2a). We also expected that
co-rumination with mothers about the mothers’ problems would be a stronger predictor of depressive symptoms than co-rumination about the adolescents’ problems (Hypothesis 2b).
Regarding the opposite direction, we expected that depressive symptoms would predict bigger increases in co-rumination with peers than with mothers (Hypothesis 2c), because adolescence is a period in which there is a shift toward seeking peers for support. Based on the function of co-rumination as providing social support, and the expectation that adolescents experiencing depressive symptoms will seek social support, we also expected depressive symptoms to predict bigger increases in co-rumination about the adolescents’ problems than in co-rumination about the mothers’ problems (Hypothesis 2d).
Our last aim was to examine the role of gender. More specifically, we examined whether girls and boys differ in the reported amount of co-rumination and depressive
symptoms, and whether gender moderates the associations between co-rumination with peers and depressive symptoms. We expected girls to co-ruminate more than boys (Hypothesis 3), with peers (Hypothesis 3a) and mothers, both about the adolescents’ problems (Hypothesis 3b) and about the mothers’ problems (Hypothesis 3c). We also expected girls to report more depressive symptoms than boys (Hypothesis 3d). Based on the mixed findings regarding gender as a moderator, this was explored without specific expectations.
Method Participants
Our study used data from a larger research project called the Three City Study (Boersma, 2021). It was a longitudinal project ranging over five years, focusing on psychosocial problems in adolescents. The participants were recruited from three lower secondary schools in three different cities in central Sweden. For our study, a subset of the original sample was used, based on available data on the constructs of interest. We used data from the third and fourth data collection points (hereafter T1 and T2, respectively), due to the
interest in the reports of primary caregivers (hereafter referred to as “parents”) that were only available at these time-points.
For the purpose of this study, we only included participants who were in the age range of 15 to 17 years old (Mage= 15.87, SD = 0.70) at T1. All adolescents participating at T1 were included in the current study (n = 2914). Of those who participated at T1, 1898 participated in the next wave (T2), giving a retention rate of 65.1%. Out of the participants, 52.7% were boys (n = 1535) and 47.3% were girls (n = 1379). The adolescents were predominantly born in Sweden (86.9%), with parents born in Sweden (mothers 73.9%, fathers 73.7.0%), and exclusively spoke Swedish at home (77.0%). Most parents were married or cohabitating (64.8%). Most of the adolescents lived with both parents (67.9%), while the remaining moved between their parents (11.5%), lived with only one parent (15.4% with mothers, 3.5% with fathers), or lived with someone other than their parents (1.7%). On a scale ranging from 4 to 13, with higher values indicating higher disposable incomes, this sample was relatively well off (M = 10.76, SD = 1.64, modal response = 11).
Parents of the adolescents were recruited via mail. At T1, 44 mothers (42 reported as mothers, 1 reported as other than mother or stepmother, 1 did not report the relationship to participating adolescent) gave consent and participated. Out of the mothers who replied at T1, 33 mothers also replied at T2, giving a retention rate of 75%. Although fathers participated, we did not use their data, as the response rate was too low. At T1 the mothers ranged in age from 36 to 58 years (Mage = 45.43, SD = 4.76). The most common education level was
reported as having some college or university education (3 years or longer) (44.2%).
The 44 adolescents whose parents participated were a subset of the whole sample and also ranged from 15 to 17 years (Mage = 15.75, SD = .65). Out of these adolescents, 45.5%
were boys (n = 20) and 54.5% were girls (n = 24). The adolescents were predominantly born in Sweden (93.2%), with parents born in Sweden (mothers 88.6%, fathers 88.6%), and
exclusively spoke Swedish at home (88.6%). Most parents were married or cohabitating (81.8%). Most of the adolescents lived with both parents (79.5%) while the remaining moved between their parents (11.4%), lived with only the mother (6.8% ) or lived with someone other than their parents (2.3%). They reported a mean disposable income of 10.89 (SD = 1.71).
To test if the subsample of adolescents whose mothers replied differed from the adolescents whose mothers did not reply, we conducted independent samples t-tests for age, disposable income, co-rumination with peers and depressive symptoms. The results showed that adolescents whose mothers replied were more likely to co-ruminate more with peers at T1 (t (2912) = -2.78, p < .01) than those whose mothers did not reply. They did not
significantly differ at T1 in age (t (2912) = 1.18, p > .05), disposable income (t (2912) = -.52, p > .05) or depressive symptoms (t (2912) = .74, p > .05). For means and standard deviations, see Table 1. We also conducted chi-square analyses to test for differences between the two mentioned groups regarding all of our categorical variables. There was a significant association with parents’ marital status: the adolescents whose mothers did reply were less likely than those whose mothers did not reply to have parents who were divorced or
separated, χ2 (1, N = 2914) = 5.68, p < .05. There was no significant differences for gender χ2
(1, N = 2914) = .94, p > .05, place of birth, χ2 (1, N = 2914) = 1.55, p > .05, mothers’ place of
birth, χ2 (3, N = 2914) = 5.48, p > .05, fathers’ place of birth, χ2 (3, N = 2914) = 5.37, p > .05,
which adults they lived with, χ2 (4, N = 2914) = 4.69, p > .05, or language spoken at home, χ2
(2, N = 2914) = 3.41, p > .05. Procedure
Before data collection began, parents and adolescents received information about the study, including what steps would be taken to protect their data, that participation was
Table 1
Means and Standard Deviations for Adolescents Whose Mothers Did not Reply and for Those Whose Mothers Did Reply at T1
Mothers did not reply Mothers did reply
Variable M SD M SD
Age 15.88 .70 15.75 .65
Disposable income 10.76 1.64 10.89 1.71
Co-rumination with peers 2.98 1.13 3.46 .99
Depressive symptoms 1.80 .58 1.74 .53
opportunity to inform the researchers (by phone, email, or postage-paid card) that they did not want their children to participate. Before the first data collection, 121 parents did not consent to their children’s participation. All adolescents whose parents did not decline consent were able to participate (that is, had passive parental consent). Adolescents provided active consent if they filled in the questionnaires after a consent briefing.
Adolescent data were collected annually during the spring, in classrooms during school hours, via questionnaire. Trained research assistants administered the questionnaires and were available in case the adolescents needed help. The adolescents were each offered a snack during the break, and each class received 300 Swedish crowns (~30 €) afterward. Students who did not speak Swedish or had other difficulties understanding written language did not participate.
Parents received their questionnaires by mail to fill in at home, and to return in a self-addressed stamped envelope. The data collection of parents’ reports took place at two
different time-points, at the same time or shortly after data from adolescents was collected, with a year-long interval between.
Ethical Considerations
The Three City Study was approved by the Regional Ethics Board in Uppsala, Sweden, prior to the start of data collection. Some ethical considerations were also made regarding the procedure of the study. First, as described above, the participants received information about the study and gave their active consent, while the parents only had to give passive consent to their adolescents’ participation. While the preferred method would have been asking for active consent also from the parents, that would come with the risk of losing participants, especially adolescents experiencing more risk factors for developing
internalizing problems. Therefore, this method was considered the best way to balance these risks.
The data collection did not expose the participants for significant risks. However, some of the questions addressed topics that could be considered private or sensitive, potentially evoking negative emotions in the adolescents. To address this there were test leaders present at the data collections that the adolescents could speak to. The adolescents in need of more support could also be referred to a school counselor.
A couple of steps were also taken to secure the participants’ confidentiality. The teachers left the room during data collections in the schools. The data was also registered in a de-identified manner and kept in a secure location at Örebro University.
Missing Data and Attrition Analysis
There was a minimal amount of missing data at each time point and for each reporter (< 5%). To avoid the bias introduced by dropping observations (Enders, 2010), one of the project researchers replaced the missing data using a modern two-step method described by Kärnä et al. (2011). Specifically, the Quark package in R (Lang et al., 2015) was used to create principal components (10 linear and 3 non-linear components). These components were used as auxiliary variables to impute 100 datasets using the MICE package of R (van
Buuren & Grotthuis-Oudshoorn, 2011). The average or modal imputed value for each adolescent was used to replace their missing data point. These values represent the “best population estimate of the value needed to reproduce the population parameters” (Kärnä et al., 2011, p. 55). This method was used for adolescents’ reports at both time frames. As the number of mothers who participated was quite low, this option was not used for the mothers’ data.
To test for attrition effects, we conducted independent samples t-tests for age, disposable income, co-rumination with peers and depressive symptoms. The results showed that adolescents who dropped out were more likely to be younger (t (1987.82) = -4.01, p < .001), to report lower disposable income (t (1850.43) = -5.55, p < .001) and to report more depressive symptoms (t (1869.17) = 3.93, p < .001) than adolescents who participated at both times. Adolescents who dropped out did not significantly differ from those who participated at both time points in reported co-rumination (t (2912) = 1.51, p > .05). For means and standard deviations see Table 2. We also conducted chi-square to test for attrition effects for all of our categorical variables. There was a significant attrition effect for place of birth, χ2 (1,
N = 2914) = 39.86, p < .001, mothers’ place of birth, χ2 (3, N = 2914) = 45.84, p < .001,
fathers’ place of birth, χ2 (3, N = 2914) = 60.71, p < .001, parents’ marital status χ2 (1, N =
2914) = 4.57, p < .05, which adults they lived with , χ2 (4, N = 2914) = 19.96, p < .01, and
language spoken at home, χ2 (2, N = 2914) = 46.74, p < .001. Adolescents who participated at
only the first point were more likely than adolescents who participated at both time-points to be born in another country than Sweden, have a mother that was born outside of Sweden, have a father born outside of Sweden, have divorced or separated parents, and they were less likely to live with both their parents and to speak only Swedish at home. There was no significant attrition effect for gender χ2 (1, N = 2914) = .76, p > .05.
Table 2
Means and Standard Deviations for Adolescents who Participated at Both Time-Points and at Time 1 Only.
Both time points Time 1 only
Variable M SD M SD Age 15.91 .69 15.8 .72 Disposable income 10.89 1.55 10.52 1.78 Co-rumination 2.97 1.11 3.03 1.16 Depressive symptoms 1.77 .55 1.86 .62 Measures Adolescent Reports
Co-rumination. Adolescents reported on their co-rumination with their peers through a revised version of the Co-rumination Questionnaire, a self-report scale that was developed to capture co-rumination between peers (Rose, 2002). The current study used a shortened version consisting of 8 items, each covering a different content area: (a) frequency of discussing problems, (b) talking about problems instead of engaging in other activities, (c) consistent encouragement of the discussion of problems, (d) discussing the same problem repeatedly, (e) trying to understand causes of problems, (f) trying to understand consequences of problems, (g) trying to understand other aspects of the problem that are not understood, and (h) focusing on negative feelings. The items chosen were the same ones as those used by Waller and Rose (2010) for their study, with the decision based on which item for every content area that had the highest factor loading in the original co-rumination study (Rose, 2002). Each item was rated on a 5-point Likert scale ranging from 1 (not at all true) to 5
(really true). Participants' scores were the mean ratings of the items. This scale had good psychometric properties, Cronbach’s 𝛼t1 = .95, 𝛼t2 = .996.
Depressive Symptoms. Adolescents reported their depressive symptoms through the self-report questionnaire Center for Epidemiology Studies Depression Scale for Children (CES-DC; Olsson & von Knorring, 1997). It consists of 20 items covering depressive symptoms such as “I felt down and unhappy”. Each item was rated on a 4-point likert scale ranging from 0 (not at all) to 3 (a lot), referring to how much every symptom has been experienced over the past week. Participants’ scores were the mean ratings of the items (Cronbach’s 𝛼t1 = .92, 𝛼t2 = .92)
Mother Reported Measures
Co-rumination. Mothers reported on their co-rumination with their adolescent through a revised version of the Co-rumination Questionnaire (Rose, 2002). The CRQ was originally designed to measure the extent to which adolescents co-ruminate with their peers and was adapted for the purpose of this study to assess co-rumination between parents and their children. One set of questions addressed co-rumination about the mothers’ problems; the other set addressed co-rumination about the adolescents’ problems (Waller & Rose, 2010). Each of the areas in the adolescent-peer version of the scale was assessed, using the same response scale.
Two different strategies were used to create the T1 and T2 co-rumination scores. For T1, planned missingness was used to reduce participation burden for parents. Due to the low number of responses, the strategy described by Kärnä et al. (2011) was not used. Instead, factor analyses using Full Information Maximum Likelihood (FIML) estimation were conducted in MPlus (Muthén & Muthén, 2017), using the full sample of parents who responded (some of whom had adolescents who did not participate in these waves). The factor scores for two factors representing rumination about mothers’ problems and
co-rumination about adolescents’ problems were saved and used in our analyses. For T2, the planned missingness was dropped, so participants' scores were the mean ratings of the items covering co-ruminating about the parents’ problems, and the mean ratings of the items covering co-rumination about the adolescents’ problems, respectively. Reliability for T2 was good, for focusing on the adolescents’ problems (𝛼t2 = .82) and focusing on the mothers’ problems (𝛼t2 = .86).
Plan of Data Analysis
Data analyses were conducted using IBM SPSS Statistics Version 27. To investigate the associations between co-rumination and depressive symptoms, we used hierarchical regression, controlling for the criterion variable at T1. In total, we conducted two sets of three regressions – one for each partner and problem focus (adolescent-peers, adolescent-mother about mother, adolescent-mother about adolescent). One set used co-rumination as the criterion variable and one set used adolescents’ depressive symptoms as the criterion
variable. In each regression, two steps were entered. In the first step, the baseline (T1) value of the criterion was entered, followed by a step entering the appropriate predictor (depressive symptoms for co-rumination, co-rumination for depressive symptoms). We planned to compare sets of predictors using a Z test, as recommended by Paternoster et al. (1998).
To examine differences between girls and boys on all study variables we conducted an independent samples t-test. To test if gender moderated the associations between co-rumination with peers and depressive symptoms, we added a step to our hierarchical regressions, in which the appropriate interaction between gender and the predictor was entered (depressive symptoms and gender for co-rumination, co-rumination and gender for depressive symptoms).
Results
Our study variables correlated as expected, except for the variables covering co-rumination with mothers, which only correlated with each other, and in some ways with co- rumination with peers. Means, standard deviations and correlations for all study variables are presented in Table 3.
Co-Rumination With Peers Predicting Depressive Symptoms
The results of the hierarchical regression analysis suggested that 35.3% of the
variations in adolescents’ depressive symptoms at T2 was predicted by depressive symptoms at T1, R2 = .35, adjusted R2 = .35, F(1, 1896) = 1035.63, p < .001 (all model estimates are
reported in Table 4). Specifically, higher levels of depressive symptoms at T1 predicted increases in depressive symptoms at T2. Moreover, co-rumination with peers explained an additional 0.6% of the variance in depressive symptoms at T2, F∆(1, 1895) = 17.03, p < .001. That is, adolescents who reported more co-rumination also reported increases in depressive symptoms at T2, b = .04, β = .08, p < .001. Thus, we found support for our hypothesis that co-rumination is related to increases in depressive symptoms, at least when the partners are friends (Hypothesis 1a).
Co-Rumination With Mothers About Adolescents’ Problems Predicting Depressive Symptoms
The results of the hierarchical regression analysis suggested that 10.1% of the
variations in adolescents’ depressive symptoms at T2 was predicted by depressive symptoms at T1, R2 = .10, adjusted R2 = .08, F(1, 42) = 4.73, p < .05 (all model estimates are reported in
Table 4). Specifically, more depressive symptoms at T1 predicted more depressive symptoms at T2. However, co-rumination with mothers about adolescents’ problems at T1 did not significantly predict adolescents’ depressive symptoms at T2, F∆ (1, 41) = .18, p > .05. Thus, we did not find support for our hypothesis that co-rumination with mothers about the
Table 3
Descriptives and Correlations Among Study Variables
Time 1 Time 2
Variable n M SD Gender 1 2 3 4 5 6 7 8
Gender 2914 Boys 52.7% -
Time 1
1. Co-rumination with peers 2914 2.99 1.13 -.34** -
2. Co-rumination adolescents’ problems 44 -.03 1.71 .15 .01 -
3. Co-rumination mothers’ problems 44 -.03 .84 .04 .11 .75** -
4. Depressive symptoms 2914 1.80 .58 -.30** .06** .02 .05 -
Time 2
5. Co-rumination with peers 1864 3.04 1.10 -.34** .61** .21 .12 .06* -
6. Co-rumination adolescents’ problems 33 2.73 .84 .05 .34 .52** .45** -.27 .55** -
7. Co-rumination mothers’ problems 33 2.17 .92 .01 .34* .45** .51** -.23 .48** .80** -
8. Depressive symptoms 1898 1.83 .58 -.36** .13** -.06 .06 .59** .08** .10 .10 - * p < .05; ** p < .01.
Table 4
Summary of Hierarchical Regressions Examining Adolescent Depressive Symptoms as a Function of Co-Rumination With Peers and Mothers About Both Partners’ Problems
Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with peers
Model 1 .35** - 1035.63** Depressive symptoms T1 .63 .02 .59** Model 2 .36** .01** 17.03** Depressive symptoms T1 .62 .02 .59** Co-rumination T1 .04 .01 .08** Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with mothers, adolescents’ problems
Model 1 .10* - 4.73* Depressive symptoms T1 .36 .17 .32* Model 2 .11 .00 .18 Depressive symptoms T1 .36 .17 .32* Co-rumination T1 -.05 .13 -.06 Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with mothers, mothers’ problems
Model 1 .10* - 4.73* Depressive symptoms T1 .36 .17 .32* Model 2 .10 .00 .08 Depressive symptoms T1 .36 .17 .32* Co-rumination T1 .03 .11 .04 * p < .05 ** p < .001
adolescents’ problems were related to increases in depressive symptoms (Hypothesis 1b). Co-Rumination With Mothers About Mothers’ Problems Predicting Depressive Symptoms
The results of the hierarchical regression analysis again suggested that 10.1% of the variance in adolescents’ depressive symptoms at T2 was predicted by depressive symptoms at T1, R2 = .10, adjusted R2 = .08, F(1, 42) = 4.73, p < .05 (all model estimates are reported in
Table 4). Specifically, more depressive symptoms at T1 predicted more depressive symptoms at T2. However, co-rumination with mothers about mothers’ problems did not significantly predict adolescents’ depressive symptoms at T2, F∆ (1, 41) = .08, p > .05. Thus, our
hypothesis that co-ruminating with mothers about the mothers’ problems would be related to increases in depressive symptoms was not supported (Hypothesis 1c).
Depressive Symptoms Predicting Co-Rumination With Peers
The results of the hierarchical regression analysis suggested that 36.6% of the variations in adolescents’ co-rumination with peers at T2 was predicted by co-rumination with peers at T1, R2 = .37, adjusted R2 = .37, F(1, 1859) = 1073.79, p < .001 (see Table 5 for
all model estimates). Specifically, more rumination with peers at T1 predicted more co-rumination with peers at T2. However, depressive symptoms at T1 did not significantly predict adolescents’ co-rumination with peers, F∆ (1, 1858) = .21, p > .05. Thus, our hypothesis that depressive symptoms would be related to increases in co-rumination with peers was not supported (Hypothesis 1d).
Depressive Symptoms Predicting Co-Rumination With Mothers About Adolescents’ Problems
The results of the hierarchical regression analysis suggested that 26.8% of the
variations in adolescents’ co-rumination with mothers about the adolescents’ problems at T2 was predicted by co-rumination with mothers about the adolescents’ problems at T1, R2 =
Table 5
Summary of Hierarchical Regressions Examining Co-Rumination With Peers and Mothers About Both Partners’ Problems as a Function of Adolescent Depressive Symptoms
Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with peers
Model 1 .37*** - 1073.79*** Co-rumination T1 .60 .02 .61*** Model 2 .37*** .00 .21 Co-rumination T1 .60 .02 .60*** Depressive symptoms T1 .02 .04 .01 Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with mothers, adolescents’ problems
Model 1 .27** - 11.34** Co-rumination T1 .67 .20 .52** Model 2 .35*** .09 4.00 Co-rumination T1 .69 .19 .53*** Depressive symptoms T1 -.43 .22 -.29 Variable B SE B β R2 R2∆ F∆ in R2
Co-rumination with mothers, mothers’ problems
Model 1 .26** - 10.62** Co-rumination T1 .60 .18 .51** Model 2 .36*** .10* 4.82* Co-rumination T1 .67 .18 .56*** Depressive symptoms T1 -.52 .24 -.33* * p < .05 ** p < .01 *** p ≤ .001
.27, adjusted R2 = .24, F(1, 31) = 11.34, p < .01 (see Table 5 for all model estimates).
Specifically, more co-rumination with mothers about the adolescents’ problems at T1
predicted more co-rumination with mothers about the adolescents’ problems at T2. However, depressive symptoms at T1 did not significantly predict adolescents’ co-rumination with mothers about the adolescents’ problems at T2, F∆ (1, 30) = 4.00, p > .05. Thus, we did not find support for our hypothesis that depressive symptoms were related to increases in co-rumination with mothers about the adolescents’ problem (Hypothesis 1e).
Depressive Symptoms Predicting Co-Rumination With Mothers About Mothers’ Problems
The results of the hierarchical regression analysis suggested that 25.5% of the variations in adolescents’ co-rumination with mothers about mothers’ problems at T2 was predicted by co-rumination with mothers about mothers’ problems at T1, R2 = .26, adjusted
R2 = .23, F(1, 31) = 10.62, p < .01 (see Table 5 for all estimates). Specifically, more
co-rumination with mothers about mothers’ problems at T1 predicted more co-co-rumination with mothers about mothers’ problems at T2. Moreover, depressive symptoms at T1 explained an additional 10.3% of the variance in co-rumination with mothers about mothers’ problems at T2, F∆(1, 30) = 4.82, p < .05. Specifically, adolescents who reported more depressive symptoms at T1 decreased in co-rumination with their mothers about mothers’ problems at T2, b = -.52, β = -.33, p < .05. Thus, we did not find support for our hypothesis that
depressive symptoms were related to increases in co-rumination with mothers (Hypothesis 1f).
Gender Differences
Girls reported significantly higher results on co-rumination with peers at T1, t (2911.97) = 19.91, p < .001, and at T2, t (1858.99) = 15.47, p < .001; and on depressive
symptoms at T1 t (2725.06) = 16.59, p < .001 and T2 t (1708.64) = 16.65, p < .001. There was no significant difference between girls and boys in co-rumination with mothers about mothers’ problems at T1 t (42) = -.28, p > .05 or T2 t (31) = -.08, p > .05; or about
adolescents’ problems at T1 t (42) = -.96, p > .05 or T2 t (31) = -.25, p > .05 (see Table 6 for means and standard deviations). Thus, we found support for our hypothesis that girls would report more co-rumination with peers (Hypothesis 3a) as well as more depressive symptoms (Hypothesis 3d) than boys, but we did not find support for our hypothesis that girls would co-ruminate more than boys with mothers about their own problems (Hypothesis 3b) or the mothers’ problems (Hypothesis 3c).
Moderation Analysis
Moderated regression analysis suggested that 38.6 % of the variations in depressive symptoms at T2 was predicted by co-rumination with peers at T1, gender, depressive
symptoms at T1 and the interaction term, F(4, 1893) = 297.58, p < .001. However, the test of interaction revealed that gender did not significantly moderate the association between co-rumination with peers and depressive symptoms, F∆ (1, 1893) = .11, p > .05.
The second moderated regression analysis suggested that 38.3 % of the variation in rumination with peers at T2 was predicted by depressive symptoms at T1, gender, co-rumination with peers at T1 and the interaction term, F(4, 1856) = 288.13, p < .001. However, the test of interaction revealed that gender did not significantly moderate the association between depressive symptoms at T1 and co-rumination with peers at T2, F∆ (1, 1856) = 1.05, p > .05. Thus, we did not find support for gender being a moderator.
Summary
In sum, we found that co-rumination and depressive symptoms were significantly correlated with each other across all time-points, and that co-rumination about mothers’
Table 6
Means and Standard Deviations for Girls and Boys for Co-Rumination With Peers, Co-Rumination With Mothers (Adolescents’ Problems and Mothers’ Problems), and Depressive Symptoms.
Time 1 Time 2
Girls Boys Girls Boys
Variable n M SD n M SD n M SD n M SD
Co-rumination with peers 1379 3.40 1.0 1535 2.62 1.11 878 3.43 .97 983 2.69 1.09 Co-rumination adolescent’s problems 24 -.13 .74 20 .08 .67 17 2.70 .83 16 2.77 .88 Co-rumination mother’s problems 24 -.06 .79 20 .01 .91 17 2.16 .99 16 2.19 .87 Depressive symptoms 1379 1.98 .60 1535 1.64 .51 887 2.05 .60 1011 1.63 .49
problems was correlated to co-rumination about adolescents’ problems, also across all time-points. Co-rumination with peers and co-rumination with mothers were correlated to each other in some ways, but not in all. The results also showed that co-rumination with peers at T1 significantly predicted increases in depressive symptoms at T2 when controlling for symptoms at baseline. We also found that depressive symptoms predicted decreases in co-rumination with mothers about the mothers’ problems when controlling for co-co-rumination at baseline. Further, our results showed that girls reported significantly higher levels of co-rumination with peers and depressive symptoms than boys did.
Discussion
In this study, we examined the associations between co-rumination and depressive symptoms. We asked if co-rumination and depressive symptoms were unidirectionally or reciprocally related. We asked this for both co-rumination with peers and with mothers, as well as for co-rumination with mothers about the mothers’ problems and their own problems. We also asked if gender moderated the association between co-rumination with peers and depressive symptoms.
We only found evidence of unidirectionality, not for a reciprocal association. The direction also depended on with whom the adolescents ruminated. Specifically, co-rumination with friends was related to increases in depressive symptoms; by contrast, depressive symptoms predicted decreases in co-rumination with mothers about the mothers’ problems. This adds to evidence that the direction of relations should be considered. So, more longitudinal studies are needed, with co-rumination and depressive symptoms measured at multiple waves.
Our findings that adolescents’ co-rumination with peers predicted increases in depressive symptoms (which is consistent with our hypothesis and is further supported by results from previous research, Rose et al. 2007), should be further studied to identify the
mechanisms linking these behaviors. Previous studies have proposed that dwelling on negative affect, in contrast to reflecting on problems, might be central in explaining the associations with depressive symptoms (Bastin et al. 2018). Specifically, co-ruminating about their problems with peers might evoke and enhance negative emotions in adolescents, and the repetitive and extensive nature of co-rumination might also make it difficult to get out of the negative emotional state. This could not only be linked to triggering and sustaining depressed mood, but it may also affect processes associated with depression and similar disorders. For example, they may internalize the focus on negative affect, from interpersonal
communication processes to internal cognitive processes, such as rumination. Rumination is known to be associated with increased duration and severity of depressed mood (Nolen-Hoeksema et al., 1993; Nolen-(Nolen-Hoeksema et al., 1994). If co-rumination leads to rumination, it would expose the adolescents to even more negative affect, and make it even harder to leave a negative affective state. Perhaps it would also train them in noticing and focusing on problems rather than solutions, making them more likely to discover additional areas of distress. This could further enhance the sense of hopelessness and reinforce depressive symptoms. It is also possible that rumination acts as a moderator. Future studies should investigate if some adolescents begin to ruminate more when co-ruminating, while others do not, and if those groups differ in development of depressive symptoms.
The variance that was explained by co-rumination in our study was small, which is to be expected, because a) co-rumination is just one of many factors that might contribute to the development of depressive symptoms, b) it is an estimate linking behaviors over a one-year period, and c) depressive symptoms were relatively rank-order stable over time, affecting the amount of variation in change that can be predicted. Notably, our estimates are consistent with those found in similar longitudinal studies (e.g., Rose et al., 2007).
We also found that adolescents' depressive symptoms predicted decreases in co-rumination with mothers’ about the mothers’ problems, which was contradictory to our hypothesis. This could indicate either that adolescents with increasing depressive symptoms withdraw and talk less to their mothers, or that mothers to adolescents with depressive symptoms avoid talking about problems with them to protect them, or because they do not know how to cope with their adolescents’ distress. While it might be adaptive to decrease co-rumination about the mothers’ problems, this could also indicate that adolescents are not receiving sufficient parental support when experiencing depressive symptoms. This in turn could be linked to further deterioration of adolescents’ wellbeing.
Beyond previously discussed associations, we found no evidence that co-rumination with mothers was linked to depressive symptoms, which also contradicted our hypothesis. However, this lack of significance is likely due to Type II error, as our sample size was quite small. Moreover, we were not able to examine fathers as co-rumination partners, as the number of fathers participating precluded analyses. We suggest then, that these questions not be abandoned, but raised in the future, where researchers can also look at whether the waning influence of parents (Buhrmester & Furman, 1987) is a contributing factor.
In contrast to our study, most similar studies only looked at co-rumination within same-gender best friend dyads (eg. Rose, 2002, Schwartz-Mette & Smith, 2018). Some have argued that this method might underestimate boys’ co-rumination, since boys might be more likely to disclose personal information to female friends. When looking at co-rumination with confidants of choice, boys have been found more likely to report a female friend than girls to report a male friend, and with that method previously stated gender differences in reported co-rumination disappeared (Barstead et al., 2013). Since participants in our study were not limited to consider co-rumination with a same-gender friend, our results differ from those of Barstead et al. (2013). Our results indicate that adolescent boys and girls in our sample
actually differ in the amount they co-ruminate, even when the gender of their friend is not considered.
Although the preliminary analyses showed that girls reported more depressive
symptoms and more co-rumination with friends than did boys, we found no evidence that the processes linking co-rumination with depressive symptoms are gender-specific. This is in line with some studies stating the same (Hankin et al., 2010; Rose et al., 2014; Schwartz-Mette & Rose, 2012; Schwartz-Mette & Smith, 2018; Stone et al., 2010, 2011; Stone & Gibb, 2015), while it contradicts others suggesting that co-rumination only predicts depressive symptoms among girls, but not among boys (Bastin et al., 2018; Rose et al., 2007). Future research should try to give more clarity to the issue.
Preliminary analyses indicated that co-ruminating with friends and co-ruminating with mothers was largely orthogonal. This would indicate that adolescents do not necessarily co-ruminate with both friends and mothers, but with one of them. However, care has to be taken to replicate this non-significant result, as it may be due to Type II error.
We did find, however, a relatively strong association between the two focal problems of co-ruminating with mothers. That is, when adolescents co-ruminate with their mothers about one’s problem, they tend to co-ruminate about the other’s problem as well. Putting adolescents in the position of dealing with their mothers’ problems and negative emotions can be viewed as inappropriate and outside of normative parent-child roles. As argued in the introduction, it can also be viewed as a form of emotional parentification. Considering the detrimental aspects that are associated with this phenomenon, we believe our results deserve further attention. This might include person-centered approaches that could identify the characteristics of adolescents and mothers who tend to co-ruminate about both types of problems, compared to those that only co-ruminate about adolescents’ problems, and identify when such overlap is related to adolescents’ adjustment.
Strengths
Our study had several strengths, including a large, longitudinal sample of adolescents, using established measures with good psychometric properties. The longitudinal study
design, with analyses controlling for baseline symptoms, allowed us to examine prospective changes in our outcome variables, with sufficient power to have some confidence in our findings.
Another strength of our study was the ability to examine co-rumination across different partners and across different problems. Few studies have looked at the role of co-rumination with mothers, and even fewer have examined co-co-rumination about the
adolescents' problems and mothers’ problems separately. Only one has examined
adolescents’ co-rumination with friends and mothers in the same sample, however using only cross-sectional data. By looking at all these conditions of co-rumination in the same study, our study contributes to extending the knowledge on how different conditions of co-rumination might contribute to depressive symptoms. In this respect, there is still much to learn.
Further, our community sample of adolescents in middle to late adolescence comes with the advantage of providing information that generalizes to a non-clinical population, and how these processes work for the great majority in this age group. However, future studies might want to extend this research by examining what the associations look like in a clinical sample, to guide potential clinical interventions.
Limitations and Future Directions
Despite the contributions of this study, there are some limitations to consider. First, we did not find that co-rumination with mothers significantly predicted increases in
depressive symptoms. As previously mentioned, the sample of mothers who replied was small, decreasing our power and increasing the risk of Type II error. Having a bigger sample
would also allow for performing more analyses using co-rumination with mothers. Future research should examine what the associations with depressive symptoms would look like if all three co-rumination conditions are investigated simultaneously in the same analysis, considering that the shared variance might change the separate associations when looked at in relation to each other. Future studies should also look at gender as a moderator of the
associations between co-rumination with mothers and depressive symptoms, to see if the associations differ between girls and boys.
Further, only mothers’ reports of co-ruminating with their adolescents were used in this study. Previous research has found that adolescents’ reports of co-rumination with their mothers were associated with depressive symptoms, while mothers’ reports were not (Waller & Rose, 2010). Future research could therefore use both adolescents’ and mothers’ reports to prevent getting one sided information that could be biased from individual responders’ perspectives.
As we have stated, it was not possible to examine co-rumination with fathers, due to the low response rates. We believe that looking into adolescents’ co-rumination with fathers, the prevalence as well as its correlates, is of great importance in future research, given that previous research suggests that their influence could be of special importance for
development of internalizing problems (Ioffe et al., 2020; Lamb & Lewis, 2013; Lansford et al., 2014). We would also propose that looking at it in relation to co-rumination with mothers would shed further light on the dynamics of co-rumination within families. Co-ruminating with one parent may be less problematic than co-ruminating with the other or with both. We speculate that co-ruminating with mothers may be less problematic than co-ruminating with fathers. We base this speculation of research that shows that adolescents tend to report feeling closer to their mothers than their fathers (Lamb & Lewis, 2013), that open communication with fathers is negatively associated with internalizing symptoms (Ioffe et al., 2020), and that
co-rumination with fathers predicts increases in depressive symptoms, whereas co-rumination with mothers predicts decreases (Ioffe et al., 2020). These somewhat contradictory findings warrants a further examination of potential different pathways.
Another aspect to consider is that the adolescents were asked to report co-rumination with a friend, but were not asked to consider a specific friend, or peers in general. It is possible that individuals can have a personal tendency to co-ruminate more or less across relationships. However, friendships are known to have different functions (Rose & Rudolph, 2006) and it is therefore probable that adolescents do not co-ruminate in all of their
friendships, but only in some. It is still unclear if it is more detrimental to co-ruminate intensely with one friend or to co-ruminate at moderate levels with many friends. While previous studies have examined both co-rumination within specific friend dyads (eg. Rose et al., 2007) and with peers in general (eg. Hankin et al., 2010), future studies should investigate both of them in the same design, to examine if one of them is a stronger predictor of
depressive symptoms than the other. Conclusions
In this study we have investigated the role of co-rumination with peers and mothers in relation to depressive symptoms. While these associations have been studied before, this is one of the first studies to examine co-rumination within both relationships in the same sample, as well as comparing co-rumination about the mothers’ problems to co-rumination about the adolescents’ problems.
Our results contribute to the field by adding to the research base suggesting that adolescents’ co-rumination with peers is associated with increases in depressive symptoms. While future studies need to replicate the findings in a clinical population before generalizing the results to adolescents with clinically significant symptoms, we believe that this provides preliminary reason to consider co-rumination with peers when trying to decrease depressive
symptoms in youth. Helping adolescents become aware of when their adaptive problem talk becomes maladaptive (i.e., turns into co-rumination), and to find more adaptive ways to seek emotional support than through co-rumination, could be considered as potential interventions.
Our findings suggest there are a few areas that might be relevant for care-providers to take into consideration when meeting with clients. Considering the potential risks associated, care-providers should be observant of a) adolescents co-ruminating with peers at high levels, b) mothers co-ruminating about their own problems with their adolescents, and c) mothers reducing their communications with their depressed adolescents. Identifying and addressing these processes might be helpful in preventing increases in adolescents’ internalizing problems.
Overall, the study provides further information about the importance of co-rumination in adolescents in developing depressive symptoms over time. Co-rumination with peers seems to be especially important, but there is still much to learn about the processes linked to co-rumination with parents. Future research should continue to examine the area to broaden the knowledge and to identify areas for intervention.
References
Angold, A., Erkanli, A., Silberg, J., Eaves, L., & Costello, E. J. (2002). Depression scale scores in 8-17-year-olds: Effects of age and gender. Journal of Child Psychology and Psychiatry, 43(8), 1052–1063. https://doi.org/10.1111/1469-7610.00232
Barstead, M. G., Bouchard, L. C., & Shih, J. H. (2013). Understanding gender differences in co-rumination and confidant choice in young adults. Journal of Social and Clinical Psychology, 32(7), 791–808. https://doi-org.db.ub.oru.se/10.1521/jscp.2013.32.7.791 Bastin, M., Mezulis, A. H., Ahles, J., Raes, F., & Bijttebier, P. (2015). Moderating effects of
brooding and co-rumination on the relationship between stress and depressive symptoms in early adolescence: A multi-wave study. Journal of Abnormal Child Psychology, 43(4), 607–618. https://doi.org/10.1007/s10802-014-9912-7
Bastin, M., Vanhalst, J., Raes, F., & Bijttebier, P. (2018). Co-brooding and co-reflection as differential predictors of depressive symptoms and friendship quality in adolescents: Investigating the moderating role of gender. Journal of Youth and Adolescence, 47(5), 1037–1051. https://doi.org/10.1007/s10964-017-0746-9
Boersma, K. (14 January 2021). Trestadsstudien. Örebro Universitet. https://www.oru.se/forskning/forskningsproj/jps/trestadsstudien/
Borowski, S. K., & Zeman, J. (2018). Emotional competencies relate to co‐rumination: Implications for emotion socialization within adolescent friendships. Social Development, 27(4), 808–825. https://doi.org/10.1111/sode.12293
Buhrmester, D., & Furman, W. (1987). The development of companionship and intimacy. Child Development, 58(4), 1101–1113. https://doi.org/10.2307/1130550
Burton, S., Hooper, L. M., Tomek, S., Cauley, B., Washington, A., & Pössel, P. (2018). The mediating effects of parentification on the relation between parenting behavior and
well-being and depressive symptoms in early adolescents. Journal of Child and Family Studies, 4044-4059. https://doi.org/10.1007/s10826-018-1215-0
Calmes, C. A., & Roberts, J. E. (2008). Rumination in interpersonal relationships: Does co-rumination explain gender differences in emotional distress and relationship
satisfaction among college students? Cognitive Therapy and Research, 32(4), 577– 590. https://doi.org/10.1007/s10608-008-9200-3
Champion, J. E., Jaser, S. S., Reeslund, K. L., Simmons, L., Potts, J. E., Shears, A. R., & Compas, B. E. (2009). Caretaking behaviors by adolescent children of mothers with and without a history of depression. Journal of Family Psychology, 23(2), 156–166. https://doi.org/10.1037/a0014978
Cohen, D., Greene, J. A., Toyinbo, P. A., & Siskowski, C. T. (2012). Impact of family caregiving by youth on their psychological well-being: A latent trait analysis. The Journal of Behavioral Health Services & Research, 39(3), 245–256.
https://doi.org/10.1007/s11414-011-9264-9
Costello, J. E., Erkanli, A., & Angold, A. (2006), Is there an epidemic of child or adolescent depression? Journal of Child Psychology and Psychiatry, 47(12), 1263-1271.
https://doi.org/10.1111/j.1469-7610.2006.01682.x
Enders, C. (2010). Applied missing data analysis. New York, NY: Guildford Press.
Fan, C., Chu, X., Wang, M., & Zhou, Z. (2016). Interpersonal stressors in the schoolyard and depressive symptoms among Chinese adolescents: The mediating roles of rumination and co-rumination. School Psychology International, 37(6), 664–679.
https://doi.org/10.1177/0143034316678447
Felton, J. W., Cole, D. A., Havewala, M., Kurdziel, G., & Brown, V. (2019). Talking together, thinking alone: Relations among co-rumination, peer relationships, and
