What factors influence romantic attraction and mate selection in the United States?

Child Dev. Author manuscript; available in PMC 2012 Aug 16.

Published in final edited form as:

PMCID: PMC3420070

NIHMSID: NIHMS69248

Abstract

This prospective study examined romantic partner selection and socialization among a sample of 78 young adolescents (6th–8th graders). Independent assessments of adolescent and romantic partner adjustment were collected before and after relationships initiated via peer nomination and self-report. Prior to their relationship, adolescents and partners were significantly alike on popularity, physical attraction, and depressive symptoms. Controlling for initial similarity, partners' popularity, depressive symptoms, relational aggression and relational victimization significantly predicted changes in adolescents' functioning in these areas over time. However, the magnitude and direction of change varied according to adolescents' and partners' pre-relationship functioning. In general, adolescents who dated high-functioning partners changed more than those who dated low-functioning partners, and partner characteristics predicted greater change among low versus high-functioning adolescents. Results were consistent even when controlling for best friend characteristics. The current findings are among the first to demonstrate unique contributions of romantic partner characteristics to adolescents' psychosocial functioning.

The emergence of romantic relationships is among the most significant psychosocial developments of adolescence. Romantic experiences appear to be associated with a diverse range of psychosocial outcomes. Some studies document enhancements in social status, romantic self-concept, self-esteem, and feelings of belonging, while others link romantic involvement to higher levels of depression, mood swings, conflict, and antisocial behavior (for reviews see Collins, 2003; Furman & Shaffer, 2003). The processes by which romantic relationships could affect adolescent adjustment are not well understood. Characteristics of romantic partners are believed to be an important factor, but little is known about adolescents' choice of romantic partners or the effect of romantic partner characteristics on psychosocial functioning (Collins, 2003). The current study addresses this important gap in the literature by examining partner selection and socialization in adolescents' romantic relationships.

At all ages, partner selection and socialization processes appear to promote homophily between relationship partners (McPherson, Smith-Lovin, & Cook, 2001). A general tendency to be attracted to similar others increases the odds of choosing partners who share commonalities. Once the relationship is formed, partners tend to shape and reinforce similarities over time. Among youth, homophily has been documented in friendships and peer groups (Espelage, Holt, & Henkel, 2003; Hogue & Steinberg, 1995). In adulthood, homophily has been reported among dating, cohabitating, and married couples (Blackwell & Lichter, 2004; McPherson et al., 2001).

To the extent that homophily pervades social relationships, we might also expect to find homophily within adolescents' romantic relationships. If true, adolescents and partners would be similar to one another prior to relationship formation (selection) and become more similar over time (socialization). Yet we know of no published studies on this topic. Findings of adult romantic homophily are suggestive; however, there are likely to be developmental differences in the partner characteristics that are salient to adolescent and adult relationships. Research on peer influence offers insights about which partner characteristics could be salient during adolescence; yet friend, peer group, and romantic relationships have distinctive features (e.g., sexuality) and are thus not inherently redundant. Studies of adolescents' romantic partner preferences are informative regarding the characteristics adolescent find appealing (e.g., Hansen, 1977; Regan & Joshi, 2003), but preferences do not necessarily predict adolescents' actual choices or behavior. Understanding how adolescents select romantic partners and the contributions of which romantic partner characteristics to adolescents' psychosocial functioning would provide important information about individual differences in links between romantic involvement and adjustment. In the current study, we assessed pre-relationship similarities among prospective romantic partners and examined whether romantic partners' pre-relationship characteristics predicted adolescents' subsequent psychosocial functioning.

Romantic Partner Selection

Studies of interpersonal attraction and adult romantic homophily have consistently identified status dimensions as important to partner selection (McPherson et al., 2001; Regan & Joshi, 2003). Status dimensions include sociodemographic factors (e.g., age, ethnicity) as well as ascribed characteristics, such as social standing or physical attractiveness. The current study focused on two ascribed status variables that are salient to young adolescents – physical appearance and popularity. Adolescents rate appearance as important to romantic partner selection, and adult partners' self-ratings of attractiveness tend to be positively related (Feingold, 1988; Regan & Joshi, 2003; Roscoe et al., 1987). Accordingly, we predicted that adolescents would select romantic partners who are similar to themselves in physical attractiveness and body appeal. Just as friends tend to be similar in social status, we expected that adolescents would also select partners who shared similar levels of popularity (Haselager, Hartup, van Lieshout, & Walraven, 1998).

The function and structure of young adolescents' peer groups could promote further similarity between romantic partners. Because friendships and romantic relationships serve comparable functions, adolescents could use similar criteria to select friends and romantic partners (Furman, Brown, & Feiring, 1999). In addition, early romantic relationships are initiated within mixed-sex peer groups (Connolly, Craig, Goldberg, & Pepler, 2004), which are themselves subject to homophily effects at both the friend and peer group level (Espelage et al., 2003; Hogue & Steinberg, 1995). Hence, similarity could be particularly evident during early adolescence, when adolescents and partners are more likely to both belong to peer groups that already share similar features. Friends could also encourage romantic partner similarity through their roles as judges and brokers of romantic relationships. By defining what makes for suitable partners, friends could actively discourage the selection of partners who are discrepant in developmentally salient social or behavioral norms, such as social status or aggression (Brown, 1999). Such counsel could be especially prescriptive for young adolescents who are nearing peak susceptibility to peer conformity (Berndt, 1979; Krosnick & Judd, 1982; Steinberg & Silverberg, 1986).

For these reasons, we expected that adolescents would be similar to their romantic partners on dimensions that are important to friendships and peer groups at this age, including popularity, physical appearance, depressive symptoms, peer aggression, and peer victimization (Espelage, et al., 2003; Kandel, Davies, & Baydar, 1990; Mariano & Harton, 2005; Prinstein, 2007; Stevens & Prinstein, 2005). However, similarity at the time of partner selection could be greater for observed than for psychological characteristics. After all, overt similarities are more easily identified than internal psychological characteristics during the initial attraction phase of relationships (Kandel, 1978; Urberg, Degirmencioglu, & Tolson, 1998). Accordingly, we predicted that prospective partners would be more similar in their physical appearance, peer popularity, peer aggression, and peer victimization than in their levels of depressive symptoms.

Implicit in these hypotheses is the suggestion that friends and romantic partners are alike. If adolescents select friends who are similar to themselves and romantic partners serve similar functions as friends, we would expect adolescents to choose romantic partners who are similar to their friends. Accordingly, we predicted that friends' and romantic partners' levels of popularity, physical appearance, peer aggression, and peer victimization would be significantly correlated. Yet the particular salience of physical appearance to romantic attraction led us to anticipate that adolescents would be more similar to prospective romantic partners than friends on this dimension.

Romantic Partner Socialization

Adolescents may select peers to whom they are similar, but over time these peers may also influence adolescents' behavior. For instance, several studies indicate that friend characteristics predict changes in adolescents' depressive symptoms, peer aggression, and peer victimization over time, even after controlling for similarity (Espelage et al., 2003; Hogue & Steinberg, 1995; Jaccard, Blanton, & Dodge, 2005). In the current study, we examine whether romantic partner characteristics are predictive of similar changes in adolescents' psychosocial functioning. Such direct evidence of romantic partner socialization is currently absent from the literature. However, research by Adams and colleagues indicates that young adolescents rate romantic partners to be at least as influential as friends on their thoughts, feelings, and behavior (Adams, Laursen, & Wilder, 2001). These findings, coupled with the time and emotion adolescents invest in romantic relationships, suggest that partners could be important socializing agents.

Researchers typically estimate peer socialization by assessing whether peers' functioning at one time significantly predicts youths' functioning at a later time. The current study employed a similar design. We assessed whether partners' psychosocial functioning prior to the relationship predicted change in adolescents' psychosocial functioning after the two began dating. Changes in adolescents' psychosocial functioning predicted by partners' pre-relationship functioning are interpreted as evidence of peer socialization.

We expected that some partner characteristics would be important to selection or socialization, while others would be important to both. As noted earlier, appearance and popularity should each be important to partner selection. Yet only partners' popularity was expected to exert a significant socialization effect when predicting adolescents' functioning over time. To the extent that status grading is an important aspect of early romantic relationships, partners' pre-relationship popularity should be positively associated with adolescents' subsequent popularity (Skipper & Nass, 1966). Prior findings of significant peer socialization when predicting adolescents' depressive symptoms, peer aggression, and peer victimization suggested that romantic partners' functioning in these areas could also be important predictors. Furthermore, we expected that some partner characteristics would predict adolescents' subsequent functioning even in the absence of initial partner similarity. For instance, covert characteristics, such as internalizing symptoms, could be less important to partner selection but then gain influence as partners spend time together (Baker, Milich, & Manolis, 1996; Stevens & Prinstein, 2005).

In the majority of homophily studies, peer socialization is examined in an additive model where peer characteristics are tested as main effects under the assumption that all peers influence all adolescents in a roughly equivalent manner. More recently, Hartup (1999, 2005) has argued that peer socialization is best conceived as an interaction between characteristics of the socializing agent and the socialized individual. In other words, some partners could be more influential than others, and some adolescents could be more open to influence than others. Relatively few studies have examined peer influence from this perspective; however, those that have suggest that healthy and negative peer influences are contingent on both partners' initial level of functioning (e.g., Adams, Bukowski, & Bagwell, 2005; Dishion & Owen, 2002).

Similarly, we hypothesized that high and low functioning romantic partners might predict different patterns of change for high and low functioning adolescents. For example, dating a partner with few psychosocial problems (i.e., high functioning) could be especially helpful for adolescents experiencing more psychosocial problems (i.e., low functioning) but could be less useful for adolescents with few problems. In support of this idea, having a friend who is low on aggression predicts decreases in adolescents' aggression over time, but only for adolescents who are initially more aggressive (Adams et al., 2005). In the current study, we expected that adolescents with more pre-relationship problems (i.e., low functioning) who dated high functioning partners (i.e., those with few pre-relationship problems) would show improvements in psychosocial functioning over time. Adolescents with few pre-relationship problems who paired with similarly high functioning partners were not expected to change.

In contrast, the characteristics of low functioning partners (i.e., those with more pre-relationship problems) were not expected to predict much change in adolescents' psychosocial functioning. High functioning adolescents, who have few psychosocial problems, could be resistant to the problems of their low functioning partners (Adams et al., 2005; Dishion & Dodge, 2005). On the other hand, low functioning partners could reinforce the problems of low functioning adolescents once the two begin dating (Duncan, Boisjoly, Kremer, Levey, & Eccles, 2005; Poulin, Dishion, & Burraston, 2001). In either case, we expected to see little change among adolescents dating low functioning partners.

The Current Study

Although romantic partners are presumed to affect adolescents' development, we know little about how adolescents select romantic partners or the consequences of their choices. This could be partly due to methodological challenges, as isolating selection from socialization effects to predict change as a function of partner characteristics requires longitudinal designs that can identify individuals' future romantic partners. In the current study, we identified adolescent couples within a longitudinal school-based sample. Using a follow-back design, we traced adolescents and romantic partners to a prior data collection to examine pre-relationship similarities. This allowed for a relatively pure estimation of selection effects, as neither adolescents nor partners identified themselves as being romantically involved at that time. Pre-relationship characteristics of romantic partners were then used to estimate the socialization effects of romantic partner characteristics in predicting changes in adolescents' psychosocial functioning over the time prior to (Time 1) and after relationships were established (Time 2).

Young adolescents' friend and romantic relationships are frequently developed at school, and our school-based design allowed us to examine emerging romantic relationships in their broader social context. Although the data points are necessarily anchored around the initiation of romantic relationships, estimates of friendship similarity prior to the initiation of romantic relationships provide important information about the role of friends in romantic partner selection and socialization. Comparing the characteristics of adolescents' extant friends to those of prospective romantic partners as well as comparing the degree of similarity within friend and romantic dyads would offer new data on the relative importance of general and relationship-specific peer selection criteria in early adolescence. The inclusion of best friends also permitted us to examine romantic partner socialization while accounting for co-occurring socialization by best friends. We expected to find similarity but not redundancy across the two relationships. Romantic relationships are a unique facet of adolescent social development (see Collins, 2003; Furman & Shaffer, 2003 for reviews). Adolescents themselves report different expectations for friend and romantic relationships (Connolly & Goldberg, 1999), and empirical associations between the two relationships are significant but modest (Furman, Simon, Shaffer, & Bouchey, 2002).

Methods

Participants

Participants included 78 adolescents (48% female) who were in the sixth (32 %), seventh (35%), and eighth (33%) grade at the beginning of the study. The ethnic composition of the sample included 87% European American, 2% African American, 4% Asian American, 2% Latino American, and 6% from mixed ethnic backgrounds. Participants were enrolled in public schooling within a town of fairly homogenous middle-class socioeconomic status (per capita income = $32,301). According to school records, 11% of children were eligible for free/reduced lunch. These participants were part of a larger study (n = 520) designed to examine developmental trajectories of depression during adolescence. All sixth through eighth grade students were invited to participate in the first phase of data collection for the larger study (Time 1). Consent forms were returned for 92% of families (n = 784); of these 80% of parents gave consent for their child's participation (n = 637, 74% of total population). Students who were absent one of the testing days (n = 10), provided incomplete data (n = 15), or refused to assent (n = 4) were excluded from analyses, yielding a final sample of 598 participants at Time 1. A total of 520 (87%) of these participants were available for testing eleven months later at Time 2, when students were in grades seven through nine. Attrition was due to participants moving away from the area (n = 36), absenteeism (n = 7), incomplete data (n = 30), and refusal to participate (n = 5). No significant difference was found between adolescents who participated in two versus one time points.

The current sample included adolescents who indicated on the Networks of Relationship Inventory (NRI; Furman & Buhrmester, 1985) that they began a romantic relationship between Time 1 and Time 2. Specifically, participants were asked whether they had a boy/girlfriend, the name of this person, and how long this person had been their boy/girlfriend. On average, these romantic relationships lasted 13.63 weeks (SD = 19.10). Because romantic partner and best friend data were needed to examine selection and socialization effects, only those adolescents whose best friend and romantic partner were also participants in the study were included in our sample. To avoid dependency in the data, one member of any reciprocally nominated romantic dyad was randomly dropped from the sample. This resulted in a data set in which each adolescent served as only a target participant or as a romantic partner. Similarly, no adolescent appeared as both a friend and a romantic partner in the data set. We did not drop any target adolescent who was named as another target adolescent's best friend because friendships were not the focus of the study and doing so could have resulted in a biased sample of adolescents whose best friends were not dating.

The 78 target participants were compared to 62 participants who reported a romantic relationship at Time 2 but who did not meet other study criteria (i.e., the best friend or romantic partner did not also participate in the study). No significant group differences emerged for gender, grade, or any Time 1 primary study variables. Similar analyses were conducted to compare adolescents in the larger sample who were and were not dating. Compared to non-dating adolescents, students in romantic relationships were seen by peers as more physically attractive, t (518) = 6.93, p < .001, M = .55, SD = 1.29 vs. M = −.10, SD = .89; more popular, t (518) =6.68, p < .001, M = .53 SD= .98 vs. M = −.10, SD = .95; less sad, t (518) = −4.78, p < .001, M = −.27, SD = .46 vs. M = .06, SD = 1.08; and less physically victimized by peers, t (518) = −2.36, p < .05, M = −.19, SD = .77 vs. M = .04, SD = 1.06.

Procedure

A letter introducing the study was mailed to the homes of all potential participants, and a consent form was sent home with each student. Parents were asked to either grant or deny their consent for their child's participation, and adolescents were asked to return the consent form regardless of their parent's decision. At both time points, questionnaires were administered to adolescents in their classrooms over two days. Each participant received a small token of appreciation (e.g. a key chain), a $5 gift certificate, and a raffle entry for a Sony Playstation 2 or Microsoft Xbox.

Measures

Popularity

Sociometric peer nominations were used to obtain measures of adolescents' peer perceived popularity at both time points. Students at this school were organized into teams about twice the size of traditional academic classrooms. Using an alphabetized roster of all classmates within their academic team, participants were asked to select an unlimited number of peers who were “most popular”. The order of names was counterbalanced on these rosters (e.g. Z through A) to control for possible effects of alphabetization on nominee selection. A continuous measure of perceived popularity was computed by standardizing the number of most popular votes with higher scores indicating greater popularity (Rose, Swenson, & Waller, 2004).

Physical appearance

The peer nomination procedure described above also was used at both time points to obtain peer ratings of physical attractiveness and body appeal. The question for physical attractiveness was “Who is good looking?” and the question for body appeal was “Who is physically fit?” As described above, the number of nominations that each participant received for each question was summed and standardized relative to their teammates.

Peer-rated sadness

Peer nomination procedures were used to assess participants' sadness at each time point from the question “Who looks sad and unhappy most of the time?” The number of nominations received by each participant was summed and standardized relative to their teammates.

Self-rated depressive symptoms

The Children's Depression Inventory was used to obtain self-reports of depressive symptoms at both time points (CDI; Kovacs, 2003). The CDI contains 27 items that assesses affective, cognitive, motivational, and somatic symptoms of depression. For each item, youth choose one of three statements, scored 0 through 2, that best describe their level of depressive symptoms over the prior two weeks. Mean scores were computed for each student, with higher scores reflecting greater levels of depressive symptoms. In the current sample, Cronbach's alpha was .87 at Time 1 and .86 at Time 2.

Peer-rated aggression

Peer nomination procedures were used to assess participants' relational and physical aggression from the questions “Who ignores others or spreads rumors about others when they are mad at them? “ (relational) and “Who starts fights, says mean things, and picks on others?” (physical). The number of nominations received by each participant for each item was summed and standardized relative to their teammates.

Peer-rated victimization

Peer nomination procedures were used to assess participants' relational and physical victimization from the questions “Who has lies, rumors, or mean things said about them?” (relational) and “Who gets picked on or teased by other kids?” (physical). The number of nominations received by each participant for each item was summed and standardized relative to their teammates.

Results

Preliminary Analyses

Ratings of depressive symptoms were obtained from adolescents as well as peers, and the two were significantly related (r = .43, p < .001). Prior to data analysis, scores for self-reported depressive symptoms were standardized to be consistent with sociometric measures and allow for meaningful comparisons across domains of functioning. Means and standard deviations for all measures are presented in Table 1 for target adolescents, romantic partners, and best friends. No significant mean differences were found between target adolescents', romantic partners', and best friends' corresponding values across any of these domains.

Table 1

Means and Standard Deviations for All Primary Time 1 Variables.

Partner Selection

Our first goal was to estimate romantic partner selection effects by examining pre-relationship similarities in developmentally salient domains. Intraclass correlations between adolescents' and romantic partners' scores on each Time 1 variable indicated significant pre-relationship similarities in four domains: popularity, body appeal, physical attractiveness, and self-rated depressive symptoms (see Table 2). To test the hypothesis that similarity would be greater for more observable characteristics, Fisher r to Z tests were calculated to compare the magnitude of similarity correlations across domains (Cohen, 1988). Among the four significant correlations, adolescent to romantic partner similarities were significantly larger for popularity than for physical attractiveness (p < .001) and self-rated depressive symptoms (p < .05). In addition, the similarity correlation for each of the four significant domains was significantly larger than each similarity correlation where no significant correlation was found (i.e., peer rated sadness, physical aggression, relational aggression, physical victimization, and relational victimization; all ps < .01).

Table 2

Correlations among Time 1 Variables for Adolescents, Best Friends, and Romantic Partners.

Parallels between adolescents' friend and romantic relationships had led us to hypothesize that these two relationship partners would share similar characteristics. The intraclass correlations between adolescents and their best friends on Time 1 variables paralleled those reported above for adolescents and prospective romantic partners, with the notable exception of body appeal (see Table 2). Specifically, adolescents were similar to both their best friend and romantic partner on popularity, attractiveness, and self rated depressive symptoms. In addition, adolescents and best friends were significantly alike on Time 1 relational aggression. Comparing the magnitude of the similarity correlations across relationships revealed that adolescents were significantly more like friends than romantic partners on popularity, Williams' t (75) = −2.65, p < .01 and significantly more like romantic partners than friends on body appeal, Williams' t (75) = 2.85, p < .01. No significant differences in adolescents' similarity to romantic partners versus best friends were found for any of the other Time 1 variables. To further explore links between adolescents' friends and romantic partners, we computed Pearson correlations between the Time 1 characteristics of the two relationship partners (see Table 2). Adolescents' extant best friends and prospective romantic partners were significantly similar in their popularity, body appeal, attractiveness, and self rated depressive symptoms at Time 1.

Partner Socialization

The second goal of this study was to estimate romantic partner socialization effects by testing whether partners' pre-relationship characteristics predicted change in adolescents' functioning over time. We predicted that romantic partner characteristics would significantly predict adolescents' functioning at Time 2, even after accounting for co-occurring socialization by best friends. However, we expected that the strength and direction of the effects would vary according to both partners' and adolescents' pre-relationship functioning.

Data analytic strategy

To test these hypotheses, we conducted a series of regression analyses to test the main and interactive effects of adolescents' and romantic partners' pre-relationship characteristics on adolescents' psychosocial functioning at Time 2 (Baron & Kenny, 1986; Holmbeck, 2002). Prior to the analyses, all predictors were centered to reduce multicollinearity (Aiken and West, 1991). For each Time 2 outcome, adolescents' corresponding Time 1 scores were entered in step 1 to assess stability over time. Best friends' Time 1 scores were entered in the second step to control for cooccurring socialization by friends. Romantic partners' Time 1 scores were entered in a third step to assess the main effect of romantic partner characteristics. The product term of adolescents' and romantic partners' Time 1 scores was entered in the final step to assess whether adolescents' and romantic partner's Time 1 functioning interacted to predict adolescents' Time 2 functioning.

All significant adolescent by partner interactions were subsequently probed following procedures described by Holmbeck in which slope estimates are calculated and examined at high (M + 1 SD) and low (M − 1 SD) levels of the moderator (Aiken & West, 1991; Holmbeck, 2002). Significant interactions were probed twice, once with romantic partners' pre-relationship functioning as the moderator and a second time with adolescents' pre-relationship functioning as the moderator. This strategy allowed us to determine how change in adolescents' functioning over time varied according to partners' pre-relationship functioning and by adolescents' pre-relationship functioning. The two are related, though not identical ways of decomposing and understanding the observed moderator effects.

Examining the moderating effects of romantic partner characteristics identifies whether change in adolescents' psychosocial functioning is greater when partners are high versus low functioning. Specifically, post-hoc analyses assess the association between adolescents' functioning at Time 1 and Time 2 (i.e., stability) when partners are high and low functioning. Thus, change (i.e., lack of stability) in adolescents' characteristics over time as a function of romantic partners' characteristics is indicated by a nonsignificant slope (i.e., unstandardized beta). Examining the moderating effects of adolescents' pre-relationship characteristics identifies whether romantic partner characteristics predict different amounts or types of change for high versus low functioning adolescents. Specifically, the post-hoc analyses assess the association between partners' functioning at Time 1 and adolescents' functioning at Time 2 when adolescents were initially high versus low functioning. For these analyses, significant effects as a function of adolescents' initial functioning are indicated by a significant slope.

Controlling for co-occurring socialization by adolescents' best friends in step 2 of the analyses provided a conservative test of romantic partner socialization. To the extent that friend and partner effects are confounded, this strategy could underestimate romantic partner socialization. To examine this possibility, we re- ran the regressions described above without controlling for best friend characteristics (i.e., eliminating Step 2 in the regressions). The pattern of results was very similar to that obtained when controlling for best friend characteristics, with almost identical beta weights and Δ R2 values. No additional significant direct effects emerged for romantic partners' Time 1 characteristics. The similarity of results across the two sets of analyses suggested that romantic partner predictors were not redundant with friend predictors. Therefore, the results presented below are those from the analyses that include best friend values as a covariate in Step 2.

Popularity

Romanic partners' Time 1 popularity was not directly associated with adolescents' Time 2 popularity. However, as hypothesized, the interaction between adolescents' and partners' Time 1 popularity was a significant predictor (see Table 3). Results from post hoc probing of the moderating effects of romantic partners' popularity are plotted in Figure 1, where the lines depict the association between adolescents' popularity at Time 1 and Time 2 when partners' Time 1 popularity was high versus low. The beta coefficients were significant whether adolescents dated either high or low popular partners, (b = .75, p < .001 and b = 1.08, p < .001, respectively), suggesting little change in adolescents' popularity over time. However, the beta coefficient was smaller for adolescents who dated high popular partners, suggesting that these adolescents experienced more change in popularity than those who dated low popular partners.

Graph of Simple Slopes for Interactions between Adolescent and Romantic Partner Time 1 Popularity in Predicting Adolescents' Time 2 Popularity with Romantic Partners' Popularity as the Moderator.

Table 3

Summary of Hierarchical Regression Analyses Examining Main and Moderating Effects of Romantic Partner Characteristics on Adolescents' Popularity, Body Appeal, and Physical Attractiveness at Time 2

Results from post hoc probing of the moderating effects of adolescents' pre-relationship popularity are plotted in Figure 2, where the lines depict the association between partners' popularity at Time 1 and adolescents' popularity at Time 2 when adolescents' Time 1 popularity was high versus low. Here it can be seen that romantic partners' initial popularity predicted the subsequent popularity of low but not high popular adolescents (b = .15, p < .05 and b = −.05, p = ns respectively). Among adolescents who were initially low in popularity, those who dated a high popular partner were more popular at Time 2 than those who dated a low popular partner.

Graph of Simple Slopes for Interactions between Adolescent and Romantic Partner Time 1 Popularity in Predicting Adolescents' Time 2 Popularity with Adolescents' Popularity as the Moderator.

Considered together, these two sets of post hoc analyses facilitate a better understanding of how the combination of adolescents' and partners' initial popularity predict change in adolescents' popularity over time. Figure 3 illustrates these patterns of change for low and high popular adolescents when they dated low versus high popular partners. High and low functioning are defined in relation to the sample mean (M +/− 1 SD). Already high popular adolescents remained fairly stable over time, regardless of their partners' popularity. In contrast, low popular adolescents changed, but this was limited to those who dated high popular partners. Low popular adolescents who dated a high popular partner became more popular while low popular adolescents who dated a low popular partner did not change.

Graph of Change Patterns in Adolescents' Popularity from Time 1 to Time 2 by Combination of Adolescent and Partner Popularity Levels at Time 1.

Physical Appearance

Consistent with our hypotheses, romantic partners' physical appearance (i.e., body appeal or physical attractiveness) did not exert a significant main effect on adolescents' peer rated appearance nor were there interactive effects between romantic partners' and adolescents' physical appearance in predicting Time 2 ratings of physical appearance (see Table 3).

Depressive symptoms

Romanic partners' peer-rated sadness at Time 1 was not directly associated with adolescents' Time 2 peer-rated sadness. As hypothesized, however, the interaction between adolescents' and partners' Time 1 sadness was a significant predictor (see Table 4). Results from post hoc probing of the moderating effects of adolescents' romantic partners are illustrated in Figure 4, and the results are illustrative of partner-moderated effects for other “problem-oriented” outcomes reported subsequently. Figure 4 plots the association between adolescents' Time 1 and Time 2 sadness when romantic partners' Time 1 sadness was high versus low. Adolescents' Time 1 sadness predicted their Time 2 sadness when partners' initial sadness was high (b = 1.06, p < .001) but not low (b = −.10, ns). These findings suggest that adolescents who dated partners who were initially low on sadness changed more than those who dated partners who were initially high on sadness.

Graph of Simple Slopes for Interactions between Adolescent and Romantic Partner Time 1 Peer-Rated Sadness in Predicting Adolescents' Time 2 Peer-Rated Sadness with Romantic Partners' Sadness as the Moderator.

Table 4

Summary of Hierarchical Regression Analyses Examining Main and Moderating Effects of Romantic Partner Characteristics on Adolescents' Depression, Aggression, and Victimization at Time 2

Results from post hoc probing of the moderating effects of adolescents' initial sadness are illustrated in Figure 5, and the findings are illustrative of adolescent-moderated effects for subsequently reported “problem-oriented” outcomes. Here the association between partners' sadness at Time 1 and adolescents' sadness at Time 2 is plotted when adolescents' initial sadness was high versus low. Romantic partners' Time 1 sadness predicted adolescents' Time 2 sadness, but only for adolescents who were initially high on peer-rated sadness, b = .84, p < .01 for high sad and b = −.11, ns for low sad. Among adolescents who were initially high on sadness, those who dated a partner low on sadness were less sad at Time 2 than those who dated a partner high on sadness.

Graph of Simple Slopes for Interactions between Adolescent and Romantic Partner Time 1 Peer-Rated Sadness in Predicting Adolescents' Time 2 Peer-Rated Sadness with Adolescents' Sadness as the Moderator.

Considered together, the two sets of post hoc analyses facilitate a better understanding of how the combination of adolescents' and partners' initial sadness predict change in adolescents' sadness over time. Figure 6 illustrates these patterns of change for low and high functioning adolescents when they dated low versus high functioning partners. It is also illustrative of the results for depressive symptoms, relational aggression, and relational victimization. High and low functioning are defined in relation to the sample mean (M +/− 1 SD). High functioning adolescents (i.e., low on sadness, depressive symptoms, aggression, victimization) remained stable over time, regardless of their partners' level of functioning. In contrast, low functioning adolescents changed over time, but only when they dated a high functioning partner. Low functioning adolescents (i.e., high on sadness, depressive symptoms, aggression, victimization) who dated a high functioning partner were indistinguishable at Time 2 from those high functioning adolescents.

Graph of Change Patterns in Adolescents' Peer-Rated Sadness from Time 1 to Time 2 by Combination of Adolescent and Partner Sadness Levels at Time 1.

For the regression predicting adolescents' self-reported depressive symptoms at Time 2, the interaction between adolescents' and partners' Time 1 depressive symptoms was again a significant predictor (see Table 4). The main effect for partners' Time 1 depressive symptoms was not significant. Results from post hoc probing of the moderating effects of romantic partners are similar to those for peer-rated sadness. Adolescents' Time 1 symptoms predicted their Time 2 symptoms when partners had high (b = .57, p < .001) but not low (b = .28, ns) levels of depressive symptoms at Time 1. These findings suggest that adolescents who dated partners who were low on depressive symptoms changed more than those who dated partners who were high on symptoms.

Results from post hoc probing of the moderating effects of adolescents' initial symptom levels were similar to those for peer-rated sadness. Romantic partners' Time 1 depressive symptoms predicted adolescents' Time 2 depressive symptoms, but only for adolescents who were initially high on symptoms, b = .75, p < .05 for high symptoms and b = −.52, ns for low symptoms. Among adolescents who were initially high on depressive symptoms, those who dated a low symptom partner reported fewer depressive symptoms at Time 2 than those who dated a high symptom partner.

Aggression

Romanic partners' relational aggression at Time 1 was not directly associated with adolescents' Time 2 relational aggression. However, the interaction between adolescents' and partners' Time 1 relational aggression was a significant predictor (see Table 4). Results from the post hoc probing of the moderating effects of romantic partners' Time 1 relational aggression were similar to those found for self and peer-rated depressive symptoms. Adolescents' relational aggression at Time 1 predicted their aggression at Time 2 when partners were high but not low on relational aggression, b = .20, p < .05 and b = −.15, ns respectively. These findings suggest that adolescents who dated partners who were initially low on relational aggression change more than those who dated partners who were initially high on relational aggression.

Results from post hoc probing of the moderating effects of adolescents' Time 1 relational aggression were similar to those reported for self and peer-rated depressive symptoms. Romantic partners' Time 1 relational aggression predicted adolescents' Time 2 relational aggression, but only for adolescents who were initially high on aggression, b = .21, p < .05 for high aggression and b = −.13, ns for low aggression. Among adolescents who were initially high on relational aggression, those who dated a low aggressive partner were less aggressive at Time 2 than those who dated a high aggressive partner.

Neither romantic partners' physical aggression at Time 1 nor the interaction between romantic partner and adolescents' physical aggression was associated with participants' Time 2 physical aggression (see Table 4).

Victimization

Although romantic partners' relational victimization at Time 1 was not directly associated with adolescents' Time 2 relational victimization, the interaction between adolescents' and partners' Time 1 relational victimization was a significant predictor (see Table 4). Adolescents' relational victimization at Time 1 predicted their victimization at Time 2 when partners were high but not low on relational victimization, b = .36, p < .05 and b = −.24, ns respectively. These findings suggest that adolescents who dated partners who were initially low on relational victimization changed more than those who dated partners who were initially high on relational victimization.

Results from post hoc probing of the moderating effects of adolescents' Time 1 relational victimization indicated that romantic partners' Time 1 relational victimization predicted adolescents' Time 2 relational victimization, but only for adolescents who were initially high on victimization, b = .36, p < .05 for high and b = −.22, ns for low victimization. Among adolescents who were initially high on relational victimization, those who dated a partner low on victimization were less victimized at Time 2 than those who dated a partner high on victimization.

As with physical aggression, neither romantic partners' physical victimization at Time 1 nor the interaction between romantic partner and adolescents' physical victimization was associated with participants' Time 2 physical victimization after controlling for target participants' and best friends' Time 1 levels of victimization (see Table 4).

Discussion

This study is among the first to demonstrate the significance of partner selection and socialization processes in adolescents' romantic relationships. Using a longitudinal design, pre-relationship similarities between young adolescent dating partners were isolated to examine patterns of partner selection. Next, we assessed romantic partner socialization by predicting changes in adolescents' psychosocial functioning over time as a function of partners' pre-relationship characteristics. The findings suggest that romantic selection and socialization processes are operating even as youth are just beginning to participate in romantic activities, develop cross-sex interaction skills, and construct their identities as romantic partners (Brown, 1999; Connolly et al., 2004).

Partner Selection

Pre-relationship similarities detected across both peer and self-report provide compelling evidence of young adolescents' attraction to partners who share comparable levels of social standing, appearance, and depressive symptoms. These results extend prior experimental findings that similarity is predictive of interpersonal attraction (Byrne, 1971). Similarities on peer-rated attractiveness and body appeal suggest that prior reports of similarity in partners' self-rated appearance reflect more than correspondence in self-perceptions (e.g., Feingold, 1988). Likewise, similarities in self-rated depressive features suggest that young romantic partners view themselves as comparably depressed.

The pattern of selection findings was not entirely as predicted and could underscore a distinction between two types of attraction involved in relationship formation: attraction based on similarity (Byrne, 1997; Feingold, 1988) and attraction based on broad appeal (Newcomb, Bukowski, & Pattee, 1993). Although high levels of physical attractiveness and social status have broad-based appeal, the current findings suggest that adolescents, like adults, typically pair with romantic partners who share similar levels on these dimensions (Feingold, 1988; Folkes, 1982; Hendrick & Hendrick, 1992). Adolescents also paired with partners who shared similar levels of depressive features. Whereas status and attractiveness have broad-based appeal, depressive features may only be attractive to those sharing similar characteristics (Rosenblatt & Greenberg, 1988; Wenzlaff & Prohaska, 1989). Swann and colleagues have speculated that pairing with similarly depressed partners serves to verify individuals' self-image (Geisler & Swann, 1999; Swann, 2004). The aggregation of similarly depressed youth has been noted in peer groups and close friendships during early to middle adolescence (Hogue & Steinberg, 1995; Mariano & Harton, 2005; Stevens & Prinstein, 2005) and roommate and dating relationships during late adolescence (Joiner & Katz, 1999; Katz, Beach, & Joiner, 1999). The current study extends these findings to early romantic development. Additional work is needed to pinpoint the basis of similarity. Given the parallels between young adolescents' friend and romantic relationships and the nesting of both relationships within peer groups, pre-relationship similarities in depressive features could reflect peer group as well as dyadic influences.

We had expected both depressive and aggressive features would be differentially attractive to youth and result in pre-relationship similarities on both characteristics. However, we found no evidence that either physical or relational aggression led to assortative pairing. One explanation for these results is that aggression may have broad-based rather than differential appeal. Some studies suggest that aggression becomes increasingly attractive to young adolescents regardless of their own levels of aggression (Bukowski, Sippola, & Newcomb, 2000; Pelligrini & Long, 2003). Perhaps this also includes romantic attraction, in which case the general attractiveness of aggression would make selective pairing less likely to occur. The apparent absence of selective pairing on physical aggression could also reflect general developmental declines in the prominence of physical aggression in adolescent peer relationships (Nagin & Tremblay, 2005; Tremblay, Hartup, & Archer, 2005). That friend dyads in this study were similar on relational but not physical aggression supports this idea. In addition, physical aggression is more common among males and could be more appealing to females seeking male partners than vice versa (Pelligrini & Long, 2003; Tremblay et al., 2005). Sample limitations precluded testing for gender-specific selection criteria, but their presence could have attenuated any selection effects for aggression.

Overall, the pattern of findings on partner selection suggests that similarity is important to young adolescents' romantic pairing, but only for certain characteristics, and that the pattern of attraction is similar to but not identical to that among friends. The developmental course of these selection criteria within and across relationships warrants additional attention. Patterns of romantic selection could vary according to developmental shifts in social and personal needs (Bukowski et al., 2000). Some features, such as physical attractiveness, could remain important whereas others, such as intimacy and conflict skills, could gain significance. As the nature of adolescents' relationships change, the allure of some partner characteristics could wane in favor of others that better serve their needs.

Partner Socialization

Romantic partners' pre-relationship characteristics predicted change in adolescents' functioning in various domains, including peer popularity, depression, relational aggression, and relational victimization. Of these, only popularity and depressive symptoms were important to partner selection. This pattern of findings suggests that partners need not be similar to adolescents in order to be influential.

Overall, the results from analyses examining romantic partner socialization add to the literature on peer influence in two important ways. First, they provide evidence for the significance of both friends and romantic partners to adolescent adjustment. Although best friendships were already formed at the time of our initial assessment, friend characteristics remained important predictors of adolescents' physical attractiveness, depressive symptoms, relational aggression, and relational victimization. Yet even with a conservative data analytic strategy that controlled for co-occurring best friend socialization, romantic partner characteristics emerged as significant predictors of changes in adolescents' psychosocial functioning. These findings lend credible support to theoretical assertions that romantic partners are unique and significant socializing agents (Collins, 2003).

Second, our results suggest that whether and how romantic partners affect adolescent functioning depends on characteristics of both adolescents and their partners. No partner characteristics were directly predictive of adolescents' functioning over time. This could partly reflect the stability of the characteristics assessed in this study. The small amounts of observed change in adolescents' functioning could have made it difficulty to detect any direct effects of romantic partner characteristics. Perhaps other behaviors, such as drug and alcohol use, are more variable and open to socialization. Although romantic partners are not the most likely to offer youth drugs, they are among the most difficult to resist, with approximately half of all offers accepted (Trost, Lange, & Kellar-Guenther, 1999). Nonetheless, even when health risk is the focal behavior, estimates of additive peer socialization typically range from nonsignificant to modest. Recent research on deviant peer influence suggests that these main effect models are overly simplistic, as youth vary in both their power to influence and their susceptibility to be influenced (e.g., Dishion & Dodge, 2005; Hartup, 2005). The current findings are consistent with this tenet.

Across multiple outcomes, adolescents who dated high functioning romantic partners tended to change more than those who dated low functioning partners, and partner characteristics were more predictive of change for low versus high functioning adolescents. Detecting these patterns required separate analyses of the moderating effects of adolescents and partners pre-relationships functioning. Putting the two sets of analyses together illustrated a consistent pattern in which low functioning adolescents appeared to benefit from dating high functioning partners. Low popular adolescents who dated a high popular partner gained more popularity over time than the other three groups of adolescents. This particular finding is consistent with prior assertions that early romantic relationships serve to establish adolescents' status and popularity in the peer group (Brown, 1999). Interestingly, already popular adolescents did not seem to suffer much for dating lower popular partners, suggesting that high popular partners could be particularly important to status grading, at least among young adolescents.

A similar pattern was found for the more “problem-oriented” outcomes. Only adolescents who dated high functioning partners showed significant change over time, and romantic partner characteristics only predicted change for low functioning adolescents. When predicting depression, sadness, relational aggression, or relational victimization, adolescents who coupled with partners who had fewer problems showed more positive changes in these areas than those who coupled with partners with more problems. Among adolescents who dated high functioning partners, those who initially had more problems were indistinguishable at Time 2 from those who initially had few problems. These findings raise the interesting possibility that high functioning partners could help mitigate the symptoms of more poorly adjusted youth. Similar findings have been noted among adult couples where patterns of childhood conduct disorder appear to be disrupted by the presence of supportive, nondeviant partners (Laub, Nagin, & Sampsom, 1998; Quinton, Pickles, Maughan, & Rutter, 1993; Werner & Smith, 2001). In general, however, findings indicative of buffering effects are rare in the peer literature and highlight the potential for healthy peer influence at an important juncture in psychosocial development.

Additional research is needed to explain how certain adolescents with psychosocial problems pair with and benefit from high functioning partners. It seems logical that “mismatches” would be more likely to occur on characteristics for which assortative pairing is less likely. Even within the group of characteristics for which assortative pairing is common, there could be individual differences in their relative importance, leading to less similarity on certain characteristics than others. Regardless of how mismatches occur, partner socialization could be stronger in areas that are salient to interpersonal functioning at a given age. Specifying the underlying pathways and mechanisms of these processes over the course of romantic development is an important task for future studies.

The significant stability found for adolescents who dated low functioning partners also merits further inquiry. Within this group, the reasons for stability could vary between those who were initially high and low functioning. High functioning adolescents could be relatively unaffected by the problems of their low functioning partners (Adams et al. 2005). On the other hand, the stability of low functioning adolescents with low functioning partners echoes findings in the deviancy training literature which indicate that psychosocial problems thrive in dyads where both youth are experiencing similar problems (Dishion, McCord, & Poulin, 1999). Our findings are consistent with these interpretations but did not directly test them. Hence, additional research is needed to examine differential reasons for stability among youth who date low functioning partners.

This study is among the first to offer evidence that romantic partner characteristics affect the psychosocial functioning of young adolescents. The potential for partner socialization across diverse domains of functioning at this age is striking given that early romantic relationships tend to be relatively short-lived and low in intimacy. Longer-term studies with larger samples are needed to replicate the current findings, assess their duration, and examine potential carry over to subsequent romantic relationships. This information is critical to understanding trajectories of romantic development and their interface with psychosocial functioning. Assessing indices of positive adaptation (e.g., self-esteem, academic achievement, prosocial activity) will also be useful for understanding how romantic partners could promote development and well-being. Addressing these questions will require more detailed knowledge about the partner characteristics that are salient at different phases of romantic development. As noted earlier, there are likely to be age-related shifts in the salience of selection criteria. Similarly, the relative strength of various partner characteristics is likely to change with romantic development. For example, as romantic relationships become more intimate, partners' ways of seeking and providing support or of managing disagreements could become more potent socializing characteristics. Changes in which characteristics predict changes in functioning could also reflect shifts in adolescents' social activities, including age-related trends to engage in substance use or other health-risk behavior.

Finally, it is important to note that there are likely multiple ways in which romantic partner socialization could occur. The current study focused on estimating socialization effects within single domains, such as predicting changes in adolescents' depressive symptoms from partners' initial levels of depressive symptoms. Other, more complex processes could also be operating. For example, a given partner characteristic could influence a different adolescent characteristic. Similar findings have been noted in the adult literature, where partner support has been shown to disrupt patterns of childhood conduct disorder (Laub, Nagin, & Sampson, 1998; Quinton, Pickles, & Rutter, 1993). Similarly, coupling with an aggressive partner predicts increases in one's depressive symptoms (Simonelli & Ingram, 1998), and these problems could even co-occur in ways that shape both partners' behaviors (Kim & Capaldi, 2004). Such findings suggest a need for additional research to address the socialization effects of partner characteristics across domains of functioning.

Limitations and Future Directions

Several factors should be considered when interpreting the findings from this study. First, we have discussed our longitudinal findings in ways that suggest that romantic partner characteristics influence adolescents' subsequent functioning. Although our results are consistent with peer socialization processes, the data are non-experimental and thus inconclusive with respect to causal explanations. For example, some adolescents may already be headed toward change, and their selection of certain types of partners may be a signal rather than a cause of change.

Other limitations of the current study pertain to the nature of the sample. First, only 30% of participants from the larger data collection met our inclusion criteria of being in a romantic relationship at Time 2. This was a select group of youth, who, on average were more physically fit, good looking, and popular, as well as less depressed and less victimized than their non-dating peers. Their prevalence, however, is similar to base rate estimates of romantic relationships in this age group (e.g., Carver & Udry, 2003; Giordano, Manning, & Longmore, 2006). Although involvement in romantic relationships is not necessarily the norm for middle school students, our findings could still be relevant to the broader population of young adolescents. After all, adolescents are romantically active even when not in relationships. They are preoccupied with romantic issues and intensely aware of romantic relationship dynamics (Connolly et al., 2004).

Our sample was also restricted to middle school students who dated other middle school students at their school. Although this inclusion criterion was necessary to identify partner characteristics, the exclusion of adolescents dating peers from another school or age group could have influenced our findings. Likewise, our sample was limited to youth with same-sex friendships and other-sex romantic partners. Little is known about friend and peer group influences on romantic relationship formation in sexual minority youth. However, it seems reasonable that differences in the availability and identification of potential romantic partners for sexual minority adolescents would have implications for partner selection and socialization processes.

Overall, sample constraints in this study are balanced against numerous methodological strengths, including the assessment of diverse aspects of psychosocial functioning prior and subsequent to romantic relationship initiation; utilization of multiple reporters; isolation of friend and partner characteristics; ; and assessment of multiple moderators of romantic partner socialization. These assets allowed us to identify discrete patterns of selective pairing and identify ways in which the pairing of certain adolescent and partner characteristics predict changes in adolescent adjustment. It is hoped that these findings will stimulate additional research on romantic partner characteristics, as they seem important for understanding individual differences in links between romantic involvement and adjustment.

Acknowledgments

This research was supported by grants from NIMH (R01-MH59766) and the American Foundation of Suicide Prevention awarded to Mitch Prinstein. We would like to thank Annie Fairlie, Robin M. Carter, Daryn David, Carrie Hommel, and Erica Foster for their assistance with data collection and to all of the adolescents and families who participated in this project.

References

• Adams RE, Bukowski WM, Bagwell C. Stability of aggression during adolescence as moderated by reciprocated friendship status and friend's aggression. International Journal of Behavioral Development. 2005;29:139–145. [Google Scholar]
• Adams RE, Laursen B, Wilder D. Characteristics of closeness in adolescent romantic relationships. Journal of Adolescence Special Issue: Adolescent Romance: From Experiences to Relationships. 2001;24:353–363. [PMC free article] [PubMed] [Google Scholar]
• Aiken LS, West SG. Multiple regression: Testing and interpreting interactions. Sage Publications; Thousand Oaks, CA: 1991. [Google Scholar]
• Baker M, Milich R, Manolis MB. Peer interactions of dysphoric adolescents. Journal of Abnormal Child Psychology. 1996;24:241–255. [PubMed] [Google Scholar]
• Baron R, Kenny D. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology. 1986;51:1173–1182. [PubMed] [Google Scholar]
• Berndt T. Developmental changes in conformity to peers and parents. Developmental Psychology. 1979;15:608–616. [Google Scholar]
• Blackwell DL, Lichter DT. Homogamy among dating, cohabitating, and married couples. Sociological Quarterly. 2004;45:719–737. [Google Scholar]
• Brown BB. “You're going out with who?” Peer group influences on adolescent romantic relationships. In: Furman W, Brown BB, Feiring C, editors. The development of romantic relationships in adolescence. Cambridge University Press; New York, NY: 1999. pp. 291–329. [Google Scholar]
• Bukowski WM, Sippola LK, Newcomb A. Variations in patterns of attraction of same-and other-sex peers during early adolescence. Developmental Psychology. 2000;36:147–154. [PubMed] [Google Scholar]
• Byrne DS. An overview (and underview) of research and theory within the attraction paradigm. Journal of Social and Personal Relationships. 1997;14:417–431. [Google Scholar]
• Carver JR, Udry JR. National estimates of adolescent romantic relationships. In: Florsheim P, editor. Adolescent romantic relations and sexual behavior: Theory, research, and practical implications. Lawrence Erlbaum Associates Publishers; Mahwah, NJ: 2003. pp. 23–56. [Google Scholar]
• Caspi A, Herbener ES. Continuity and change: Assortative marriage and the consistency of personality in adulthood. Journal of Personality and Social Psychology. 1990;58:250–258. [PubMed] [Google Scholar]
• Cillessen AHN, Mayeux L. Sociometric status and peer group behavior: Previous findings and current directions. In: Kupersmidt JB, Dodge KA, editors. Children's peer relations: From development to intervention. American Psychological Association; Washington, DC: 2004. pp. 3–20. [Google Scholar]
• Cohen J. Set correlations and contingency tables. Applied Psychological Measurement. 1988;12:425–434. [Google Scholar]
• Collins WA. More than Myth: The Developmental Significance of Romantic Relationships During Adolescence. Journal of Research on Adolescence. 2003;13:1–24. [Google Scholar]
• Connolly J, Craig W, Goldberg A, Pepler D. Mixed-gender groups, dating, and romantic relationships in early adolescence. Journal of Research on Adolescence. 2004;14:185–207. [Google Scholar]
• Connolly J, Goldberg A. Romantic relationships in adolescence: The role of friends and peers in their emergence and development. In: Furman W, Brown BB, Feiring C, editors. The development of romantic relationships in adolescence. Cambridge University Press; New York, NY: 1999. pp. 266–290. [Google Scholar]
• Dishion T, Dodge KA. Peer contagion in interventions for children and adolescents: Moving towards an understanding of the ecology and dynamics of change. Journal of Abnormal Child Psychology. 2005;33:395–400. [PMC free article] [PubMed] [Google Scholar]
• Dishion T, McCord J, Poulin F. When interventions harm: Peer groups and problem behavior. American Psychologist. 1999;54:755–764. [PubMed] [Google Scholar]
• Duncan G, Boisjoly J, Kremer M, Levy DM, Eccles J. Peer effects in drug use and sex among college students. Journal of Abnormal Child Psychology. 2005;33:375–385. [PubMed] [Google Scholar]
• Espelage DL, Holt MK, Henkel RR. Examination of peer-group contextual effects on aggression during early adolescence. Child Development. 2003;74:205–220. [PubMed] [Google Scholar]
• Feingold A. Matching for attractiveness in romantic partners and same-sex friends: A meta-analysis and theoretical critique. Psychological Bulletin. 1988;104:226–235. [Google Scholar]
• Folkes VS. Forming relationships and the matching hypothesis. Personality and Social Psychology Bulletin. 1982;8:631–636. [Google Scholar]
• Furman W, Brown BB, Feiring C. The development of romantic relationships in adolescence. Cambridge University Press; New York, NY, US: 1999. [Google Scholar]
• Furman W, Buhrmester D. Children's perceptions of the personal relationships in their social networks. Developmental Psychology. 1985;21:1016–1024. [Google Scholar]
• Furman W, Shaffer L. The role of romantic relationships in adolescent development. In: Florsheim P, editor. Adolescent romantic relations and sexual behavior: Theory, research, and practical implications. Lawrence Erlbaum Associates Publishers; Mahwah, NJ: 2003. pp. 3–22. [Google Scholar]
• Furman W, Simon VA, Shaffer L, Bouchey HA. Adolescents' working models and styles for relationships with parents, friends, and romantic partners. Child Development. 2002;73:241–255. [PubMed] [Google Scholar]
• Giesler RB, Swann WB., Jr. Striving for confirmation: The role of self-verification in depression. In: Joiner T, Coyne JC, editors. The interactional nature of depression: Advances in interpersonal approaches. American Psychological Association; Washington, DC: 1999. pp. 189–217. [Google Scholar]
• Giordano PC, Manning WD, Longmore MA. Adolescent romantic relationships: An emerging portrait of their nature and developmental significance. In: Crouter AC, Booth A, editors. Romance and sex in adolescence and emerging adulthood: Risks and opportunities. Lawrence Erlbaum; Mahwah, NJ: 2006. pp. 127–150. [Google Scholar]
• Hansen SL. Dating choices of high school students. The Family Coordinator. 1977;26:133–138. [Google Scholar]
• Hartup W. Constraints on peer socialization: Let me count the ways. Merrill-Palmer Quarterly Special Issue: Peer Influences in Childhood and Adolescence. 1999;45:172–183. [Google Scholar]
• Hartup W. Peer Interaction: What Causes What? Journal of Abnormal Child Psychology. 2005;33:387–394. [PubMed] [Google Scholar]
• Haselager GJ, Hartup W, van Lieshout CF, Riksen-Walraven JM. Similarities between friends and nonfriends in middle childhood. Child Development. 1998;69:1198–1208. [Google Scholar]
• Hendrick ST, Hendrick C. Romantic love. Sage Publications,Inc; Thousand Oaks, CA, US: 1992. [Google Scholar]
• Hogue A, Steinberg L. Homophily of internalized distress in adolescent peer groups. Developmental Psychology. 1995;30:897–906. [Google Scholar]
• Holmbeck GN. Post-hoc probing of significant moderational and mediational effects in studies of pediatric populations. Journal of Pediatric Psychology. 2002;27:87–96. [PubMed] [Google Scholar]
• Jaccard J, Blanton H, Dodge T. Peer Influences on Risk Behavior: An Analysis of the Effects of a Close Friend. Developmental Psychology. 2005;41:135–147. [PubMed] [Google Scholar]
• Joiner TE, Katz J. Contagion of depressive symptoms and mood: Meta-analytic review and explanations from cognitive, behavioral, and interpersonal viewpoints. Clinical Psychology: Science and Practice. 1999;6:149–164. [Google Scholar]
• Kandel DB. Homophily, selection, and socialization in adolescent friendships. American Journal of Sociology. 1978;84:427–436. [Google Scholar]
• Kandel DB, Davies M, Baydar N. The creation of interpersonal contexts: Homophily in dyadic relationships in adolescence and young adulthood. In: Robins LN, Rutter M, editors. Straight and devious pathways from childhood to adulthood. Cambridge University Press; New York, NY: 1990. pp. 221–241. [Google Scholar]
• Katz JW, Beach SRH, Joiner TE., Jr. Contagious depression in dating couples. Journal of Social and Clinical Psychology. 1999;18:1–13. [Google Scholar]
• Kim HK, Capaldi DM. The association of antisocial behavior and depressive symptoms between partners and risk for aggression in romantic relationships. Journal of Family Psychology. 2004;18:82–96. [PubMed] [Google Scholar]
• Kovacs M. Children's Depression Inventory Manual. MultiHealth Systems, Inc; North Tonawanda, New York: 1992. [Google Scholar]
• Krosnick JA, Judd CM. Transitions in social influence at adolescence: Who induces cigarette smoking? Developmental Psychology. 1982;18:359–368. [Google Scholar]
• LaFontana K, Cillessen AH. Children's perceptions of popular and unpopular peers: A multimethod assessment. Developmental Psychology. 2002;38:635–647. [PubMed] [Google Scholar]
• Laub JH, Nagin DS, Sampson RJ. Trajectories of change in criminal offending: Good marriages and the desistence process. American Sociological Review. 1998;63:225–238. [Google Scholar]
• March JS, Parker JD, Sullivan K, Stallings P. The Multidimensional Anxiety Scale for Children (MASC): Factor structure, reliability, and validity. Journal of the American Academy of Child and Adolescent Psychiatry. 1997;36:554–565. [PubMed] [Google Scholar]
• Mariano KA, Harton H. Similarities in aggression, inattention/hyperactivity, depression, and anxiety in middle childhood friendships. Journal of Social and Clinical Psychology. 2005;24:471–496. [Google Scholar]
• McPherson M, Smith-Lovin L, Cook JM. Birds of a Feather: Homophily in Social Networks. Annual Review of Sociology. 2001;27:415–444. [Google Scholar]
• Moffitt TE. Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review. 1993;100:674–701. [PubMed] [Google Scholar]
• Nagin DS, Tremblay RE. What has been learned from group-based trajectory modeling? Examples from physical aggression and other problem behaviors. Annals of the American Academy of Political and Social Science. 2005;602:82–117. [Google Scholar]
• Newcomb AF, Bukowski WM, Pattee L. Children's peer relations: A meta-analytic review of popular, rejected, neglected, controversial, and average sociometric status. Psychological Bulletin. 1993;113:99–128. [PubMed] [Google Scholar]
• Pelligrini A, Long JD. A sexual selection theory longitudinal analysis of sexual segregation and integration in early adolescence. Journal of Experimental Child Psychology. 2003;85:257–278. [PubMed] [Google Scholar]
• Poulin F, Dishion TJ, Burraston B. 3-year iatrogenic effects associated with aggregating high-risk adolescents in cognitive-behavioral preventive interventions. Applied Developmental Science. 2001;5:214–224. [Google Scholar]
• Prinstein MJ. Moderators of peer contagion: A longitudinal examination of depression socialization between adolescents and their best friends. Journal of Clinical Child and Adolescent Psychology. 2007;36:159–170. [PubMed] [Google Scholar]
• Quinton D, Pickles A, Maughan B, Rutter M. Partners, peers, and pathways: Assortative pairing and continuities in conduct disorder. Development and Psychopathology. 1993;5:763–783. [Google Scholar]
• Regan PC, Joshi A. Ideal partner preferences among adolescents. Social Behavior and Personality. 2003;31:13–20. [Google Scholar]
• Roscoe B, Diana MS, Brooks RH. Early, middle, and late adolescents' views on dating and factors influencing partner selection. Adolescence. 1987;22:59–68. [PubMed] [Google Scholar]
• Rose AJ, Swenson LP, Waller EM. Overt and relational aggression and perceived popularity: Developmental differences in concurrent and prospective relations. Developmental Psychology. 2004;40:378–387. [PubMed] [Google Scholar]
• Rosenblatt A, Greenberg J. Depression and interpersonal attraction: The role of perceived similarity. Journal of Personality and Social Psychology. 1988;55:112–119. [PubMed] [Google Scholar]
• Skipper JK, Nass G. Dating behavior: A framework for analysis and an illustration. Journal of Marriage and the Family. 1966;28:412–420. [Google Scholar]
• Simonelli CJ, Ingram KM. Psychological distress among men experiencing physical and emotional abuse in heterosexual dating relationships. Journal of Interpersonal Violence. 1998;13:667–681. [Google Scholar]
• Steinberg L, Silverberg SB. The vicissitudes of autonomy in early adolescence. Child Development. 1986;57:841–851. [PubMed] [Google Scholar]
• Stevens E, Prinstein MJ. Peer contagion of depressogenic attributional styles among adolescents: A longitudinal study. Journal of Abnormal Child Psychology. 2005;33:25–37. [PubMed] [Google Scholar]
• Swann WB. The trouble with change: Self-verification and allegiance to the self. In: Kowalski RM, Leary MR, editors. The interface of social and clinical psychology: Key readings. Psychology Press; New York, NY: US: 2004. pp. 349–356. [Google Scholar]
• Tremblay RE, Hartup WW, Archer J, editors. Developmental origins of aggression. Guilford Press; NewYork, NY, US: 2005. [Google Scholar]
• Trost M, Langan EJ, Kellar-Guenteher Y. Not everyone listens when you “just say no”: Drug resistance in relational context. Journal of Applied Communication Research. 1999;27:120–138. [Google Scholar]
• Urberg KA, Degirmencioglu SM, Tolson JM. Adolescent friendship selection and termination: The role of similarity. Journal of Social and Personal Relationships. 1998;15:703–710. [Google Scholar]
• Werner EE, Smith RS. Journeys from childhood to midlife: Risk, resilience, and recovery. Cornell University Press; Ithaca, NY, US: 2001. [PubMed] [Google Scholar]
• Wenzlaff RM, Prohaska ML. When misery prefers company: Depression, attributions, and responses to others' moods. Journal of Experimental Social Psychology. 1989;25:220–233. [Google Scholar]