|Home | About | Journals | Submit | Contact Us | Français|
Using a daily diary method, this study assessed daily episodes of family and peer conflict among 578 adolescents in the ninth grade in order to examine potential bidirectional associations between the family and peer domains. Adolescents completed a daily diary checklist at the end of each day over a fourteen day period to report events of conflict and their emotional states for a given day. Overall, our within-person models provided evidence for the bidirectional nature of family-peer linkages across gender and ethnicity. Adolescents experienced more peer conflict on days in which they argued with parents or other family members, and vice versa. Effect of family conflict further spilled over into peer relationships the next day and two days later, whereas peer conflict predicted only the following day family conflict. Adolescents’ emotional distress partially explained these short term spillovers between family and peer conflict.
The current study is one in a series of papers examining connections across multiple components of adolescents’ daily lives in interpersonal settings (see Chung, Flook, & Fuligni, 2009; Flook & Fuligni, 2008; Flook, in press). In a previous paper, we examined daily spillover of conflict within the family setting and found that adolescents were more likely to argue with their parents on days in which their parents argued with each other (Chung et al., 2009). We also found that the effect of interparental argument was mediated in part by adolescents’ emotional distress. In another paper, we examined day-to-day spillover between family stressors and school problems and reported that family and school experiences reciprocally predicted adolescents’ functioning in each setting (Flook & Fuligni, 2008). Finally, we found that positive and negative daily interpersonal events were reciprocally associated with adolescents’ daily mood (Flook, in press). Although these collective findings suggest spillover of negative interpersonal events within and across social settings in daily life (e.g., family and school), less is known about spillover involving the peer domain and the potential role of affective mediators of such spillover. A next step is to examine how adolescents’ experiences are linked across family and peer settings and whether emotional distress is a potential explanatory mechanism for the family-peer linkage in daily life. Thus, the current study builds on the previous studies by assessing how conflict spreads beyond the boundary of family and shapes adolescents’ later interpersonal experience in peer group, and vice versa.
Consistent with our previous studies (Chung et al., 2009; Flook & Fuligni, 2008), the spillover model guided the current study. This model posits that negative emotion triggered by a negative interpersonal event in one setting spills over to affect a later interaction in the other setting within a relatively short time frame (Almeida, Wethington, & Chandler, 1999; Chung et al., 2009; Flook & Fuligni, 2008; Lehman & Repetti, 2007). Existing research, however, has relied primarily on global or aggregate measures of family and peer interactions collected across several months to years (e.g., McCloskey & Lichter, 2003). Thus, little is known about how adolescents’ experiences in peer and family contexts are interconnected in daily life. Drawing from the spillover model, the current study examines whether an argument with parents or other family members on a given day leads to changes in the adolescent’s peer interactions, and vice versa, the following day and even two days later.
Daily spillover of conflict or tension has been found among adolescents in our previous studies (e.g., Chung et al., 2009; Flook & Fuligni, 2008) as well as adults (e.g., Almeida et al., 1999) and younger children (e.g., Lehman & Repetti, 2007). Less is known about a pathway for such connections across social contexts. Alhough there is evidence that anxiety is a partial mediator among younger children (Lehman & Repetti, 2007), we know nothing about it during adolescence. We thus begin an explorative mapping process by considering adolescent emotional distress as one possible way in which the daily spillover of conflict could take place across family and peer contexts. This is consistent with the spillover model positing that conflict spreads from family to peer, and vice versa, through a transmission of negative emotions. It also is consistent with the idea that adolescents respond with more extreme and negative emotions than do preadolescents or adults because they experience multiple developmental transitions that might be potentially stressful (e.g., changing schools, puberty, and dating) and interpret them more negatively than adults (Arnett, 1999). Given this tendency among adolescents, emotional distress might potentially explain the proposed family-peer spillover of conflict.
Research suggests that females tend to experience greater psychological distress as compared to males in the context of interpersonally stressful events (e.g., Almeida & Kessler, 1998). For instance, adolescent girls are more likely than boys to report feeling sad or upset after conflict with peers (e.g., Noakes & Rinaldi, 2006). Similarly, we previously found that adolescent girls reported significantly more emotional distress than boys on days in which they argued with mother or father (Chung et al., 2009; Flook, in press). Thus, drawing on the spillover model that recognizes a role of emotional tension through which conflict spills over, we expect that if adolescent girls argue with parents or other family members on a given day, they will be more likely than boys to engage in conflict with peers the next day and two days later, and vice versa.
Norms and values that guide family relationships vary across cultures. According to Cooper, Baker, Polichar, and Welsh (1991), adolescents from Asian backgrounds reported higher levels of familial values and were less comfortable in expressing ideas with their parents than their peers from Latino and European backgrounds. Phinney, Kim-Jo, Osorio, and Vilhjalmsdottir (2005) also found that compared to adolescents from European backgrounds, their peers from Korean, Mexican, and Armenian backgrounds were more likely to handle disagreements by complying to their parents in order to avoid open conflict. In this light, when conflict does occur in these families, its impact on adolescents’ emotions may be more pronounced due to cultural emphasis on compliance and family harmony (Chao & Tseng, 2002; García-Coll & Vázquez García, 1995). In parallel with our earlier prediction for the role of emotional distress in the spillover of conflict, it is likely that arguing with parents or family members would trigger greater amount of emotional distress among adolescents from Chinese and Mexican backgrounds because such behaviors would be equated with violating a traditional cultural value of promoting harmonious family interactions and being deferent to authority figures in their families. Thus, we hypothesize that family-to-peer conflict spillover would be more likely for adolescents from Chinese and Mexican backgrounds as compared to European American adolescents.
A daily diary method has been successfully used in our previous studies to examine daily spillover in varying interpersonal contexts (e.g., Chung et al., 2009; Flook, in press; Flook & Fuligni, 2008). In the current study, we consider how daily variation within adolescents in the frequency of experiencing family conflict is associated with their daily conflictual interactions with peers, and vice versa, across the fourteen-day period. The within-person daily design offers several benefits that cannot be easily achieved through using standard designs (Bolger, Davis, & Rafaeli, 2003). First, by having adolescents report on episodes of conflict in each of the family and peer domains each day over a two week period reduces a reporter’s retrospective recall bias because this daily reporting allows them to report events closer to the time that they had occurred. Second, obtaining multiple data points within a person across time (e.g., 14 days in this study) reduces potentially confounding influences of unmeasured variables, such as individual differences in temporally more stable personality attributes and chronic familial or individual circumstances (Bolger et al., 2003). Third, related to the previous point, the micro longitudinal aspect of the design can provide a more accurate assessment of the temporal sequencing in which conflict carries over from one domain to another as compared to the traditional longitudinal design because it is easier to control for intervening variables with a relatively shorter interval between the two time points (Almeida et al., 1999). In this study, we control for prior day levels of conflict in all models tested.
The following four questions guide the current study: (1) Does family conflict covary with peer conflict on a given day, and vice versa? (2) Does conflict with peers carry over to adolescents’ family domain, and vice versa, the next day and two days later? (3) Do these daily associations vary according to adolescents’ gender and ethnicity? (4) Does emotional distress mediate the spillover of conflict between the family and peer domains?
Ninth grade students were recruited from three public high schools in the Los Angeles metropolitan area. Approximately 65% of the recruited students participated in the study, resulting in a total sample of 783 ninth-grade students from various ethnic, socioeconomic, and immigrant backgrounds. The target sample for the current study included 578 adolescents from Mexican (40%; n = 235), Chinese (30%; n = 172), and European (30%; n = 171) backgrounds who completed daily diary checklists on time in the ninth grade (M = 14.87 years, SD = 0.38). Our sample was fairly evenly distributed by girls (52%) and boys (48%).
Students completed an initial background questionnaire at school and then a daily diary checklist at the end of each day over fourteen days, reporting their emotions and whether various events had occurred that day including arguing with parents and friends. A thorough description of the daily diary procedure can be found in our previously published papers (e.g., Chung et al., 2009; Flook & Fuligni, 2008).
Adolescents indicated each day whether the following had occurred: (1) Argued with your mother about something, (2) Argued with your father about something, and (3) Argued with another family member about something. All scores were summed to create an index of daily family conflict calculated separately for each day for each participant (M = .27, SD = .57, Range = 0-3).
Adolescents reported each day whether they argued with a close friend, boyfriend, or girlfriend. Adolescents checked the box if an event had occurred (coded as “1”) but otherwise left it blank (coded as “0”). All marked items each day were summed as an indicator of daily peer conflict for each participant (M = .08, SD = .28, Range = 0-1).
Consistent with our previous studies (e.g., Chung et al., 2009; Flook, in press), daily emotional distress was assessed using the anxiety and depression subscales of the Profile of Moods States (POMS; Lorr & McNair, 1971). Participants rated their anxious (on edge, nervous, uneasy, and unable to concentrate) and depressed mood (sad, hopeless, and discouraged) each day on a scale from 0 “Not at all” to 4 “Extremely.”
Consistent with our previous studies (Chung et al., 2009; Flook & Fuligni, 2008; Flook, in press), multilevel modeling was applied using HLM statistical software (Raudenbusch & Bryk, 2002). Results on family to peer associations are presented first followed by peer to family associations.
The following equation shows the basic model for the same day family conflict to peer conflict association.
Peer conflict on a given day (i) for a particular adolescent (j) was modeled by each individual’s intercept (b0j) and family conflict that same day (b1j). Prior day peer conflict (b2j) was included to parse out the effects of any peer conflict from the previous day and to capture the unique effect of family conflict. In order to reduce possible confounds resulting from effects of the repeated-measures diary method, the week of the study (effect coded -1 for week one, days 1 to 7, and 1 for week two, days 8 to 14) was entered as a control variable in all equations (b3j). The error term in the equation represents unexplained variance (eij). Family conflict significantly predicted peer conflict the same day even after controlling for peer conflict the prior day, although the effect was very small and prior day peer conflict was a stronger predictor of the current day peer conflict (see Table 1).
For one-day lag spillover, the basic model (Equation 1) was slightly modified so that peer conflict was predicted by prior day family conflict, while controlling for peer conflict the prior day. For two-day lagged spillover, prior day family conflict and peer conflict from the previous day and two days before were included as control variables. Very small but significant spillover effects were detected one and two days after the occurrence of the initial peer conflict, while controlling for prior levels of conflict (see Table 1). As before, previous peer conflict predicted the current peer conflict more strongly than previous family conflict did.
The following daily level equation is the basic model used to predict family conflict from same-day peer conflict, controlling for family conflict on the prior day:
This model is similar to the model testing spillover in the reverse direction, except here the variables are reversed, with family conflict as the outcome and peer conflict as the predictor. As shown in Table 2, peer conflict significantly predicted an increase in family conflict the same day. However, the effect was small and family conflict predicted itself more strongly than peer conflict did.
Similarly, in the models testing one day and two-day lagged spillovers of peer to family conflict, prior levels of family conflict were entered as control variables. According to Table 2, peer conflict significantly predicted an increase in family conflict the following day, controlling for prior day family conflict. However, the effect was no longer significant two days later.
In order to examine whether average level of peer conflict, the same day family to peer association, and the spillover of family conflict on to peer conflict varied by gender or ethnicity, the following equations were mapped on to Equation 1.
Gender was effect coded, −1 for males and 1 for females. Ethnicity was dummy-coded, with adolescents from European backgrounds designated as the baseline group for comparison with adolescents from Mexican and Chinese backgrounds. Comparisons between adolescents from Mexican and Chinese backgrounds were made by changing the baseline group in Equations 3 and 4 to adolescents from Mexican backgrounds. Error terms contributing to unexplained variance are represented by u0j and u1j.
Girls reported more peer conflict (b = .06) than boys (b = .02), whereas ethnicity did not predict daily peer conflict. Family conflict made peer conflict significantly more likely for girls, but there was no effect for boys (bgirl = .03, SEgirl=.01, p=.00; bboy = −.01, SEboy = .01, p=.09). No moderation effects were found for the one-day and two-day lagged associations. No significant gender by ethnicity interactions emerged.
For similar analyses in reverse direction, equations 3 and 4 above were mapped onto Equation 2. Girls reported more family conflict than boys (bboys = .08, bgirls = .20, p<.01). Adolescents from Chinese backgrounds reported less family conflict on average as compared to those from European and Mexican backgrounds (bCA = .16, bEA = .26, bMA = .24, p<.01), and these ethnic differences remained significant even after generation was taken into account. There were no significant differences in the spillover of peer conflict on to family conflict as a function of gender or ethnicity. No significant gender by ethnicity interaction was found.
The extent to which same day and one-day lagged family to peer conflict associations were mediated by adolescent’s daily distress as follows using guidelines suggested by Kenny, Korchmaros, and Bolger (2003):
Family conflict predicted the next day emotional distress after controlling for prior day level of distress (b = .03, p<.05) and emotional distress predicted the same day peer conflict even after prior day level of peer conflict was accounted for (b = .07, p<.05). A partial mediation effect was observed where the variance of emotional distress explained 11% of the variation in the effect of family conflict on the next day peer conflict. In addition, the variance of intervening emotional distress explained 2% of the variation in the two-day lagged spillover from family conflict to peer conflict.
The extent to which daily adolescent distress mediated the daily association between peer and family conflict was examined as follows:
Daily level mediational analyses (Kenny et al., 2003) indicated that peer conflict did not predict the following day emotional distress (b = .04, p>.10), thus no mediation effect was observed for the one day lagged association between peer conflict and family conflict. Because two day prior peer conflict was not a significant predictor of family conflict, emotional distress from one day prior was not tested as a mediator.
Same day associations in this study indicate co-occurrence of conflict in family and peer relationships, whereas one and two-day lagged spillovers show that conflict in one setting leads to higher levels of conflict in another. Though the non-experimental nature of this study prevents us from being conclusive about a causal relationship between family and peer conflict, our confidence in the directionality of the family-peer linkages observed was improved by controlling for prior levels of conflict in all analyses.
We found that the effect of family conflict persisted longer than that of peer conflict. On the one hand, peer conflict may be more quickly resolved and as a result perceived more favorably by adolescents, thus its effect might be less enduring. According to Adams and Laursen (2001), adolescents reported relatively friendlier affect and used more positive conflict resolution tactics after peer conflict as compared to family conflict. On the other hand, because our measure of peer conflict had a much narrower possible range of scores (e.g., 0 – 1;“yes” or “no”) than the measure of family conflict (e.g., 0 – 3; argued with mother, father, and/or other family member), there was less variability in peer conflict scores and it could affect the study’s results. For instance, our measure of peer conflict does not capture a possibility that a teenager could experience conflict with multiple friends on a given day.
As expected, on days when teenagers argued with parents or other family members, girls experienced more peer conflict than did boys. This finding provides additional support for greater reactivity to interpersonally stressful events among girls. In addition, the fact that no gender difference was detected in the reverse association suggests that arguing with parents or other family members (as opposed to friends) may be a distinctly more stressful event for girls during this period. It is possible that compared to boys, adolescent girls tend to use cooperative and prosocial tactics to solve conflict with peers more quickly and in a more pleasant manner (Noakes & Rinaldi, 2006; Rose & Asher, 1999) than conflict with family. This could be due to greater importance of intimacy and loyalty with friends than with parents during this period, especially among girls (Steinberg, 2008).
Contrary to our expectation, the daily family-peer linkage operated similarly across ethnicities. However, more research is needed to confirm our finding. Previous studies documented ethnic difference in adolescents’ response to conflict when disagreements involved mundane issues (e.g., chores) versus fundamental issues (e.g., family values and cultural expectations) (Phinney, Ong, & Madden, 2000; Ying, Lee, Tsai, Lee, & Tsang, 2001). Thus, asking what each argument was about could potentially capture ethnic difference in family-peer linkage. Further, tapping finer nuances of ethnicity (e.g., differences in acculturation status and dynamics of argument) rather than using self-reported ethnic labels might help ethnic differences to emerge.
The present study found somewhat weak support for adolescents’ daily emotional distress as a partial explanation for the daily spillover between family and peer conflict. Our small-sized mediation effect indicates that other things could be operating. For instance, adolescents’ anger, rather than depressed or anxious feelings, may be a stronger mediator of the conflict spillover. It will be important for future research to identify other daily level mechanisms through which the spillover between family and peer conflict might occur. Despite the small effect, conflict led to increase in emotional distress within the same adolescent, partially accounting for the daily spillover of conflict from family to peer.
Findings reported in this article must be interpreted with limitations of the study in mind. While adolescents might have multiple arguments with the same individual on any given day, our measurement of conflict variables did not reflect this possibility (i.e., we had no frequency measure). Daily experience of adolescents who had only one versus two or more arguments on a given day is likely to be qualitatively different. Another limitation is our reliance solely on adolescents’ self-reports to assess conflict, which may partially explain the spillover effects detected in the study. Further, given that “argument” was not defined for participants of the study and its definition might vary across gender, ethnicity, or emotional maturity, it is unclear to what extent adolescents’ perceptions of conflict accurately represent actual interactions. Lastly, we did not assess intensity of conflict. It is possible that the more intense the argument was, the more likely the spillover might be.
Overall, our understanding of family-peer linkage was advanced by a within subjects model, which examined changes in conflict that can be observed within the same adolescent over fourteen days. Adolescents’ interactions in the home as well as with peers shape each other on a daily basis in part through emotional distress.
1The study was supported by the Russell Sage Foundation and the preparation of this paper was supported by National Institute of Mental Health through Family Research Consortium IV postdoctoral fellowship awarded to the first author. We are grateful to anonymous reviewers for valuable comments on this manuscript and the first author would like to also thank Research Institute of Human Ecology at Seoul National University for additional support.
Grace H. Chung, Seoul National University Lisa Flook University of Wisconsin at Madison.
Andrew J. Fuligni, University of California, Los Angeles.