Table
1 presents descriptive statistics for all variables included in the analyses. Mean scores suggest that the girls in this study reported, on average, internalizing symptoms within a normal range. The average participant received approximately two popularity nominations from peers and one rumor nomination at each wave. At the beginning of 6th grade, the average self-perceived pubertal development score was 2.13 (SD = .63), indicating that girls, on average, reported their physical development as just beginning. This average score is similar to that reported among 10–12 year old African American females (Ge et al.
2003,
2006) and slightly lower than that reported among 7th grade white girls (Ge et al.
2001). There were no significant ethnic group differences in average PDS scores,
F (3, 908) = 1.49, MSE = .40,
n.s.
Table 1
Means and standard deviations for study variables separated by wave and source
Self-report |
Pubertal timing | 2.14 | .64 | 912 | | | | | | |
Self-worth | | | | 3.25 | .71 | 787 | 3.22 | .75 | 705 |
Depressive symptoms | | | | .26 | .32 | 798 | .27 | .34 | 710 |
Teacher-report |
Attractiveness | 4.80 | 1.48 | 912 | | | | | | |
Popularity among boys | 4.01 | 1.42 | 884 | 3.99 | 1.51 | 799 | | | |
Popularity among girls | 4.87 | 1.38 | 895 | 4.84 | 1.35 | 794 | | | |
Peer-report |
Popularity among boys | .22 | .60 | 912 | .21 | .64 | 858 | | | |
Popularity among girls | 1.50 | 1.74 | 912 | 1.57 | 2.03 | 858 | | | |
Rumors | .88 | 1.47 | 912 | 1.03 | 1.81 | 857 | | | |
Perceived popularity and Early Maturation
The first hypothesis states that more advanced pubertal development compared to peers at the start of middle school will be associated with peer- and teacher-reported perceived popularity in both the fall and the spring of 6th grade. Two sets of hierarchical multiple regression analyses were computed to predict indicators of same-sex and opposite-sex popularity based on peer nominations and teacher-ratings while controlling ethnicity and teacher-reported attractiveness at Step 1 of the analyses. Ethnicity was dummy coded such that Latinas (the largest group) served as the reference point for the contrasts.
Peer-Nominated Popularity
Popularity nominations received from boys and girls in the fall served as the outcome variables, respectively, in two regression analyses. As shown in Table
2, girls’ more advanced physical development in the fall of 6th grade was associated with greater perceptions of popularity both among boys (
β = .14,
p < .001) and among girls (β = .14,
p < .001) in the fall of 6th grade. Results of two similar hierarchical multiple regression analyses indicated a prospective association between more advanced pubertal development at the start of the school year and perceived popularity among boys (β = .09,
p < .01) and among girls (β = .09,
p < .01) at the end of the school year, controlling ethnicity and attractiveness (see Table
2).
Table 2
Summary of hierarchical regression analyses predicting peer-nominated and teacher-reported popularity among boys and girls in the fall and spring of 6th grade
Peer nominated popularity |
Asian | −.06 | .09 | −.03 | .10 | .01 | .10 | −.01 | .11 |
Black | −.02 | .06 | .02 | .07 | .00 | .07 | −.01 | .07 |
White | −.02 | .10 | −.02 | .11 | .05 | .11 | −.01 | .12 |
Attractiveness | .10** | .02 | .23*** | .02 | .16*** | .02 | .15*** | .02 |
Pubertal timing | .14*** | .03 | .14*** | .03 | .09** | .03 | .09** | .03 |
R
2
| .03 | | .07 | | .04 | | .03 | |
Teacher reported popularity |
Asian | −.05 | .14 | .04 | .14 | −.11** | .16 | −.02 | .15 |
Black | .03 | .09 | .02 | .10 | −.05 | .11 | −.04 | .10 |
White | .02 | .14 | .07* | .15 | −.02 | .18 | .04 | .16 |
Attractiveness | .53*** | .03 | .43*** | .03 | .44*** | .03 | .40*** | .03 |
Pubertal timing | .15*** | .04 | .08** | .04 | .14*** | .05 | .07* | .04 |
R
2
| .30 | | .20 | | .21 | | .16 | |
Teacher-Reported popularity
The concurrent association between pubertal maturation and teacher-reported popularity among boys and girls in the fall of 6th grade was tested via two hierarchical multiple regression analyses. When entered at step two in each model, more advanced pubertal development was associated with higher teacher-reported popularity among boys (β = .15,
p < .001) and among girls (β = .08,
p < .01) in the fall of 6th grade, controlling ethnicity and teacher-reported attractiveness at step one (see Table
2). Similarly, in regression analyses modeling advanced pubertal maturation as a prospective predictor of teacher-reported popularity in the spring of 6th grade, more advanced development in the fall predicted greater perceptions of popularity among boys (β = .14,
p < .001) and among girls (β = .07,
p < .05) in the spring after controlling ethnicity and attractiveness.
In support of the first hypothesis, results suggest that being more physically developed compared to other girls in school at the start of 6th grade is associated, via converging views, with greater popularity among both boys and girls at the beginning and end of 6th grade. Thus, earlier maturation seems to confer some concurrent and subsequent benefits. Beyond the variance accounted for by stability in perceived popularity from fall to spring of 6th grade, earlier pubertal timing continued to significantly predict only teacher-reported popularity among boys (β = .07, p < .05). According to teacher perceptions, then, early maturing girls increased in popularity with boys across the first year of middle school.
Rumors and Early Maturation
The second hypothesis states that relatively advanced pubertal development will be associated with rumors among peers. Two hierarchical multiple regression analyses, controlling ethnicity and attractiveness at step one, tested the concurrent and prospective association between advanced pubertal maturation and peer-nominated rumors. As shown in Table
3, more advanced physical development compared to peers in the fall of 6th grade predicted rumors concurrently in the fall (β = .11,
p < .01) and prospectively in the spring of that same year (β = .14,
p < .001), controlling ethnicity and attractiveness. Compared to Latina girls, black and white girls had a stronger reputation as targets of rumors in the fall of 6th grade (β = .13,
p < .001; β = .10,
p < .01, respectively); however, ethnicity was not significantly related to rumors in the spring of that year. Greater physical attractiveness, as rated by teachers, was associated with a lower degree of rumors in the fall (β = −.09,
p < .01) and spring of 6th grade (β = −.07,
p < .05).
Table 3
Summary of hierarchical regression analyses predicting rumors in the fall and spring of 6th grade
Asian | .01 | .10 | .02 | .11 |
Black | .13*** | .07 | .05 | .07 |
White | .10** | .11 | .06 | .12 |
Attractiveness | −.09** | .02 | −.07* | .02 |
Pubertal timing | .11** | .03 | .14*** | .03 |
R
2
| .04 | | .03 | |
In support of the second hypothesis, girls who reported more advanced development compared to peers at the start of middle school had a stronger reputation as the target of rumors both concurrently in the fall of 6th grade, as well as a semester later in the spring. Furthermore, change models suggest that earlier pubertal maturation continues to predict rumors in the spring of 6th grade (β = .08, p < .01) even after fall rumors are controlled in statistical analyses. Thus, earlier maturing girls experienced increases in peer nominations of rumors across the first year of middle school.
Social Benefits and Social Costs
The third hypothesis predicted that popularity among boys, in particular, would serve as a partial mediator of the association between advanced pubertal development and future rumors. As presented above and in Table
2, analyses revealed that more advanced pubertal development compared to peers in the fall of 6th grade was significantly associated with a stronger reputation of popularity among boys at that same wave (β = .14,
p < .001). Table
4 shows that popularity among boys, in turn, predicted rumors in the spring of 6th grade, after controlling pubertal timing, ethnicity, and attractiveness (β = .11,
p < .01). Next, as reported above, more advanced pubertal development in the fall significantly predicted rumors in the spring of 6th grade (β = .14
p < .001). Finally, the association between pubertal timing and rumors decreased when popularity among boys was taken into account (β = .13,
p < .01), indicating that having a reputation among boys as being more popular partially helps explain the link between advanced pubertal development at the start of middle school and rumors at the end of the first school year.
Table 4
Summary of hierarchical regression analyses testing peer- and teacher- reported popularity among boys as a mediator of the link between pubertal timing and rumors
Step 1 |
Asian | .02 | .11 | | .02 | .11 | |
Black | .05 | .08 | | .05 | .08 | |
White | .05 | .12 | | .05 | .12 | |
Attractiveness | −.08* | .02 | .009 | −.08* | .02 | .009 |
Step 2 |
Asian | .02 | .11 | | .02 | .11 | |
Black | .05 | .07 | | .05 | .07 | |
White | .06 | .12 | | .06 | .12 | |
Attractiveness | −.07* | .02 | | −.07* | .02 | |
Pubertal timing | .14*** | .03 | .02*** | .14*** | .03 | .02*** |
Step 3 |
Asian | .03 | .11 | | .01 | .11 | |
Black | .05 | .07 | | .05 | .08 | |
White | .06 | .12 | | .06 | .12 | |
Attractiveness | −.09* | .02 | | −.13* | .03 | |
Pubertal timing | .13** | .03 | | .12** | .03 | |
Boy popularity | .11** | .04 | .01** | .10* | .03 | .01* |
R
2
| .04 | | | .04 | | |
Teacher-reported popularity among boys in the fall was also examined at a multivariate level via hierarchical regression as a potential mediator of the link between earlier pubertal maturation and rumors. Results indicated that more advanced pubertal development in the fall of 6th grade was associated with teacher reports of popularity among boys at the same wave (β = .15,
p < .001), which were in turn associated with rumors in the spring (β = .10,
p < .05), controlling pubertal timing (see Table
4). Similar to the findings based on peer nominations, the strength of the association between earlier pubertal maturation and rumors in the spring (β = .14,
p < .001) decreased when teacher-reported popularity among boys was included as a predictor in the model (β = .12,
p <
.01), providing evidence from another perspective that popularity among boys is a partial mediator of the link between relatively advanced timing in the fall and the receipt of more rumor nominations from peers in the spring of sixth grade.
A bootstrapping method of testing the significance and effect-size of mediation empirically generates sampling distributions of the indirect effects by taking a sample (with replacement) of size
n from the full data set and calculating the indirect effects in each of the samples (Preacher and Hayes
2004,
2008). Point estimates and 95% confidence intervals for the indirect effects are generated and significant mediation is implied when the 95% confidence interval for the indirect effect does not contain zero. Employing this procedure (with
n = 5,000 bootstrap samples) to examine mediation in the model based on peer-reports of popularity among boys yielded a 95% bias corrected and accelerated confidence interval (BCa
CI) of .003 through .030. Because this interval does not contain 0, the indirect effect was significant, indicating that a stronger reputation among boys partially explains the link between pubertal maturation and later rumors. Repeating this procedure for the model focusing on teacher-reported popularity among boys also revealed the indirect effect to be significantly different from zero (BCa
CI of .003 through .029). Thus, being relatively earlier maturing than other girls at the start of middle school predicts popularity among boys (as reported by both peers and teachers), which in turn helps explain the link between early pubertal timing and rumors at the end of 6th grade.
Social Reputation and Emotional Adjustment
The final hypothesis stated that peer-nominated rumors would be a partial mediator of the link between early pubertal maturation and the internalizing symptoms of depression and low self-worth. First focusing on depressive symptoms, it was predicted that rumors in the spring of 6th grade would partially mediate the link between pubertal timing measured in the fall of 6th grade and depressive symptoms reported both in the spring of 6th grade, as well as in the spring of the following school year. As can be seen in Table
5, pubertal timing was significantly associated with depressive symptoms reported in the spring of 6th grade (β = .14,
p < .001). Pubertal timing was also shown, in Table
3, to significantly predict rumors in the spring of 6th grade (β = .14,
p < .001), which in turn are associated with depressive symptoms reported in the spring of 6th grade, controlling for pubertal timing (β = .14,
p < .001). Finally, the effect of pubertal timing on depressive symptoms in the spring of 6th grade was lowered when rumors were taken into account (β = .12,
p < .01). More advanced development compared to peers in the fall of 6th grade was also linked with greater reports of depressive symptoms in the spring of the following school year (β = .10,
p < .01). The strength of the association between pubertal development and depressive symptoms reported in the spring of 7th grade decreased to non-significance when rumors were included in the model (β = .07,
ns).
Table 5
Summary of hierarchical regression analyses testing rumors as a mediator of the link between pubertal timing and internalizing symptoms reported in the spring of 6th and 7th grades
Step 1 |
Asian | −.01 | .04 | | .00 | .09 | | .03 | .04 | | −.11** | .09 | |
Black | −.05 | .03 | | .05 | .06 | | −.11** | .03 | | .13** | .07 | |
White | .00 | .04 | | .04 | .09 | | −.08* | .05 | | .10* | .11 | |
Attractiveness | −.05 | .01 | .005 | .08* | .02 | .01 | .03 | .01 | .02* | −.04 | .02 | .04*** |
Step 2 |
Asian | .00 | .04 | | −.01 | .09 | | .04 | .04 | | −.11** | .09 | |
Black | −.05 | .03 | | .05 | .06 | | −.11* | .03 | | .13** | .07 | |
White | .01 | .04 | | .03 | .09 | | −.07 | .05 | | .10* | .10 | |
Attractiveness | −.04 | .01 | | .07 | .02 | | .03 | .01 | | −.04 | .02 | |
Pubertal timing | .14*** | .01 | .02*** | −.14*** | .03 | .02*** | .10** | .01 | .01* | −.12** | .03 | .01** |
Step 3 |
Asian | −.01 | .04 | | .00 | .09 | | .02 | .04 | | −.11** | .09 | |
Black | −.05 | .03 | | .06 | .06 | | −.11** | .03 | | .13** | .07 | |
White | .00 | .04 | | .04 | .09 | | −.08 | .05 | | .10* | .10 | |
Attractiveness | −.03 | .01 | | .06 | .02 | | .04 | .01 | | −.04 | .02 | |
Pubertal timing | .12** | .01 | | −.13** | .03 | | .07 | .01 | | −.10* | .03 | |
Rumors | .14*** | .01 | .02*** | −.11** | .03 | .01** | .12** | .02 | .01** | −.09* | .03 | .01* |
R
2
| .04 | | | .04 | | | .04 | | | .06 | | |
A bootstrapping method (with n = 5,000 bootstrap samples) was employed to examine the significance of the reported mediation. The model examining depressive symptoms in the spring of 6th grade (BCa CI of .002 through .013) and the model examining depressive symptoms in spring of 7th grade (BCa CI of .002 through .014) both indicated that the indirect effect was significantly different from zero. Thus, rumors in the spring of 6th grade partially account for the association between pubertal timing and depressive symptoms. This was true when symptoms were measured in the spring of 6th grade, shortly after pubertal timing was assessed, and when symptoms were measured an entire year later.
Mediation analyses were repeated with self-worth as the dependent variable instead of depressive symptoms (see Table
5). More advanced development in the fall of 6th grade was associated with lower self-worth in the spring of that year (β = -.14,
p < .001), as well as the following spring (β = −.12,
p < .01). Rumors in the spring of 6th grade were also associated with low self-worth that same wave (β = −.11,
p < .01), and a year later (β = −.09,
p < .05), controlling pubertal timing. Finally, the strength of the relationship between pubertal timing and low self-worth decreased when rumors were accounted for in the regression model predicting low self-worth in the spring of 6th grade (β = −.13,
p < .001), and the spring of 7th grade (β = −.10,
p < .01).
A bootstrapping method of testing the significance of mediation (with n = 5,000 bootstrap samples) was employed to examine the model predicting self-worth in the spring of 6th grade (BCa CI: −.024 to −.003) and the spring of 7th grade (BCa CI: −.024 to −.002). Indirect effects were found to differ significantly from zero in both models, indicating partial mediation. Thus, rumors in the spring of 6th grade partially accounted for the association between pubertal timing and low self-worth, both when self-worth was measured in the spring of 6th grade, shortly after pubertal timing was assessed, and when it was measured an entire year later.