328 STEIGER ALLEMAND RORINS AND FEND adolescent self-estcem (level or change)on depr n over de as rel m eaminin the fpe sion over years (cf.So 0&0th.2013 Method ng age 12 th ntal stag Participants cific evaluation as p ceived physical Grob,La ch.2002)3 Participants (N or 13T aspect 2.003) 把 1980:N= 1.952).and 16 year data from a large,nat T5tnhodw 1,0 or this study.wefir investigated level and change in globa 1.527 em during ado cenc .other have reported the oppo origin,urban ve us rura place o r no chang at all from 12to16 hool levels ace to thei d to fin ool)(see ed in the d itial level of self-esteem and in the diffe from the original lescent sample on any of the key nd r physica al.2009) ve effects of both leve nd char in adolesce we Measures Global self-est Self-e nd tho decline in self-esteem ove r the Self-Esteem Scal (RSES dep le (0 199 each item on a di The d t rify the prosp of self-esteem devel score (potential =0-8).Kuder-Ric (1937 nd 7 for th Gordor 2003 The 2000 rded Domain-pficsPhysical appearance and ac pared to others I'm retty attractiv (pe ter to m d phy trying hard. an elational status a low self- The s are to the Pe .C .2012:Th0 et al..2010).As cale (-not true for me. true for me n ad The item to create scores ranging from 0 to 6.The reliability totest for pe re included tim ovariates that are directly linked to the h suggests,anegative Scha 2012 O'De ond,as previo trates 200 from L Craven.206).we included the participants'school grades as an 断 nsadolescent self-esteem (level or change) on depression over de￾cades. Previous research has relied on short-term longitudinal studies, examining the link between low self-esteem and depres￾sion over years rather than decades (cf. Sowislo & Orth, 2013).2 Instead, we tested the effects of adolescent self-esteem on adult depressive symptoms two decades later, spanning age 12 through age 35, and thus two different developmental stages. Third, most previous research has focused on global self-esteem and neglected domain-specific evaluations such as perceived physical appear￾ance or academic competence. Because these domains are highly consequential during the transitional period of adolescence, we also tested models for these aspects of domain-specific self￾esteem. Fourth, the present study uses data from a large, nationally representative sample for the used variables. For this study, we first investigated level and change in global and domain-specific self-esteem during adolescence. Although most prior studies have revealed a mean-level decrease in self￾esteem during adolescence, other studies have reported the oppo￾site pattern, raising the possibility that we might find an increase, a decrease, or no change at all from age 12 to 16 years. More important, we expected to find substantial individual differences in level and change of self-esteem, suggesting that adolescents differ both in their initial level of self-esteem and in the shape and direction of their development. Second, we expected to find gender differences in self-esteem, with boys exhibiting higher levels of global self-esteem and more positive perceptions of their physical appearance and academic ability. Third, we examined the prospec￾tive effects of both level and change in adolescent self-esteem on depressive symptoms in adulthood. We expected to replicate the vulnerability effect, such that adolescents with low self-esteem, and those showing declines in self-esteem over the course of adolescence, would be more prone to depressive symptoms two decades later. To further clarify the prospective effects of self-esteem devel￾opment on adult depression, we included three time-varying co￾variates: peer popularity, body mass index (BMI), and school grades (Cohen, Kasen, Chen, Hartmark, & Gordon, 2003). The rationale was as follows. According to the sociometer theory (Leary & Baumeister, 2000), global self-esteem can be regarded as a sociometer to monitor how much a person is valued by signifi￾cant others. Thus, when members of desirable groups or social relationships value one’s relational status as low, self-esteem should be low too. Indeed, research supports the assumption that peer popularity is related to higher self-esteem (e.g., Litwack, Aikins, & Cillessen, 2012; Thomaes et al., 2010). As peers play an increasingly important role in adolescence, we included peer-rated popularity as a time-varying covariate (at all measurement occa￾sions in adolescence) in our analyses to test for peer influences on global self-esteem (see Figure 1). For domain-specific self-esteem, we included time-varying covariates that are directly linked to the respective domains. First, as previous research suggests, a negative association between BMI and both perceived physical appearance and depression (Mustillo, Hendrix, & Schafer, 2012; O’Dea, 2006), we controlled for the potential influence of body mass index (BMI) when examining the effects of physical appearance on depressive symptoms. Second, as previous research demonstrates a positive association between academic achievement and perceived academic competence (Guay, Marsh, & Boivin, 2003; Marsh & Craven, 2006), we included the participants’ school grades as an annual covariate when examining the effects of perceived aca￾demic competence. Method Participants Data came from the German LifE-study (Fend, Georg, Berger, Grob, & Lauterbach, 2002).3 Participants (N
2,054) were as￾sessed five times during adolescence at the age of 12 (Time 1 [T1]: 1979), 13 (Time 2 [T2]: 1980; N
2,047), 14 (Time 3 [T3]: 1981; N
2,003), 15 (Time 4 [T4]: 1982; N
1,952), and 16 years (Time 5 [T5]: 1983; N
1,790). A follow-up assessment was conducted in adulthood when participants were 35 years old (Time 6 [T6]: 2002). From the initial study participants, 74% (n
1,527) participated at T6. The adolescent participants are broadly repre￾sentative of the Western German population with regard to socio￾economic status, gender, ethnic origin, urban versus rural place of residence, and education level (in Germany, pupils are divided in separate school levels according to their academic performance in primary school) (see Fend, Berger, & Grob, 2009, for details). Study members who participated in the age 35 assessment did not differ from the original adolescent sample on any of the key variables of this study, including depressive symptoms at age 16 or global and domain-specific self-esteem at ages 12–16 (see Fend et al., 2009). Measures Global self-esteem. Self-esteem in adolescence was measured with eight items from the Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1979). Participants rated each item on a dichotomous scale (0
disagree, 1
agree). The items were summed to create a total score (potential range
0 – 8). Kuder-Richardson (1937) reliability estimates (KR-20) ranged between .72 and .77 for the five measurement occasions. Domain-specific self-esteem: Physical appearance and aca￾demic competence. Each self-esteem domain was measured with six items. “Compared to others, I’m pretty attractive” (per￾ceived physical appearance) and “Even when trying hard, I can’t achieve what others can do easily” (perceived academic compe￾tence). The items are conceptually comparable to the Perceived Competence Scale for Adolescents (Harter, 1982; Wünsche & Schneekind, 1989). Participants rated each item on a dichotomous scale (0
not true for me, 1
true for me). The items were summed to create scores ranging from 0 to 6. The reliability 2 One exception is the study by Schafer, Wickrama, and Keith (1998), who found adult self-esteem level predicting depression in a small sample of 98 married couples over 13 years. However, this study only included two measurement occasions and did not focus on adolescent development as a predictor of adult depression. 3 Lebensverläufe von der späten Kindheit ins frühe Erwachsenenalter (LifE). Die Bedeutung von Erziehungs erfahrungen und Entwicklungs￾prozessen für die Lebensbewältigung – Follow-Up zur Konstanzer Ju￾gendlängsschnittstudie, Entwicklung im Jugendalter‘ (Authors: Fend, Georg, Berger, Grob, & Lauterbach, 2002). [Pathways from Late Child￾hood to Adulthood. Context and Development in Adolescence as Predic￾tors of Productive Life-Courses (Lebensverläufe ins frühe Erwachsenenal￾ter: LifE)]. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 328 STEIGER, ALLEMAND, ROBINS, AND FEND