Introduction

Social interactions and the need to belong with peers are essential components of development that are present from early childhood and persist in their importance through adulthood (Baumeister & Leary, 1995; Harris, 1995). Peer rejection is associated with a host of negative outcomes (Beeson et al., 2020; Dodge et al., 2003; Nesdale & Zimmer-Gembeck, 2013; Sentse et al., 2017; Zimmer-Gembeck et al., 2010), including both internalizing and externalizing problems (Beeson et al., 2020). Critically, there is evidence that peer rejection not only occurs concurrently with negative outcomes such as depression, anxiety, and aggression, but may also temporally precede and succeed these issues (Kim & Cicchetti, 2010; Prinstein & Aikins, 2004; Trentacosta & Shaw, 2009). These patterns render unclear whether peer rejection is a risk factor for psychopathology, whether psychopathology is a risk factor for peer rejection, or both. Separate theoretical models reviewed below have emerged for each of these possibilities, and some researchers have conducted longitudinal studies to investigate the relationships between these issues over time. Nevertheless, to our knowledge, there are no published studies examining these relationships across the whole of development from childhood through the critical period of adolescence into young adulthood; further, investigators tend to study internalizing or externalizing problems without including both in the same model. The current study aimed to test the direction of effects by applying a developmental cascade model (Beeson et al., 2020; Masten & Cicchetti, 2010) with cross-lagged paths between peer rejection, depression, anxiety, and externalizing problems in two longitudinal samples spanning two decades of development beginning in childhood and extending through young adulthood.

Developmental Role of Peer Rejection

Children begin developing their earliest friendships during the toddler and preschool ages, but social comparisons based on shared feelings, values, and loyalty (Rose-Krasnor, 1997; Trentacosta & Shaw, 2009) render middle childhood (i.e., ages 6–12) a period of high significance for peer relations, and thus potentially higher consequences of rejection. In fact, transitioning to school is a time in which children may initially find themselves in a formal peer group. This setting allows critical chances to develop and practice social skills along with learning social norms (Gooren et al., 2011). Thus, the development of adequate peer relations during middle childhood is a critical developmental process. Unfortunately, a notable proportion of children are unable to successfully navigate this process and experience peer rejection. Peer rejection can be defined as a state of being actively disliked by peers, which is typically hurtful (Beeson et al., 2020; Coie et al., 1982). Though peer rejection can be limited to particular incidents, experiencing peer rejection remains fairly stable longitudinally and across genders, although some studies differ in these findings (Krygsman & Vaillancourt, 2017; Rubin et al., 2007). Higher levels of peer rejection generally result in social isolation or exclusion from social groups (Nesdale & Zimmer-Gembeck, 2013).

Researchers have posited that, due to reduced social standing, children who experience peer rejection have fewer opportunities to develop and practice social skills as well as to accrue knowledge about social norms and rules (Gooren et al., 2011). Consequently, the likelihood grows that these children will behave inappropriately or choose not to engage in social situations and will receive further negative responses from peers (Hartup, 1992; Rubin et al., 2007). There is evidence that children who experience higher levels of peer rejection are liable to develop negative biases in social cognitions and, therefore, are more likely to engage in negative maladaptive behaviors (Dodge et al., 2003). Thus it is sadly predictable that, as mentioned above, experiencing peer rejection relates to concurrent depression, anxiety, aggression, and other negative consequences (Beeson et al., 2020; Dodge et al., 2003; Morales et al., 2022; Sentse et al., 2017; Zimmer-Gembeck et al., 2010), although the directionality of the relationships are contested. In fact, competing models of the relationship between peer rejection and both internalizing and externalizing symptoms have emerged (reviewed by Beeson et al., 2020).

Models of Relations Between Peer Rejection and Psychopathology

The most commonly modeled version of these relationships maintains that experiencing peer rejection increases prospective risk for psychopathology, For example, within internalizing psychopathology research, this is referred to as the interpersonal risk model, whereas in externalizing problems research, this has been called the peer socialization model (Beeson et al., 2020; Rose & Rudolph, 2006; Sentse et al., 2010). There is a wealth of evidence supporting the increase in prospective risk model for both internalizing (e.g., Nolan et al., 2003; Zimmer-Gembeck et al., 2010) and externalizing (Haltigan & Vaillancourt, 2014; Ostrov, 2010) symptoms across development, but especially childhood compared to adolescence (Cillessen & Mayeux, 2004; Ladd, 2006). Notably, the vast majority of research on peer rejection in internalizing psychopathology have reported on depression and relatively few studies have included anxiety (Will et al., 2016).

In contrast, a separate body of studies has found that psychopathology increases prospective risk for peer rejection. This is less commonly studied for internalizing psychopathology, for which this model is referred to as the symptoms-driven model (Joiner, 2001; Vaillancourt et al., 2013). Individuals studying externalizing problems have referred to this psychopathology increasing risk for peer rejection as the social process model (Boivin & Hymel, 1997; Ostrov, 2008; Ostrov et al., 2011).

The final model, referred to as the transactional model (Sameroff, 2009) posits a bidirectional relationship between psychopathology and peer rejection such that increases in either one over time increases the risk for the other and can lead to feedback loops in which risk reciprocally increases. Reasonable support exists for this model across adolescence for both internalizing and externalizing problems (Beeson et al., 2020; Ferguson et al., 2016; Zimmer-Gembeck et al., 2016). There is also some support for a model in which conduct problems result in peer rejection, which ultimately predicts internalizing symptoms (Gooren et al., 2011).

It is also worth noting that although typically there are not meaningful gender differences found in mean levels of peer rejection (Krygsman & Vaillancourt, 2017; Rubin et al., 2007), peer relations are often found to have differing levels of importance between genders (Rose & Rudolph, 2006) for development. There are also meta-analytic findings that the relation between social support and depression in youth does not differ across genders (Rueger et al., 2016). For peer rejection specifically, although this area is understudied, there is some evidence that peer rejection may have a stronger effect on biological stress response in females (Stroud et al., 2017) and that emotion dysregulation mediates in the relation of peer rejection to depression in older boys (Fussner et al., 2016).

Given these conflicting models and the proposed complex relationships between these variables over time, developmental cascade models are a relevant and appropriate method of approaching longitudinal studies of these variables. Developmental cascades are the relationships among levels, domains, and systems across development that aggregate into resulting symptoms or phenomena over time (Beeson et al., 2020; Masten & Cicchetti, 2010). Crucially, cascade models can test the direction of particular effects and whether those effects are direct or indirect while taking into account stability in constructs along with concurrent relationships between constructs (Beeson et al., 2020; Masten & Cicchetti, 2010).

Current Study

The importance of peer rejection across development has been established by the reviewed literature, yet to our knowledge, few if any studies have examined its stability, correlates, and prospective relationships across the wide developmental span of middle childhood through young adulthood. Understanding the full arc of the role of peer rejection, particularly in relation to internalizing and externalizing problems, is crucial to providing context to previous research findings as well as illuminating potential key points of intervention and warning signs of future suffering. In particular, many researchers posit that the middle childhood period is a critical period for the development of adequate peer relations, and thus peer rejection in this period is significantly more deleterious than in adolescence. Further, the competing models reviewed above render unclear whether social or psychopathological interventions are more likely to disrupt the developmental trajectories of peer rejection and/or psychopathology. Consequently, in the current project, we studied two longitudinal community samples of individuals from middle childhood to young adulthood.

Aims of the current study were to: (1) ascertain mean levels and stability of peer rejection across this developmental span, (2) measure the concurrent relationships of peer rejection with demographic variables (i.e., age, gender, parental socioeconomic status, parental educational attainment), (3) evaluate the concurrent relationships of peer rejection with depression, anxiety, and externalizing problems, and (4) fit developmental cascade models to test directionality and examine patterns of relationships between peer rejection, depression, anxiety, and externalizing problems across development. We then used these results to evaluate; (1) the importance of middle childhood peer rejection compared to other time periods, and (2) patterns of temporal precedence of peer rejection versus psychopathological symptoms.

Method

Participants

This study is part of a broader research project on the relationships among emotional problems, behavior problems, and learning difficulties in 2nd, 4th, and 6th grades in public schools in the Coimbra region of Portugal (LoParo et al., 2022). For the current study, data for the 2nd and 4th grade cohorts were employed because they were the only samples with longitudinal data. At the time of initial data collection, there were 445 participants in the 2nd grade cohort (46.0% female as reported by parents) and 448 participants in the 4th grade cohort (47.9% female as reported by parents). Given the demographic makeup of the Coimbra region, the sample almost exclusively comprises White individuals. According to the Portuguese SES classification system of Simões et al. (1995), the majority of participants came from the upper middle and lower middle class (76.2%), with 15.6% and 9.2% coming from the upper and lower classes, respectively. The 2nd grade cohort participated in five data collection points (Wave 1: 1993 [N = 445, mean age = 7.5, SD = 0.81], Wave 2: 1997 [N = 445, mean age = 11.9, SD = 0.96], Wave 3: 2000 [N = 426, mean age = 14.7, SD = 0.91], Wave 4: 2003–2004 [N = 410, mean age = 18.1, SD = 1.19], Wave 5: 2011–2013 [N = 417, mean age = 26.6, SD = 1.13]) and the 4th grade cohort participated in 3 data collection points (Wave 1: 1993 [N = 448, mean age = 9.7, SD = 0.81], Wave 4: 2003–2004 [N = 393, mean age = 18.6, SD = 1.44], Wave 5: 2012–2013 [N = 330, mean age = 29.5, SD = 1.05]).

Procedure

The study was approved by school boards, and consent and assent forms were obtained, respectively, from parents and participants. The sample was selected at Wave 1 (1993) using simple randomization and stratification techniques. First, schools in the Coimbra region were selected with probability proportional to the number of enrolled students. Second, classes within each school were randomly selected from the 2nd and 4th grades. Participating students were given a packet of questionnaires and a letter explaining the project in a sealed envelope to take home to their parents. Teachers of the selected classes were asked to complete questionnaires regarding the behavior of students. The students completed self-report measures of behavior problems. After this initial data collection, participants were contacted and assessed, in the follow-up, generally in small groups at their schools, except for the last wave for whom in-person interviews were conducted and they also completed several questionnaires assessing several aspects of their functioning, including anxiety and depression.

Measures

Achenbach System of Empirically Based Assessments (ASEBA)

Teacher Report Form (TRF) (Achenbach, 1991; Achenbach et al., 2001)

The TRF is a 113-item questionnaire completed by teachers who were asked to indicate how often the behavior described in each item is true of a student using a 3-point scale. TRF scoring includes multiple scales comprised of items from the questionnaire. For this study, peer rejection was measured by averaging responses to four items available on the ASEBA forms (“Feels or complains that no one loves him/her”, “feels others are out to get him/her”, “gets teased a lot” “not liked by others”). At Wave 1, we used scores from these items for the TRF in both cohorts. For Waves 2 through 5, we used responses to the self-report measures due to data availability (YSR and ASR). Items were also drawn from the DSM Depression and Anxiety scales, created to reflect DSM diagnostic categories (Achenbach & Dumenci, 2001) to measure participant’s depression and anxiety. Only items from the DSM Depression and DSM Anxiety scales that were common across all measured timepoints were used; this produced 9 items measuring depression (enjoys little, cries a lot, deliberately tries to hurt self, feels worthless, feels guilty, feels overtired, thinks about suicide, lacks energy, feels unhappy) and 4 items measuring anxiety (fears certain animals/situations/places, nervous, too fearful or anxious, worries). the total raw score of the Externalizing scale was used as well.

Youth Self Report Form (YSR) (Achenbach, 1991; Achenbach et al., 2001)

The YSR is a 112-item questionnaire completed by youth 11 years to 18 years old. Its format and scoring is identical to the TRF.

Adult Self Report Form (ASR) (Achenbach et al., 2001)

The ASR is a 110 item questionnaire completed by young adults between the ages of 18 and 30. Its format and scoring is identical to the TRF and YSR; however, some items are different so the items comprising each scale differ slightly (e.g., the youth forms include questions about fearing school, which are not part of the adult scale).

Analyses

All analyses were conducted with MPlus Version 8 and SPSS version 28. To maximize the number of participants with useable data at each wave (i.e., there was no YSR at Wave 1 and many less TRFs and CBCLs available in subsequent waves), data were drawn from the Wave 1 TRF, the Waves 2–4 YSR, and the Wave 5 ASR (N’s are available in Table 1). Little’s Missing Completely At Random (MCAR) test was conducted separately for each sample and both the 2nd grade (p = 0.318) and 4th grade (p = 0.828) were deemed to be MCAR. This means that there were no systematic differences between individuals with missing values (i.e., the probability of having missing data was not significantly associated with any study variables), thus Full Information Maximum Likelihood (FIML) was employed to manage missing data due to meeting MCAR assumptions.

Correlation analyses were conducted using Spearman’s rho due to the non-normal distribution of the variables. All depression, anxiety, externalizing, and peer rejection items were modeled as count variables with a Poisson distribution in the developmental cascade models to account for their low frequency of endorsement. This means that that in the developmental cascade analyses, these regressions are Poisson regressions and thus expect the distributions of the variables to follow a Poisson distribution. The reported standardized coefficients are interpreted the same way as common correlation coefficients.

Psychopathology variables were modeled as latent variables drawn from previous analyses (for details, see LoParo et al., 2022). Developmental cascade modeling used a series of nested models, and statistical fit between the models was assessed at each step; for all models Chi Square, Root Mean Square Error of Approximation (RMSEA), Confirmatory Fit Index (CFI), Tucker–Lewis Index (TLI), Standardized Root Mean Square Residual (SRMR), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). The Satorra-Bentler scaled chi-square test was used to compare nested models.

Results

Peer Rejection Across Development

Psychometric properties of peer rejection variables in both samples at all assessment time points are available in Table 1. Notably, in all instances scores are skewed toward 0, and this led us analytically to treat these as count variables. Further, mean rejection levels were quite stable other than a spike at age 12 and mild increase in young adulthood (see Fig. 1).

Table 1 Peer Rejection Psychometric Properties
Full size table
Fig. 1
figure 1

Mean Peer Rejection over Time

Full size image

Next, we analyzed concurrent correlations of peer rejection with demographic variables (see Table 2). Generally, rejection was mildly correlated with older age within early waves of data collection, but not during later waves. Gender was not correlated with rejection at any age. Lower parental SES and parental years of education were mildly correlated with higher peer rejection in pre-adolescence but not in adolescence or young adulthood.

Table 2 Peer Rejection Spearman’s Rho Correlation with Demographic Variables
Full size table

Finally, we analyzed correlations of peer rejection scores over each wave of data collection (see Table 3). There were mild to moderate correlations between peer rejection scores that were closer together in time, and the level of correlation attenuated as timepoints that were further apart.

Table 3 Spearman’s rho statistics of association between Peer Rejection Variables across Ages
Full size table

Peer Rejection and Psychopathology

While controlling for age and gender by partialling them out of the variables of interest, we analyzed the association of peer rejection with concurrent depression, anxiety, and externalizing problems. In the 2nd Grade Cohort, at age 7.5, peer rejection was associated with depression (Beta = 0.29, p = 4 × 10–8), anxiety (Beta = 0.30, p = 8X10−9), and externalizing symptoms (Beta = 0.43, p = 1.2 × 10–12). At age 12, peer rejection was associated with depression (Beta = 0.41, p = 1 × 10–15), anxiety (Beta = 0.30, p = 1 × 10–9), and externalizing symptoms (Beta = 0.50, p = 1.1 × 10–20). At age 15, peer rejection was associated with depression (Beta = 0.43, p = 1 × 10–17), anxiety (Beta = 0.31, p = 4 × 10–10), and externalizing problems (Beta = 0.40, p = 1.1 × 10–14). At age 18, peer rejection was associated with depression (Beta = 0.46, p = 1 × 10–18), anxiety (Beta = 0.31, p = 2 × 10–10), and externalizing problems (Beta = 0.36, p = 2.2 × 10–12). At age 27, peer rejection was associated with depression (Beta = 0.35, p = 7 × 10–12), anxiety (Beta = 0.37, p = 4 × 10–13), and externalizing problems (Beta = 0.34, p = 2.2 × 10–12). Mean differences between genders in these samples have been previously reported (LoParo et al., 2022) such that depression and anxiety levels are equivalent between genders at Wave 1 and females have higher levels at all future timepoints, whereas externalizing symptoms were higher for males at Wave 1 and equivalent at future timepoints. As an exploratory analysis, we also tested the effect of interactions between gender and peer rejection on psychopathology variables (see Supplementary Table 1). While most interactions were nonsignficiant, the effect of peer rejection on externalizing psychopathology appeared stronger for males than females at age 7.5, and the effect of peer rejection of depression was stronger in females at ages 15 and 27.

In the 4th grade Cohort, at age 9.7, depression (Beta = 0.44, p = 1.1 × 10–16), anxiety (Beta = 0.36, p = 2 × 10–12), and externalizing problems (Beta = 0.45, p = 4.6 × 10–15). At age 18.5, depression (Beta = 0.43, p = 3 × 10–16), anxiety (Beta = 0.31, p = 7 × 10–10), and externalizing problems (Beta = 0.38, p = 1.8 × 10–12). At age 29.5. depression (Beta = 0.43, p = 3 × 10–16), anxiety (Beta = 0.31, p = 7 × 10–10), and externalizing problems (Beta = 0.41, p = 1.7034 × 10–14). We also tested the effect of interactions between gender and peer rejection on psychopathology variables in this sample (see Supplementary Table 1). There were no significant interactions in these analyses.

Developmental Cascade Model

Using a developmental cascade approach (Masten & Cicchetti, 2010), we tested alternative models of associations between depression, anxiety, externalizing problems, and peer rejection. The baseline model included only within-time covariances between the variables. In Model 2, we added all one-year across-time pathways. Model 3 included cross-lagged paths between the four variables at adjacent time points. Model 4 removed nonsignificant pathways from Model 4. These analyses were conducted separately in the two cohorts, and in both cases, Model 4 was chosen as the best fitting model due to the fit statistics available in Table 4. The standardized model parameters for these two models are available in Figs. 2 and 3.

Table 4 Fit Statistics for Developmental Cascade Models
Full size table
Fig. 2
figure 2

Developmental Cascade Model of Depression, Anxiety, Peer Rejection, and Externalizing Problems in 2nd Grade Cohort. Note: Dep = Depression, Anx = Anxiety, Rej = Peer Rejection, Ext = Externalizing Problems. Standardized parameter estimates are presented and are significant at p < 0.05. All correlations between variables are significant but not presented for legibility

Full size image
Fig. 3
figure 3

Developmental Cascade Model of Depression, Anxiety, Peer Rejection, and Externalizing Problems in 4th Grade Cohort. Note: Dep = Depression, Anx = Anxiety, Rej = Peer Rejection, Ext = Externalizing Problems. Standardized parameter estimates are presented and are significant at p < 0.05. All correlations between variables are significant but not presented for legibility

Full size image

In the best fitting models, we found that all four variables were positively correlated concurrently at each time point, matching the results presented above. As a general trend in the 2nd grade cohort, most across-time paths were significant, meaning that higher levels of depression, anxiety, peer rejection, or externalizing problems at one timepoint were associated with higher levels of these same issues at the following timepoint. There were also several significant cross-lagged paths. Notably, higher anxiety at each timepoint was associated with higher levels of all four variables at each subsequent timepoint. Age 7 depression predicted age 12 anxiety and peer rejection. Age 12 peer rejection predicted age 15 depression. Age 15 depression predicted age 18 peer rejection and externalizing problems. Finally, age 18 depression predicted age 26 anxiety. Finally, we fit a model in which paths that were not significant at an alpha of p < 0.001 were removed; the results are available in Supplemental Figs. 1 and 2.

In the 4th grade cohort, likely due to the increased time between data collection points, there were fewer significant across-time paths. As for cross-lagged paths, Age 9.7 peer rejection predicted age 18.5 depression, age 9.7 externalizing problems predicted age 18.5 anxiety, and Age 18.5 depression predicted all four variables at age 29.5.

Discussion

This study used two longitudinal community samples recruited from Portuguese schools with data from middle childhood through early adulthood to investigate peer rejection, its correlates, and its developmental relationships with depression, anxiety, and externalizing problems. Specifically, we measured mean levels and assessed stability of peer rejection across development, measured peer rejection’s association with demographic and psychopathology variables concurrently across development, and fit developmental cascade models of peer rejection, depression, anxiety, and externalizing problems.

Peer rejection variables, as assessed by four items available on the ASEBA system of measurement, had acceptable internal consistency at each timepoint in both samples. Mean levels of endorsement of peer rejection were low (ranging between ~0.6 and ~1.2 on a scale with a maximum of 8) and were skewed toward 0. Further, mean peer rejection levels remained relatively stable over time, although there was a spike at age 12 and a mild increase in young adulthood. We also found that peer rejection scores were mildly to moderately correlated at measurement points closer together in time, but these correlations attenuated at timepoints that were temporally further apart. Taken together, these findings indicate that peer rejection occurs rarely for the average individual and the tendency to experience peer rejection is not highly stable across the entirety of development, although it is mildly to moderately stable over shorter periods of time. Even though previous studies have used a variety of timespans and measures of peer rejection, these results are quite consistent with the distributions and correlations found in other studies with similar methodologies (Beeson et al., 2020; Krygsman & Vaillancourt, 2017). Further work in this area is needed to clarify to what degree the stability of peer rejection is due to static social groups with established peer dynamics versus an individual liability to experience peer rejection inherent to individuals.

When evaluating the concurrent correlations between peer rejection and demographic variables, we found that prior to adulthood, there was a slight tendency for individuals older at the time of assessment to experience more peer rejection. This is somewhat consistent with findings that mean peer rejection levels appear to increase slightly as participants age. There was no evidence of gender differences in mean peer rejection at any timepoint in either sample. Previous studies have found inconsistent evidence for gender differences in peer rejection, both in its significance and direction (Beeson et al., 2020; Veenstra et al., 2010). These differences may be attributable to the particular measures used, age at assessment, or cultural differences in samples. Despite similar rates of peer rejection, the importance of peer relations may differ across genders and thus the impact of peer rejection on psychopathology or other aspects of adjustment may differ as well (DeRosier et al., 1994; van Lier et al., 2005). Low parental socioeconomic status and education were mildly and positively correlated with more peer rejection in preadolescence but not in adolescence or adulthood. To our knowledge, correlations of peer rejection and parental SES or education have not been previously reported. These results may indicate that familial standing has some impact on rejection earlier in life, but by adolescence individuals’ peer relations transcend family background.

Next we calculated the concurrent correlation between peer rejection and depression, anxiety, and externalizing problems. We found that peer rejection was significantly correlated at the r = ~0.3–0.5 level with all three variables at each timepoint in both samples. Thus, this medium-to-large effect was remarkably consistent across both time and form of psychopathology. This is also quite consistent with previously reported findings (Krygsman & Vaillancourt, 2017; Will et al., 2016; Wood et al., 2010). These results indicate that at any given point in an individual’s development, rejection by peers is likely to co-occur with psychopathology and, regardless of causal direction, seems to be a nonspecific correlate of psychopathology. Further, it does not appear that peer rejection in middle childhood (i.e., Wave 1 in both cohorts and Wave 2 in the 2nd grade cohort) indicates higher concurrent risk of psychopathology than other time points, which may indicate that theories regarding the importance of peer rejection in middle childhood (Gooren et al., 2011) are not supported in terms of concurrent risk.

We tested alternative developmental cascade models in both samples to increase our capability of making longitudinal, directional inferences. The best fitting model in both samples allowed cross-lagged paths and pruned nonsignificant paths. Peer rejection significantly but weakly predicted psychopathology at a later timepoint only once in each sample (age 12 peer rejection to age 15 depression in the 2nd grade cohort and age 9.7 peer rejection to age 18.5 depression in the 4th grade cohort). In this case, the interpersonal risk model was supported. It is not difficult to imagine that during pre-adolescence an individual who experiences peer rejection is more likely to have more depressive symptoms in adolescence. Thus, this pattern of relationships provides some support for the importance of peer rejection in middle childhood in the case of prospective risk for depression. Nevertheless, the correlation is weak in both samples and did not extend to other forms of psychopathology, indicating that peer rejection may be best described as a stronger indicator of concurrent psychopathology at any point in development rather than of significant importance to prospective general psychopathology risk during middle childhood.

In contrast, there was more robust support for the symptoms-driven model. Anxiety significantly predicted future peer rejection at each timepoint in the 2nd grade cohort, and depression predicted future peer rejection from age 7.5 to 12 and from age 15 to 18 in the 2nd grade cohort and age 18.5 to 29.5 in the 4th grade cohort. There is ample empirical support that both depression and anxiety are associated with reduced social skills development (Angélico et al., 2013; Segrin, 2000; Yao & Enright, 2021). It may be the case that internalizing psychopathology, whether through reduction in social experiences or shared etiology, increases social deficits which ultimately lead to peer rejection further along in time. In addition, social withdrawal due to mood concerns or avoidance due to distress may evoke rejection from peers in the future, consistent with models posited in the literature (DeRosier et al., 1994; Ladd, 2006).

Support for the transactional model includes the indirect paths that pass reciprocally through depression and peer rejection from ages 7.5 to 18 in the 2nd grade cohort and age 9.7 to 29.5 in the 4th grade cohort. It may be the case that for some individuals, depression and peer rejection reciprocally influence each other over time, creating a “downward spiral” effect that is often noted clinically. According to these findings, improving social relationships early in development (Arnarson & Craighead, 2009) may play an important preventative role in the development of depression.

Other notable results from the developmental cascade models include the robustness of anxiety as a precedent for peer rejection as well as depression and externalizing problems in the 2nd grade cohort. These results indicate that increased anxiety may be a valuable warning sign for future broad psychopathological and social risk that warrants further investigation (Merikangas et al., 2009). In addition, we found that despite empirical support for a longitudinal connection between peer rejection and externalizing problems (Haltigan & Vaillancourt, 2014; Ostrov et al., 2011), these paths were not significant in either sample. It may be the case that such connections in other studies are due to indirect influence through internalizing psychopathology—notably, another study using a developmental cascade model including depression, aggression, and peer rejection had similar findings (Beeson et al., 2020). This highlights the importance of multivariate longitudinal models in developmental research.

Two goals of this research were to evaluate: (1) the importance of middle childhood peer rejection compared to other time periods, and (2) patterns of temporal precedence of peer rejection versus psychopathological symptoms. First, looking across the findings in our samples it did not appear that peer rejection in middle childhood indicated higher concurrent risk nor generally higher prospective risk for psychopathology, although there was some limited support for peer rejection in middle childhood predicting future depression. Second, although there was no clear pattern across all levels of the developmental cascades, generally there was more support for the symptoms-driven model of prospective risk, and even transactional effects were more likely to “start” with depression than peer rejection. Thus, we posit that peer rejection is more likely to be a concurrent and developmental consequence of psychopathology than vice versa, and that when considering the full course of development, the degree to which peer rejection contributes to the development of psychopathology is more often an exacerbation of psychopathology than as a starting point. This is further supported by our findings that psychopathology is more developmentally stable than peer rejection.

Finally, it is worth considering the clinical implications of these findings. First, given that peer rejection may be more apparent to school staff, parents, or even youth themselves, than psychopathology, those experiencing peer rejection should be screened for current psychopathology. Second, in children seeking treatment for depression and anxiety social skills training, promotion of social engagement, and reducing avoidance of social distress may be important treatment targets to reduce risk of future peer rejection, which may reinforce or exacerbate psychopathology. Third, although peer rejection and externalizing psychopathology are oft linked, clinicians should be sure to examine the role of concurrent internalizing psychopathology when attempting to reduce risk for further social or psychopathological issues.

There are limitations of the current study. First, the relative cultural and demographic homogeneity of the Coimbra region of Portugal may limit the generalizability of the results. In fact, there is some evidence that the prevalence and importance of poor social relations and their association with social-behavioral outcomes can differ by culture across American, Portuguese, and South Korean samples, for example (Oh et al., 2021). Portugal is considered to have a moderate level of expected adherence to social norms compared to samples of English speaking (low expected adherence) and Asian speaking (high expected adherence) (Oh et al., 2021). Thus, it is plausible that cultural norms in Portugal (as compared to other cultures) regarding the importance of peer relationships might lead to a different transactional relationship with psychopathology if those relationships are disrupted. Second, gaps in time between waves of assessment suggest the possibility that fluctuations in mean levels of variables and the paths between them may occur between waves and were not captured, particularly in the 4th grade sample. Relatedly, the lack of self-report data available at the first assessment for Wave 1 forced the use of teacher-reported data. Third, biological sex was not assessed, and only binary gender was assessed by parent report. Further, although we found some evidence that the effect of peer rejection may differ by gender on concurrent psychopathology, it was outside the power and scope of this study to more comprehensively assess gender differences in the developmental cascade models given the lack of consistency in the interaction analyses that we conducted.

Beyond replicating these results in more diverse samples, future studies can expand on these findings in several ways. First, there may be particular forms of peer rejection that we were not able to parse that have particular relations to psychopathology. More robust measures of peer rejection should be developed to investigate these possibilities. Second, longitudinal clinical samples may be able to elucidate whether the concurrent and prospective relations of these variables differ at clinical levels of psychopathology. Third, given findings that social relationships are often found to function differently between genders (Rose & Rudolph, 2006), more research should examine gender differences in both concurrent and longitudinal relations among these variables. Finally, clinical researchers should investigate whether interventions aimed at either reducing psychopathology or improving peer relations have a prospective effect on peer relations or psychopathology, respectively, at various points of development. This could help identify and prioritize critical periods and targets of intervention with the broadest impact.