Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Child Development, xxxx 2020, Volume 00, Number 0, Pages 1–18

Pathways From Childhood Sociomoral Sensitivity in Friendship, Insecurity,


and Peer Rejection to Adult Friendship Quality
Annekatrin Steinhoff Monika Keller
University of Zurich Max Planck Institute for Human Development

Across the life span, friendship is an important component of people’s support networks. This article explores
the developmental roots of adult friendship intimacy and satisfaction, taking into consideration the early inter-
play between sociomoral sensitivity in friendship, insecurity in peer contexts, and peer rejection. Data
(N = 176) came from the longitudinal study “Individual Development and Social Structure.” Respondents
were surveyed repeatedly from age 7 to 37 years. Autoregressive cross-lagged panel models show that socio-
moral sensitivity in friendship protects adolescents from peer rejection and is reciprocally associated with inse-
curity. Childhood and adolescent sociomoral sensitivity antecede early adult friendship intimacy, which, in
turn, antecedes friendship satisfaction in mid-adulthood. The findings indicate a sequence of pathways from
sociomoral sensitivity developed early in life to friendship quality in adulthood.

Across the life span, friendship is an important Bokhorst, Sumter, & Westenberg, 2010). Given the
social resource with the potential to support indi- ongoing significance of friendship for humans thriv-
vidual growth and self-realization. During child- ing, it is important to understand how individual
hood and adolescence, friendships are main differences in friendship quality emerge and carry
contexts for the development of sociomoral compe- on over time. To date longitudinal studies on
tencies, including perspective taking, the under- friendship development from childhood to adult-
standing of rules and norms, and empathy hood are largely missing.
(Bukowski & Sippola, 1996; Caravita, Sijtsema, The aim of this study was to address this gap by
Rambaran, & Gini, 2014; Keller, 1996). These com- examining pathways of friendship development
petencies, in turn, likely affect friendship quality in over three decades of life taking into consideration
sequential periods of life, thus enabling individuals age-related developmental tasks in the realm of pla-
to establish intimate social bonds. The latter are tonic peer relationships. We conceptualize these
essential for identity development (Erikson, 1959; tasks as (a) the development of a moral sense in
Havighurst, 1976) and are typically prerequisites close friendship (i.e., a sense of agency and respon-
for psychological well-being (Ryff & Singer, 2008). sibility), mainly progressing during the period from
Indeed, supportive friendships can help young peo- childhood to mid-adolescence and likely to be inter-
ple navigate the potentially stressful transition to related with experiences and behaviors in the
adulthood (Bagwell et al., 2005), are consistently broader peer context; (b) the formation of intimate
associated with better mental health (Bagwell et al., friendship by emerging adulthood; and (c) the real-
2005; Cable, Bartley, Chandola, & Sacker, 2013) and ization of satisfactory friendship in mid-adulthood.
remain a vital component of adult social networks, We therefore explore stability and change in indi-
particularly during times of personal hardship (e.g., vidual differences in friendship patterns by investi-
gating a pathway from childhood and adolescent
The data used in this study come from the project “Individual sociomoral development to early adult and mid-
Development and Social Structure,” also known as the “Iceland adult friendship characteristics. Data came from the
Study,” which was directed by Wolfgang Edelstein, director
emeritus at the Max Planck Institute of Human Development in Icelandic longitudinal study “Individual Develop-
Berlin. We thank Wolfgang Edelstein and Matthias Grundmann ment and Social Structure (IDSS)” (Edelstein, Keller,
for fruitful discussions and support. We are grateful to all our & Schr€ oder, 1990), assessing individual develop-
Icelandic cooperation partners, including interviewers, teachers,
and interviewees. ment in the cognitive, personality, and sociomoral
Correspondence concerning this article should be addressed to
Annekatrin Steinhoff, Jacobs Center for Productive Youth Devel-
opment, University of Zurich, Andreasstrasse 15, 8050 Zurich, © 2020 Society for Research in Child Development
Switzerland. Electronic mail may be sent to steinhoff@jacobscente All rights reserved. 0009-3920/2020/xxxx-xxxx
r.uzh.ch. DOI: 10.1111/cdev.13381
2 Steinhoff and Keller

domain from ages 7 to 37 years. When examining of the diverging expectations, interests, and action
friendship development, we consider that behaviors capacities of the self and others in relationships
and perceptions in particular relationships may be (here: friendship). Based on developmental inter-
shaped by broader relationship representations. The views from childhood to adolescence in the longitu-
IDSS project offers a rich database with information dinal study IDSS, Keller and her colleagues
on various aspects of individual development and described four developmental levels: (0) no differen-
socialization. We make use of these rich data when tiation between the subjective perspectives of rela-
testing whether there are unique associations among tionship partners (friends); (1) beginning
indicators of friendship development across the life understanding of different subjective perspectives,
span. informed by individual needs, interests, and expec-
tations for friendship; (2) taking into account an
intersubjective perspective on shared standards in
Theoretical Considerations
the relationship; (3) a third person’s perspective on
We present our theoretical considerations in three the relationship which integrates generalized norms
steps which describe the hypothesized sequence of with the requirements of particular situations (Kel-
pathways from child to adult friendship develop- ler, 1996, p. 132; Keller & Edelstein, 1991, p. 102;
ment. First, we introduce the concept of sociomoral Keller & Reuss, 1984). Being a morally sensitive
sensitivity (Keller, 1984, 1996) and explore whether person and friend would mean to consider not only
interrelations with (a) insecurity in peer contexts and both the self’s and the friend’s subjective perspec-
(b) peer rejection can help explain individual differ- tives, including desires, feelings, expectations, and
ences in friendship-related sociomoral sensitivity action orientations, but also the rules and norms
during childhood and adolescence (for related stud- that underlie the friendship bond.
ies in the IDSS project, e.g., Hart, Hofmann, Edel- Development from childhood to adolescence. Evi-
stein, & Keller, 1997; Hofmann, 1991; Schellhas, dence from the IDSS project has shown that late
1993). Second, we explore the significance of adoles- childhood to mid-adolescence is a formative period
cent socio-moral sensitivity for friendship intimacy for the successive sophistication of sociomoral sensi-
in emerging adulthood, considering links between tivity in friendship (Keller, 1996). Interindividual
social-cognitive and socioemotional development differences in friendship-related sociomoral sensitiv-
and friendship experiences (e.g., Bukowski & Sip- ity are remarkably stable from ages 7 to 15 years
pola, 1996; Rubin, Wojslawowicz, Rose-Krasnor, (Keller, 1996; Keller, Malti, & Dravenau, 2007). This
Booth-LaForce, & Burgess, 2006). As a complemen- stability of interindividual competencies has also
tary framework, we consider how major life course been shown in other research (Colby, Kohlberg,
transitions shape young people’s development Gibbs, & Lieberman, 1983). Nevertheless, stability
(Buchmann & Steinhoff, 2017). Third, we explore dis- coefficients from previous investigations indicate
tal effects of early adult friendship intimacy on mid- that about 50%–60% of young people change rank
adult friendship satisfaction, considering the role of orders on sociomoral sensitivity across the years
intimacy for satisfaction in symmetrical relation- (see Keller et al., 2007). Here we argue that behav-
ships, including friendship (Finkenauer, Engels, iors in the broader peer context, and the peers’
Branje, & Meeus, 2004; Sanderson, Rahm, & Beigbe- responses, can play a vital role in these processes of
der, 2005), and age-graded social support networks. stability and change. We take insecurity expressed
in peer contexts as a first example and peer rejection
as a second example.
Sociomoral Sensitivity and Friendship
We assume that sociomoral sensitivity is a basis
The concept of sociomoral sensitivity (Keller, of how adolescents interact with friends and peers
1984, 1996; Keller & Reuss, 1984) integrates social- in general (see Keller, 1996). A child with a rela-
cognitive, affective, and moral processes in the indi- tively poor understanding of peers’ perspectives
vidual development of interpersonal understanding and the rules underlying their relationships with
and the interpretation of social and moral rules. one another may shy away from interactions with
Sociomoral sensitivity is part of the self-concept peers in order to avoid feelings of being or behav-
(Keller & Edelstein, 1991, 1993), including a sense ing in ways that are “inadequate.” Similar argu-
of agency and responsibility in close relationships. ments have been made with regard to
It is based on social-cognitive abilities such as per- developmental risks associated with social anxiety
spective taking, and its development proceeds and withdrawal from peers (Biggs, Vernberg, &
toward an increasingly sophisticated understanding Wu, 2012; Rubin, Burgess, Kennedy, & Stewart,
Life-Span Perspective on Friendship Development 3

2003; Rubin, LeMare, & Lollis, 1990). Lacking When assessing the longitudinal associations
friendship-related sociomoral competencies may between sociomoral sensitivity and friendship inti-
also increase risk for peer rejection; partly because, macy, it is important to consider that major life
on the dyadic level, poor sociomoral competencies course transitions after compulsory schooling often
could make the child a rather unattractive relation- demand the formation of new friendships. On the
ship partner for many peers, particularly for those one hand, transitions can offer opportunities to
with a more sophisticated sociomoral understand- renegotiate one’s position in the peer context (Shell,
ing. In turn, both insecurity and peer rejection are Gazelle, & Faldowski, 2014) and thus limit the
likely to entail poorer opportunities to form close potential impact of, for example, adolescent peer
dyadic bonds and further learn social competencies rejection on early adult friendship quality. On the
relevant in friendship. When insecure children are other hand, forming new peer relationships, includ-
generally quiet in peer interactions and do not put ing friendships, can be particularly challenging for
forward their sociomoral reasoning, they can also those who feel insecure (Oh et al., 2008). Existing
not receive an authentic peer response to their inter- evidence also suggests that particularly after a
personal understanding and their sociomoral devel- (school) transition, adolescents’ social and emo-
opment may slow down further. Indeed, negative tional dispositions are formative for their friend-
peer experiences are associated with young people’s ships (Goodwin, Mrug, Borch, & Cillessen, 2012),
social and emotional development (e.g., Ladd, 2006; for example, because they influence the selection of
Storch & Masia-Warner, 2004), and poorer interper- new friends with similar dispositions. We therefore
sonal understanding is linked with different aspects assume that the level of sociomoral sensitivity
underlying insecurity in peer contexts, such as feel- achieved by mid-adolescence, when compulsory
ings of inferiority, anxiety, and shyness with nega- schooling typically ends, antecedes intimacy in
tive affect (Banerjee & Henderson, 2001)—albeit friendships at the beginning of adulthood. We
there is considerable heterogeneity among insecure know of one other prospective longitudinal study
children in regards to their social skills (Gazelle, that has assessed the impact of earlier peer rejection
2008). Overall, these arguments suggest bidirec- and friendship status on early adult friendships
tional associations between sociomoral sensitivity in (Bagwell, Newcomb, & Bukowski, 1998). No signifi-
friendship and insecurity in peer contexts on the cant effects were found; but in contrast to our
one hand, and peer rejection on the other hand. approach, friendship status was conceptualized as
Associations with early adult friendship inti- having (or not having) a mutual friendship (rather
macy. When adolescents grow older, they increas- than focusing on friendship-related competencies),
ingly focus on mutual disclosure, trust, and with assessments made in preadolescence.
emotional comfort in friendship (e.g., Hartup & Ste-
vens, 1997), all of which we conceive as components
Friendship Intimacy in Early Adulthood and Mid-Adult
of friendship intimacy. Early adult women tend to
Friendship Satisfaction
perceive and expect more intimacy than men, but for
both genders intimacy seems to be an important pre- Until adulthood, friendships are linked with
requisite for friendship satisfaction (Felmlee, Sweet, identity formation, lifestyle, and individual func-
& Sinclair, 2012; Jones, 1991). The motivation and tioning within the broader social structures (Eve,
capacity to resolve conflicts and reach consensus 2002). Existing evidence also shows positive correla-
while taking into consideration the particularities of tions between well-integrated friendship networks
the friend’s perspective besides one’s own are mark- and adult psychological well-being (e.g., Cable
ers of a highly developed sociomoral sensitivity (Kel- et al., 2013; Gillespie, Lever, Frederick, & Royce,
ler & Edelstein, 1993) and, at the same time, are a 2015). However, the importance people ascribe to
likely prerequisite for establishing intimacy in friend- close friendships, and the expectations they have
ship (e.g., Buhrmester, 1990). Indeed, associations for friendship, may somewhat decrease and depend
among sociomoral competencies and friendship inti- on individual life circumstances, particularly mari-
macy persist until early adulthood: In a sample of 21- tal status (Carbery & Buhrmester, 1998; Pahl &
year-olds, perceived support and intimacy in friend- Pevalin, 2005; Perilloux & Buss, 2008; Tashiro &
ships were positively correlated with moral motiva- Frazier, 2003). While for the late adolescent and
tion—including moral decision making, emotion early adult years, friendship intimacy might be seen
attributions, and justifications (Malti & Buchmann, as a developmental milestone, the developmental
2010). However, these associations have not been achievements typically related to mid-adult friend-
investigated longitudinally. ship seem to be less generalizable. The concept of
4 Steinhoff and Keller

relationship satisfaction offers a framework to assess broad field of attachment research has shown that
the quality of relationships based on subjective eval- individual differences in relationship practices and
uations (Schumm et al., 1986, p. 385). These evalua- perceptions can be stable across long periods of the
tions of relationships will be based on different human life span (e.g., Waters, Merrick, Treboux,
criteria depending on the individual life circum- Crowell, & Albersheim, 2000). Attachment represen-
stances, needs, and interests that may be unknown tations are often generalized (Hazan & Shaver,
to the researcher. Nevertheless, some characteristics 1987) and attachment functions are typically trans-
that are likely to enhance the satisfactoriness of sym- ferred from parents to friends, and then romantic
metrical relationships can be identified. They include partners, when young people grow older (Fraley &
sensitive conflict management (see Bukowski & Sip- Davis, 1997). In order to explore the associations
pola, 1996; Cramer, 2000; Crockett & Randall, 2006; among friendship-related competencies and experi-
Hartup & Stevens, 1997), the experience of trust, and ences over long periods of time, we take into
opportunity for self-disclosure (e.g., Finkenauer account confounding effects of global relationship
et al., 2004; Jones, 1991; Meeks, Hendrick, & Hen- representations or relationship status. The role of
drick, 1998). In turn, these aspects are closely linked children’s attachment representation (secure or inse-
to the concept of intimacy (see Sanderson et al., cure attachment style, for details about the assess-
2005), although the framework of friendship satisfac- ments in IDSS, see Jacobsen, Edelstein, & Hofmann,
tion makes no assumptions about the exact degree 1994; Jacobsen & Hofmann, 1997) is taken into
of intimacy needed in order to make a friendship account, because research has shown correlations
satisfactory. Rather, we assume that people with a with children’s behaviors in peer contexts (e.g.,
higher potential to establish intimacy will evaluate Booth-LaForce & Oxford, 2008) and with socio-
their friendships more positively. Indeed, previous moral sensitivity in friendship (evidence from IDSS:
research indicates that friendship intimacy con- Keller et al., 2007). When predicting early adult
tributes to psychological need fulfillment and happi- friendship intimacy, we consider the role of parent-

ness (Demir & Ozdemir, 2010). Given that early child relationship quality perceived at the same age
adult friendship experiences offer a “training (see Steinhoff, 2015; for a more general perspective
ground” for general relationship conceptions and on family socialization in the IDSS project, see also
behaviors (e.g., Barry, Madsen, Nelson, Carroll, & Grundmann, Dravenau, Bittlingmayer, & Edelstein,
Badger, 2009; Erikson, 1959; Sullivan, 1953), these 2006). Finally, we will adjust effects on mid-adult
arguments suggest that people with intimate friend- friendship satisfaction for marital status (Pahl &
ships at the beginning of adulthood will be more Pevalin, 2005).
satisfied with their friendships later in life. We also consider potential emotional bias in rela-
To date, long-term investigations of these associa- tionship representations and friendship formation,
tions are largely missing. One exception is a 30-year particularly during transition periods when young
longitudinal study by Marion, Laursen, Zettergren, people need to establish new relationships (Good-
and Bergman (2013) who explored the effects of ado- win et al., 2012). Methodologically we therefore
lescent friendship (age 15) on mid-adulthood rela- adjust path coefficients to adult friendship charac-
tionship quality including, but not limited to, teristics for depressive symptoms.
friendship satisfaction. They found that friendship
moderates the link between adolescent peer rejection
Hypotheses
and mid-adult relationship quality, but there was no
direct effect of the previous friendship experience. Three hypotheses link child and adolescent socio-
However, similar to the Bagwell et al. (1998) study moral sensitivity and its associations with (a) inse-
cited earlier, friendship experiences were opera- curity in peer contexts and (b) peer rejection to
tionalized by asking whether the adolescents had at friendship experiences during adulthood. A fourth
least one friend, whereas the quality of friendship hypothesis refers to the impact of broader relation-
during early adulthood was not considered. ship representations and emotional bias.

Hypothesis 1: Previous analyses in the IDSS project (Kel-


The Role of Broader Relationship Representations,
ler, 1996; Keller & Wood, 1989; Keller
Relationship Status, and Emotional Bias
et al., 2007) revealed significant rank-order
While knowledge about the sequential pathways stability of sociomoral sensitivity in friend-
of friendship development from childhood to adult- ship across childhood and adolescence (ages
hood is largely missing from the literature, the 7–15 years). We hypothesize that there will
Life-Span Perspective on Friendship Development 5

also be bidirectional effects between socio- Figure 1 provides a graphical summary of the
moral sensitivity and insecurity as well as expected developmental pathways (i.e., stability of
peer rejection: the higher the level of socio- sociomoral sensitivity, insecurity in peer contexts and
moral sensitivity is, the lower is the subse- peer rejection, and cross-lagged paths among the
quent level of insecurity/peer rejection and concepts; and then predictive effects on subsequent
vice versa. adult friendship experiences). The figure puts these
developmental pathways in a context of life phases
Hypothesis 2: We expect that individual differences in and associated social transitions, which may offer
mid-adolescent sociomoral sensitivity opportunities for the renewal of peer relationships.
(age 15 years) are antecedents of differ-
ences in friendship intimacy experienced
at the beginning of adulthood (around Method
age 22 years): the higher the respondents’
The IDSS Project
level of sociomoral sensitivity is in mid-
adolescence, the more friendship intimacy Data are taken from the Iceland longitudinal
they experience in early adulthood. study “IDSS” (for detailed information on the
study, see especially Edelstein et al., 1990). The goal
Hypothesis 3: We hypothesize that the experience of of the study was to follow-up on cognitive, social-
friendship intimacy in early adulthood is cognitive, and sociomoral competencies; emotional
a precursor of relationship satisfaction development; and personality from childhood to
later in life: the higher the level of inti- adulthood. One particular focus is on the role of
macy is in friendship at the beginning of social structures and socialization processes for
adulthood (around age 22 years), the young people’s development across the life span
higher is the likelihood of being highly (Grundmann, 1998; Grundmann, Steinhoff, & Edel-
satisfied with friendship in one’s late stein, 2011). Research based on the data from the
thirties (around age 37 years). IDSS project has been widely published and is ref-
erenced only selectively in this article (see also
Hypothesis 4: We assume that the predictive effects Steinhoff, 2015).
hypothesized in H1–H3 hold when indica-
tors of broader relationship representa-
Sample
tions/emotional bias are controlled for.
Additional assumptions include: higher Data were collected from respondents aged 7 to
levels of perceived closeness in the par- around 37 years between 1976 and 2007. The full
ent–child relationship are associated with sample includes 193 respondents (45% female)
higher intimacy in friendships at the growing up in an urban (n = 121) or rural area of
onset of adulthood (22 years); depressive Iceland. There were three phases of data collection:
symptoms impair perceived intimacy with (a) data were collected at the ages of 7, 9, 12, and
friends. 15 years to represent the period from childhood to

Individual Development and Social Transitions

Childhood Adolescence Emerging Adulthood Adulthood


Age 7 – 9 Age 12 – 15 Age 22 Age 37

Socio-Moral Sensitivity in Friendship

Friendship Friendship
Intimacy Satisfaction

Peer–Related Behaviors and Experiences

Figure 1. Conceptual model: Associations among sociomoral sensitivity in friendship and peer-related behaviors and experiences, and
their joint impact on friendship quality across the individual life span.
Note. Ages refer to the assessment points in our study.
6 Steinhoff and Keller

mid-adolescence; (b) the respondents provided data (i.e., based on their occupation). The individuals
once at the beginning of adulthood at an age were assigned to one of three groups (low SES: e.g.,
between 21 and 23 years; (c) respondents were sur- unskilled workers; medium SES: e.g., technicians;
veyed once again at an age between 36 and 38. high SES: e.g., academic professionals, managers
Sample attrition mainly occurred after adolescence, typically with university degree). Two raters agreed
which is a common observation in longitudinal on these assessments. Twenty-eight percent of the
studies (see Denissen, Asendorpf, & van Aken, respondents had a low SES, 48% had a medium
2008): for 87% of the original sample data are avail- SES, and 24% had a high SES. At that time, 46% of
able across the first seven years until age 15. The the respondents were married and 33% were in a
sample size then dropped to 112 participants at the steady partnership.
beginning of adulthood and increased again to 125
in mid-adulthood. For 41% of respondents, com-
plete data are available on all of the study variables Variables
at all of the study waves. Data were collected using
Main Study Variables
questionnaires, tests, and personal interviews.
At the onset of the study, the urban sample was Sociomoral sensitivity in friendship was assessed by
stratified by parental socioeconomic status (SES; six means of semistructured interviews at the ages of 7,
groups), children’s general ability level rated by a 9, 12, and 15 years (the concept was not included
teacher (high/low), and gender. Groups of approxi- in the survey waves during adulthood). A scale is
mately equal size were created that exhibit all pos- used that was developed by Keller and colleagues
sible combinations of parameter values on the (Keller, 1996; Keller & Edelstein, 1991; Keller &
category variables gender (male/female), SES (six Reuss, 1984). It describes the level of interpersonal
groups from high to low), and general ability understanding articulated when discussing a hypo-
(high/low). This sampling design was used in thetical dilemma based on the cognitive–structural
order to disentangle the influences of sociocultural approach of friendship development and the
environments and individual abilities on children’s dilemma proposed by Selman (1980). The story is
developmental trajectories. For more detailed infor- about an everyday action dilemma occurring
mation on the study design and the urban subsam- between friends. It addresses conflicting expecta-
ple in particular, see Edelstein et al. (1990) and tions and interests of a protagonist, a good friend,
Grundmann et al. (2006). In our analyses, we take and a child who is new in class, as well as general
the sampling design into consideration by control- friendship obligations (e.g., keeping a promise). The
ling for gender, the SES of origin and a measure of story can be found in the Appendix (A1).
the 7-year-olds’ cognitive ability that was assessed The protagonist’s gender was matched with that
in both subsamples (i.e., the rural and the urban of the respondent, and the story’s content was
sample) and was positively correlated with the tea- slightly modified over time so as to address age-
cher ratings of ability used in the urban-sample typical interests. Five components of sociomoral
stratification procedure (see Variables section for sensitivity were derived from the discussions (Kel-
more details). ler, 1996; Keller & Reuss, 1984): (a) understanding
Altogether, the sample of the IDSS study is of the situation, (b) reasoning about the action alter-
diverse with regard to childhood and adulthood natives from which the protagonist can choose, (c)
SES. Using a 6-point scale assessing fathers’ occupa- anticipation of the consequences these different
tion (see Bj€ ornsson, Edelstein, & Kreppner, 1977), actions have for the persons involved in the story,
25% of the children had a low SES (Category 1; (d) moral evaluation of the action alternatives, (e)
e.g., unskilled workers), 27%, 11%, 15%, and 12% strategies for the resolution of conflicts that may
were in Categories 2–5, respectively, and 10% had arise in the situation described. In addition to these
high SES (Category 6; e.g., academic professionals). aspects, the assessments were based on the respon-
With regard to parental educational background, dents’ general conceptualization of “close friend-
18% of fathers had a postsecondary or academic ship” (Selman, 1980) and their understanding of the
educational degree, 25% had an upper secondary meaning of “promises” (Colby et al., 1983). Topics
degree, and 57% had lower secondary education or were scored for levels of development. A more
less. At the last survey wave (age 37), an attempt detailed description of the assessments and the rat-
was made to assess the respondents’ achieved SES ing procedures can be found in Keller (1996; see
in a way comparable to the first wave’s assessment also Keller & Wood, 1989), where also good
Life-Span Perspective on Friendship Development 7

internal consistency among the different compo- satisfied” (63%) from those who reported some
nents is reported (a ≥ .82 at all time-points). In this lower level of satisfaction (37%). In preparation of
study, we use the individual mean scores that were our analyses, we attempted to evaluate the reliabil-
computed across the topics and domains. On a ity and validity of this measure by exploring its
scale from 0 to 1, the sample mean of sociomoral associations with related concepts, including satis-
sensitivity in friendship is M (SD) = 0.29 (0.08), 0.41 faction with other types of relationships and global
(0.07), 0.51 (0.05), and 0.61 (0.06) at ages 7, 9, 12, life satisfaction (Steinhoff, 2015). Confirmatory fac-
and 15 respectively. tor analyses revealed that friendship satisfaction
For insecurity in peer contexts and peer rejection, loaded significantly on a latent factor shared with
we use teacher reports which offer an observational partnership satisfaction and satisfaction with the
assessment of the adolescents’ regular interactions relationships with one’s own children (both
with familiar peers in the classroom context at the assessed at age 37 using a single item, as with
ages of 9, 12, and 15 years. The teachers rated the friendship satisfaction). This shared factor (relation-
frequency of their students’ behaviors on a 5-point ship satisfaction) in turn predicted global life satis-
scale from 1 = almost never to 5 = almost always. For faction at b = .39 (p < .01). However, friendship
insecurity, four items from Quay and Peterson’s satisfaction alone was uncorrelated with global life
(1975) “Behavior Problem Checklist” are used (i.e., satisfaction. We therefore deem the variable as an
“[student] is shy“, “introvert”, “feels inferior”, “in- adequate indicator of relationship satisfaction in a
secure”), with a = .86 at age 9, a = .84 at age 12, particular social domain (i.e., friendship).
and a = .81 at age 15. For peer rejection, four items
from Kohn’s (1977) “Competence Scale” are used
Affective Relationship Representations
(e.g., “Others are unwilling to work with [stu-
dent]”), with Cronbach’s alpha at a = .77 at age 9, The 7-year-olds’ attachment representation was
a = .77 at age 12, and a = .70 at age 15. We com- assessed based on their discussion of a picture-story
puted mean scales with higher values indicating addressing the separation of parent and child
more insecurity/peer rejection. M (SD) = 1.87 (Chandler, 1973). Validation of using the measure
(0.93), 2.07 (0.88), and 2.29 (0.91) for insecurity in to assess children’s attachment representation is
peer contexts at ages 9, 12, and 15; M (SD) = 2.36 provided by Jacobsen et al. (1994; Jacobsen & Hof-
(0.86), 2.48 (0.86), and 2.52 (0.80) for peer rejection mann, 1997; see also Ziegenhain, Rottmann, &
at ages 9, 12, and 15. Rauh, 1988). We recognize, however, that the mea-
For assessing friendship intimacy at an age around sure also relies on the child’s social-cognitive under-
22 years, we use a six-item scale (e.g., “I can fully standing of relationships (Keller et al., 2007, pp.
rely on my friends”; “My friends console me when 132, 137). We use a binary variable to distinguish
I’m in trouble”). Cronbach’s alpha is adequate children who were securely attached from those
(a = .84). Based on exploratory as well as confirma- with insecure attachment, including ambivalent,
tory factor analyses (comparative fit index anxious-avoidant, and disorganized attachment rep-
[CFI] = .97, root mean square error of approxima- resentations. Thirty-nine percent of the respondents
tion [RMSEA] = .06, standardized root mean square were securely attached. This proportion is within
residual [SRMR] = .04), the scale was derived from the range typically reported in the literature on
a larger set of items addressing experiences in peer mid-childhood attachment styles (Del Giudice,
and friendship contexts (Steinhoff, 2015). The 2008; Solomon, George, & De Jong, 1995).
respondents rated their agreement with the six Closeness in the parent–child relationship at around
statements on a 5-point scale from total disagreement age 22 was measured using a six-item scale. The
to total agreement. The average score is M = 4.08 respondents were asked to think back to their ado-
(SD = 0.33). lescent years and assess their level of agreement
Satisfaction in friendship was evaluated at an age with statements regarding parental support (e.g.,
around 37 years. As with the assessment of friend- “My parents were interested in my concerns”) on a
ship intimacy, this variable refers to generic friend- Likert-type 5-point scale and about their parents’
ships (“How satisfied are you with your techniques when dealing with a parent–child con-
relationships with your friends?”) with ratings on a flict (e.g., “My parents tried to explain why their
5-point scale from completely unsatisfied to very satis- opinion was different from mine”), with assess-
fied. We dichotomized the assessment due to its ments made on a 4-point scale (never to often). Con-
highly skewed value distribution, thereby distin- firmatory factor analyses showed that the six items
guishing respondents who reported to be “very are indicators of one latent factor (model fit:
8 Steinhoff and Keller

v2 = 11.968 [p = .215], CFI = .993, RMSEA = .054; wave, at age 37. Respondents who still participated
see Steinhoff, 2015). In order to standardize the had significantly higher levels of cognitive ability
responses given on the different scales, we used z- (age 7) than drop-outs: t(182) = 2.05, p = .042,
transformations. Cronbach’s alpha for the six items d = .32, which is a medium effect size, following
is a = .82, and we computed a mean scale where Cohen’s rules of thumb (1988, pp. 24–27). Respon-
higher values indicate closer parent–child relation- dents with a secure attachment style in Wave 1
ships (M = 2.00, SD = 0.72). were more likely to keep participating in the study
Marital status at around age 37 is a binary vari- than insecurely attached respondents (v2(1,
able that takes the value 1 when respondents were n = 181) = 7.81, p = .005, OR = 2.57). Study partici-
currently in a steady partnership, including married pation in the last survey wave was not related to
and unmarried couples (78%), and 0 when they SES of origin, gender, or childhood sociomoral sen-
were not (22%). sitivity.
Depressive symptoms at around age 22 were Assuming that data are “missing at random,” as
assessed using the widely used “Center for Epi- opposed to “missing completely at random” or
demiologic Studies Depression Scale.” This is an 18- “not at random” (see Allison, 2002), we handled
item self-report measure used to assess current missingness using the Full Information Maximum
depressive symptomatology in nonpathological Likelihood (FIML) estimation when testing our
samples. It was proven reasonably valid and inter- models. In preparation of bivariate correlation anal-
nally consistent (Radloff, 1977). On a scale from 0 yses, we carried out multiple imputation. All vari-
to 1, the mean score was M = 0.31 (SD = 0.19). ables included in this study were used to formulate
the imputation model, and 30 data sets were cre-
ated. Multiple imputation and FIML are highly rec-
Additional Project Variables
ommended features to handle missing data in
Gender (0 = male, 1 = female), SES of origin and developmental research in order to reduce potential
cognitive ability will be included as control vari- bias that may arise due to attrition mechanisms (see
ables. The SES of origin was measured as a multidi- Enders, 2013; Schafer & Graham, 2002). Since FIML
mensional configuration of the parents’ (mostly is applied to endogenous variables only, a small
fathers’) occupation reported in Wave 1. Relative number of random missing values on exogenous
income, educational prerequisites, and the degree of variables lead to a sample size of N = 176 in our
responsibility and authority over others were taken final models.
into account (Bj€ ornsson et al., 1977). We use a
dichotomous variable dividing the sample into two
Statistical Procedure
groups of respondents with lower SES (49%) and
higher SES (51%) of origin. In order to examine the developmental pathway
The 7-year-olds’ general cognitive ability was from child and adolescent sociomoral sensitivity
assessed using Raven’s (1960) Progressive Matrices. and peer experiences to early and mid-adult friend-
The variable informs about the capability to derive ship quality, we specified autoregressive cross-
meaning from complex figures and to reproduce lagged path models within a structural equa-
information. We use the Raven score instead of the tion modeling framework. Two versions of the
teacher ratings which were used in the original same model were specified, including insecurity in
IDSS urban sample stratification procedure because peer contexts (model A) and peer rejection (model
the teacher ratings are not available for the rural B) as an outcome and, in turn, antecedent of socio-
sample. However, the two measures were closely moral sensitivity during child and adolescent devel-
linked in the urban sample (r = .57, p = .000). On a opment.
scale from 0 to 1, the sample mean of cognitive Stability of individual differences in sociomoral
ability at age 7 is M = 0.59 (SD = 0.33). sensitivity was estimated by specifying autoregres-
sive paths between adjacent time points from
school entry to the end of compulsory schooling
Attrition Mechanisms and Handling of Missing Data
(i.e., from ages 7 to 9, 9 to 12, and 12 to 15).
We analyzed missing data patterns by compar- Although the longitudinal stability of insecurity in
ing the groups of respondents and nonrespondents peer contexts and peer rejection was not at the core
in the last survey wave with regard to the Wave 1 of our theoretical argumentation, we thought it was
variables used in our investigation. Sixty-three per- also mandatory to specify autoregressive paths for
cent of the initial sample still participated in the last these variables, given the longitudinal rank-order
Life-Span Perspective on Friendship Development 9

stability reported in the literature for these variables

.35***(**)

deviated from the MI-based analysis, the level of the p-values from pairwise present data analyses is given in parentheses (note that in all of these cases the correlation coefficients
Note. Correlations are based on data with prior multiple imputation and the imputation model included all of the variables used in our study. The Maximum Likelihood Robust
estimator was used. Comparison with coefficients based on pairwise present data did not alter the interpretation of the results; where p-values from pairwise present data analyses
11.
(e.g., Ladd, 2006). Friendship intimacy at around
age 22 was specified as an outcome of sociomoral
sensitivity at age 15, and this effect was adjusted
for the potential impact of insecurity/peer rejection

.04
.11
10.
at the same age. The sequence of pathways is
completed by the estimation of an effect of early

.44***
adult friendship intimacy on friendship satisfaction

.08
.02
9.
in mid-adulthood at around age 37 (see also
Figure 1).
The residuals of endogenous variables that were

.60***
.38***
8.
concurrently assessed were allowed to covary. All

.01
.06
possible paths from control and context variables to
variables that were assessed at the same time point,

.45***
.46***
.61***

from the MI-based analyses were within the respective 90% confidence intervals of those from an analysis with pairwise present data).
or at a later time point, were estimated in a first

.13
.05
7.
step. These paths were then omitted from the mod-
els if they were not significant (p > .10) in order to

.44***
.38***
.53***
.35***
avoid overcontrol (Little, 2013). Control variables

.08
.04
6.
assessed in adulthood were included as endoge-
nous variables, with paths from variables assessed
at adjacent prior time points estimated freely if sig-

.56***
.49***
.44***
.39***
.39***
nificant (see also Steinhoff, 2015).

.01
.15
5.
We used the Maximum Likelihood Robust
(MLR) estimator implemented in MPlus V7
(Muthen & Muthen, 2012), taking into account that

.28***(**)

.26***(*)

.22*(†)
.36***

.35***

.28***
some variables in the models were skewed. Model

.21**
4.

.06
fit was evaluated based on chi-square (adequate
model fit is indicated when this test has a non-
significant result), RMSEA, CFI, and SRMR. Com-
.30***(**)
mon cut-off criteria proposed by Hu and Bentler
.45***
.30***
.27***

.39***
.35***
.39***
3.

(1999) were used as a guideline to assess the ade-

.15†
.11†
quacy of the models (RMSEA ≤ .06, CFI ≥ .95, and

Number of cases with valid data. †p < .10. *p < .05. **p < .01. ***p < .001.
SRMR ≤ .08). Because these fit indices are not avail-
able with categorical outcomes and using MLR, a
.28***(**)

.30***(**)

.23*(**)
linear model was estimated in a first step (i.e., treat- .16†(*)
.60***
.42***

.28***

.38***
.27***
.42***
2.

ing the mid-adulthood outcome as if it was contin-


uous) in order to explore model fit. In a
supplementary analysis, the mid-adulthood out-
.29***(**)

come was treated as categorical variable in order to


Bivariate Correlations Among the Main Study Variables

.55***
.41***
.52***

.44***
.36***
.36***
.50***
.34***
1.

.24*

check the robustness of the results and logistic


.08

regression odds ratios (OR) and confidence inter-


vals (CI) are reported.
169
162
182
176

169
181
165

112
125
169
181
165
Na

Results
6. Insecurity in peer context 12
7. Insecurity in peer context 15
5. Insecurity in peer context 9

12. Friendship satisfaction 37

Bivariate Correlations
3. Sociomoral sensitivity 12
4. Sociomoral sensitivity 15

11. Friendship intimacy 22


1. Sociomoral sensitivity 7
2. Sociomoral sensitivity 9

Bivariate correlations among the main study


10. Peer rejection 15

variables and autocorrelations for sociomoral sensi-


9. Peer rejection 12
8. Peer rejection 9

tivity in friendship, insecurity in peer contexts, and


peer rejection across the years are provided in
Table 1. Several findings are noteworthy. The auto-
Table 1

correlations of sociomoral sensitivity, insecurity,


and peer rejection suggest that, in all three
a
10 Steinhoff and Keller

domains, stability is higher between the ages of 9 sociomoral sensitivity and also in insecurity and
and 12 than between 12 and 15. During the mid- peer rejection. The results of both models show that
adolescent years, there is apparently more room for sociomoral sensitivity in friendship influences such
change (in terms of interindividual differences) in changes. Higher levels of sociomoral sensitivity
peer experiences and related learning. There are apparently encourage children and adolescents to
also negative correlations between sociomoral sensi- approach others and interact with them openly
tivity and insecurity/peer rejection at all time (model A). Higher levels of sociomoral sensitivity
points. As expected, sociomoral sensitivity at age 15 also protect children and adolescents from being
is positively correlated with early adult friendship rejected by their classmates (model B). Interestingly,
intimacy at age 22. Interestingly enough, even while peer rejection does not have a feedback effect
sociomoral sensitivity at the ages of 7 and 9 years on sociomoral development, the adolescents’ inse-
is positively correlated with early adult friendship curity in the peer context at age 12 apparently
intimacy and the sizes of these correlations are sim- slows down further growth of sociomoral sensitiv-
ilar to that for sociomoral sensitivity at age 15. ity until age 15. The findings therefore support our
There are no significant correlations among early hypothesis (H1) that sociomoral sensitivity in
adult friendship intimacy and prior insecurity/peer friendship influences children’s and adolescents’
rejection, although there is, by trend, a negative interactions with peers, which, in turn, foster socio-
relation with insecurity in peer contexts. Finally, moral development. However, unlike insecurity,
friendship intimacy at an age of around 22 years is peer rejection is not a unique risk factor for a
positively correlated with friendship satisfaction slower development of sociomoral sensitivity.
assessed 15 years later. In line with our expectations (H2), higher levels
of mid-adolescent sociomoral sensitivity in friend-
ship are followed by more intimacy experienced
Path Models
with friends in early adulthood (age 22), after the
Figure 2 shows standardized path coefficients young people have graduated from compulsory
from the cross-lagged path models. The illustration school. It is important to note again that these asso-
is limited to the pathways of main interest. Model ciations hold while global relationship representa-
A addressed the interplay of sociomoral sensitivity tions are controlled for.
and insecurity in peer contexts across childhood The results also support our hypothesis (H3) that
and adolescence, whereas model B focused on early adult friendship experiences are an important
sociomoral sensitivity and peer rejection. Both mod- antecedent of the quality of friendships in mid-
els show adequate fit to the data. All stability coef- adulthood. The higher the level of friendship inti-
ficients between adjacent time points reach high macy reported at around age 22, the higher is the
levels of significance, but, across time, more than likelihood of being very satisfied with one’s friend-
45% of the respondents change their rank orders in ships at an age around 37.

Socio-Moral Socio-Moral Socio-Moral Socio-Moral


.48*** .47*** .37***
Sensitivity Sensitivity Sensitivity Sensitivity
(.48***) (.47***) (.43***)
Age 7 Age 9 Age 12 Age 15 .17*
(.20*)

-.27***
Friendship Friendship
.37**
-.38*** -.13* Intimacy Satisfaction
(.37**)
(-.37***) Age 22 Age 37
-.14†
(-.26***)

Insecurity in Peer Insecurity in Peer Insecurity in Peer


.55*** .35***
Context Context Context
(.49***) (.34***)
(Peer Rejection) (Peer Rejection) (Peer Rejection)
Age 9 Age 12 Age 15

Figure 2. Standardized estimates for the joint pathways from child and adolescent sociomoral sensitivity in friendship and insecurity in
the peer context (model A)/peer rejection (model B) to adult friendship quality. Results from model B are in parentheses.
Note. The illustration is limited to significant pathways of main interest; consult main text for a detailed description of model specifica-
tions. Model fit A: v = 77.03, df = 68, p = .212, root mean square error of approximation (RMSEA) = .03, comparative fit index
(CFI) = .98, standardized root mean square residual (SRMR) = .07. Model fit B: v2 = 79.32, df = 66, p = .126, RMSEA = .03, CFI = .97,
SRMR = .07.

p < .10. *p < .05. **p < .01. ***p < .001.
Life-Span Perspective on Friendship Development 11

Significant coefficients for control and context omitted that path and entered one from age 9
variables (H4) can be found in the Appendix (A2). instead; and, in a third model, a path from age 12
Most importantly, these coefficients reveal that was included). The level of sociomoral sensitivity
secure attachment representations not only foster achieved by the age of 9 was indeed the best
sociomoral development (Keller et al., 2007), but unique predictor of early adult friendship intimacy
also protect children, to some degree, from peer (b = .18 [.22], p = .040 [.014]; results for model B are
rejection. Securely attached children also tend to be in parentheses). These exploratory findings strongly
less insecure in the peer context. Similarly, cognitive encourage further research on the significance of
ability as assessed at the onset of schooling early developmental achievements for adult life
decreases peer rejection and adolescent insecurity in outcomes.
peer contexts, and supports sociomoral growth. Finally, in addition to models A and B, we tested
This is in line with previous research showing, for a full model including cross-lagged paths among
example, that higher cognitive ability is related to sociomoral sensitivity in friendship, insecurity in
better social comprehension (Pellegrini, Masten, peer contexts, and peer rejection, in order to deter-
Garmezy, & Ferrarese, 1987). As expected, per- mine whether there were unique associations
ceived intimacy with friends at the onset of adult- among each pair of constructs. Results are pre-
hood is decreased by depressive symptoms and sented in the Appendix (A3). All significant cross-
increased by closeness with the parents. It is also lagged paths from models A and B remained signif-
noteworthy that females report higher levels of inti- icant, with one exception: the path from sociomoral
macy with friends at age 22, but there were no gen- sensitivity at age 9 to insecurity in peer contexts at
der differences in friendship satisfaction at age 37. age 12. This latter finding suggests that sociomoral
Marital status was uncorrelated with friendship sat- sensitivity and peer rejection in late childhood share
isfaction. some of the variance that predicts insecurity in peer
contexts at age 12. With regard to the control vari-
ables, the results from models A and B were also
Robustness Checks and Follow-Up Analyses
replicated, with the exception of the paths from
We ran a set of follow-up analyses to check the cognitive ability to insecurity in peer contexts at
robustness of our results and to further explore age 15 and to peer rejection at ages 9 and 12 (ns) as
some surprising findings. First, in order to calculate well as the path from secure attachment to peer
absolute fit indices, the results presented in Figure 2 rejection at age 12 (ns).
were based on MLR estimations and the final out-
come (friendship satisfaction) was treated as if it
was continuous rather than categorical. We ran a
Discussion
second version of the model, with the outcome trea-
ted as categorical, in order to check the robustness Our study provides novel insights into the develop-
of the results. The positive effect of early adult ment of child and adolescent sociomoral sensitivity
friendship intimacy on friendship satisfaction in in friendship and its distal impact on friendship
mid-adulthood was replicated (OR = 2.72 [2.70], experiences in adulthood. Taking a life-span perspec-
95% CI = 1.25–5.93 [1.25–5.86]; results for model B tive, and using 30-year longitudinal data, the study
are in parentheses). therefore advances knowledge about the role of early
A second concern was that the sizes of the sociomoral development for adult life outcomes. Our
bivariate correlations between sociomoral sensitivity analyses revealed a sequence of developmental path-
at the ages of 7 and 9 and early adult friendship ways involving friendship experiences and compe-
intimacy were as high as the size of the correlation tencies over three decades of life.
among sociomoral sensitivity at age 15 and friend-
ship intimacy. In order to identify the age at which
Development of Sociomoral Sensitivity in Friendship
the achieved level of sociomoral sensitivity best pre-
From Childhood to Adolescence
dicts early adult friendship intimacy, we ran an
exploratory set of three models where sociomoral Our first aim was to explore the reciprocal influ-
sensitivity at age 15 and at another one of the pre- ences among sociomoral development and (a) inse-
vious time points were included as predictors (i.e., curity in peer contexts and (b) peer rejection (H1).
in the first model, we included an additional path- Going beyond previous research, we were able to
way from sociomoral sensitivity at age 7 to early show that individual differences in sociomoral sen-
adult friendship intimacy; in the second model, we sitivity are not only remarkably stable across time
12 Steinhoff and Keller

(Keller, 1996; Keller et al., 2007), but also driven by itself, partly a behavioral manifestation of previous
feedback effects of young people’s interactions with sociomoral development—our findings show persis-
their peers. Higher early levels of sociomoral sensi- tent and cumulative developmental disadvantages
tivity in friendship can apparently (a) decrease chil- for children and adolescents whose sociomoral sen-
dren and adolescents’ insecurity when engaging in sitivity in friendship grows relatively slowly. To
peer relationships, perhaps by increasing their self- our surprise, peer rejection does not play a forma-
confidence in peer interactions. Sociomoral compe- tive role in this process, given that the level of peer
tencies may also equip children with better social rejection did not predict the level of sociomoral sen-
comprehension in peer contexts, and therefore sitivity once the autoregressive effects of sociomoral
advance their social interaction skills. These could, sensitivity were controlled for. Apparently, aspects
in turn, decrease insecurity (Banerjee & Henderson, of peer rejection that are independent from one’s
2001). A similar interpretation may apply to our previous sociomoral sensitivity (and can therefore
finding that higher sociomoral competencies (b) not be interpreted as the peers’ “response”) do not
decrease risk for peer-rejection. Better understand- necessarily impair sociomoral learning. However, it
ing and more sensitive handling of complex social is important to keep in mind that our measures of
situations (as was assessed in the friendship peer rejection were based on teachers’ observations
dilemma) may lead to more peer acceptance, and in in the classroom context. Adolescents who experi-
turn, less rejection. It is an intriguing finding of our ence rejection by their classmates could still have
study that less insecure behavior then helps the more positive peer relationships in other social con-
young people to further sophisticate their socio- texts (Kiesner, Poulin, & Nicotra, 2003). Further
moral sensitivity. Perhaps, open and authentic research is therefore needed in order to illuminate
engagement in social interactions with peers the associations among sociomoral development
enabled the adolescents to learn from their peers’ and peer rejection in different social contexts. It is
responses to their sociomoral understanding. The also noteworthy that in our data peer rejection was
finding is in line with other researchers’ findings significantly and reciprocally associated with inse-
that children with relatively poor social skills asso- curity in peer contexts. Future research with larger
ciated with sociomoral competencies are less likely sample sizes may use more complex models, and
than their age mates to get involved in peer rela- model comparisons, in order to disentangle the
tionships which foster their social skills develop- associations among the triad of sociomoral growth,
ment (Rubin et al., 2003, 2006). insecurity in peer contexts, and peer rejection in
Another interesting finding of our study is that more detail. In particular, person-centered analytic
sociomoral sensitivity in friendship constantly approaches could complement our variable-cen-
decreases insecurity in peer contexts across the ado- tered approach to help understand how sociomoral
lescent years (in the full model: from ages 7 to 9, sensitivity interacts with insecurity and peer experi-
and then from ages 12 to 15), but insecurity does ences in different groups of individuals (Gazelle,
not have a feedback effect before mid-adolescence. 2008), and how these groups then differ in their
An explanation could be that, when adolescents get friendship experiences across the life span.
older, the influence of their parents and teachers on
the quality of their peer contacts decreases and it
Adolescent Sociomoral Development Antecedes Early
becomes more important that the adolescents
Adult Friendship Intimacy
actively approach their peers in order to bond with
and learn from them. This would mean that insecu- One of the main contributions of our study is to
rity becomes increasingly formative for the quality link adolescent development of sociomoral sensitiv-
of one’s peer relationships, and the association with ity in friendship to adult friendship quality. The
sociomoral learning therefore strengthens across the findings confirm our hypothesis (H2) that the level
years. In order to support young people’s socio- of sociomoral sensitivity attained by the end of
moral growth during the mid-adolescent years, it compulsory schooling predicts intimacy in friend-
may therefore be promising to plan interventions ship at the onset of adulthood. While previous
that foster social skills development during child- research showed relations between moral develop-
hood or increase insecure children’s opportunities ment and intimate friendship bonds based on data
for easy social interactions with familiar peers. from children and adolescents only, and/or based
Given the significant rank-order stability of on cross-sectional data (Buhrmester, 1990; Malti &
sociomoral sensitivity across time, and the addi- Buchmann, 2010), our longitudinal study design
tional impact of insecurity—which seems to be, in reveals distal effects of 15-year-olds’, and even
Life-Span Perspective on Friendship Development 13

stronger effects of 9-year-olds’, sociomoral sensitiv- of particular friends studied by Pahl & Pevalin,
ity on early adult friendship experiences. Interven- 2005), there was no correlation with adult friend-
tions that support child and adolescent sociomoral ship satisfaction in our data.
development may therefore eventually help the The distal effect of early adult friendship inti-
young people form supportive intimate bonds with macy on mid-adult friendship satisfaction may, at
friends during the potentially stressful transition least partly, be driven by the close associations
period to adulthood. between intimacy and relationship satisfaction (e.g.,
It is noteworthy that neither mid-adolescent inse- Finkenauer et al., 2004; Jones, 1991; Meeks et al.,
curity in peer contexts nor the experience of peer 1998). However, the individual criteria that
rejection added to the predictive effect of socio- informed the respondents’ subjective evaluations of
moral sensitivity on early adult friendship intimacy. their friendships at age 37 remain unknown. There-
Consistent with prior research (Shell et al., 2014), it fore, further investigations are needed in order to
seems that life course transitions after compulsory better understand why people who experienced
schooling can help adolescents overcome some of higher levels of friendship intimacy at the onset of
their previous disadvantages in the realm of peer- adulthood are more likely to be satisfied with their
related behaviors and experiences and the associ- friendships later in life.
ated risks with regard to friendship formation.
However, this does not mean that peer-related
Limitations and Future Perspectives
behaviors and experiences during the adolescent
years are meaningless with regard to early adult Some limitations of our study deserve attention.
friendship quality. At least insecurity in peer con- One is a lack of information on the quality of close
texts may play an indirect role, as it affects adoles- friendships during childhood and adolescence.
cent sociomoral sensitivity, which, in turn, fosters Autoregressive effects of sociomoral sensitivity in
friendship intimacy. friendship may indicate feedback loops among
sociomoral development and friendship experi-
ences; but further research is needed in order to
Early Adult Friendship Intimacy Antecedes Mid-Adult
disentangle these interrelations and their links with
Friendship Satisfaction
the more general peer experiences we addressed
As hypothesized (H3), higher levels of early here (see Bagwell et al., 1998; Sette, Baumgartner,
adult friendship intimacy increase the likelihood of Laghi, & Coplan, 2016).
being highly satisfied with one’s friendships about The assessments of insecurity in peer contexts
15 years later. Here our study adds to the scarce and peer rejection were based on teacher ratings.
longitudinal evidence of distal effects of adolescent Such observational measures provide valuable
friendship experiences on adult relationship satis- insights into the behavioral development of stu-
faction (Marion et al., 2013). While the previous dents. Nonetheless, it has been an important obser-
study by Marion and colleagues revealed that hav- vation in previous research that data provided by
ing, versus not having, a friend in adolescence does different informants (e.g., parents or students’ self-
not directly affect mid-adult relationship satisfac- reports) could provide different pieces of informa-
tion, our investigation suggests that early adult tion about an individual’s development and behav-
friendship experiences can set the stage for need ior in the various social contexts of growing up (de

fulfillment in friendship (Demir & Ozdemir, 2010) Los Reyes, 2011; Zuffian o, Sette, Colasante, Buch-
until mid-adulthood. As with other aspects of close mann, & Malti, 2018). We therefore see our find-
social bonds, such as attachment representations ings, being based on teacher observations in the
(Waters et al., 2000), these findings indicate that classroom context, as one important piece of a lar-
friendship patterns can be stable over long periods ger puzzle.
of the life span. Yet, the sequential pathways Another caveat is that the assessment of mid-
described here, and the medium effect sizes, also adult friendship satisfaction was based on a dichot-
indicate that friendship experiences change consid- omized single item. In research on domain-specific
erably between childhood and mid-adulthood, and global life satisfaction, the use of single item
which may, at least partly, result from the various measures is well established (see Hsieh, 2003), but
social transitions pending during this period the skew of such measures, especially in the realm
(Arnett, 2004). However, while other studies have of relationship satisfactions, calls for caution when
shown correlations among marital status and some interpreting the respective results. Further research
characteristics of adult friendships (e.g., the choice is needed in order to zoom in on the various facets
14 Steinhoff and Keller

of friendship quality that people deem relevant for Bagwell, C. L., Bender, S. E., Andreassi, C. L., Kinoshita,
themselves. T. L., Montarello, S. A., & Muller, J. G. (2005). Friend-
It is also important to keep in mind that our ship quality and perceived relationship changes predict
assessments of adult friendship quality were both psychosocial adjustment in early adulthood. Journal of
Social and Personal Relationships, 22, 235–254. https://
based on self-reports about generic friends. In order
doi.org/10.1177/0265407505050945
to make the picture more complete, future research
Bagwell, C. L., Newcomb, A. F., & Bukowski, W. M. (1998).
is needed that focuses on particular friendship Preadolescent friendship and peer rejection as predictors
dyads or “best friends,” ideally assessing relation- of adult adjustment. Child Development, 69, 140–153.
ship characteristics from both friends’ perspectives. https://doi.org/10.1111/j.1467-8624.1998.tb06139.x
Given that one of the two subsamples used in Banerjee, R., & Henderson, L. (2001). Social-cognitive fac-
this study was stratified by childhood SES and cog- tors in childhood social anxiety: A preliminary investi-
nitive ability, the effects of these variables shall be gation. Social Development, 10, 558–572. https://doi.
interpreted with caution and our findings need to org/10.1111/1467-9507.00180
be replicated, ideally in research based on large, Barry, C. M., Madsen, S. D., Nelson, L. J., Carroll, J. S., &
representative community samples. Badger, S. (2009). Friendship and romantic relationship
qualities in emerging adulthood: Differential associa-
Finally, given that some variables were skewed or
tions with identity development and achieved adult-
binary, and due to the relatively small sample size,
hood criteria. Journal of Adult Development, 16, 209–222.
parts of our study are exploratory; and although https://doi.org/10.1007/s10804-009-9067-x
there is path dependency among friendship experi- Biggs, B. K., Vernberg, E. M., & Wu, Y. P. (2012). Social
ences from childhood to adulthood, the sizes of all of anxiety and adolescents’ friendships: The role of social
the path coefficients were far from indicating perfect withdrawal. Journal of Early Adolescence, 32, 802–823.
correlations. We therefore regard our findings as https://doi.org/10.1177/0272431611426145
encouragement for further investigations of the ornsson, S., Edelstein, W., & Kreppner, K. (1977). Explo-
Bj€
dynamics of friendships across the life span. rations in social inequality. Stratification dynamics in social
and individual development in Iceland. Berlin, Germany:
Max-Planck-Institut f€ ur Bildungsforschung.
Conclusion Bokhorst, C. L., Sumter, S. R., & Westenberg, P. M.
Our study is among the first ones exploring the (2010). Social support from parents, friends, classmates,
interlinked pathways of friendship-related competen- and teachers in children and adolescents aged 9 to 18
years: Who is perceived as most supportive? Social
cies, behaviors, and experiences across childhood, ado-
Development, 19, 417–426. https://doi.org/10.1111/j.
lescence, and the adult years. The distal effects are 1467-9507.2009.00540.x
remarkable, given the long-time interval between Booth-LaForce, C., & Oxford, M. L. (2008). Trajectories of
some of the measurement time points. However, the social withdrawal from grades 1 to 6: Prediction from
sizes of the coefficients were moderate and there was early parenting, attachment, and temperament. Develop-
no direct association between mid-adolescent socio- mental Psychology, 44, 1298–1313. https://doi.org/10.
moral competencies and mid-adult friendship experi- 1037/a0012954
ences. Our study therefore also shows that the ways in Buchmann, M., & Steinhoff, A. (2017). Social inequality,
which individuals bond with friends change consider- life course transitions, and adolescent development:
ably during all of the life periods considered here. Introduction to the special issue. Journal of Youth and
Thus, much remains to be learned about potential Adolescence, 46, 2083–2090. https://doi.org/10.1007/
s10964-017-0740-2
mediators and factors that interrupt the sequence of
Buhrmester, D. (1990). Intimacy of friendship, interper-
friendship development. Better knowledge about how sonal competence, and adjustment during preadoles-
social and individual factors interact longitudinally, cence and adolescence. Child Development, 61, 1101–
and how they shape people’s friendship networks, 1111. https://doi.org/10.2307/1130878
could inform effective interventions that help young Bukowski, W. M., & Sippola, L. K. (1996). Friendship and
people to bond with others in ways that improve their morality: (How) are they related? In W. M. Bokowski,
satisfaction and well-being across the life span. A. F. Newcomb, & W. W. Hartup (Eds.), The company
they keep. Friendship in childhood and adolescence (pp.
References 238–261). Cambridge, UK: Cambridge University Press.
Cable, N., Bartley, M., Chandola, T., & Sacker, A. (2013).
Allison, P. D. (2002). Missing data. London, UK: Sage. Friends are equally important to men and women, but
Arnett, J. J. (2004). Emerging adulthood. The winding road family matters more for men’s well-being. Journal of
from the late teens through the twenties. Oxford, UK and Epidemiology & Community Health, 67, 166–171. https://
New York, NY: Oxford University Press. doi.org/10.1136/jech-2012-201113
Life-Span Perspective on Friendship Development 15

Caravita, S. C. S., Sijtsema, J. J., Rambaran, A. J., & Gini, Eve, M. (2002). Is friendship a sociological topic? European
G. (2014). Peer influences on moral disengagement in Journal of Sociology, 43, 386–409. https://doi.org/10.
late childhood and early adolescence. Journal of Youth 1017/S0003975602001157
and Adolescence, 43, 193–207. https://doi.org/10.1007/ Felmlee, D., Sweet, E., & Sinclair, H. C. (2012). Gender rules:
s10964-013-9953-1 Same- and cross-gender friendships norms. Sex Roles, 66,
Carbery, J., & Buhrmester, D. (1998). Friendship and need 518–529. https://doi.org/10.1007/s11199-011-0109-z
fulfillment during three phases of young adulthood. Finkenauer, C., Engels, R. C. M. E., Branje, S. J., & Meeus,
Journal of Social and Personal Relationships, 15, 393–409. W. (2004). Disclosure and relationship satisfaction in
https://doi.org/10.1177/0265407598153005 families. Journal of Marriage and Family, 66, 195–209.
Chandler, M. J. (1973). Egocentrism and antisocial behav- https://doi.org/10.1111/j.0022-2445.2004.00013.x-i1
ior: The assessment and training of social perspective- Fraley, R. C., & Davis, K. E. (1997). Attachment formation
taking skills. Developmental Psychology, 9, 326–332. and transfer in young adults’ close friendships and
https://doi.org/10.1037/h0034974 romantic relationships. Personal Relationships, 4, 131–
Cohen, J. (1988). Statistical power analysis for the behavioral 144. https://doi.org/10.1111/j.1475-6811.1997.tb00135.x
sciences (2nd ed.). New York, NY: Psychology Press, Gazelle, H. (2008). Behavioral profiles of anxious solitary
Taylor & Francis Group. children and heterogeneity in peer relations. Develop-
Colby, A., Kohlberg, L., Gibbs, J., & Lieberman, M. mental Psychology, 44, 1604–1624. https://doi.org/10.
(1983). A longitudinal study of moral judgment. Mono- 1037/a0013303
graphs of the Society for Research in Child Development, Gillespie, B. J., Lever, J., Frederick, D., & Royce, T. (2015).
48, 1–124. https://doi.org/10.2307/1165935 Close adult friendships, gender, and the life cycle. Jour-
Cramer, D. (2000). Relationship satisfaction and conflict nal of Social and Personal Relationships, 32, 709–736.
style in romantic relationships. The Journal of Psychology: https://doi.org/10.1177/0265407514546977
Interdisciplinary and Applied, 134, 337–341. https://doi. Goodwin, N. P., Mrug, S., Borch, C., & Cillessen, A. H.
org/10.1080/00223980009600873 N. (2012). Peer selection and socialization in adolescent
Crockett, L. J., & Randall, B. A. (2006). Linking adolescent depression: The role of school transitions. Journal of
family and peer relationships to the quality of young Youth and Adolescence, 41, 320–332. https://doi.org/10.
adult romantic relationships: The mediating role of con- 1007/s10964-011-9723-x
flict tactics. Journal of Social and Personal Relationships, 23, Grundmann, M. (1998). Norm und Konstruktion: Zur Dialek-
761–780. https://doi.org/10.1177/0265407506068262 tik von Bildungsvererbung und Bildungsaneignung [Norm
de Los Reyes, A. (2011). Introduction to the special sec- and construction: the dialectic of the transmission and attain-
tion: More than measurement error: Discovering mean- ment of education]. Opladen, Germany: Leske & Budrich.
ing behind informant discrepancies in clinical Grundmann, M., Dravenau, D., Bittlingmayer, U. H., &
assessments of children and adolescents. Journal of Clin- Edelstein, W. (2006). Handlungsbef€ahigung und Milieu.
ical Child and Adolescent Psychology, 40, 1–9. https://doi. Zur Analyse milieuspezifischer Alltagspraktiken und ihrer
org/10.1080/15374416.2011.533405 Ungleichheitsrelevanz [Capability and milieu. The analysis
Del Giudice, M. (2008). Sex-biased ratio of avoidant/am- of milieu specific daily practices and their relevance for
bivalent attachment in middle childhood. British Journal inequality]. Berlin, Germany: LIT.
of Developmental Psychology, 26, 369–379. https://doi. Grundmann, M., Steinhoff, A., & Edelstein, W. (2011).
org/10.1348/026151007X243289 Social class, socialization and capabilities in a modern

Demir, M., & Ozdemir, M. (2010). Friendship, need satis- welfare state: Results from the Iceland longitudinal
faction and happiness. Journal of Happiness Studies, 11, study. In O. Leßmann, H.-U. Otto, & H. Ziegler (Eds.),
243–259. https://doi.org/10.1007/s10902-009-9138-5 Closing the capabilities gap. Renegotiating social justice for
Denissen, J. J. A., Asendorpf, J. B., & van Aken, M. A. G. the young (pp. 233–252). Opladen & Farmington Hills,
(2008). Childhood personality predicts long-term trajec- MI: Barbara Budrich.
tories of shyness and aggressiveness in the context of Hart, D., Hofmann, V., Edelstein, W., & Keller, M. (1997).
demographic transitions in emerging adulthood. Journal The relation of childhood personality types to adoles-
of Personality, 76, 67–99. https://doi.org/10.1111/j.1467- cent behavior and development: A longitudinal study
6494.2007.00480.x of Icelandic children. Developmental Psychology, 33, 195–
Edelstein, W., Keller, M., & Schr€ oder, E. (1990). Child 205. https://doi.org/10.1037//0012-1649.33.2.195
development and social structure: A longitudinal study Hartup, W. W., & Stevens, N. (1997). Friendships and
of individual differences. In P. Baltes, D. L. Feather- adaptation in the life course. Psychological Bulletin, 121,
man, & R. M. Lerner (Eds.), Life-span development and 355–370. https://doi.org/10.1037//0033-2909.121.3.355
behavior (Vol. 10, pp. 151–185). Hillsdale, NJ: Erlbaum. Havighurst, R. (1976). Developmental tasks and education.
Enders, C. K. (2013). Dealing with missing data in devel- New York, NY: McKay.
opmental research. Child Development Perspectives, 7, 27– Hazan, C., & Shaver, P. (1987). Romantic love conceptual-
31. https://doi.org/10.1111/cdep.12008 ized as an attachment process. Journal of Personality and
Erikson, E. H. (1959). Identity and the life cycle: Selected Social Psychology, 52, 511–524. https://doi.org/10.1037/
papers. New York, NY: International University Press. 0022-3514.52.3.511
16 Steinhoff and Keller

Hofmann, V. (1991). Die Entwicklung depressiver Reaktionen Keller, M., & Reuss, S. (1984). An action-theoretical recon-
in Kindheit und Jugend. Eine entwicklungspsychopathologis- struction of the development of social-cognitive compe-
che L€angsschnittuntersuchung [The development of depres- tence. Human Development, 27, 211–220. https://doi.
sive reactions in childhood and adolescence. A developmental org/10.1159/000272911
psychopathological Investigation]. Berlin, Germany: Max- Keller, M., & Wood, P. (1989). Development of friendship
Planck-Institut f€ ur Bildungsforschung. reasoning: A study of interindividual differences in
Hsieh, C.-M. (2003). Counting importance: The case of life intraindividual change. Developmental Psychology, 25,
satisfaction and relative domain importance. Social Indi- 820–826. https://doi.org/10.1037/0012-1649.25.5.820
cators Research, 61, 227–240. https://doi.org/10.1023/A: Kiesner, J., Poulin, F., & Nicotra, E. (2003). Peer relations
1021354132664 across contexts: Individual-network homophily and net-
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit work inclusion in and after school. Child Development, 74,
indexes in covariance structure analysis: Conventional 1328–1343. https://doi.org/10.1111/1467-8624.00610
criteria versus new alternatives. Structural Equation Kohn, M. L. (1977). Social competence, symptoms and under-
Modeling: A Multidisciplinary Journal, 6, 1–55. https:// achievement in childhood. Washington, DC: Winston & Sons.
doi.org/10.1080/10705519909540118 Ladd, G. W. (2006). Peer rejection, aggressive or with-
Jacobsen, T., Edelstein, W., & Hofmann, V. (1994). A longi- drawn behavior, and psychological maladjustment
tudinal study of the relation between representations of from ages 5 to 12: An examination of four predictive
attachment in childhood and cognitive functioning in models. Child Development, 77, 822–846. https://doi.
childhood and adolescence. Developmental Psychology, 30, org/10.1111/j.1467-8624.2006.00905.x
112–124. https://doi.org/10.1037//0012-1649.30.1.112 Little, T. D. (2013). Longitudinal structural equation modeling.
Jacobsen, T., & Hofmann, V. (1997). Children’s attach- New York, NY: Guilford.
ment representations: Longitudinal relations to school
Malti, T., & Buchmann, M. (2010). Socialization and indi-
behavior and academic competency in middle child-
vidual antecedents of adolescents’ and young adults’
hood and adolescence. Developmental Psychology, 33,
moral motivation. Journal of Youth and Adolescence, 39,
703–710. https://doi.org/10.1037//0012-1649.33.4.703
138–149. https://doi.org/10.1007/s10964-009-9400-5
Jones, D. C. (1991). Friendship satisfaction and gender: An
Marion, D., Laursen, B., Zettergren, P., & Bergman, L. R.
examination of sex differences in contributors to friend-
(2013). Predicting life satisfaction during middle adult-
ship satisfaction. Journal of Social and Personal Relationships,
hood from peer relationships during mid-adolescence.
8, 167–185. https://doi.org/10.1177/0265407591082002
Journal of Youth and Adolescence, 42, 1299–1307. https://
Keller, M. (1984). Resolving conflicts in friendship: The
doi.org/10.1007/s10964-013-9969-6
development of moral understanding in everyday life.
In W. M. Kurtines & J. L. Gewirtz (Eds.), Morality, Meeks, B. S., Hendrick, S. S., & Hendrick, C. (1998). Com-
moral behavior, and moral development (pp. 140–158). munication, love and relationship satisfaction. Journal of
New York, NY: Wiley. Social and Personal Relationships, 15, 755–773. https://
Keller, M. (1996). Moralische Sensibilit€at: Entwicklung in doi.org/10.1177/0265407598156003
Freundschaft und Familie [Moral sensitivity. Development Muthen, L. K., & Muthen, B. O. (2012). Mplus user’s guide
in friendship and family]. Weinheim, Germany: Beltz/ (7th ed.). Los Angeles, CA: Author.
Psychologie VerlagsUnion. Oh, W., Rubin, K. H., Bowker, J. C., Booth-LaForce, C.,
Keller, M., & Edelstein, W. (1991). The development of Rose-Krasnor, L., & Laursen, B. (2008). Trajectories of
socio-moral meaning-making: Domains, categories, and social withdrawal from middle childhood to early ado-
perspective-taking. In W. M. Kurtines & J. L. Gewirtz lescence. Journal of Abnormal Child Psychology, 36, 553–
(Eds.), Handbook of moral behavior and development: Vol. 566. https://doi.org/10.1007/s10802-007-9199-z
2. Research (Vol. 2, pp. 89–114). Hillsdale, NJ: Erlbaum. Pahl, R., & Pevalin, D. J. (2005). Between family and
Keller, M., & Edelstein, W. (1993). The development of the friends: A longitudinal study of friendship choice. The
moral self from childhood to adolescence. In G. G. Noam, British Journal of Sociology, 56, 433–451. https://doi.org/
T. E. Wren, G. Nunner-Winkler, & W. Edelstein (Eds.), 10.1111/j.1468-4446.2005.00076.x
The moral self (pp. 310–336). Cambridge, MA: MIT Press. Pellegrini, D. S., Masten, A. S., Garmezy, N., & Ferrarese,
Keller, M., Malti, T., & Dravenau, D. (2007). Bindung, M. J. (1987). Correlates of social and academic compe-
Sozialisation und die Entwicklung sozio-moralischen tence in middle childhood. Journal of Child Psychology
Denkens von der Kindheit bis zum Jugendalter: Befunde and Psychiatry, 28, 699–714. https://doi.org/10.1111/j.
einer L€angsschnittuntersuchung [Attachment, socializa- 1469-7610.1987.tb01553.x
tion, and the development of socio-moral thinking from Perilloux, C., & Buss, D. M. (2008). Breaking up romantic
childhood to adolescence: findings from a longitudinal relationships: Costs experienced and coping strategies
study]. In C. Hopf & G. Nunner-Winkler (Eds.), Fr€ uhe deployed. Evolutionary Psychology, 6, 164–181. https://
Bindungen und moralische Entwicklung: Aktuelle Befunde zu doi.org/10.1177/147470490800600119
psychischen und sozialen Bedingungen moralischer Eigen- Quay, H. C., & Peterson, D. R. (1975). Manual for the
st€andigkeit [Early attachment and moral development: recent behavior problem checklist. Unpublished manuscript.
findings on psychological and social conditions of moral inde- Radloff, L. S. (1977). The CES-D scale: A self-report
pendence] (pp. 125–150). Weinheim, Germany: Juventa. depression scale for research in the general population.
Life-Span Perspective on Friendship Development 17

Applied Psychological Measurement, 1, 385–401. https:// Solomon, J., George, C., & De Jong, A. (1995). Children
doi.org/10.1177/014662167700100306 classified as controlling at age six: Evidence of disorga-
Raven, J. C. (1960). Guide to the standard progressive matri- nized representational strategies and aggression at
ces. London, UK: H. K. Lewis. home and at school. Development and Psychopathology, 7,
Rubin, K. H., Burgess, K. B., Kennedy, A. W., & Stewart, 447–463. https://doi.org/10.1017/S0954579400006623
S. L. (2003). Social withdrawal in childhood. In E. J. Steinhoff, A. (2015). Emerging adults’ commitment to effec-
Mash & R. A. Barkley (Eds.), Child psychopathology (pp. tively supportive friendships: A life-span perspective on
372–406). New York, NY: Guilford. social cognitive and behavioral antecedents, and implications
Rubin, K. H., LeMare, L. J., & Lollis, S. (1990). Social for adult relationship satisfaction. Doctoral thesis, Univer-
withdrawal in childhood: Developmental pathways to sity of Muenster.
peer rejection. In S. R. Asher & J. D. Coie (Eds.), Peer Storch, E. A., & Masia-Warner, C. (2004). The relationship
rejection in childhood (pp. 217–249). Cambridge, UK: of peer victimization to social anxiety and loneliness in
Cambridge University Press. adolescent females. Journal of Adolescence, 27, 351–362.
Rubin, K. H., Wojslawowicz, J. C., Rose-Krasnor, L., https://doi.org/10.1016/j.adolescence.2004.03.003
Booth-LaForce, C., & Burgess, K. B. (2006). The best Sullivan, H. S. (1953). The interpersonal theory of psychiatry.
friendships of shy/withdrawn children: Prevalence, sta- New York, NY: W. W. Norton.
bility, and relationship quality. Journal of Abnormal Child Tashiro, T., & Frazier, P. (2003). “I’ll never be in a rela-
Psychology, 34, 143–157. https://doi.org/10.1007/ tionship like that again”: Personal growth following
s10802-005-9017-4 romantic relationship breakups. Personal Relationships,
Ryff, C. D., & Singer, B. H. (2008). Know thyself and 10, 113–128. https://doi.org/10.1111/1475-6811.00039
become what you are: A eudaimonic approach to psy- Waters, E., Merrick, S., Treboux, D., Crowell, J., & Alber-
chological well-being. Journal of Happiness Studies, 9, 13– sheim, L. (2000). Attachment security in infancy and
39. https://doi.org/10.1007/s10902-006-9019-0 early adulthood: A twenty-year longitudinal study.
Sanderson, C. A., Rahm, K. B., & Beigbeder, S. A. (2005). Child Development, 71, 684–689. https://doi.org/10.
The link between the pursuit of intimacy goals and sat- 1111/1467-8624.00176
isfaction in close same-sex friendships: An examination Ziegenhain, U., Rottmann, U., & Rauh, H. (1988). Test-
of the underlying processes. Journal of Social and Per- barkeit von 1 1/2 jahrigen Kindern innerhalb des vertrauten
sonal Relationships, 22, 75–98. https://doi.org/10.1177/ haeuslichen settings in Abhangigkeit von der Bin-
0265407505049322 dungssicherheit zur Mutter [Testability of 1.5-year-old chil-
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our dren in the familiar home setting dependent on the mother’s
view of the state of the art. Psychological Methods, 7, attachment security]. Paper presented at the 35th Con-
147–177. https://doi.org/10.1037//1082-989X.7.2.147 gress of the German Psychological Association, Berlin,

Schellhas, B. (1993). Die Entwicklung der Angstlichkeit in Germany.
Kindheit und Jugend: Befunde einer L€angsschnittstudie u€ber Zuffian o, A., Sette, S., Colasante, T., Buchmann, M., &

die Bedeutung der Angstlichkeit f€
ur die Entwicklung der Malti, T. (2018). Cross-informant assessment of chil-
Kognition und des Schulerfolgs [The development of anxiety dren’s sympathy: Disentangling trait and state agree-
in childhood and adolescence: findings from a longitudinal ment. Frontiers in Applied Mathematics and Statistics, 4,
study on the significance of anxiety for the development of https://doi.org/10.3389/fams.2018.00008
cognition and school success]. Berlin, Germany: Max-
Planck-Institut f€ur Bildungsforschung, Ed. Sigma.
Schumm, W. R., Paff-Bergen, L. A., Hatch, R. C., Obiorah, Appendix
F. C., Copeland, M., Meens, L. D., & Bugaighis, M. A.
(1986). Concurrent and Discriminant Validity of the Kan- A1: The Friendship Dilemma
sas Marital Satisfaction Scale. Journal of Marriage and the
Family, 48, 381–387. https://doi.org/10.2307/352405 The protagonist (actor) promised to meet his or
Selman, R. L. (1980). The growth of interpersonal under- her best friend on their special meeting day.
standing: The developmental and clinical analyses. New Later, the protagonist receives a more attractive
York, NY: Academic Press. invitation from a third child (movie or pop con-
Sette, S., Baumgartner, E., Laghi, F., & Coplan, R. J. (2016). cert depending on age) who has only recently
The role of emotion knowledge in the links between
moved into the neighborhood. This invitation
shyness and children’s socio-emotional functioning at
preschool. The British Journal of Developmental Psychology,
happens to be at the same time the protagonist
34, 471–488. https://doi.org/10.1111/bjdp.12144 had promised to meet the best friend. Various
Shell, M. D., Gazelle, H., & Faldowski, R. A. (2014). Anx- psychological details are mentioned that compli-
ious solitude and the middle school transition: A cate matters further, for example, that the best
diathesis 9 stress model of peer exclusion and victim- friend has problems he or she wants to talk
ization trajectories. Developmental Psychology, 50, 1569– about and that he or she does not like the new
1583. https://doi.org/10.1037/a0035528 child. (Keller & Edelstein, 1991, p. 100)
18 Steinhoff and Keller

A2: Coefficients for Control and Context Variables

Table A2
Significant Path Coefficients (b) From Control and Context Variables

Independent variable ? Dependent variable Model A Model B

Depressive symptoms 22 Friendship intimacy 22 .20* .21*


Gender (female) .32*** .32***
Closeness with parents .34*** .35***
Gender (female) Sociomoral sensitivity 15 .11† .11†
Cognitive ability 7 Sociomoral sensitivity 12 .20*** .19**
Secure attachment 7 .19** .18**
Socioeconomic status 7 Sociomoral sensitivity 9 .20** .19**
Cognitive ability 7 Insecurity in peer contexts 15 .17*
Secure attachment 7 Insecurity in peer contexts 9 .11†
Cognitive ability 7 Peer rejection 12 .12†
Secure attachment 7 .16*
Cognitive ability 7 Peer rejection 9 .19**
Secure attachment 7 .18**

Note. Numbers next to variables indicate age (years).



p < .10. *p < .05. **p < .01. ***p < .001.

A3: The Full Model Including Cross-Lagged Relations Among Sociomoral Sensitivity, Insecurity in peer
Contexts, and Peer Rejection

Socio-Moral .49*** Socio-Moral Socio-Moral Socio-Moral


.46*** .39***
Sensitivity Sensitivity Sensitivity Sensitivity
Age 7 Age 9 Age 12 Age 15
.20*
-.25***
-.38*** Friendship Friendship
.36**
-.31*** -.13† Intimacy Satisfaction
-.43*** Age 22 Age 37
Insecurity in Peer .51*** Insecurity in Peer .23** Insecurity in Peer
Context Context Context
Age 9 Age 12 Age 15

.14†

.27**

.53*** .29***
Peer Rejection Peer Rejection Peer Rejection
Age 9 Age 12 Age 15

Figure A3. Standardized estimates for the joint pathways from child and adolescent sociomoral sensitivity in friendship, insecurity in
the peer context, and peer rejection to adult friendship quality.
Note. The illustration is limited to significant pathways of main interest. Model fit: v2 = 131.67, df = 97, p = .011, root mean square error
of approximation = .05, comparative fit index = .95, standardized root mean square residual = .07.

p < .10. *p < .05. **p < .01. ***p < .001.

You might also like