Within-Person Reciprocal Associations between Peer Victimization and Need for Approval

Abstract

Prior research links need for approval (NFA; the extent to which self-worth is contingent on peer approval or disapproval) to critical developmental outcomes, but little is known about how NFA develops over time or within social contexts. To address this gap, the present study used a sophisticated analytic approach (autoregressive latent trajectory modeling with standardized residuals) to examine dynamic associations between one salient social experience – peer victimization – and two dimensions of NFA, conceptualized in terms of approach motivation (NFA_approach; enhanced self-worth based on peer approval) and avoidance motivation (NFA_avoid; depleted self-worth based on peer disapproval). Following 636 youth (338 girls; M_age = 7.96 years at Wave 1; 66.7% White; 35.0% subsidized school lunch) from second to seventh grade, analyses revealed that peer victimization predicts subsequent increases in NFA_avoid, which in turn predicts subsequent increases in victimization. Findings also revealed that although mean levels of NFA_avoid decrease during childhood, increases or decreases in NFA become more entrenched. Thus, childhood peer victimization may disrupt normative decreases in NFA_avoid and contribute to a cycle in which negative peer judgments increasingly foster low self-worth and further peer difficulties. Preventing this cycle may require encouraging peer-victimized youth to base their self-worth on internal standards rather than peer feedback while helping them develop positive relationships that promote self-worth.

Social belonging is a fundamental human need that scaffolds how individuals process social cues and respond to the environment (Baumeister & Leary, 1995). According to sociometer theory (Leary et al., 1995), state self-esteem serves as a subjective marker of one’s level of social belonging, suggesting that self-worth might be contingent in part on receiving approval or disapproval from others. Because the peer group serves as a primary context of socialization during childhood and early adolescence (Masten et al., 2009; Rubin et., 2005), peer belonging may be particularly influential for self-esteem. One manifestation of this contingent self-worth is need for approval (NFA), conceptualized as the extent to which self-worth is contingent on positive and negative peer judgments (Rudolph et al., 2005).

A growing body of research implicates NFA within the peer context as a critical aspect of contingent self-worth, with accompanying consequences for children’s social and emotional development (for a review, see Rudolph, 2021). Although integrating the appraisals of peers into one’s self-concept is part of a normative developmental process (Harter, 1998), individual differences may emerge in the extent to which youth rely on peers’ approval for their self-worth. However, little is known about how these individual differences develop over time or intersect with youths’ social experiences. The goal of the present study was to track the development of NFA from middle childhood through early adolescence and to understand the dynamic interchange between one salient social experience (i.e., peer victimization) and NFA from peers.

Conceptualization of Need for Approval

Drawing from theories of contingent self-worth (Crocker & Wolfe, 2001; Deci & Ryan, 1995), NFA is conceptualized as the extent to which positive peer judgments augment one’s sense of self-worth (e.g., feeling proud of oneself in the face of social approval) versus the extent to which negative peer judgments weaken one’s sense of self-worth (e.g., feeling ashamed of oneself in the face of social disapproval). Consistent with theories of social motivation (Gable, 2006; Rudolph, 2021), the former dimension is viewed as approach-oriented (NFA_approach) and the latter dimension is viewed as avoidance-oriented (NFA_avoid). Whereas approach-oriented individuals are sensitive to social rewards, such as belonging, approval, and acceptance, avoidance-oriented individuals are sensitive to social punishments, such as non-belonging, disapproval, and rejection (Gable, 2006; Rudolph, 2021). Prior research supports this two-dimensional structure and suggests that NFA is distinct from global self-worth (Liu & Huang, 2018; Rudolph et al., 2005). Although NFA_approach and NFA_avoid are moderately positively correlated (Rudolph & Bohn, 2014; Rudolph et al., 2005) and share variance associated with general sensitivity to social cues (Rudolph, 2021), these two dimensions have distinct and often opposing implications for emotional and behavioral adjustment (Liu & Huang, 2018; Rudolph & Bohn, 2014; for a review, see Rudolph, 2021).

At the core of NFA is the idea that external experiences are interpreted in self-relevant ways—positive experiences are internalized as self-promoting whereas negative experiences are internalized as self-threatening. This self-relevant processing is presumed to energize approach/avoidance motivations, which may then prompt approach/avoidance-oriented cognitions and emotions (e.g., rejection sensitivity), goals and strategies (e.g., conflict avoidance), and behaviors (e.g., social withdrawal) (Park, 2010; Rudolph, 2021).

Reciprocal Association between Peer Victimization and NFA

During childhood and adolescence, peer relationships become more salient (Parker et al., 2006), and youth become increasingly concerned about peer evaluation and belonging (Laursen, 1996; Rubin et al., 2005). Thus, youth who experience frequent or severe peer victimization likely have unmet belonging needs. Over time, they may internalize negative peer judgments into their self-concept, leading them to become increasingly sensitive to disapproval. In addition, peer victimization triggers low self-esteem (van Geel et al., 2018), which may cause youth to adopt self-protective (e.g., avoidance) goals to prevent their self-esteem from further decreasing (Park, 2010). Conversely, victimized youth may struggle to integrate positive social feedback into their sense of self, resulting in ignoring or discounting peers’ approval because such positive responses are inconsistent with their previous experiences and self-concept (Rudolph, 2021; Skymba et al., 2022). Thus, victimization may cause youth to develop higher levels of NFA_avoid and lower levels of NFA_approach over time.

In turn, individual differences in NFA may predict subsequent peer victimization. Youth with high NFA_approach strive to gain peers’ approval to enhance their sense of worth. This focus on peer approval may motivate them to engage in behaviors (e.g., cooperation) that enhance relationships and to suppress behaviors (e.g., overt aggression, withdrawal) that jeopardize relationships, reducing their likelihood of being victimized (Rudolph & Bohn, 2014; Sugimura et al., 2017). Youth with high NFA_avoid strive to avoid disapproval from peers to protect their self-worth, potentially putting them at risk for avoidant behavior that leaves them at risk for victimization (Rudolph & Bohn, 2014). Youth with high NFA_avoid also may present self-deprecating signals of vulnerability (e.g., sadness, fear, cautiousness), attracting negative attention from bullies (Egan & Perry, 1998; Rudolph & Bohn, 2014).

Consistent with the idea that peer victimization may predict subsequent NFA, youth who suffer frequent peer victimization or general peer stress show heightened rejection sensitivity (tendency to anxiously expect and overreact to interpersonal rejection; Zimmer-Gembeck et al., 2014) and negative relational self-views (beliefs that one is incompetent and unworthy while interacting with peers; Caldwell et al., 2004). Moreover, self-reports of peer adversity (Skymba et al., 2022) are concurrently associated with higher levels of NFA_avoid and lower levels of NFA_approach among adolescent girls. Consistent with the idea that NFA may predict subsequent peer victimization, analyses in a subset of children from the present sample revealed that NFA_approach predicts less victimization whereas NFA_avoid predicts more victimization from second to third grade (Rudolph & Bohn, 2014).

Despite this preliminary evidence, prior studies were largely based on cross-sectional designs, and the one longitudinal study examining NFA as a predictor of peer victimization focused on a very limited temporal frame, one specific developmental stage, and a single direction of effect rather than reciprocal associations. Moreover, studies used analytic approaches that did not distinguish between-person change (shifts in rank order of one’s score within the overall sample) and within-person change (shifts in the deviation of one’s score from one’s average score over time). This constraint leads to interpretational difficulties, often resulting in the misattribution of parameter estimates as indicating within-person increases or decreases (Berry & Willoughby, 2017). This is particularly problematic when studying within-person associations between peer victimization and NFA, as increases in negative peer experiences could lead to decreases in NFA_approach and increases in NFA_avoid regardless of whether between-person changes occur. Indeed, Schacter and Juvonen (2019) proposed that although some children might suffer repeated victimization over time, most children’s negative peer experiences “come and go” unsystematically from year to year, highlighting the importance of understanding the within-person effect of peer victimization (Schacter & Juvonen, 2019; Zeiders et al., 2015). Thus, studies are needed to examine how NFA develops over time and how social experiences, such as peer victimization, shape individual differences in NFA or vice versa over longer developmental periods, using analytic approaches that clearly test for within-person development (Hamaker et al., 2015).

Study Overview

This study aimed to explore the normative patterns of change in NFA and the within-person reciprocal associations between peer victimization and NFA from middle childhood to early adolescence, a developmental time frame during which youth show increasing sensitivity to peer evaluation and belonging (Laursen, 1996; Rubin et al., 2005). Specifically, we used an auto-regressive latent trajectory model with structured residuals (ALT-SR; Berry & Willoughby, 2017) to disaggregate within-person and between-person effects. This approach allowed us to investigate whether children who experienced increased victimization one year as compared with their typical level would be more or less inclined to integrate judgments from peers into their sense of self the following year, and whether an increase in NFA predicted more or less frequent peer victimization the following year. Based on theory and prior research, we hypothesized that peer victimization and NFA_avoid would reciprocally intensify each other, such that peer victimization would predict within-person increases in subsequent NFA_avoid, and NFA_avoid would predict within-person increases in subsequent peer victimization. Conversely, we hypothesized that peer victimization would predict within-person decreases in subsequent NFA_approach, and NFA_approach would predict within-person decreases in subsequent peer victimization. To examine whether effects were robust across different informants and types of victimization, we investigated these hypotheses using both self- and teacher report of overt and relational victimization.

We also conducted exploratory analyses to examine gender differences. During middle childhood to early adolescence, girls show more concerns about social evaluation and intimate relationships compared to boys (Rudolph et al., 2005; Rudolph & Dodson, 2022) as well as more psychological (Rueger & Jenkins, 2014) and neural (Guyer et al., 2012) sensitivity to social threats, which may lead victimization to have a stronger impact on girls. Supporting the possibility of gender differences, one study (using a subsample of the present sample) revealed that NFA_avoid predicted lower peer acceptance in girls but not boys across a one-year period; however, NFA_approach predicted higher peer acceptance and lower peer victimization and exclusion in boys but not girls (Rudolph & Bohn, 2014). Given this mixed evidence, the lack of studies investigating gender differences in the reverse direction of effect, and the fact that prior research focused on between-person rather than within-person changes over very limited time frames and developmental stages, analyses examining gender differences were exploratory. In addition, developmental shifts may occur such that youth become increasingly sensitive to the effects of peer victimization on contingent self-worth and vice versa during early adolescence. Therefore, analyses also tested for age-related changes in the reciprocal associations between NFA and victimization.

Method

Participants and Procedures

Participants were 636 children (298 boys, 338 girls; M_age at recruitment = 7.96 years, SD_age = 0.35 years; 66.7% White, 21.7% Black, 11.6% other; 34.7% of them received a subsidized school lunch). In the second grade, parents provided written consent, and children provided oral assent. Of the 725 eligible children, 576 (80%) received parental consent. Participants and nonparticipants at Wave 1 did not significantly differ in gender, χ²(1) = 0.15, ns; ethnicity (White vs. minority), χ²(1) = 0.59, ns; age, t(723) = 0.63, ns; or school lunch status (full pay vs. subsidized), χ²(1) = 0.35, ns. In the third grade, 60 classmates of the participating children were recruited, yielding a total of 636 participants.

All procedures were approved by the University of Illinois, Urbana-Champaign Institutional Review Board (IRB Protocol #05244, Peer Victimization and Children’s Development). In the winter of each year, questionnaires were administered during two classroom sessions to small groups (3 – 4 students) in elementary school (2^nd – 5^th grades) and larger groups (15 – 20 students) in middle school (6^th and 7^th grades). Teachers completed the peer victimization questionnaire each year and returned their surveys to a locked box at their school, in person, or by mail. Children received a small gift; teachers received a monetary reimbursement. Each participating elementary classroom received a monetary honorarium, and middle schools received a school-wide honorarium.

Measures

Table 1 provides descriptive and reliability data for the measures.

Table 1.

Descriptive Statistics and Internal Reliabilities of Measures

Variable	Overall				Boys			Girls			Gender comparison
	N	M	SD	α	M	SD	α	M	SD	α	t	d
2nd grade OPV (S)	576	2.17	0.85	0.87	2.15	0.84	0.87	2.18	0.87	0.88	−0.42	−0.03
3rd grade OPV (S)	597	1.97	0.75	0.88	1.99	0.75	0.87	1.96	0.75	0.89	0.58	0.05
4th grade OPV (S)	573	1.85	0.71	0.89	1.94	0.78	0.90	1.77	0.64	0.88	2.89**	0.24
5th grade OPV (S)	557	1.81	0.68	0.88	1.86	0.67	0.87	1.77	0.68	0.89	1.50	0.13
6th grade OPV (S)	532	1.81	0.66	0.89	1.88	0.68	0.90	1.75	0.64	0.89	2.18*	0.19
7th grade OPV (S)	474	1.72	0.62	0.89	1.80	0.62	0.89	1.66	0.60	0.89	2.52*	0.23
2nd grade RPV (S)	575	2.09	0.82	0.84	2.03	0.77	0.81	2.15	0.87	0.87	−1.68	−0.14
3rd grade RPV (S)	597	1.97	0.77	0.88	1.88	0.72	0.85	2.05	0.81	0.89	−2.71**	−0.22
4th grade RPV (S)	573	1.84	0.73	0.89	1.79	0.72	0.89	1.88	0.73	0.89	−1.49	−0.12
5th grade RPV (S)	557	1.74	0.71	0.90	1.66	0.66	0.89	1.82	0.75	0.91	−2.77**	−0.24
6th grade RPV (S)	532	1.66	0.65	0.90	1.60	0.62	0.89	1.71	0.68	0.91	−1.92	−0.17
7th grade RPV (S)	474	1.57	0.59	0.90	1.49	0.51	0.87	1.63	0.64	0.91	−2.67**	−0.24
2nd grade OPV (T)	576	1.61	0.58	0.94	1.72	0.64	0.95	1.52	0.51	0.92	4.04***	0.34
3rd grade OPV (T)	596	1.65	0.60	0.94	1.78	0.66	0.95	1.54	0.51	0.93	5.01***	0.41
4th grade OPV (T)	571	1.67	0.67	0.95	1.84	0.72	0.95	1.52	0.57	0.94	5.88***	0.50
5th grade OPV (T)	567	1.64	0.62	0.95	1.72	0.64	0.94	1.56	0.59	0.95	3.26**	0.27
6th grade OPV (T)	548	1.51	0.57	0.94	1.66	0.60	0.95	1.37	0.50	0.93	6.03***	0.52
7th grade OPV (T)	516	1.44	0.50	0.92	1.53	0.51	0.92	1.37	0.47	0.92	3.68***	0.33
2nd grade RPV (T)	575	1.76	0.67	0.95	1.70	0.68	0.95	1.82	0.66	0.94	−2.00*	−0.17
3rd grade RPV (T)	596	1.80	0.69	0.96	1.74	0.67	0.95	1.85	0.71	0.96	−1.98*	−0.16
4th grade RPV (T)	571	1.80	0.73	0.96	1.75	0.71	0.95	1.84	0.74	0.96	−1.44	−0.12
5th grade RPV (T)	567	1.70	0.69	0.96	1.62	0.63	0.95	1.77	0.74	0.97	−2.51*	−0.21
6th grade RPV (T)	547	1.53	0.62	0.96	1.56	0.61	0.95	1.50	0.62	0.96	1.14	0.10
7th grade RPV (T)	515	1.42	0.52	0.96	1.37	0.46	0.92	1.46	0.57	0.95	−1.95	−0.17
2nd grade NFAapproach	576	3.70	1.02	0.78	3.70	1.02	0.77	3.71	1.02	0.79	−0.08	−0.01
3rd grade NFAapproach	594	3.76	0.96	0.79	3.70	1.00	0.78	3.82	0.92	0.79	−1.46	−0.12
4th grade NFAapproach	573	3.76	0.99	0.85	3.71	0.98	0.84	3.81	0.99	0.85	−1.31	−0.11
5th grade NFAapproach	557	3.65	0.99	0.87	3.54	1.03	0.87	3.74	0.95	0.87	−2.36*	−0.20
6th grade NFAapproach	528	3.48	1.06	0.89	3.38	1.10	0.90	3.55	1.02	0.88	−1.83	−0.16
7th grade NFAapproach	470	3.34	1.07	0.90	3.21	1.10	0.90	3.45	1.04	0.90	−2.40*	−0.22
2nd grade NFAavoid	576	2.31	1.10	0.76	2.25	1.12	0.77	2.37	1.08	0.75	−1.31	−0.11
3rd grade NFAavoid	594	2.11	1.04	0.82	2.03	1.05	0.82	2.18	1.03	0.81	−1.75	−0.14
4th grade NFAavoid	573	1.96	0.96	0.83	2.00	1.00	0.82	1.92	0.93	0.85	1.01	0.08
5th grade NFAavoid	557	1.84	0.98	0.89	1.83	0.93	0.88	1.86	1.03	0.90	−0.35	−0.03
6th grade NFAavoid	528	1.81	0.98	0.90	1.79	0.94	0.88	1.82	1.02	0.89	−0.37	−0.03
7th grade NFAavoid	470	1.72	0.90	0.90	1.67	0.86	0.91	1.75	0.93	0.89	−0.91	−0.08

Note. OPV = overt peer victimization; RPV = relational peer victimization; T = teacher-reported, S = self-reported; NFA_approach = approach-oriented need for approval, NFA_avoid = avoidance-oriented need for approval.

^*p < .05.

^**p < .01.

^***p < .001.

Peer Victimization

Youth and teachers completed a revised version (Rudolph et al., 2011) of the Social Experiences Questionnaire (SEQ; Crick & Grotpeter, 1995). Eleven items were added to the original measure to provide a more comprehensive assessment. The revised SEQ contains 11 overt victimization items (e.g., “How often do you get pushed or shoved by another kid?”) and 10 relational victimization items (e.g., “How often do other kids leave you out on purpose when it’s time to play or do an activity?”). Items on the child and teacher report versions were identical other than altering the wording as relevant (e.g., substituting “you” with “this child”). Youth checked a box and teachers rated how often the youth experienced each type of victimization on a 5-point scale (1 = Never to 5 = All the Time). Four scores were computed separately by informant as the mean of the items within each subscale, yielding four scores: self-reported overt, self-reported relational, teacher-reported overt, and teacher-reported relational. The revised version of the SEQ has strong reliability and predictive validity (Rudolph et al., 2011; Troop-Gordon et al., 2015).

Need for Approval

Youth completed the Need for Approval Questionnaire (Rudolph et al., 2005), which contains two subscales: NFA_approach, assessing the extent to which peer approval enhances self-worth (4 items; e.g., “I feel like a good person when other kids like me”) and NFA_avoid, assessing the extent to which peer disapproval diminishes self-worth (4 items; e.g., “I feel ashamed of myself when other kids don’t like me”). Youth rated each item on a 5-point scale (1 = Not at All to 5 = Very Much). Scores were calculated as a mean of each subscale, with higher scores indicating higher levels of NFA. Prior research supports this two-factor structure of NFA and establishes convergent and discriminant validity (Rudolph, 2021; Rudolph & Bohn, 2014; Rudolph et al., 2005).

Analytic Approach

Prior to testing hypothesized associations between peer victimization and need for approval, preliminary analyses were conducted to provide descriptive information on the distributional properties of the variables, identify gender differences, and examine correlations among the study variables. Rates of missingness and any systematic patterns of missing data were also identified. These analyses were followed by a test of the ALT-SR model depicted in Figure 1. Model testing occurred separately for self-reported and teacher-reported overt and relational victimization, yielding a total of four models. For each model, parameter estimation occurred in four stages. All models were tested using Mplus 7.4 (Muthén & Muthén, 2012) with full information maximum likelihood estimation (FIML; Enders & Bandalos, 2001), allowing parameters to be estimated based on all available data from the 636 youth. To examine gender differences, multi-group comparison analyses and chi-square difference tests were employed.

Figure 1.

Auto-Regressive Latent Trajectory Model with Structured Residuals (ALT-SR) Testing the Within-Person Associations between Peer Victimization (PV) and Need for Approval (NFA)

Note. Paths and coefficients can be found in Table 2–5. PV = peer victimization; NFA_approach = approach-oriented need for approval; NFA_avoid = avoidance-oriented need for approval.

In the first stage of model development, unconditional latent growth curve models were tested for each variable. Accurate estimation of structured residuals (i.e., within-person deviations from one’s developmental trajectory) necessitates identification of the best-fitting growth curve. Accordingly, linear and quadratic slopes were tested, and the quadratic slope was retained if it improved model fit, as determined by smaller BIC and AIC scores and a significant log-likelihood ratio test. In addition to determining the proper parameterization of the growth curves, this stage of analysis was used to identify gender differences in the means and variances of the retained growth parameters (i.e., intercept, linear slope, and quadratic slope).

In the second stage, latent growth curves for the target peer victimization variable (e.g., self-reported overt victimization), NFA_avoid, and NFA_approach were combined into a single model, and latent structured residuals were estimated for each one. Any equality constraints identified in the first step were retained. Additional constraints were included as needed to achieve model convergence (e.g., constraining variance of quadratic slopes to 0) and are noted in the supplement. Covariances between the latent growth parameters and within-wave covariances between structured residuals were estimated. Gender differences in these covariances and in the magnitude of the structured residuals were tested. This model also allowed for testing whether the unadjusted structured residuals (i.e., prior to accounting for variance explained by predictors at the previous timepoint) and within-wave covariances were invariant across time.

In the third stage, the autoregressive paths were added to the model and tested for invariance across age and gender. A model in which all autoregressive paths were freely estimated was tested against one model in which the paths were set to be equivalent across time and one in which the paths were set to be equivalent across gender. Autoregressive paths were then constrained to be equivalent across time or gender as warranted. If paths were found to be variant across time, follow-up analyses were conducted to determine the exact time points at which autoregressive paths increased or decreased. Specific pairs of autoregressive paths were sequentially tested for invariance beginning with the path from Wave 1 to Wave 2. If a pair of paths (e.g., Wave 1 to Wave 2 and Wave 2 to Wave 3) were found to be invariant, they were set to be equal, and equality with the next autoregressive path was tested. No gender differences in the autoregressive paths were found.

In the final model stage, all cross-lagged paths were added to the model identified in Stage 3. Tests for differences across time and gender in the cross-lagged paths followed the same steps as those for testing the autoregressive paths. As a final check, structured residuals and within-wave covariances between structured residuals were tested for invariance across time and gender, and equality constraints were incorporated into the final model if invariance was established.

To evaluate fit of the final model, several fit indices were examined, including the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). Acceptable model fit was indicated by χ²/df ≤ 3 (Mennin et al., 2007), CFI values ≥ .95, RMSEA ≤ .06, and SRMR values ≤ .08 (Hu & Bentler, 1999).

Transparency and Openness

We have complied with the requirements of the Transparency and Openness Promotion (TOP) guidelines at Level 1. First, all the methods developed by others have been appropriately cited. Second, the raw data, processing data, computer syntax, and materials in the Method section are available upon request. We used APA 7^th style across the manuscript and did not pre-register the present analyses.

Results

Gender Differences, Intercorrelations, and Missing Data

Boys generally had higher overt victimization scores than girls, whereas girls generally had higher relational victimization scores than boys, although this pattern was stronger for teacher report than self-report (see Table 1 for more details). In addition, girls reported higher NFA_approach than did boys in 5^th grade, t(555) = −2.36, p = .019, and 7^th grade, t(468) = −2.40, p = .017 (Table 1).

Tables S1–S4 present correlations between peer victimization and NFA, separately by subtype and informant. Correlation separately by gender are available upon request. Victimization was generally positively and significantly correlated with NFA_avoid but not NFA_approach. Correlations between NFA_avoid and NFA_approach were also positive and increased in magnitude with age. Increases in stability coefficients over time were observed for all variables. For NFA_approach, stability increased from .26 between 2^nd and 3^rd grade to .52 between 6^th and 7^th grade. For NFA_avoid, stability increased from .28 between 2^nd and 3^rd grade to .49 between 6^th and 7^th grade.

Of the 636 participants, 510 had 7^th grade teacher reports of peer victimization, 474 had 7^th grade self-reports of peer victimization, and 470 had 7^th grade self-reports of need for approval. Little’s (1998) missing completely at random (MCAR) test was significant, χ²(1186) = 1339.44, p = .001, suggesting that the data were not MCAR. Follow-up analyses indicated that youth with complete vs. incomplete data in 7^th grade did not significantly differ in gender, χ²(1) = 0.48, ns; age, t(634) = −1.26, ns; ethnicity, χ²(1) = 0.56, ns; or 2^nd grade scores on key variables, |t|s ≤ 1.36, ns. They did differ on family income (marginally), t(428) = −1.92, p = .055, mother’s educational level, t(429) = 2.55, p = .011, and father’s educational level, t(360) = 2.73, p = .007. Therefore, we used full information maximum likelihood estimation (FIML), one of the most reliable methods for addressing non-MCAR missing values (Enders & Bandalos, 2001; Young & Johnson, 2013), to estimate parameters based on all available data. Furthermore, to optimize the accuracy of parameter estimates, five auxiliary variables were included in all subsequent analyses based on correlating with having missing data (family income, mother’s educational level, father’s educational level) or with some of the indexes of peer victimization at particular waves (Wave 1 teacher ratings of academic performance, Wave 1 aggression).

ALT-SR Models

Detailed analyses for each step of model building are presented in the supplement. All four models provided converging evidence of a reciprocal association between within-person changes in peer victimization and NFA_avoid, with the exception of a nonsignificant path from within-person changes in teacher-reported overt victimization to within-person changes in NFA_avoid^. In contrast, in none of the four models were longitudinal associations found between within-person changes in peer victimization and within-person changes in NFA_approach. To summarize the analyses leading up to our final models, we first review the results of the unconditional latent growth curve analyses. We then present the findings for the four models (i.e., each assessment of peer victimization) separately.

Unconditional Latent Growth Curves

Table S5 and S6 in the supplement present the model fit statistics, gender comparisons, and final parameter estimates for the unconditional growth curve models. The inclusion of a quadratic slope improved model fit for all variables (all log-likelihood ratio tests were significant at p < .001). Using guidelines proposed by Raftry (1995), across models, there was positive to very strong evidence in support of the inclusion of a quadratic term, although the difference for self-reported relational victimization was just below the 2.0 cutoff (ΔBIC = 1.97).

Children reported overall declines in peer victimization (for overt victimization M_LinearSlope = −.15, p < .001; for relational victimization, M_LinearSlope = −.14, p < .001), and girls reported higher levels of relational victimization in 2^nd grade than boys. The declines in self-reported overt victimization abated with age (M_{QuadraticSlope} = .015, p < .001), but those for relational victimization did not (M_{QuadraticSlope} = .006, p = .104). In contrast, teacher-reported relational victimization increased for all children (M_LinearSlope = .05, p = .023), and teacher-reported overt victimization increased for boys (M_LinarSlope = .10, p < .001), but not girls (M_LinearSlope = .024, p = .297), a statistically significant gender difference. For both girls and boys, and for teacher-reported overt and relational victimization, increases abated over time, and peer victimization scores declined beginning in 5^th grade (all M_{QuadraticSlopes} ≤ −.012, ps ≤ .008). This decline was steeper for boys than for girls. Plots of these trajectories can be found in the supplement (see Figure S1).

For NFA_avoid and NFA_approach, constraining the means and variances of the intercept, linear slope, and quadratic slope equal for girls and boys did not reduce model fit, indicating that overall patterns of growth were equivalent across gender. During middle childhood, children reported a decline in NFA_avoid (M_LinearSlope = −.20, p < .001) and an increase in NFA_approach (M_LinearSlope = .07, p = .031) with age. However, starting in 5^th grade, declines in NFAavoid plateaued, and NFAapproach steadily decreased (for NFAavoid, MQuadraticSlope = .02, p = .004; for NFA_approach, M_{QuadraticSlope} = −.03, p < .001). As this was the first study to track the developmental course of NFA for an extended period, plots of these trajectories are presented in Figure 2.

Figure 2.

Plots of Unconditional Latent Growth Curves for NFA_avoid and NFA_approach

ALT-SR Model for Self-Reported Overt Victimization

Table S7 presents the results of a model in which growth curves and structured residuals were simultaneously estimated for self-reported overt victimization, NFA_avoid, and NFA_approach. No gender differences were found in the covariances between the growth parameters. Of note, the intercept of overt victimization was positively associated with the intercept of NFA_approach, but not NFA_avoidance. Within-wave covariances between structured residuals were invariant across time and gender. However, the size of the structured residuals were not. An overall pattern emerged such that residuals became smaller over time, suggesting less deviation from one’s trajectory. Gender differences in the size of the structured residuals followed no discernable pattern and likely reflect random fluctuations. Table S8 presents the results after autoregressive paths were added to the model. Although autoregressive lags were invariant across gender, they differed over time. Specifically, the autoregressive paths for overt victimization increased and remained steady starting in 3^rd grade. The autoregressive paths for NFA_avoid increased and remained constant beginning in 5^th grade, and there was a temporary jump in the autoregressive path for NFA_approach from 5^th to 6^th grades.

Table S9 presents the final model in which cross-lagged paths were added and tested. All cross-lagged paths were invariant across time and gender, as were the within-wave covariances. As in the previous step, the structured residuals remained variant across time and gender. The final model fit the data well, χ²(278) = 373.63, χ²/df = 1.34, p < .001, CFI = .97, RMSEA = .03, SRMR = .07. Table 2 presents the final parameter estimates. With regard to within-person effects, overt victimization significantly predicted within-person increases in NFA_avoid (b = .10, SE = .02, p = .014); in turn, NFA_avoid significantly predicted within-person increases in overt victimization (b = .07, SE = .01, p < .001). Overt victimization did not predict within-person changes in NFA_approach (b = −.06, SE = .05, p = .185), and NFA_approach did not predict within-person changes in overt victimization (b = −.01, SE = .02, p = .542).

Table 2.

ALT-SR Model: Reciprocal Associations Between Self-Reported Overt Victimization and Need for Approval

ALT-SR effects	b (SE)
Cross-Lags (Within-Person)
OV (t−1) → NFAavoid (t)	.11(.04)*
OV (t−1) →NFAapproach (t)	−.06(.05)
NFAavoid (t−1) → OV (t)	.10(.02)***
NFAapproach (t−1) → OV (t)	−.01(.02)
NFAavoid (t−1) →NFAapproach (t)	.08(.03)**
NFAapproach (t−1) → NFAavoid (t)	.04(.03)
Residual Variances (Within-Person)
OV(ϵit2–ϵit6)	.17(.03)***– 32(.03)***/.09(.02)***– .33(.03)***
NFAavoid (ϵit2–ϵit6)	.31(.07)***– 75(.07)***/.42(.07)***– .88(.08)***
NFAapproach (ϵit2–ϵit6)	.54(.13)***– .61(.07)***/.41(.13)***– .68(.07)***
Concurrent covariances (Within-Person)
OV with NFAavoid (wave 2–6)	.10(.01)***
OV with NFAapproach (wave 2–6)	.01(.01)
NFAapproach with NFAavoid (wave 2–6)	.17(.02)***
Autoregressive paths (Within-Person)
OV: Wave 1 to Wave 2	.01(.07)
OV: Waves 2 through Wave 6	.18(.04)***
NFAavoid: Waves 1 through Wave 4	.06(.04)
NFAavoid: Waves 4 through Wave 6	.21(.06)***
NFAapproach: Waves 1 through Wave 4	−.01(.04)
NFAapproach: Wave 4 to Wave 5	.27(.07)***
NFAapproach: Wave 5 to Wave 6	.07(.12)

Note. OV = overt victimization; NFA = Need for Approval. ϵit = structured residual for individual i at time t; int = latent intercept. The parameters with significant gender differences are separated by a forward slash, with boys’ parameters on the left, and girls’ parameters on the right.

^*p < .05.

^**p < .01.

^***p < .001.

ALT-SR Model for Self-Reported Relational Victimization

Table S10 presents the results of a model in which growth curves and structured residuals were simultaneously estimated for self-reported relational victimization, NFA_avoid, and NFA_approach. Only two small gender differences emerged in the covariances between the growth parameters, reflecting slightly stronger positive associations between the intercept and quadratic slope of relational victimization and between the intercept and quadratic slope of NFA_avoid for boys than girls. Positive associations emerged for the intercepts of relational victimization and NFA_avoid and the intercepts of NFA_avoid and NFA_approach. Structured residuals were variant across time, reflecting a decrease in overall magnitude of within-person change. Within-wave covariances also varied across time, but structured residuals and within-wave covariances did not differ by gender. Table S11 presents the results after autoregressive paths were added to the model. Although autoregressive lags were invariant across gender, they differed over time. Specifically, the autoregressive paths for relational victimization increased and remained steady starting in 3^rd grade. The autoregressive paths for NFA_avoid and NFA_approach temporarily increased from the 5^th to 6^th grade.

Table S12 presents the final model in which cross-lagged paths were added and tested. All cross-lagged paths were invariant across time and gender. Structured residuals and within-wave covariances were variant across time but invariant across gender. The final model fit the data well, χ²(282) = 373.36, χ²/df = 1.32, p < .001, CFI = .97, RMSEA = .03, SRMR = .07. Table 3 presents the final parameter estimates. With regard to within-person effects, relational victimization significantly predicted within-person increases in NFA_avoid (b = .09, SE = .04, p = .030); in turn, NFA_avoid significantly predicted within-person increases in relational victimization (b = .08, SE = .02, p < .001). Relational victimization did not predict within-person changes in NFA_approach (b = −.06, SE = .05, p = .231), and NFA_approach did not predict within-person changes in relational victimization (b = −.01, SE = .02, p = .693).

Table 3.

ALT-SR Model: Reciprocal Associations Between Self-Reported Relational Victimization and Need for Approval

ALT-SR effects	b (SE)
Cross-Lags (Within-Person)
RV (t−1) → NFAavoid (t)	.09(.04)*
RV (t−1) →NFAapproach (t)	−.06(.05)
NFAavoid (t−1) → RV (t)	.08(.02)***
NFAapproach (t−1) → RV (t)	−.01(.02)
NFAavoid (t−1) →NFAapproach (t)	.08(.03)**
NFAapproach (t−1) → NFAavoid (t)	.04(.03)
Residual Variances (Within-Person)
RV(ϵit2–ϵit6)	.06(.02)*** – .32(.03)***
NFAavoid (ϵit2–ϵit6)	.06(.10) – .83(.06)***
NFAapproach (ϵit2–ϵit6)	.39(.13)** – .64(.14)***
Concurrent covariances (Within-Person)
RV with NFAavoid (wave 2–6)	.01(.03) – .14(.02)***
RV with NFAapproach (wave 2–6)	−.01(.03) – .05(.03)
NFAapproach with NFAavoid (wave 2–6)	.05(.05) – .21(.03)***
Autoregressive paths (Within-Person)
RV: Wave 1 to Wave 2	−.05(.08)
RV: Waves 2 through Wave 6	.09(.04)*
NFAavoid: Waves 1 through Wave 4	.06(.04)
NFAavoid: Waves 4 to Wave 5	.17(.06)**
NFAavoid: Waves 5 to Wave 6	−.07 (.12)
NFAapproach: Waves 1 through Wave 4	.01(.04)
NFAapproach: Wave 4 to Wave 5	.25(.06)***
NFAapproach: Wave 5 to Wave 6	.02(.13)

Note. RV = relational victimization; NFA = Need for Approval. ϵit = structured residual for individual i at time t; int = latent intercept.

^*p < .05.

^**p < .01.

^***p < .001.

ALT-SR Model for Teacher-Reported Overt Victimization

Table S13 presents the results of a model in which growth curves and structured residuals were simultaneously estimated for teacher-reported overt victimization, NFA_avoid, and NFA_approach. Only two small gender differences emerged in the covariances between the growth parameters. Although overt victimization was positively associated with the intercept of NFA_avoid for girls and boys, a positive association emerged between the intercepts of overt victimization and NFA_approach for girls only. An inverse association emerged for the intercept and linear slope of NFA_approach that was stronger for girls than boys. In addition, for boys and girls, the linear slopes of NFA_avoid and NFA_approach were positive and significant. Structured residuals were variant across time, reflecting a decrease in overall magnitude of within-person change, but were invariant across gender. Within-wave covariances varied across time and gender. Table S14 presents the results after autoregressive paths were added to the model. Although autoregressive lags were invariant across gender, they differed over time. Specifically, the autoregressive paths for overt victimization increased and remained steady starting in 3^rd grade. In this model, the autoregressive paths for NFA_avoid and NFA_approach changed across each lag, predominantly reflecting increases over time.

Table S15 presents the final model in which cross-lagged paths were added and tested. All cross-lagged paths were invariant across time and gender. However, the structured residuals and within-wave covariances between structured residuals were variant across time and gender. The final model fit the data well, χ²(246) = 337.09, χ²/df = 1.37, p < .001, CFI = .95, RMSEA = .03, SRMR = .07. Table 4 presents the final parameter estimates. With regard to within-person effects, overt victimization did not predict within-person increases in NFA_avoid (b = .03, SE = .04, p = .479), but NFA_avoid significantly predicted within-person increases in overt victimization (b = .05, SE = .02, p = .002). Overt victimization did not predict within-person changes in NFA_approach (b = −.02, SE = .04, p = .708), and NFA_approach did not predict within-person changes in overt victimization (b = −.01, SE = .02, p = .339).

Table 4.

ALT-SR Model: Reciprocal Associations Between Teacher-Reported Overt Victimization and Need for Approval)

ALT-SR effects	b (SE)
Cross-Lags (Within-Person)
OV (t−1) → NFAavoid (t)	.03(.04)
OV (t−1) →NFAapproach (t)	−.02(.04)
NFAavoid (t−1) → OV (t)	.05(.02)**
NFAapproach (t−1) → OV (t)	−.01(.02)
NFAavoid (t−1) →NFAapproach (t)	.06(.03)*
NFAapproach (t−1) → NFAavoid (t)	.04(.03)
Residual Variances (Within-Person)
OV(ϵit2–ϵit6)	.19(.02)***– .36(.04)***/.12(.02)***– .25(.02)***
NFAavoid (ϵit2–ϵit6)	.35(.08)***– .75(.10)***/.45(.08)***– .83(.10)***
NFAapproach (ϵit2–ϵit6)	.67(.08)***– .73(.07)***/ .57(.08)***– .75(.07)***
Concurrent covariances (Within-Person)
OV with NFAavoid (wave 2–6)	−.02(.04) – .05(.03) / −.04(.03) – .07(.03)*
OV with NFAapproach (wave 2–6)	−.07(.03)*– .05(.04)/−.06(.03)*–.04(.02)
NFAapproach with NFAavoid (wave 2–6)	.04(.06) – .29(.06)***/.09(.06) – .23(.05)***
Autoregressive paths (Within-Person)
OV: Wave 1 to Wave 2	−.15(.07)*
OV: Waves 2 through Wave 6	.07(.03)*
NFAavoid: Wave 1 to Wave 2	−.08(.09)
NFAavoid: Wave 2 to Wave 3	.12(.05)*
NFAavoid: Wave 3 to Wave 4	.27(.05)***
NFAavoid: Wave 4 to Wave 5	.29(.06)***
NFAavoid: Wave 5 to Wave 6	.19(.10)
NFAapproach: Wave 1 to Wave 2	−.05(.10)
NFAapproach: Wave 2 to Wave 3	.17(.06)*
NFAapproach: Wave 3 to Wave 4	.14(.05)**
NFAapproach: Wave 4 to Wave 5	.35(.06)***
NFAapproach: Wave 5 to Wave 6	.25(.09)**

Note. OV = overt victimization; NFA = Need for Approval. ϵit = structured residual for individual i at time t; int = latent intercept. The parameters with significant gender differences are separated by a forward slash, with boys’ parameters on the left, and girls’ parameters on the right.

^*p < .05.

^**p < .01.

^***p < .001.

ALT-SR Model for Teacher-Reported Relational Victimization

Table S16 presents the results of a model in which growth curves and structured residuals were simultaneously estimated for teacher-reported relational victimization, NFA_avoid, and NFA_approach. Three gender differences emerged in the covariances between the growth parameters. Although relational victimization was positively associated with the intercept of NFA_avoid for girls and boys, a positive association emerged between the intercepts of relational victimization and NFA_approach for girls only. Inverse associations emerged for the intercept and linear slope of relational victimization, and for the linear and quadratic slopes of NFA_avoid, that were stronger for boys than girls. Structured residuals were variant across time, reflecting a decrease in overall magnitude of within-person change, but were invariant across gender. Within-wave covariances were invariant across time and gender. Table S17 presents the results after autoregressive paths were added to the model. Although autoregressive lags were invariant across gender, they differed over time. Specifically, the autoregressive path for teacher-reported relational victimization was negative from 2^nd to 3^rd grade and nonsignificant after that. For this model, the autoregressive paths for NFA_avoid increased between 3^rd and 4^th grade and again between 4^th and 5^th grade. The autoregressive paths for NFA_approach increased between 3^rd and 4^th grade and again between 4^th and 6^th grade.

Table S18 presents the final model in which cross-lagged paths were added and tested. All cross-lagged paths were invariant across time and gender. Structured residuals were variant across time but not gender, and within-wave covariances between structured residuals were invariant across time and gender. The final model fit the data well, χ²(295) = 378.81, χ²/df = 1.28, p < .001, CFI = .95, RMSEA = .03, SRMR = .07. Table 5 presents the final parameter estimates. With regard to within-person effects, relational victimization significantly predicted within-person increases in NFA_avoid (b = .08, SE = .04, p = .019); in turn, NFA_avoid significantly predicted within-person increases in relational victimization (b = .07, SE = .02, p < .001). Relational victimization did not predict within-person changes in NFA_approach (b = −.04, SE = .04, p = .238), and NFA_approach did not predict within-person changes in relational victimization (b = .00, SE = .02, p = .951).

Table 5.

ALT-SR Model: Reciprocal Associations Between Teacher-Reported Relational Victimization and Need for Approval

ALT-SR effects	b (SE)
Cross-Lags (Within-Person)
RV (t−1) → NFAavoid (t)	.08(.04)*
RV (t−1) →NFAapproach (t)	−.04(.04)
NFAavoid (t−1) → RV (t)	.07(.02)***
NFAapproach (t−1) → RV (t)	−.00(.02)
NFAavoid (t−1) →NFAapproach (t)	.05(.03)
NFAapproach (t−1) → NFAavoid (t)	.02(.03)
Residual Variances (Within-Person)
RV(ϵit2–ϵit6)	.17(.02)***– .43(.03)***
NFAavoid (ϵit2–ϵit6)	.47(.05)***–.78(.08)***
NFAapproach (ϵit2–ϵit6)	.61(.06)***–.76(.07)***
Concurrent covariances (Within-Person)
RV with NFAavoid (wave 2–6)	.04(.01)**
RV with NFAapproach (wave 2–6)	.01(.01)
NFAapp with NFAavoid (wave 2–6)	.19(.02)***
Autoregressive paths (Within-Person)
RV: Wave 1 to Wave 2	−.23(.12)
RV: Waves 2 through Wave 4	.07(.04)
RV: Waves 4 through Wave 6	−.04(.04)
NFAavoid: Wave 1 to Wave 2	−.06(.09)
NFAavoid: Wave 2 to Wave 3	.12(.05)*
NFAavoid: Wave 3 through Wave 6	.28(.05)***
NFAapproach: Wave 1 to Wave 2	−.05(.09)
NFAapproach: Waves 2 through Waves 4	.16(.05)***
NFAapproach: Waves 4 through Wave 6	.34(.06)***

Note. RV = relational victimization; NFA = Need for Approval. ϵit = structured residual for individual i at time t; int = latent intercept. The parameters with significant gender differences are separated by a forward slash, with boys’ parameters on the left, and girls’ parameters on the right.

^*p < .05.

^**p < .01.

^***p < .001.

Discussion

Peer group belonging becomes increasingly important for youth as they progress through middle childhood and adolescence (Masten et al., 2009). Consequently, self-esteem may become tied, in part, to approval or disapproval by peers. Despite this normative process, youth may differ in the extent to which they rely on peer approval for determining their self-worth. Because individual differences in the strength of NFA predict critical outcomes (Rudolph & Bohn, 2014), it is important to elucidate how they develop. To identify mechanisms of NFA development, the present study explored the within-person dynamic interchange between peer victimization and NFA from second through seventh grades. Analyses revealed that peer victimization predicted increases in NFA_avoid one year later, which predicted subsequent increases in peer victimization, creating a self-perpetuating cycle. Furthermore, as youth became older, increases in NFA and peer victimization were less likely to be transitory. Rather, heightened NFA and peer victimization one year forecast similarly heightened NFA and peer victimization the following year. Thus, cycles of NFA_avoid and peer victimization may become increasingly entrenched as youth transition to early adolescence.

Peer Victimization and NFA_avoid

Using the ALT-SR model to disaggregate within-person and between-person effects, we found that peer victimization had a within-person longitudinal effect on NFA_avoid, such that experiencing an increase in peer victimization predicted increases in NFA_avoid the following year. Therefore, the longitudinal prediction from peer victimization to NFA_avoid cannot be exclusively attributed to a trait-like association, but reflects within-individual change over time. This pattern is consistent with prior research suggesting that peer victimization predicts avoidance-oriented tendencies, such as social withdrawal (Wang et al., 2020) and helpless behavior (Rudolph et al., 2014), but extends these findings to incorporate avoidance motivation. In turn, NFA_avoid predicted subsequent within-person increases in peer victimization. Youth with avoidance-oriented social tendencies may aim to reduce conflict with peers. Thus, their behavior may signal that they will not defend themselves against peer harassment, which might attract victimization from bullies (Perry et al., 1988), consistent with the presence of “passive victims” (Olweus, 1994). NFA_avoid also may interfere with positive approach behaviors as youth may be concerned about meeting with peer disapproval; thus, youth with high NFA_avoid may not have supportive peers to protect them against victimization (Rudolph & Bohn, 2014). This reciprocal association emerged consistently across types of victimization (overt vs. relational) as well as informants (self vs. teacher), as well as a composite victimization variable (see description in the supplemental file and Tables S19 and S20), with the exception of the path from teacher-reported overt victimization to NFA_avoid. It is possible that overt forms of victimization (e.g., punching, threatening), which reflect clear violations of acceptable norms, sometimes occur outside of teachers’ presence, resulting in less predictive value of these reports relative to reports of relational victimization.

Collectively, these findings suggest that experiencing peer victimization interferes with healthy self-concept development, due to an excessive reliance on appraisals of others, particularly disapproval, to determine self-worth. According to symbolic interactionist theories (Cooley, 1902; Mead, 1934), appraisals of significant others are incorporated into one self-concept as a normative part of development. However, over time, self-concept typically becomes distinct from its social origins (Harter, 1998). Variations in this process may lead to individual differences in the strength of NFA. Results of this study suggest that peer victimization may set youth upon a trajectory toward an increasing likelihood of interpreting negative experiences as self-threatening. Unfortunately, when self-worth is depleted by peer disapproval, youth become increasingly victimized over time, resulting in a self-perpetuating cycle wherein youth elicit the very experiences that they seek to avoid.

Peer Victimization and NFA_approach

Contrary to hypotheses, analyses did not reveal within-person associations between peer victimization and NFA_approach. We speculated that peer victimization may lead to selective processing of external feedback, causing youth to attend to feedback that reinforces their pre-existing, albeit negative, self-views, as well as to an inability or a reluctance to integrate positive appraisals into their self-worth. In turn, we anticipated that decreases in NFA_approach would inhibit approach behaviors, enhancing the likelihood of future victimization.

The absence of a significant contribution of victimization to subsequent NFA_approach is at odds with prior findings that peer victimization predicts lower levels of other dimensions of approach (e.g., prosocial behavior; Rudolph et al., 2014). Similarly, the absence of a significant contribution of NFA_approach to subsequent victimization is inconsistent with prior between-subjects analyses with a subsample of this data set (Rudolph & Bohn, 2014) revealing that NFA_approach predicts less victimization in boys from second to third grade. Moreover, other research reveals that general social sensitivity (e.g., caring about peers’ judgments; Chen et al., 2018) and NFA_approach (Rudolph et al., 2005) are concurrently associated with higher levels of social competence. These contrasting findings could be due to various design differences, including the age and cultural backgrounds of participants, the time frames of analysis, the precise constructs measured, and the analytic approaches (e.g., the disaggregation of between- and within-subject effects in the present study).

It is also possible that NFA_approach is a complex construct with diverse and sometimes paradoxical implications for development. Although youth who strive to obtain peer approval to enhance their self-worth may show more positive engagement, NFA_approach also may confer risks for peer victimization. Children with strong approach tendencies may disregard potential consequences of their actions to meet their goals, particularly if they have poor self-regulatory skills, which may prompt more aggressive behavior (Rudolph et al., 2013). Therefore, some children with high NFA_approach may have trouble ingratiating themselves with peers, potentially leading to unskillful attempts to join activities or befriend higher status peers, and in turn, eliciting victimization. These opposing pathways (both inhibitory and amplifying effects of NFA_approach on subsequent victimization) may have obscured identification of significant effects, and warrant further investigation.

With regard to the nonsignificant effect of victimization on subsequent NFA_approach, perhaps victimization leads some youth to show a general sensitivity to social cues, such that they experience increases in NFA_approach. Research also will need to consider other contextual variables that may contribute to the development of NFA_approach. The present study focused on adverse peer experiences, which sensitized children to internalizing negative feedback into their sense of self. Perhaps positive peer experiences (e.g., peer acceptance, close friendships) contribute more strongly to NFA_approach. Identifying pathways to the development of NFA_approach should be on the future research agenda.

Gender and Developmental Differences

Despite evidence for gender differences in previous research testing associations between NFA and social competence (Rudolph et al., 2005; Rudolph & Bohn, 2014), the present study did not detect gender differences in the within-person reciprocal association between peer victimization and NFA across middle childhood to early adolescence. Similarly, the present study did not reveal any developmental differences in the reciprocal association between NFA and peer victimization across informants or types of peer victimization, indicating that the observed self-perpetuating cycle continues from middle childhood through early adolescence. However, other findings suggest some important developmental shifts. During middle childhood, NFA_approach remained relatively high and stable, while NFA_avoid steadily decreased. These trends presumably reflect a healthy pattern and may provide insights into why children’s self-esteem often increases in middle childhood (Orth et al., 2018). However, during early adolescence, NFA_avoid plateaued while NFA_approach steadily declined, implicating early adolescence as a time when youth may be less inclined to integrate positive peer judgments into their sense of self. Furthermore, as children got older their likelihood of evidencing a within-person change in NFA decreased, as indicated by the diminishing structured residuals, but the odds that change in NFA one year would be followed by a parallel change the following year increased, as indicated by shifts in the autoregressive paths. Thus, although NFA may increasingly crystallize for most youth, those who evidence an increase in NFA_avoid and a decrease in NFA_approach may be at heightened risk for further detriments in their contingent self-worth as they get older.

Implications, Limitations, and Future Directions

This research provides novel information about the development of NFA in the peer context across middle childhood and adolescence, and implicates one salient social experience, namely peer victimization, that contributes to individual differences in NFA. Although the need to belong to social groups may be universal (Baumeister & Leary, 1995), when this need becomes too strong or determines one’s sense of self, it may become detrimental. Specifically, this overreliance on the judgments of others may foster heightened levels of psychological distress and, perhaps, make adolescents more susceptible to peer pressure to secure approval and disapproval. Although the extent to which youth base their self-worth on peer judgments generally became less malleable during adolescence, increases in negative experiences such as victimization prompted increases in NFA_avoid, which then elicited further victimization. Preventing this dangerous cycle may require encouraging victimized youth to base their self-worth on internal standards rather than peer feedback while at the same time helping them develop positive peer relationships that promote self-worth. In addition, the increasing stability of both peer victimization and NFA from middle childhood through early adolescence highlights the importance of early prevention programs that break the self-perpetuating cycle between victimization and NFA_avoid while they are relatively more malleable.

Despite these contributions, some limitations of this research should be noted. First, this study used a variable-centered approach to examine the association between peer victimization and the two dimensions of NFA. Although we adjusted for the alternate dimension of NFA, another interesting approach to understanding approach- and avoidance-oriented contingent self-worth involves a person-centered perspective in which combinations of these tendencies are considered. Indeed, one study found that NFA_approach and NFA_avoid interacted to predict social-evaluative concerns, such that NFA_avoid was correlated more strongly with social-evaluative concerns for children with low (vs. high) levels of NFA_approach (Rudolph et al., 2005). Future research examining different profiles of NFA_approach and NFA_avoid would be helpful for better understanding the antecedents and consequences of peer approval based contingent self-worth.

Second, because the goal of the present study was to understand the evolution of NFA, analyses focused specifically on the reciprocal associations between NFA and peer victimization. In future research it would be helpful to understand the unique contribution of NFA to future peer victimization beyond other possible contributors, such as other motivational (e.g., general behavioral activation and inhibition), cognitive-emotional (e.g., rejection sensitivity), or behavioral (e.g., prosocial or withdrawn behavior) aspects of approach and avoidance. As noted earlier, it also would be beneficial to consider the role of positive peer experiences (e.g., peer popularity) as either antecedents or consequences of NFA.

Third, the nonsignificant association between peer victimization and NFA_approach suggests the possibility that individual or contextual factors may moderate the reciprocal associations between peer victimization and NFA. For example, research suggests that the impact of social approach and avoidance tendencies may differ depending on youths’ level of inhibitory control, such that self-regulatory capacity may determine whether these tendencies are translated into more or less adaptive functioning (Rudolph et al., 2013). In addition, given the use of a community sample in the US, we need to be cautious about generalization of our results. More research is needed to examine the association between peer victimization and NFA in other socioeconomic and cultural groups. For instance, the meaning and function of NFA may differ across cultures (Crocker & Wolfe, 2001). Social approach tendencies may be less highly valued or appreciated in collective (e.g., Eastern Asian) societies because they may undermine the harmony and cohesiveness of the group (Chen & French, 2008). Likewise, social avoidance tendencies may not be as maladaptive in group-oriented contexts because peers may view social inhibition as acceptable and appropriate (Chen, 2019).

Fourth, this research did not identify the mechanisms underlying the reciprocal associations between victimization and NFA. Furthering our understanding of this developmental process will require understanding how this process unfolds. One possible direction may be the integration of social neuroscience methods into this line of research. Emerging research has identified brain networks involved in processing social reward and social punishment (Guyer et al., 2016; Rudolph et al., 2016). Moreover, research suggests that chronic victimization (Rudolph et al., 2016) and other forms of peer adversity (Rudolph et al., 2020) predict individual differences in neural processing of social reward and punishment. It would be beneficial for future studies to bridge psychological and neuroscience perspectives to elucidate possible pathways underlying the dynamic interchange between peer victimization and NFA.

Conclusion

Building on a limited database of prior research examining correlates or short-term consequences of NFA, this research provides a novel perspective on the development of NFA from mid-childhood through early adolescence and the role of social context in shaping individual differences in NFA. Using the ALT-SR model to disaggregate within-person and between-person effects as well as dual-informant reports of victimization, we documented a cycle wherein peer victimization predicted increases in NFA_avoid over time, which then elicited subsequent increases in peer victimization. This research provides important insights regarding possible points of intervention when working with peer-victimized youth.