Internalizing Symptoms and Friendship Stability: Longitudinal Actor-Partner Effects in Early Adolescent Best Friend Dyads

Abstract

The present study investigated the stability of friendship nominations over the course of a school year as a function of early adolescents’ and their classroom best friends’ internalizing symptoms (i.e., depression, anxiety, and somatization). Sample consisted of 156 early adolescents (57.1% female; $\bar{X}$ age = 12.62; SD = 0.62) involved in 78 same-sex best friendship dyads. We assessed best friendship (classroom) nominations at beginning (T1) and end (T2) of the school year. Results of longitudinal analyses performed with the Actor-Partner Interdependence Model indicated adolescents’ and their classroom best friends’ depressive symptoms predicted lower stability of best friendships over time, whereas best friends’ somatization emerged as a predictor of higher friendship stability. In addition, positive dyadic friendship quality predicted greater stability over time. These findings highlight the importance of employing a dyadic framework when examining the role of internalizing symptoms in friendship stability.

Keywords

friendships dyadic relationships peer relationships internalizing symptoms

Researchers have widely demonstrated that involvement in stable friendships plays a key role in the psychosocial adjustment in childhood and adolescence (Berndt, 2002; Berndt & Keefe, 1995; Ladd, Kochenderfer, & Coleman, 1996; Ladd & Price, 1987; Parker & Seal, 1996; Tani, Tomada, Guarnieri, & Tonci, 2006; Vitaro, Boivin, & Bukowski, 2009; Wojslawowicz, Rubin, Burgess, Booth-LaForce, & Rose-Krasnor, 2006). During early adolescence, from about 12 to 16 years of age, dyadic friendships become more important and especially significant for the psychosocial adjustment (Berndt, 1982), partly due to the biological, cognitive, and social changes related to puberty. Preadolescents tend to emphasize the importance of many positive qualities in their friendships, such as closeness, intimacy, loyalty, and security (Bukowski, Hoza, & Boivin, 1993; Fonzi & Tani, 1996; Hartup, 1996). High-quality friendships tend to predict high stability of friendship relations over time (Berndt, 1986; Bowker, 2004; Branje, Frijns, Finkenauer, Engels, & Meeus, 2007; Bukowski, Hoza, & Boivin, 1994; Ciairano, Rabaglietti, Roggero, Bonino, & Beyers, 2007; Fonzi, Schneider, Tani, & Tomada, 1997; Schneider, Fonzi, Tani, & Tomada, 1997).

Among the individual predictors of friendship stability, internalizing symptoms have been shown to play a relevant role in adolescence (Poulin & Chan, 2010). Indeed, given that internalizing symptoms may be manifested and expressed within interpersonal contexts, they are prone to affect interpersonal partners directly. This may be, in particular, the case for depressive symptoms. In line with the symptom-driven theoretical perspective advocated by Kochel, Ladd, and Rudolph (2012), depressive symptoms can have a disruptive effect on relationships because of intensification of perceptions of interpersonal difficulties (Rudolph, Flynn, Abaied, Groot, & Thompson, 2009) and depression-linked behaviors such as social withdrawal from peers (Boivin, Hymel, & Bukowski, 1995; Flynn & Rudolph, 2014). Consistent with interpersonal theories (Coyne, 1976; Joiner & Timmons, 2009; Rudolph, Flynn, & Abaied, 2008), depressed individuals may also elicit rejection from close others due to their proneness to engage in dysfunctional interpersonal behaviors, such as excessive reassurance seeking, negative feedback seeking, or social withdrawal, which may ultimately undermine the stability of their friendships (Chan & Poulin, 2009; Klima & Repetti, 2008) as well as their quality (Rubin, Wojslawowicz, Rose-Krasnor, Booth-LaForce, & Burgess, 2006).

At the same time, recent findings suggest that a mild degree of internalizing symptoms may improve, rather than compromise, adolescents’ dyadic frienships (Hill & Swenson, 2014). In particular, when compared with depressive symptoms, anxiety has been shown to be characterized by a more ambigous relationship with friendship characteristics. Anxiety disorders have been associated with an increase in friendship problems (Biggs, Vernberg, & Wu, 2012; Goodyer, Wright, & Altham, 1990). However, nonclinical anxiety has been shown to predict higher friendship stability, positive friendship quality, and a larger number of friends (Rose et al., 2011). Indeed, while anxious adolescents may engage in socially dysfunctional behaviors, they may nonetheless be prone to maintain their current friendship when worried about rejection and lack of acceptance by other peers. Similarly, literature concerning somatization symptoms and relationships with peers is mixed: Findings indicate both negative (Rhee, Holditch-Davis, & Miles, 2005) and nonsignificant associations between somatic complaints and social status among peers (Jellesma, Rieffe, & Terwogt, 2008).

Although many studies have investigated the role of internalizing symptoms as a predictor of friendship stability at the individual level, in a dyadic friendship, friends’ behaviors may be better conceptualized as the result of an interactive process in which the personal characteristics of an individual affect the personal characteristics and choices of his or her partner (Cillessen, Jiang, West, & Laszkowski, 2005; Cook & Kenny, 2005; Parker & Asher, 1993). Thus, in a dyadic perspective, friendship choices and maintenance are expected to be affected by traits, behaviors, and attributes of both partners (Giletta et al., 2012). The importance of employing a dyadic framework when studying factors influencing friendship stability is consistent with recent findings demonstrating that individual characteristics tend to be less important than dyadic characteristics in predicting friendship dissolution (Hartl, Laursen, & Cillessen, 2015).

In light of these considerations, the aim of the present study is to investigate the influence of internalizing symptoms (i.e., depression, anxiety, and somatization) as assessed in adolescents involved in a same-sex friendship dyad on the stability of their friendship over time. We have chosen to analyze the data by applying the Actor-Partner Interdependence Model (APIM; Cook & Kenny, 2005; Kashy & Kenny, 2000; Kenny & Cook, 1999) due to the dyadic, interdependent nature of friendship nominations. By accounting for dyadic interdependence among actors’ and partners’ measures, the APIM allows for the investigation of individual actor and partner effects influencing the actors’ outcomes. By using this approach, in our study we evaluated the role of both early adolescents’ and their very best friends’ internalizing symptoms as actor and partner effects influencing the stability of early adolescents’ friendship choices over time. Gender and dyadic measures of perceived friendship quality were additionally included in the model as control variables. Findings suggest the existence of gender differences in the stability of same-sex friendships during early adolescence (Benenson & Christakos, 2003). In addition, our choice to include friendship quality as a control in the model relates to its strong theoretical and empirical connections with friendship stability in pre- and early adolescence (Betts & Stiller, 2014; Bukowski, Hoza & Boivin, 1994; Schneider et al., 1997).

Method

Participants

Our initial sample consisted of 340 seventh-grade students (53% females; age: $\bar{X}$ = 12.62, SD = 0.62) attending four schools located in an urban area of central Italy. In compliance with the Italian law and the ethical code of the Association of Italian Psychologists, students and parents were informed about the characteristics of the study and provided written consent several weeks prior to the initial data collection. Parents were mailed a letter describing the study and a consent form, which they were asked to sign and return via mail. Students with consent were approached in class, with research assistants describing the study and informing students that their answers would be kept confidential. All students agreed to participate in the research; parental consent rate was 100%. Questionnaires took approximately 30 minutes to complete. Teachers were not present in the classroom during the administration of the questionnaire. No incentives were offered for participation. Questionnaire data was collected at the beginning (T1, October) and at the end (T2, May) of the school year; at each of the two time points, students were asked to nominate their same-sex best friend in the classroom. About 9% of students (n = 32) were either not present at both time points or had missing data (i.e., skipped items) on the study measures (i.e., friendship quality, internalizing symptoms). Mean imputation was used for six participants who had two or less items missing per subscale.

For the purpose of this study, analyses were performed on a subsample of 156 participants (57.1% females; age: $\bar{X}$ = 12.61 years, SD = .62) which, based on classroom best friend nomination, were found to be part of a reciprocal same-sex best friendship dyad at Time 1 (78 dyads, 45 female dyads). In order to investigate possible biases due to significant differences between actual participants and the 184 respondents who were not included in the final sample, we conducted t tests on the internalizing symptoms and quality of friendship measures: No significant differences emerged between the two samples.

Measures

Internalizing symptoms

Participants’ internalizing symptoms were assessed at the beginning of the school year (T1) by administering the Italian adaptation of the Revised Seattle Personality Inventory (Greenberg & Kusche, 1990; Tani & Schneider, 1998). We administered three subscales—Anxiety (six items), Depression (ten items), and Somatization (five items). Example items include “Do you often feel unhappy?” (Depression); “Are you worried about what other people think of you?” (Anxiety); and “Do you often have stomach ache?” (Somatization). Participants were asked to rate the frequency of symptoms on a 4-point scale ranging from never or almost never (1) to always or almost always (4). For each of the subscales, a score was generated by summing items. Cronbach’s alpha for Anxiety was .77, for Somatization it was .65; and for Depression it was .79.

Friendship stability

Friendship stability is generally operationalized as a dyadic measure (Ellis & Zarbatany, 2007; Rose et al., 2011), that is, stability is established when friendship nominations are found to be reciprocated by each member across two or more time points. In the present study, however, friendship stability was operationalized at the individual level: Friendship stability was determined for same-sex members of a reciprocal dyad by evaluating whether their very best friend nomination was consistent over the course of a school year. Specifically, individual friendship stability was assessed using the following procedure. As a first step, reciprocal best friend dyads were identified by administering a classroom-level friendship nomination procedure ((Bukowski et al., 1994; Cillessen et al., 2005) at the beginning of the school year (T1). In the administered peer nomination procedure, participants were asked (a) to nominate up to five friends in the classroom, (b) to indicate who their same-sex best friend was among them, and (c) to rate their best friendship by completing the Friendship Quality Questionnaire (FQQ). Participants were considered involved in a reciprocal best friendship dyad if they indicated each other as classroom best friend in the peer nomination procedure described above. We chose to limit the nomination procedure to classroom peers due to the specific characteristics of the Italian middle school education system. In Italy, middle school students generally remain in the same classroom (i.e., classrooms in which peers do not change) during their middle school years (Grades 6-8), and stay in the classroom all day (up to 8 hours/day) while teachers of different subjects rotate among the different classrooms. Using this approach, 78 reciprocal best friend dyads were identified at T1. Given that participants were allowed to indicate only one best friend, all best friend dyads were unique, that is, no member of any dyad was also a member of another dyad. Eventually, the peer nomination procedure was administered again at the end of the school year (T2). By comparing participants’ best friend nominations at T2 with those reported at T1, we computed a dichotomous indicator of individual friendship stability where 1 indicated stability in the nomination between the two time points, and 0 indicated a change in nomination at T2.

Overall, we found 48 of the 78 reciprocal dyads to be stable across the two time points (T1, T2). The remaining 30 dyads consisted of best friend dyads in which either one member (n = 18) or both (n = 12) did not confirm their best friend nomination to the other member at T2. The relatively low frequency of stable best friendships is consistent with findings indicating that more than half of middle school friendships do not last an academic year (Bowker, 2004; Poulin & Chan, 2010).

Dyadic friendship quality

Participants completed the FQQ (Parker & Asher, 1993), a self-report questionnaire designed to assess multiple qualitative aspects of a child’s “very best friendship,” at the beginning (T1) and end (T2) of the school year. The questionnaire consists of 40 items and an initial “warm-up” item, each in the form of a statement about their best friendship. Items can be combined to form six subscales: Companionship and Recreation (five items), Validation and Caring (ten items), Conflict and Betrayal (seven items) Conflict Resolution (three items), Help and Guidance (nine items), and Intimate Exchange (six items). Example items include “My friend and I always sit together at lunch”(Companionship and Recreation), “My friend makes me feel good about my ideas” (Validation and Caring), “My friend and I argue a lot” (Conflict and Betrayal), “My friend and I make up easily when we have a fight” (Conflict Resolution), “My friend helps me so I can get done quicker” (Help and Guidance),” and “My friend and I always tell each other our problems” (Intimate Exchange). At each time point (T1, T2), participants rated on a 5-point scale (1 = not all true, 5 = really true) how true each statement was for their current classroom best friendship. Thus, participants who nominated different best friends at T1 and T2 answered the FQQ by reporting about different best friendship relationships at the two time points. For this reason, in the present study, analyses were performed on participants’ ratings as collected at T1, while ratings collected at T2 were not examined. As done by other authors (Rose & Asher, 2004), subscales were combined to obtain two total scores representing positive features of friendship (i.e., all subscales except for Conflict and Betrayal) and friendship conflict (i.e., Conflict and Betrayal). Scores were computed by summing students’ answers to items as assessed at T1. Cronbach’s alpha for positive friendship quality score was .93, while for conflict it was .77. Correlation of friendship quality measures between reciprocal dyad members was found to be quite high (conflict: r = .44, p < .001; positive friendship quality: r = .63, p < .001). Eventually, for the purpose of APIM analyses, two dyad-level indicators of friendship quality (dyadic positive friendship quality, and conflict) were computed by averaging the scores reported at T1 by participants involved in the same best friend dyad.

Data Analysis

As a first step, Pearson’s correlation coefficients were computed for all the study measures. Prior to the analyses with the APIM, the pairwise intraclass correlation coefficient (PICC; Zou & Donner, 2004) was used to investigate the within-dyad degree of nonindependence on the friendship stability variable. For binary data, the PICC may be computed using the standard Pearson correlation coefficient; for large numbers of clusters (k > 50), it performs well as an approximation of the standard intraclass correlation coefficient (ICC; Zou & Donner, 2004). Unlike the standard intraclass correlation, however, the PICC does not provide an estimate of the variance in the outcome attributable to cluster membership but only an estimate of interdependence between dyad partners. ICC was also computed on internalizing symptoms in order to investigate within-dyad nonindependence on the predictor variables.

Prior to analyses with the APIM, all continuous variables were standardized to help interpretation and comparison of parameter estimates. The APIM was then implemented on the data using a generalized estimating equations (GEE) analytic approach. When modeling dyadic data with binary outcomes, the GEE approach can be viewed as an extension of the logistic regression model for independent data accounting for the nonindependence of dyadic outcomes. Our choice to implement the APIM using GEE instead of other approaches (e.g., structural equation modeling, Olsen & Kenny, 2006; multilevel linear modeling) is consistent with recent findings indicating the GEE approach as being able to provide more accurate estimates of actor and partner effects when modeling binary outcomes, and also when the sample size is small (Loeys & Molenberghs, 2013). For the purpose of this study, dyads were treated in the analyses as indistinguishable dyads (i.e., dyads within which a specific ordering of members cannot be established), as is typical when analyzing same-sex best friendships (Olsen & Kenny, 2006). A conceptual formulation of the APIM for indistinguishable dyads for binary outcomes is the following:

\begin{array}{l} log it (pr (Outcome = 1)) = β 0 + β 1 \times {Predictor}_{Actor} + β 2 \times {Predictor}_{Partner} + \\ β 3 \times {Covariate}_{Dyadic} . \end{array}

That is, using the logit link function, the probability of observing the actors’ outcome is modeled in the APIM using both actor and partner effects, and dyad-level covariates (also often referred as between-dyad variables). In the case of the APIM for indistinguishable dyads, members of the same dyad are generally modeled as repeated observations in order to control for dyadic nonindependence. In our study, the APIM was used to test the effect of each actor’s and his or her partner’s symptoms of anxiety, depression, and somatization on the stability of the actor’s best friend nomination over time. We performed analyses employing a two-step approach. As a first step, actor and partner effects were examined in two separate models. Then, actor and partner effects were included in a single APIM. Placing both actor and partner effects in the model simultaneously allowed us to examine the relative importance of partners’ effects, while controlling for actors’ effects, and vice versa. Two measures of dyadic friendship quality at T1 (within-dyad average positive quality and conflict), and a dichotomous indicator of dyadic gender (1 = female; 0 = male) were included in the models as dyad-level covariates. Analyses were performed using the PROC GENMOD procedure in SAS by specifying the working correlation matrix as unstructured, as suggested by Loeys and colleagues (2014) in the case of nonindependent binary outcomes. The Wald chi-square (score) test was used to test significance of the parameter estimates.

Results

Descriptive Statistics and Correlations

Table 1 shows the main descriptive statistics and correlations for the employed measures (T1). A weak but significant negative correlation was found between the positive friendship quality and conflict scores. Significant correlations were also found between the internalizing symptoms. Specifically, anxiety revealed positive correlations with both depression and somatization symptoms; in addition, somatization was found to be positively correlated with depression.

Table 1.

Descriptive Statistics and Correlation Among the Study Measures (T1) (N = 156).

		1	2	3	4	M	SD
1	Positive friendship quality					115.10	24.33
2	Friendship conflict	−.22**				20.26	5.77
3	Anxiety	.04	.10			13.59	4.14
4	Somatization	.07	.18*	.20*		6.67	1.90
5	Depression	−.09	.17*	.36**	.56**	19.27	5.23

p < .05. **p < .01.

Significant correlations emerged also between the friendship quality measures and internalizing symptoms. The conflict score was found to be weakly correlated with depression and somatization symptoms. No correlations emerged between the positive friendship quality score and internalizing symptoms.

Analysis of Nonindependence

Prior to the analyses with the APIM, we investigated within-dyads nonindependence in the stability of best friend nomination. Friendship nomination stability had a PICC value of .41, with an asymptotic standard error of .12 and confidence limits of .18 and .64, overall revealing medium-to-large nonindependence between members of the same dyad on the outcome measure. This result indicates that, for members of reciprocated dyads, the decision to confirm (or alternatively, not confirm) the very best friend status to the other member through nomination is more likely to be reciprocal than unilateral. The presence of nonignorable interdependence on the outcome measure supported the adequacy of the APIM for the analysis of the dyadic data. Internalizing symptoms revealed only small interdependence in the dyads (Depression: ICC = .08; Anxiety: ICC = .09; Somatization: ICC = .08).

Actors’ and Partners’ Internalizing Symptoms as Predictors of Actors’ Friendship Stability

As a preliminary step to the estimation of the APIM model, two separate analyses were performed respectively evaluating the role of actors’ and partners’ internalizing symptoms on actors’ stability in nominating their partner across the two time points. In both analyses, dyadic friendship quality measures (positive, conflict) and gender were included as controls.

Tables 2 and 3 show the results of these preliminary analyses: Actor depression emerged as a significant negative predictor of actor’s stability of friendship nomination at T2; in addition, partners’ somatization emerged as a positive predictor of actors’ choice to confirm their best friendship nomination at T2. In both models, dyadic positive friendship quality as assessed at T1 emerged as a significant positive predictor of actor’s nomination stability at T2.

Table 2.

Actors’ Internalizing Symptoms as Predictors of Actors’ Friendship Nomination Stability, Controlling for Dyadic Friendship Quality and Gender.

	B	SE	OR	Wald χ²	df
Actor–Anxiety	0.40	0.22	1.50	3.23	1
Actor–Depression	−0.50	0.25	0.61	3.93*	1
Actor–Somatization	0.23	0.22	1.26	1.09	1
Dyadic positive friendship quality	0.67	0.28	1.95	5.69*	1
Dyadic friendship conflict	0.14	0.24	1.15	0.36	1
Gender (0 = female, 1 = male)	−0.44	0.46	0.64	0.93	1

p < .05.

Table 3.

Partners’ Internalizing Symptoms as Predictors of Actors’ Friendship Nomination Stability, Controlling for Dyadic Friendship Quality and Gender.

	B	SE	OR	Wald χ²	df
Partner–Anxiety	−0.08	0.23	0.92	0.13	1
Partner–Depression	−0.28	0.28	0.75	1.05	1
Partner–Somatization	0.47	0.21	1.60	5.18*	1
Dyadic positive friendship quality	0.69	0.29	2.00	5.56*	1
Dyadic friendship conflict	0.12	0.24	1.13	0.25	1
Gender (0 = female, 1 = male)	−0.38	0.45	0.69	0.71	1

p < .05.

Results of the analysis with the APIM, which are reported in Table 4, revealed both significant actor and partner effects. In particular, both actor and partner depression negatively predicted the stability of the actor best friend nomination at T2. In turn, no significant actor and partner effects of anxiety emerged. Higher partner somatization symptoms at T1 were found to increase the likelihood of stability of the actor’s best friend nomination at T2. No significant actor effect emerged for somatization. As expected, higher dyadic positive friendship quality at T1 was found to increase the likelihood of stability of the actor’s best friend nomination over time. No significant effect was found for the dyadic conflict measure. No gender effect emerged.

Table 4.

Actors and Partners’ Internalizing Symptoms as Predictors of Actors’ Friendship Nomination Stability, Controlling for Dyadic Friendship Quality and Gender.

	B	SE	OR	Wald χ²	df
Actor–Anxiety	0.51	0.26	1.66	3.83	1
Actor–Depression	−0.78	0.26	0.46	9.32**	1
Actor–Somatization	0.48	0.25	1.62	3.75	1
Partner–Anxiety	0.05	0.23	1.06	0.05	1
Partner–Depression	−0.62	0.29	0.54	4.72*	1
Partner–Somatization	0.71	0.26	2.03	7.35**	1
Dyadic positive friendship quality	0.61	0.28	1.84	4.76*	1
Dyadic friendship conflict	0.12	0.24	1.13	0.27	1
Gender (0 = female, 1 = male)	−0.58	0.49	0.56	1.39	1

p < .05. **p < .01.

Discussion

The main aim of the present study was to investigate the role of internalizing symptoms (i.e., depression, anxiety, and somatization) reported by adolescents, involved in a reciprocal very best friend relationship, in predicting the stability of their best friend choice over time. By employing a longitudinal dyadic analytic approach, for each participant, we were able to examine the role of both their individual and their best friend’s symptoms in predicting their choice to confirm, or alternatively, disconfirm, the very best friend status of their initially indicated partner. In the analyses, we also examined the role of gender and friendship quality as relevant covariates. Dyadic positive friendship quality was confirmed as a strong positive predictor of friendship continuation over time (Branje et al., 2007). In line with previous findings (Schneider et al., 1997), dyadic friendship conflict did not reveal a significant effect.

Concerning internalizing symptoms, both actor and partner effects emerged. Results from the present study indicate that symptoms of depression measured at the actor level are associated with lower stability of the actor’s friendship nomination over time. This finding is consistent with other studies (Prinstein, Borelli, Cheah, Simon, & Aikins, 2005; Rose et al., 2011) suggesting that depressive symptoms can have a disruptive effect on relationships with peers, possibly due to increased perceptions of interpersonal difficulties, and by increasing withdrawal from peers. Similarly, actors’ decision to confirm the best friend showed a significant association with their partners’ depressive symptoms: Adolescents reporting high levels of depressive symptoms were less likely to be confirmed as a best friend by their peers when compared with less depressed ones. This finding is consistent with many findings indicating the negative impact of depressive symptoms on peer relationships in adolescence: Depressed adolescents tend to be less attractive and popular among peers than adolescents reporting lower depressive symptoms (Field, Diego, & Sanders, 2001; Stice, Ragan, & Randall, 2004). However, in our study, the effect of partner depression was found to be significant only when controlling for actors’ symptoms, suggesting the existence of a suppression effect. That is, when both actor and partner effects were entered simultaneously in the model, it is possible that the presence of actors’ variables resulted in an increase in the magnitude of this relationship, inflating the relationship between the partner’s depression and actors’ friendship stability across the two time points. For this reason, this finding should be interpreted with caution.

In contrast with our findings on depression, no significant effect emerged when examining the role of adolescents’ and their friends’ anxiety symptoms in relation to the stability of their individual friendship choices. Literature on the impact of anxiety on peer relationship is somewhat inconclusive (La Greca & Harrison, 2005; Vernberg, Abwender, Ewell, & Beery, 1992). Recent findings have suggested that moderate levels of nonclinical anxiety on the actor side may predict higher friendship stability (Rose et al., 2011). As stated above, this link was not confirmed in our study. To our knowledge, no previous study has examined the impact of anxiety symptoms of both members of reciprocal dyadic friendships on the stability of their friendship. As a result, a thorough comparison with previous findings is not possible. Further studies are needed to better clarify the role of anxiety symptoms in friendship selection and maintenance among best friendship dyads.

The link between somatization symptoms and stability of friendship nomination at the actor level was not significant. Given the moderate correlation in our sample among depression and somatic symptoms, this result is not surprising and may indicate that the residual variance accounted by physical manifestations of distress in explaining individual friendship stability is negligible after controlling for negative affect. On the contrary, results regarding the effect of the partner’s somatization symptoms on the actor’s best friend nomination were unexpected. In our study, we found somatization symptoms as measured at the partner level to positively predict stability of the actor’s best friendship nomination. This could indicate that symptoms that are acted out primarily at a physical level may more likely elicit feelings of caring and support among mutual friends. Combined with results about depressive symptoms, this finding suggests that the impact of adolescents’ internalizing symptoms on the stability of their partners’ friendship nomination may be in part dependent on the nature of such symptoms.

As already stated by Hill and Swenson (2014), the association between internalizing problems and friendship is a complicated one. Nevertheless, the present study highlights that internalizing symptoms are not always detrimental to friendships; rather, in some cases, these symptoms may favor friendship stability. In fact, our findings indicate the existence of different patterns of relationships in predicting the stability of best friend choices of reciprocal friends when respectively examining the role of their own and their friends’ internalizing symptoms. Results also suggest that by investigating different facets of internalizing symptoms, as opposed to global indexes, differential roles may emerge. However, it is important to note that in our study, we examined only data coming from a nonclinical sample. Literature on clinical samples concerning the relationship among internalizing symptoms and friendship stability is lacking. Future research should address this issue.

The results of the present study should be understood in light of some limitations. First, the use of a small convenience sample indicates that caution should be applied in generalizing findings to other grade groups and the overall population. An additional limitation relates to the employed friendship nomination procedure. By limiting the very best friendship nomination procedure and friendship quality assessment to only one classroom peer, we compromised our ability to detect important information concerning the characteristics of adolescents’ best friendships network, and to evaluate if the importance of friendship quality in predicting friendship stability varies among friendships with different levels of closeness.

The present study also failed to investigate other characteristics that may be important for the stability of friendship over time, such as friendship length, and the degree of similarity (or dissimilarity) of the dyad’s members on the study measures (Hartl et al., 2015). Examining similarity effects represents a natural extension of our findings. However, given the small sample size employed in our study, statistical power concerns limited our ability to examine similarity/dissimilarity effects. For this reason, no information concerning the relative role of individual and dyadic-level (e.g., within-dyad average, dissimilarity) internalizing symptoms can be drawn from the present study. Future studies with larger samples might help clarify the relative importance of adolescents’ internalizing symptoms, and their similarity among friends, in influencing friendship continuation.

Finally, the current study relied on self-report measures about friendship dyads; the use of qualitative instruments, such as observation or interviews, in addition to the questionnaire could also have strengthened our research. In fact, qualitative instruments may help further our understanding of internalizing symptoms and friendship’s stability association.

This study also had several strengths. First, this study employed a longitudinal design. As a result, we were able to monitor specific friend’s selection behaviors over time. Secondstrength was the implementation of an APIM. By employing a dyadic analytic approach, we were able to highlight the role of internalizing symptoms as assessed from both members of a reciprocal dyad in relation to each members’ individual friendship choices. Finally, by investigating a set of internalizing symptoms instead of global composite index of distress, we were able to distinguish among different sources of effects on friendship stability. Findings from the present study highlight the relevant role of adolescents’ depression symptoms in promoting friendship dissolution. In turn, results suggest that manifesting distress at the physical level may promote friendship continuation. As a whole, this study provides novel findings highlighting the role of internalizing symptoms in friendship continuation in early adolescence.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Davide Marengo is postdoctoral fellow at the Department of Psychology of University of Turin, Italy. His main research interests are in the areas of applied psychometrics, content analysis, psychological and educational testing, and risk behaviors among adolescent and young adults. Recently, he has investigated the role of age differences in predicting different content-sharing behaviors on social media sites, as well as the relationship between activity on social media sites and self-report emotional well-being.

Emanuela Rabaglietti is associate professor in developmental psychology at Department of Psychology, University of Torino. Her research areas regard promotion of psychosocial well-being and risk prevention, particularly in relation to social relationships’ physical and cognitive activities during childhood and adolescence, and aging. She is coordinator of several research projects and is the author of numerous publications.

Franca Tani is full professor of developmental psychology in the Department of Health Sciences of the University of Florence and associate member of the International Psychoanalytical Association by virtue of associate membership in the Italian Psychoanalytical Association. The main research interests of her concern the development of social and emotional competences across lifespan with particular reference to networks of close relationships and to risk and protection factors in the intergenerational transmission.

References

Benenson

J. F.

Christakos

(2003). The greater fragility of females’ versus males’ closest same-sex friendships. Child Development, 74, 1123-1129. doi:10.1111/1467-8624.00596

Berndt

T. J.

(1982). The features and effects of friendship in early adolescence. Child Development, 53, 1447-1460. doi:10.2307/1130071

Berndt

T. J.

(1986). Children’s comments about their friendships. In Perlmutter

(Ed.), Cognitive perspectives on children’s social and behavioral development: The Minnesota symposia on child psychology (Vol. 18, pp. 189-212). Hillsdale, NJ: Lawrence Erlbaum.

Berndt

T. J.

(2002). Friendship quality and social development. Current Directions in Psychological Science, 11, 7-10. doi:10.1111/1467-8721.00157

Berndt

T. J.

Keefe

(1995). Friends’ influence on adolescents’ adjustment to school. Child Development, 66, 1312-1329. doi:10.2307/1131649

Betts

L. R.

Stiller

(2014). Centrality in children’s best friend networks: The role of social behaviour. British Journal of Developmental Psychology, 32, 34-49. doi:10.1111/bjdp.12020

Biggs

B. K.

Vernberg

E. M.

Y. P.

(2012). Social anxiety and adolescents’ friendships: The role of social withdrawal. The Journal of Early Adolescence, 32, 800-823. doi:10.1177/0272431611426145

Boivin

Hymel

Bukowski

W. M.

(1995). The roles of social withdrawal, peer rejection, and victimization by peers in predicting loneliness and depressed mood in childhood. Development and Psychopathology, 7, 765-785. doi:10.1017/S0954579400006830

Bowker

(2004). Predicting friendship stability during early adolescence. The Journal of Early Adolescence, 24, 85-112. doi:10.1177/0272431603262666

10.

Branje

S. J.

Frijns

Finkenauer

Engels

Meeus

(2007). You are my best friend: Commitment and stability in adolescents’ same-sex friendships. Personal Relationships, 14, 587-603. doi:10.1111/j.1475-6811.2007.00173.x

11.

Bukowski

W. M.

Hoza

Boivin

(1993). Popularity, friendship, and emotional adjustment during early adolescence. New Directions for Child and Adolescent Development, 1993, 23-37. doi:10.1002/cd.23219936004

12.

Bukowski

W. M.

Hoza

Boivin

(1994). Measuring friendship quality during pre-and early adolescence: The development and psychometric properties of the Friendship Qualities Scale. Journal of Social and Personal Relationships, 11, 471-484. doi:10.1177/0265407594113011

13.

Chan

Poulin

(2009). Monthly instability in early adolescent friendship networks and depressive symptoms. Social Development, 18, 1-23. doi:10.1111/j.1467-9507.2008.00461.x

14.

Ciairano

Rabaglietti

Roggero

Bonino

Beyers

(2007). Patterns of adolescent friendships and psychological adjustment and antisocial behavior: The moderating role of family stress and friendship reciprocity. International Journal of Behavioral Development, 31, 415-428. doi:10.1177/0165025407080573

15.

Cillessen

Jiang

X. L.

West

Laszkowski

(2005). Predictors of dyadic friendship quality in adolescence. International Journal of Behavioral Development, 29, 165-172. doi:10.1080/01650250444000360

16.

Cook

W. L.

Kenny

D. A.

(2005). The Actor-Partner Interdependence Model: A model of bidirectional effects in developmental studies. International Journal of Behavioral Development, 29, 101-109. doi:10.1080/01650250444000405

17.

Coyne

J. C.

(1976). Toward an interactional description of depression. Psychiatry, 39, 28-40.

18.

Ellis

Zarbatany

(2007). Explaining friendship formation and friendship stability: The role of children’s and friends’ aggression and victimization. Merrill-Palmer Quarterly, 53(1), 79-104. doi:10.1353/mpq.2007.0001

19.

Field

Diego

Sanders

(2001). Adolescent depression and risk factors. Adolescence, 36, 491-498.

20.

Flynn

Rudolph

K. D.

(2014). A prospective examination of emotional clarity, stress responses, and depressive symptoms during early adolescence. The Journal of Early Adolescence, 34, 923-939. doi:10.1177/0272431613513959

21.

Fonzi

Schneider

B. H.

Tani

Tomada

(1997). Predicting children’s friendships status from their dyadic interaction in structured situations of potential conflict. Child Development, 68, 496-506. doi:10.2307/1131674

22.

Fonzi

Tani

(1996). From childhood to early adolescence: Continuity and discontinuity in the development of friendship bonds [Dall’infanzia alla preadolescenza: continuità e discontinuità nello sviluppo dei legami di amicizia]. Evolutive Age [Età Evolutiva], 54, 89-94.

23.

Giletta

Scholte

R. H. J.

Prinstein

M. J.

Engels

R. C. M. E.

Rabaglietti

Burk

W. J.

(2012). Friendship context matters: Examining the domain specificity of alcohol and depression socialization among adolescents. Journal of Abnormal Child Psychology, 40, 1027-1043. doi:10.1007/s10802-012-9625-8

24.

Goodyer

Wright

Altham

(1990). The friendships and recent life events of anxious and depressed school-age children. The British Journal of Psychiatry, 156, 689-698. doi:10.1192/bjp.156.5.689

25.

Greenberg

M.T.

Kusche

C. A.

(1990). ‘‘Manual for seattle personality scale for children (Revised).’’ Seattle, WA: University of Washington.

26.

Hartl

A. C.

Laursen

Cillessen

A. H.

(2015). A survival analysis of adolescent friendships: The downside of dissimilarity. Psychological Science, 26, 1304-1315. doi:10.1177/0956797615588751

27.

Hartup

W. W.

(1996). The company they keep: Friendships and their developmental significance. Child Development, 67, 1-13. doi:10.2307/1131681

28.

Hill

E. N.

Swenson

L. P.

(2014). Perceptions of friendship among youth with distressed friends. Child Psychiatry & Human Development, 45, 99-109.

29.

Jellesma

F. C.

Rieffe

Terwogt

M. M.

(2008). My peers, my friend, and I: Peer interactions and somatic complaints in boys and girls. Social Science & Medicine, 66, 2195-2205. doi:10.1016/j.socscimed.2008.01.029

30.

Joiner

T. E.

Timmons

K. A.

(2009). Depression in its interpersonal context. In Gotlib

I. H.

Hammen

C. L.

(Eds.), Handbook of depression (2nd ed., pp. 322-339). New York, NY: Guilford Press.

31.

Kashy

D. A.

Kenny

D. A.

(2000). The analysis of data from dyads and groups. In Reis

H. T.

Judd

C. M.

(Eds.), Handbook of research methods in social and personality psychology (pp. 451-477). New York, NY: Cambridge University Press.

32.

Kenny

D. A.

Cook

(1999). Partner effects in relationship research: Conceptual issues, analytic difficulties, and illustrations. Personal Relationships, 6, 433-448. doi:10.1111/j.1475-6811.1999.tb00202.x

33.

Klima

Repetti

R. L.

(2008). Children’s peer relations and their psychological adjustment: Differences between close friendships and the larger peer group. Merrill-Palmer Quarterly, 54, 151-178. doi:10.1353/mpq.2008.0016

34.

Kochel

K. P.

Ladd

G. W.

Rudolph

K. D.

(2012). Longitudinal associations among youth depressive symptoms, peer victimization, and low peer acceptance: An interpersonal process perspective. Child Development, 83, 637-650. doi:10.1111/j.1467-8624.2011.01722.x

35.

Ladd

G. W.

Kochenderfer

B. J.

Coleman

C. C.

(1996). Friendship quality as a predictor of young children’s early school adjustment. Child Development, 67, 1103-1118. doi:10.2307/1131882

36.

Ladd

G. W.

Price

J. M.

(1987). Predicting children’s social and school adjustment following the transition from preschool to kindergarten. Child Development, 58, 1168-1189. doi:10.2307/1130613

37.

La Greca

A. M.

Harrison

H. M

. (2005). Adolescent peer relations, friendships, and romantic relationships: Do they predict social anxiety and depression? Journal of Clinical Child & Adolescent Psychology, 34, 49-61. doi:10.1207/s15374424jccp3401_5

38.

Loeys

Molenberghs

(2013). Modeling actor and partner effects in dyadic data when outcomes are categorical. Psychological Methods, 18, 220-236. doi:10.1037/a0030640

39.

Loeys

Cook

De Smet

Wietzker

Buysse

(2014). The actor–partner interdependence model for categorical dyadic data: A user-friendly guide to GEE. Personal Relationships, 21(2), 225-241. doi: 10.1111/pere.12028

40.

Olsen

J. A.

Kenny

D. A.

(2006). Structural equation modeling with interchangeable dyads. Psychological Methods, 11, 127-141. doi:10.1037/1082-989X.11.2.127

41.

Parker

J. G.

Asher

S. R.

(1993). Friendship and friendship quality in middle childhood: Links with peer group acceptance and feelings of loneliness and social dissatisfaction. Developmental Psychology, 29, 611-611. doi:10.1037/0012-1649.29.4.611

42.

Parker

J. G.

Seal

(1996). Forming, losing, renewing, and replacing friendships: Applying temporal parameters to the assessment of children’s friendship experiences. Child Development, 67, 2248-2268. doi:10.2307/1131621

43.

Poulin

Chan

(2010). Friendship stability and change in childhood and adolescence. Developmental Review, 30, 257-272. doi:10.1016/j.dr.2009.01.001

44.

Prinstein

M. J.

Borelli

J. L.

Cheah

C. S.

Simon

V. A.

Aikins

J. W.

(2005). Adolescent girls’ interpersonal vulnerability to depressive symptoms: A longitudinal examination of reassurance-seeking and peer relationships. Journal of Abnormal Psychology, 114, 676-688. doi:10.1037/0021-843X.114.4.676

45.

Rhee

Holditch-Davis

Miles

M. S.

(2005). Patterns of physical symptoms and relationships with psychosocial factors in adolescents. Psychosomatic Medicine, 67, 1006-1012. doi:10.1097/01.psy.0000188404.02876.8b

46.

Rose

A. J.

Asher

S. R.

(2004). Children’s strategies and goals in response to help-giving and help-seeking tasks within a friendship. Child Development, 75, 749-763. doi:10.1111/j.1467-8624.2004.00704.x

47.

Rose

A. J.

Carlson

Luebbe

A. M.

Schwartz-Mette

R. A.

Smith

R. R.

Swenson

L. P.

(2011). Predicting difficulties in youth’s friendships: Are anxiety symptoms as damaging as depressive symptoms? Merrill-Palmer Quarterly, 57, 244-262. doi:10.1353/mpq.2011.0013

48.

Rubin

K. H.

Wojslawowicz

J. C.

Rose-Krasnor

Booth-LaForce

Burgess

K. B.

(2006). The best friendships of shy/withdrawn children: Prevalence, stability, and relationship quality. Journal of Abnormal Child Psychology, 34, 139-153. doi:10.1007/s10802-005-9017-4

49.

Rudolph

K. D.

Flynn

Abaied

J. L.

(2008). A developmental perspective on interpersonal theories of youth depression. Handbook of depression in children and adolescents, 79-102.

50.

Rudolph

K. D.

Flynn

Abaied

J. L.

Groot

Thompson

(2009). Why is past depression the best predictor of future depression? Stress generation as a mechanism of depression continuity in girls. Journal of Clinical Child & Adolescent Psychology, 38, 473-485. doi:10.1080/15374410902976296

51.

Schneider

B. H.

Fonzi

Tani

Tomada

(1997). A cross-cultural exploration of the stability of children’s friendships and the predictors of their continuation. Social Development, 6, 322-339. doi:10.1111/j.1467-9507.1997.tb00109.x

52.

Stice

Ragan

Randall

(2004). Prospective relations between social support and depression: Differential direction of effects for parent and peer support? Journal of Abnormal Psychology, 113, 155-159. doi:10.1037/0021-843X.113.1.155

53.

Tani

Schneider

(1998). An instrument for the diagnosis of psychosocial risk in childhood: The Italian adaptation of the Seattle Personality Inventory (SPI-R.) of Greenberg [Uno strumento per la diagnosi del rischio psicosociale nell’infanzia: l’adattamento italiano del Seattle Personality Inventory (SPI-R.) di Greenberg]. Italian Journal of Psychology [Giornale Italiano Di Psicologia], 25, 409-428.

54.

Tani

Tomada

Guarnieri

Tonci

(2006). Well-being at school: Behavioural and relational antecedents of school participation. European Journal of School Psychology, 4, 13-31.

55.

Vernberg

E. M.

Abwender

D. A.

Ewell

K. K.

Beery

S. H.

(1992). Social anxiety and peer relationships in early adolescence: A prospective analysis. Journal of Clinical Child Psychology, 21, 189-196. doi:10.1207/s15374424jccp2102_11

56.

Vitaro

Boivin

Bukowski

W. M.

(2009). The role of friendship in child and adolescent psychosocial development. In Rubin

K. H.

Bukowski

Parker

J. G.

(Eds.), Handbook of peer interactions, relationships, and groups (pp. 568-585). New York, NY: Guilford Press.

57.

Wojslawowicz

J. C. W.

Rubin

K. H.

Burgess

K. B.

Booth-LaForce

Rose-Krasnor

(2006). Behavioral characteristics associated with stable and fluid best friendship patterns in middle childhood. Merrill-Palmer Quarterly, 52, 671-693. doi:10.1353/mpq.2006.0000

58.

Zou

Donner

(2004). Confidence interval estimation of the intraclass correlation coefficient for binary outcome data. Biometrics, 60, 807-811.