J Youth Adolescence

DOI 10.1007/s10964-013-9994-5

EMPIRICAL RESEARCH

The Development of Moral Emotions and Decision-Making From

Adolescence to Early Adulthood: A 6-Year Longitudinal Study
Tobias Krettenauer • Tyler Colasante •

Marlis Buchmann • Tina Malti

Received: 20 June 2013 / Accepted: 31 July 2013

Ó Springer Science+Business Media New York 2013

Abstract Adolescents’ emotions in the context of moral thermore, our findings suggest that moral emotions serve as
decision-making repeatedly have been shown to predict an important link between moral personality development
actual behaviour. However, little systematic information on and decision-making processes that are more proximal to
developmental change regarding these emotion expectancies everyday moral behavior.
has been available thus far. This longitudinal study investi-
gated anticipated moral emotions and decision-making Keywords Moral emotions
Moral decision-
between the ages of 15 and 21 in a representative making
Moral personality
Longitudinal study

sample of Swiss adolescents (N = 1,258; 54 % female; Adolescence
M = 15.30 years). Anticipated moral emotions and decision-
making were assessed through a semi-structured interview
procedure. Using Bernoulli hierarchical linear modeling, it Introduction
was found that positive feelings after a moral transgression
(i.e., ‘‘happy victimizer’’ responses) decreased over time, In addition to influencing the way we think about morality,
whereas positive feelings after a moral decision (i.e., ‘‘happy societal forces shape the structure of moral motivation.
moralist’’ responses) increased. However, this pattern was More than 130 years ago, Durkheim (1893/1997) noted
contingent upon the moral scenario presented. Systematic how changes in the division of labor necessitated new
relationships between anticipated moral emotions and moral forms of moral regulation, what he referred to as ‘‘organic
personality characteristics of sympathy, conscientiousness, solidarity’’ and Piaget (1932/1965) later would describe as
and agreeableness were found, even when controlling for ‘‘moral autonomy’’. There is general consensus that mod-
socio-demographic characteristics and cognitive ability. ern societies rely less on the customs and traditions of
Overall, this study demonstrates that the development of previous generations, and more so on flexible, internalized
anticipated moral emotions is not limited to childhood. Fur- forms of moral self-regulation that are better adjusted to the
complexities of modern social life. While these social
changes bear the dangers of moral decline (Jeynes 2010),
they also set the stage for the question, how children and
T. Krettenauer (&) adolescents develop moral motivation that is consistent
Department of Psychology, Wilfrid Laurier University, with an autonomous understanding of morality (e.g.,
75 University Ave W, Waterloo N2L 3C5, Canada Nunner-Winkler 2007). This question provides the back-
e-mail: [email protected]
drop for the present article. The majority of previous
T. Colasante
T. Malti research in this area has focused on the development of
Department of Psychology, University of Toronto, Toronto, ON, moral motivation and moral emotions in children (for
Canada overviews, see, Arsenio in press; Krettenauer et al. 2008;
Malti and Ongley in press). The present study expands this
M. Buchmann
T. Malti
Jacobs Center for Productive Youth Development, University of research to later development by investigating antici-
Zurich, Zurich, Switzerland pated moral emotions in adolescence. We assume that

123
 J Youth Adolescence

adolescents’ anticipated emotions influence their actual increases in anticipated moral emotions (i.e., guilt, shame,
behavior and, at the same time, reflect important aspects of sadness) are linked to initial limitations in young children’s
their moral identity. In the paragraphs to follow, we flesh cognitive ability to process opposing emotional states
out these assumptions before turning to the limitations of (Harris 1989; Wintre and Vallance 1994). As a conse-
previous research on adolescents’ moral emotion expec- quence, younger children focus on initial, positive emotions
tancies. Finally, building from these limitations, we following a moral transgression, whereas older children are
describe the leading research questions and hypotheses of more likely to consider both the gains and losses associated
the current study. with the transgression.
Self-evaluative moral emotions, such as guilt and Nonetheless, anticipated moral emotions are not a mere
shame, arise either ex-ante or ex-post. In other words, they epiphenomenon of cognitive development. If this were the
are experienced in anticipation of a moral norm violation case, the consistent relationship between moral emotions
or after the occurrence of a moral transgression (Malti and and social behavior that has been documented repeatedly
Ongley in press; Tangney et al. 2007). Anticipated moral across childhood and adolescence would be difficult to
emotions provide critical information about the desirability explain (Malti and Krettenauer 2013). As noted by
of a future action and can thus be seen as important pre- numerous authors, moral emotions are tied intimately to an
dictors of moral decision-making and associated behavior individual’s sense of self (cf. Blasi 1999). As such, they
(Baumeister et al. 2007). In support of this claim, a recent reflect the self-importance or self-relevance of moral rules
meta-analysis of over 40 experimental and correlational and values (Tracy and Robins 2007). In line with this view,
studies, with more than 8,000 participants ranging from 4 Krettenauer et al. (2013b) found a significant correlation
to 20 years of age, found significant associations between between children’s moral self-concept and anticipated
children’s and adolescents’ anticipated moral emotions and moral emotions following antisocial behavior or a lack of
both prosocial and antisocial behavioral outcomes (Malti prosocial behavior. Krettenauer and Johnston (2011)
and Krettenauer 2013). These findings resonate well with reported a substantial association between adolescents’
social information processing perspectives, which empha- moral identity and anticipated moral emotions following
size the importance of outcome expectancies in determin- moral transgressions. In a sample of young adults, Stets
ing socio-moral behavior (cf. Arsenio and Lemerise 2004). and Carter (2012) found that discrepancies between indi-
Anticipated moral emotions (also labeled moral emotion viduals’ moral identities and behaviors were associated
expectancies) have been studied predominantly in children with negative, self-evaluative emotions of guilt and shame.
between the ages of 4 and 8 years (for overviews, see From this perspective, the development of anticipated
Arsenio in press; Krettenauer et al. 2008). It is well docu- moral emotions in adolescence may reflect changes in
mented that young children spontaneously engage in pro- adolescents’ moral self and/or moral identity
social behavior as early as 2 years of age (Svetlova et al. Despite adolescence being considered a crucial period
2010; Dunfield et al. 2011; Warneken and Tomasello 2009) for moral identity development (Hardy and Carlo 2011),
and express moral emotions of shame and guilt around systematic, empirical information on age-related change in
4–5 years of age, albeit non-verbally (for an overview see moral identity throughout the adolescent years is scarce.
Eisenberg 2000). Thus, young children are far from amoral. Krettenauer (2011) reported a small, but statistically sig-
Still, relative to older children, 4- to 5-year-olds experience nificant decrease in external moral motivation throughout
marked difficulties anticipating moral emotions of shame or adolescence, suggesting that older adolescents tend to rely
guilt in the context of moral wrongdoing. At this age, on internal, self-relevant moral standards more than
children typically expect a moral wrongdoer to experience younger adolescents. This trend is consistent with research
positive emotions when transgressing a moral rule (e.g., on adolescents’ prosocial reasoning, which suggests that
happiness for having achieved a desired object). This adolescents’ reasoning behind acting morally becomes
finding has been dubbed the ‘‘happy victimizer phenome- increasingly internalized with age (Eisenberg et al. 1995;
non’’ (Arsenio in press). Typically, it is not before the age of Eisenberg et al. 2001; Gibbs et al. 2007). It also resonates
7–8 years that children begin to anticipate negative or well with research documenting a normative decrease in
mixed feelings after moral wrongdoing. moral disengagement over the adolescent years (Paciello
Interestingly, research indicates that systematic probing et al. 2008), as moral disengagement has been associated
for emotions other than happiness facilitates the anticipa- negatively with anticipated feelings of guilt and remorse
tion of negative emotions in 6- to 8-year olds, but leaves the (Bandura et al. 1996). One of the few longitudinal studies
frequency of happy victimizer responses in 4-year olds capturing the development of moral emotion expectancies
largely unaffected (Arsenio and Kramer 1992; Lourenço in adolescence is the Munich Longitudinal Study on the
1997). It has been proposed that decreases in self-focused Genesis of Individual Competencies (LOGIC, see Schnei-
emotions (i.e., happiness, pride) and corresponding der and Bullock 2009), in which moral emotions were

123
J Youth Adolescence

investigated over an extended period of time (i.e., from 4 to hand, and higher agreeableness (i.e., being friendly/com-
23 years). In this study, moral emotion expectancies passionate as opposed to cold/unkind) on the other, have
increased throughout adolescence and early adulthood, been associated with stronger moral emotion expectancies
although effects of age were less marked in later devel- in adolescence (Krettenauer et al. 2013a; Malti and
opmental periods (Nunner-Winkler 2007, 2009). Buchmann 2010). Taken together, these findings suggest
Whereas most research on children’s moral emotion that changes in adolescents’ anticipated moral emotions
expectancies has focused on set moral transgressions and might depend on individual differences in moral personality
the negatively charged moral emotions that follow, Nun- characteristics established earlier in life.
ner-Winkler (2007, 2009) asked teenagers to choose their
preferred way of action in a socio-moral situation and
recorded the emotions they anticipated as a result of their The Current Study
decision. Many adolescents preferred positive moral choi-
ces over immoral behavior. This finding is consistent with The present study investigated adolescents’ anticipated
longitudinal data reported by Malti and Keller (2010) in moral emotions in the context of moral decision-making by
which the majority of adolescents chose moral over integrating three different analytical perspectives on moral
immoral paths of action and reported feeling happy as a emotion expectancies into a single, comprehensive
result. The investigation of positive moral choices con- approach. It simultaneously examined normative age-
siderably extends the range of moral emotions under study graded change, non-normative change related to pre-
to include feelings of pride and self-satisfaction over doing existing individual differences, and task-related effects.
what is considered right. Thus, research on the develop- While previous research has established the viability of
ment of moral emotions should not be limited to the these differing perspectives, it has dealt with them mostly
decline of ‘‘happy victimizer’’ response patterns. Once in separation. Moreover, existing research on anticipated
individuals decide to act morally, the distinction between moral emotions in adolescence has utilized primarily cross-
‘‘happy’’ and ‘‘unhappy moralists’’ becomes relevant sectional data and/or small, non-representative samples. By
(Malti and Keller 2010). The longitudinal findings reported contrast, the present study employed a representative,
by Nunner-Winkler, as well as Malti and Keller, suggest longitudinal sample of 15- to 21-year-old adolescents
that the anticipation of moral emotions following positive residing in German and French speaking areas of
moral choices may be a normative developmental trend Switzerland. Data were collected at the ages of 15, 18 and
through adolescence. However, in these studies, adoles- 21 years. Our multi-faceted approach and robust sample
cents’ and young adults’ moral emotion expectancies were allowed us to investigate the development of moral emo-
combined with their emotion justifications. Thus, it is tions and decision-making from adolescence to early
unclear whether longitudinal increases in these composite adulthood in a truly comprehensive fashion.
scores reflect change in anticipated moral emotions per se.
Aside from normative change in anticipated moral Normative Age-Graded Change
emotions, past research suggests that moral emotion
expectancies mark an important dimension of individual As described above, normative change in the development
differences throughout development. As a dimension of of moral emotion expectancies in adolescence is likely
individual differences, anticipated moral emotions consis- contingent upon adolescents’ moral decision-making. As
tently have been shown to predict high levels of prosocial happy victimizer responses (i.e., making immoral decisions
behavior and low levels of antisocial behavior from 4 to and feeling positive for self-oriented reasons) are generally
20 years of age (Malti and Krettenauer 2013). In support of less frequent in adolescence, and negatively-charged moral
these dimensional findings, Krettenauer et al. (2013a) found emotions are more salient (e.g. Krettenauer and Eichler
that moral emotion expectancies were correlated system- 2006), adolescents can be expected to choose moral over
atically with morally-relevant personality traits, as defined immoral ways of action more often. The anticipated emo-
by the Five-Factor Model. This model distinguishes five tions for moral decisions can range from pride and self-
broad factors of personality (extraversion, neuroticism, satisfaction over doing what is considered to be right, or
agreeableness, conscientiousness, and openness to experi- regret for not achieving a desired object. Miller et al.
ence) that define major dimensions of individual differences (2008) demonstrated that self-satisfaction over doing what
(e.g. McCrae and Costa 2004). In particular, the two factors is considered right reflects internal moral motivation. This
of agreeableness and conscientiousness have been associ- motivation tends to increase in adolescence and early
ated repeatedly with morality (cf. Lapsley and Hill 2009). adulthood as external moral motivation decreases (Krette-
Specifically, higher conscientiousness (i.e. being efficient/ nauer 2011). In a similar vein, prosocial, moral reasoning
organized as opposed to easy-going/careless) on the one has been shown to become increasingly internalized as

123
 J Youth Adolescence

adolescents grow older (Eisenberg et al. 1995; Eisenberg but not for scenarios describing minor transgressions (e.g.,
et al. 2001; Gibbs et al. 2007). We, therefore, expected an not returning a found wallet). In the present study, we
age-related increase in positive moral emotions following investigated the extent to which normative and non-nor-
decisions to act morally. mative changes in anticipated moral emotions are consis-
tent across scenarios. Situations that are more tempting, as
Non-normative Changes in Relationship to Prior indicated by more frequent happy victimizer responses at
Individual Differences study onset, may be more challenging for younger ado-
lescents. Consequently, the effects of age may be stronger
As noted above, moral emotion expectancies reflect for these scenarios as compared to scenarios that are less
important individual differences in morally relevant challenging for adolescents’ moral sense of right and
behavioral dispositions. Accordingly, systematic relation- wrong (e.g., not returning a found wallet to the rightful
ships between personality traits of conscientiousness and owner).
agreeableness and anticipated moral emotions have been
found (Krettenauer et al. 2013a; Malti and Buchmann 2010).
However, it is unclear whether these relationships are Method
established early in the course of development or whether
personality traits continue to predict changes in moral Data were acquired from the first three waves of the Swiss
emotions through late adolescence and early adulthood. Survey of Children and Youth. This longitudinal survey
Although Krettenauer et al. (2013a) offer preliminary evi- investigates the life course and development of three age
dence in support of the continuing effects of personality cohorts (6, 15, and 21 years) using a multi-informant
traits, their study was restricted to two personality dimen- approach. The present analysis is based on the 15-year-old
sions of the Five Factor Model and did not consider indi- cohort, whose members were re-assessed at 18 and 21 years.
vidual differences in other personality characteristics that
are known to impact moral development, such as sympathy Participants
(Hoffman 2000; Eisenberg et al. 2002, 1999). Dispositional
sympathy (i.e., feelings of concern or sorrow for another A representative, random sample from German- and
person’s misfortune) represents a core aspect of the ‘‘pro- French-speaking parts of Switzerland was drawn in a two-
social personality’’ (Eisenberg et al. 1999, 2002), and has stage process in which 131 communities (broken down by
been correlated positively with prosocial behavior and community type and community size) were selected. The
negatively correlated with antisocial behavior (Eisenberg residents of each community were then sampled randomly
et al. 2006). While individual differences in sympathy on the basis of information provided by the community’s
overlap to some extent with agreeableness, the two con- official register. The final response rate was 63 % (i.e.,
structs are far from identical (cf. Graziano and Tobin 2009). based on a number of 1997 initially selected addresses).
Thus, individual differences in sympathy may contribute to For statistical analyses, the sample was weighted to correct
the development of moral emotions independently of for nonresponse, overrepresentation of some community
agreeableness and conscientiousness. Based on related types, and a moderate underrepresentation of lower edu-
previous research, we hypothesized that patterns of high cational strata, nationalities, and community types. At
moral sensitivity (i.e., happy moralist/unhappy victimizer Time 1, the sample consisted of 1,258 adolescents with an
responses) would increase over time for individuals who average age of 15.30 years (SD = 0.21; 54 % females).
scored high on conscientiousness, agreeableness, and sym- Furthermore, 1,056 primary caregivers, predominantly
pathy at the onset of the study, whereas happy victimizer mothers (89 %), were linked to the adolescent sample.
responses were expected to decrease for these individuals. Among the 15-year-old adolescents, 80 % were Swiss,
The opposite findings were expected for individuals with 18 % were of other European nationalities, and 2 % were
lower scores on these personality dimensions. non-European. Of the parents, 32 % had secondary edu-
cation or less, 44 % had vocational training or college,
Task Related Effects 15 % had a higher vocational diploma, and 9 % had a
university degree. These numbers are fairly representative
In previous research, the development of moral emotion of Swiss demographics (Swiss Federal Office of Statistics
expectancies has been contingent upon the scenarios pre- 2012). As for family composition, 82 % of the adolescents
sented to adolescents (Saelen and Markovitz 2008). For were living with both parents, and the average number of
instance, Krettenauer and Eichler (2006) found age-related siblings was 1.70 (SD = 1.12).
increases in moral emotion expectancies for scenarios At the second assessment, 952 adolescents (76 %) par-
describing severe moral offenses (e.g., hit-and-run driving) ticipated in an interview (M = 18.49 years, SD = 0.22).

123
J Youth Adolescence

At Time 3 (T3), 794 young adults (63 %) participated in an Measures

interview (M = 21.39 years, SD = 0.22).
We analyzed sample attrition for socio-demographic As the sample contained both German- and French-
variables (gender, socio-economic background, educational speaking participants, all measures were translated from
level) and study variables (pattern of moral emotions and German to French by native speakers and then back-
decision-making in two story contexts, self and other- translated to correct for ambiguous meanings. Participants
reported sympathy, agreeableness, conscientiousness, cog- were interviewed in their primary language. A pilot study
nitive ability; for a detailed description of these variables, with 236 15-year-old adolescents was conducted to test the
see below). We contrasted adolescents of the T1 sample validity of the vignettes on moral development and other
who participated in the interview at T3 (n = 794) with the measures concerning adolescent development. Findings
participants who did not participate (n = 464). For socio- indicated that two out of three vignettes on decision-mak-
demographic variables, no differences between the two ing and anticipated emotions showed sufficient inter-indi-
groups were found. The T3 non-responders evidenced a vidual variability and theoretically expected relations with
higher frequency of happy victimizer response patterns for other dimensions of adolescents’ social development, such
the purse story (9.4 % of non-responders vs. 5.3 % of as sympathy. The final two vignettes on moral decision-
responders), v2 (3) = 11.31, p \ .05, whereas no differ- making and anticipated emotions were selected on the basis
ences were found for the bike story. Moreover, non- of these results.
responders evidenced lower scores in conscientiousness,
t(1,002) = 3.89, p \ .01, Cohen’s d = 0.29, agreeable- Moral Decision-Making and Anticipated Emotions
ness, t(1,004) = 2.45, p \ .05, d = 0.15, parent-reported
sympathy, t(1,047) = 2.47, p \ .05, d = 0.14 and cogni- Moral decision-making and anticipated emotions were
tive ability, t(1,252) = 3.11, p \ .01, d = 0.17. Overall, assessed by a previously validated measure consisting of
these effects of selective attrition were small to moderate in hypothetical moral dilemmas (Malti and Buchmann 2010).
magnitude and concerned less socially desirable individual Two dilemmas involving the temptation to transgress well-
characteristics. known moral rules for personal benefit were chosen based
on the following four criteria. First, the structure and nature
Procedure of the conflict had to be a familiar, everyday-life occur-
rence for the participants. Second, the story had to involve
At the first assessment (Spring 2006), adolescents and clear moral issues. Third, the immoral action decision had
primary caregivers were individually interviewed at home to be justified easily, and the participant must have been
via a computer-assisted personal interview (CAPI). Written able to refer to the transgression’s normality. Fourth and
informed consent for testing was obtained from the par- finally, the characters and events in the stories had to be
ticipant. In addition, primary caregivers provided written free of gender stereotypes and identifiable by both sexes.
informed consent to allow their children to participate in For example, neither story involved the conflict between a
the study. prosocial moral duty and the accumulation of power. The
All participants were given a CAPI in a quiet room at sex of the characters in the story was matched to the par-
their home. Both interviews lasted about 45 min and ticipant, and the order of the stories was counterbalanced to
contained questions on the participant’s social develop- avoid order effects.
ment, as well as on the most important socialization In the first story, participants were read the following
conditions. During the interview of the adolescent, the text: ‘‘Imagine you offered your bike for sale. You want to
primary caregivers were given a questionnaire on adoles- sell it for 500 Swiss Francs. A young man is interested. He
cent social development, which they filled out and mailed bargains with you and you agree on 420 Swiss Francs.
back to the research institute. At the second and third Then he says: ‘Sorry, I don’t have the money on me; I’ll
assessments (Spring 2009 and Spring 2012, respectively), quickly run home to get it. I’ll be back in half an hour.’
a CAPI was conducted with the adolescents. Forty-two You say: ‘Agreed, I’ll wait for you.’ Shortly after he is
interviewers conducted the interviews of the 15-year-olds. gone, another customer shows up who is willing to pay the
When the participants were 18 and 21-years-old, there full price.’’ In the second story, the participants were read
were forty-one and thirty-seven interviewers, respectively. the following text: ‘‘Imagine that you have found a purse
The interviewers were recruited from a professional with 150 Swiss Francs in it and an identity card of the
research institute specializing in social-science interviews owner’’ (Krettenauer and Eichler 2006).
and had been trained by the research team on interview After reading each story, the participants were asked
techniques. (a) what they would do (action decision), (b) how they

123
 J Youth Adolescence

would feel about doing it (anticipation of emotion), and Sympathy

(c) why they would do it and why they would feel that way
(justification). All responses were probed, and the results of Adolescents’ sympathy was assessed at T1 by (a) adoles-
the probing were figured into coding. cents’ self-reports and (b) primary caregivers’ ratings. The
scale for the adolescents consisted of five items from Zhou
Coding of Action Decision and Anticipated Emotion et al. (2003); for example, ‘‘When I see another person who
is hurt or upset, I feel sorry for him or her.’’ Cronbach’s a
The action decision was coded as 1 (moral; i.e., wait for the for the sympathy scale was .72. Mean scale scores were
first customer, bring the purse to lost and found) or 0 computed (M = 4.82, SD = 0.77).
(selfish; i.e., sell the bike to the second customer, take the The items were recorded on a 6-point scale with higher
purse). The question about anticipation of emotions was scores indicating higher levels of sympathy. The primary
open-ended. Participants were asked to report the emotions caregivers responded to three items from Zhou et al. (2003)
they would anticipate after their decision. Anticipated to rate the adolescents’ sympathy; for example, ‘‘My child
emotions were coded as (1) negative or mixed (e.g., guilt, usually feels sorry for other children who are hurt or
shame, sadness, sadness and happiness) and (2) positive upset.’’ The reliability of the scale was a = .77. Mean
only (e.g., happiness, pride). scale scores were computed (M = 4.93, SD = 0.94).
At Time 1, 129 of the 1,258 interviews (10 %) were Higher scores were indicative of higher levels of sympathy.
coded by two independent coders, yielding an inter-rater
reliability of j = .97. To ensure high-quality coding, 50 Personality Characteristics
randomly chosen transcripts (i.e., 39 % of the reliability
data) from each coder at T1 were independently coded by A validated bipolar adjective checklist using three pairs of
the senior author. At Time 2, 50 of the 952 interviews contrasting adjectives for each of the five personality
(5 %) were coded by two independent coders, yielding an dimensions (e.g., ‘‘agreeable versus irritable’’) was chosen
inter-rater reliability of j = .95. At Time 3, 80 of the to assess the Big Five personality dimensions (i.e., agree-
723 interviews (11 %) were coded by two independent ableness, conscientiousness; Malti and Buchmann 2010).
coders, yielding an interrater reliability of j = .94. Dis- The items were answered on a 6-point scale with higher
agreements were discussed and the consensus solution scores indicating greater presence of the personality char-
was coded. acteristic. In accordance with the study’s research design,
primary-caregiver reports were used to assess the person-
Patterns of Decision-Making and Anticipated Emotion ality characteristics of 15-year-olds. Reliabilities of the
scales were a = .82 for conscientiousness and a = .71 for
Next, composite scores were derived to create patterns of agreeableness. Mean scale scores for agreeableness was
decision-making and anticipated emotions (see Malti and M = 4.50, SD = 0.87, and M = 4.35, SD = 1.07 for
Keller 2010; Malti et al. 2012): The happy victimizer pat- conscientiousness.
tern applied to participants who made a selfish decision (0)
and anticipated positive emotions to the self (2). The Control Variables
unhappy victimizer pattern applied to participants who
made a selfish decision (0), but anticipated negative or Cognitive Skills
mixed emotions to the self (1). The happy moralist pattern
applied to participants who made a moral decision (1) and Respondents’ cognitive competences were assessed using
anticipated positive emotions to the self (2). The unhappy the half subscale 2.4 of the Culture Fair Intelligence Test
moralist pattern applied to participants who made a moral (CFT-20; Weiß 1998). The Culture Fair Intelligence Test
decision (1), but felt negative or mixed about it (i.e., contains less verbal instructions than traditional measures
regret). and thus has the advantage of reducing cultural and edu-
For the final data analyses, we created four dummy cational biases. Given Switzerland’s high immigration
variables for each of the four patterns of decision-making rates and the representative, large-scale character of our
and anticipated emotion (i.e., happy victimizer pattern vs. design, the selection of this test seemed more appropriate
all others, unhappy victimizer pattern vs. all others, happy than a more traditional intelligence scale. Each test item
moralist pattern vs. all others, and unhappy moralist pattern was scored as 0 (incorrect) or 1 (correct), and a scale score
vs. all others). However, please note that the contrast for cognitive skills was created (ranging from 0 to 6).
between the unhappy moralist pattern versus all others was Higher scores indicate higher cognitive competences
dropped from the analyses because of its low frequency. (M = 3.31, SD = 1.47).

123
J Youth Adolescence

Level of Education at the Age of 15 Table 1 Frequencies (%) of patterns of moral decision-making and
emotions by assessment point and story context
Level of education was defined in terms of participants’ Assessment point
attained grade level in school at the age of 15 years
Time 1: Time 2: Time 3:
(ranging from 1 = special needs education to 5 = pre- 15 years 18 years 21 years
academic high school; Gymnasium/Fachmittelschule). The
B P B P B P
grand mean of the sample was M = 3.34, SD = 0.99.
Happy victimizer 14 6 16 7 10 9
Family Socioeconomic Background Unhappy victimizer 15 6 7 7 6 6
Happy moralist 61 85 70 85 76 84
The socioeconomic background of primary caregivers was Unhappy moralist 10 3 7 1 8 1
assessed: Socioeconomic status (SES) was based on coding
B bike story, P purse story
the caregiver’s current profession at T1; codes were trans-
formed into an International Socio-Economic Index of
occupational status (ISEI) score (Ganzeboom et al. 1992). impact of story context, sympathy, and personality char-
The ISEI scale is a standardized and widely used scale to acteristics on these patterns. Data waves comprised the unit
measure SES. It is derived from the International Standard of observation for the first level, story context comprised
Classification of Occupations and was constructed using the unit of observation for the second level, and partici-
comparably coded data on education, occupation, and pants comprised the unit of observation for the third level.
income for over 70,000 full-time employed individuals from Since the outcome variables were binomial (i.e., the
16 countries (Ganzeboom et al. 1992, p. 2). The final SES respective pattern of decision-making and anticipated
score was based on the caregiver with the highest ISEI score emotion versus the combined rest of the sample), Bernoulli
(M = 51.55, SD = 16.03). The possible range of scores on HLM models were run (see Hox and Roberts 2011).
the ISEI scale is 16–90. Higher scores indicate higher SES. Bernoulli models typically are used in hierarchical linear
modeling when the outcome variable is binomial (i.e.,
Treatment of Missing Data and Plan of Analysis coded as 0 and 1). We exported the 10 imputed data sets to
HLM, which can handle multiply imputed datasets.
Preliminary analyses indicated that, overall, 19 % of the We ran a series of HLM Bernoulli models for each pattern
data points were missing and they were not distributed (except for the unhappy moralist pattern as this was found to
randomly in the database. Little’s MCAR test was signifi- be rather infrequent, see Table 1). In line with our research
cant, v2(65) = 182.37 p \ .001 (Little 1988). Therefore, questions, the independent variables were time, story context,
multiple imputation was carried out to estimate the values self-reported sympathy, other-reported sympathy, agreeable-
for missing data points of the binary and continuous vari- ness, conscientiousness, and the control variables (i.e., cog-
ables using fully conditional specification in SPSS. This is nitive ability, ISEI score, educational level, and sex). In
an iterative Markov chain Monte Carlo (MCMC) method. addition, we computed two-way interactions between time
It predicts the missing values for a variable using all other and story context, time and sympathy (self report, other
available variables included in the model, and for each report), as well as time and personality characteristics
variable, the suitable model was selected (i.e., for binary (agreeableness, conscientiousness). The basis for interpreting
variables, logistic regression was specified). The number of the interactions in our analyses rested on the significance tests
iterations used was 100, and the imputed values in this for the model coefficients and the plot of the values for spe-
iteration round were used for imputation. The number of cific combinations of the independent variables. Thus, fol-
imputations was 10. lowing the procedures outlined by Preacher et al. (2006) and
Hierarchical linear modeling (HLM Version 7) was used Curran et al. (2006), the simple intercepts and simple slopes
to test our hypotheses on the impacts of story context, for the two-way interactions were probed to examine which
sympathy and personality characteristics on the develop- effects were statistically significant.
ment of moral emotions in the context of adolescents’
moral decision-making. Hierarchical linear models allow
for the analysis of cross-level, hierarchical data. Because Results
data waves were nested in story contexts, which were
nested within participants, we had a hierarchical design and Descriptive Analyses
used three-level hierarchical linear models (HLM Version
7; Raudenbush et al. 2011) to evaluate patterns of decision- Frequencies (%) of moral decision-making patterns and
making and moral emotions over time, as well as the anticipated emotions by assessment point and story context

123
 J Youth Adolescence

Table 2 Crosstabulation of patterns of moral decision-making and patterns were combined in a HLM Bernoulli model for the
emotions at Time 1 and Time 2 (% of responses) two scenarios. Table 2 shows the patterns of decision-
HV T2 UV T2 HM T2 UM T2 Total (% within T1) making and anticipated emotions for T1 and T2, and
Table 3 presents the patterns of decision-making and
HV T1 37 7 54 2 100
anticipated emotions for T2 and T3. As can be seen, many
UV T1 16 18 62 4 100 participants transitioned from the happy victimizer and
HM T1 7 5 84 4 100 unhappy victimizer patterns to the happy moralist patterns
UM T1 8 10 67 15 100 from T1 to T2, and from T2 to T3.
HV happy victimizer, UV unhappy victimizer, HM happy moralist, Table 4 displays correlations between the three patterns
UM unhappy moralist, T1 time 1, T2 time 2 of moral decision-making and moral emotions (dummy
coded) for each story context at T1–T3, and self-reported
Table 3 Crosstabulation of patterns of moral decision-making and and other-reported sympathy at T1, agreeableness at T1,
emotions at Time 2 and Time 3 (% of responses) and conscientiousness at T1. As can be seen, the happy
HV T3 UV T3 HM T3 UM T3 Total (% within T2)
victimizer pattern at T1–T3 was associated predominantly
negatively with both self-reported and other-reported
HV T2 42 12 43 3 100 sympathy at T1. Furthermore, the happy victimizer pattern
UV T2 21 23 52 4 100 at T2 and T3 tended to be associated negatively with
HM T2 10 6 80 4 100 agreeableness and conscientiousness at T1. The unhappy
UM T2 5 5 88 4 100 victimizer pattern at T1–T3 tended to be related negatively
HV happy victimizer, UV unhappy victimizer, HM happy moralist, to both self and other-reported sympathy at T1. However,
UM unhappy moralist, T2 time 2, T3 time 3 there was a positive correlation between the unhappy
victimizer pattern at T3 and sympathy at T1. In addition,
are displayed in Table 1. The happy moralist response pattern the unhappy victimizer pattern in both story contexts at T1
was by far most common across stories and assessment points, was associated negatively with agreeableness and consci-
whereas the unhappy moralist pattern occurred infrequently. entiousness at T1. At T2 and T3, the unhappy victimizer
Because of its low frequency, the unhappy moralist pattern pattern in the purse context was associated negatively with
was not considered in the HLM data analyses. agreeableness and conscientiousness. The happy moralist
Next, we looked at how many people transitioned from pattern at T1–T3 was associated predominantly positively
one pattern to another over time. For this purpose, response with both self and other-reported sympathy at T1. This

Table 4 Correlations between the moral decision-making and emotion patterns for each story context at T1–T3 with sympathy and personality
characteristics at Time 1
Time 1 Time 2 Time 3
Independent variables T1 B P B P B P

Happy victimizer (1) versus all others (0)

Sympathy (SR) -0.17*** -0.16*** -0.13*** -0.13*** -0.10*** -0.15***
Sympathy (PR) -0.10** -0.05 -0.16*** -0.08** -0.14*** -0.12***
Agreeableness -0.07* -0.02 -0.07** -0.04 -0.08** -0.08**
Conscientiousness -0.04 -0.04 -0.06* -0.11*** 0.01 -0.08**
Unhappy victimizer (1) versus all others (0)
Sympathy (SR) -0.01 -0.19*** -0.02 -0.04 0.00 0.10***
Sympathy (PR) -0.06* -0.03 0.00 -0.10** -0.03 -0.01
Agreeableness -0.12*** -0.10** 0.01 -0.08** -0.03 -0.12***
Conscientiousness -0.11*** -0.11*** 0.00 -0.10*** -0.01 -0.10***
Happy moralist (1) versus all others (0)
Sympathy (SR) 0.06* 0.23*** 0.12*** 0.12*** 0.05 0.04
Sympathy (PR) 0.10** 0.08** 0.08** 0.12*** 0.08** 0.09**
Agreeableness 0.09** 0.08** 0.05 0.08** 0.07** 0.14***
Conscientiousness 0.12*** 0.12*** 0.02 0.16*** 0.01 0.15***
B bike story, P purse story, SR self report, PR primary caregiver report
* p \ .05; ** p \ .01; *** p \ .001

123
J Youth Adolescence

pattern (happy moralist) was also related predominantly participants with low sympathy at T1 (i.e., one point lower
positively to agreeableness and conscientiousness at T1. than the grand mean on sympathy at T1) showed an increase
in this pattern from T1 to T3.
The Happy Victimizer Pattern by Time, Story Context,
Sympathy, and Personality Characteristics The Happy Moralist Pattern by Time, Story Context,
Sympathy, and Personality Characteristics
In order to examine the role of story context, sympathy and
personality characteristics for the happy victimizer pattern, a Central to the expectations of this study, the happy moralist
series of HLM Bernoulli models were run. The findings of the pattern increased from T1 to T3, OR = 1.17, p \ .001 (see
HLM Bernoulli analyses are shown in Table 5. The happy Table 5). The pattern occurred more frequently in the bike
victimizer pattern was more frequent in the bike story than in story than in the purse story, OR = 0.38, p \ .001. These
the purse story, OR = 1.90, p \ .001. In addition, the happy main effects, however, were qualified by a significant two-
victimizer pattern was associated negatively with both self- way interaction between story context and time, OR = 1.56,
reported sympathy, OR = 0.70, p \ .001, and other-reported p \ .001, indicating that the pattern occurred more fre-
sympathy, OR = 0.81 p \ .001. There was a significant quently in the bike story than in the purse story at T1 and T2
interaction effect between time and story context on the happy (p \ .001). There was no story effect on this pattern at T3.
victimizer pattern, OR = 0.67, p \ .001, indicating that this Both self-reported sympathy, OR = 1.24, p \ .001, and
pattern occurred more frequently in the bike story than in the other-reported sympathy, OR = 1.13, p \ .01, were asso-
purse story at T1 and T2 (ps \ .001). There was no story ciated the happy moralist pattern. In addition, agreeableness,
effect on this pattern at T3. A significant interaction effect OR = 1.12, p \ .05, and conscientiousness, OR = 1.14,
also was found between other-reported sympathy and time on p \ .001, predicted the happy moralist pattern significantly.
the happy victimizer pattern, OR = 0.92 p \ .05, indicating There was a significant interaction between self-reported
that participants with low levels of sympathy at T1 (i.e., one sympathy and time on the happy moralist pattern,
point lower than the grand mean on sympathy) showed a OR = 0.89, p \ .05, revealing that participants with high
slight increase in the happy victimizer pattern from T1 to T3, levels of sympathy (i.e., one point higher than the grand
whereas the happy victimizer pattern remained low from T1 mean on sympathy) were consistently high in the happy
to T3 in participants with high levels of sympathy at T1 (i.e., moralist pattern from T1 to T3. In contrast, participants with
one point higher than the grand mean on sympathy; Fig. 1). low levels of sympathy (i.e., one point lower than the grand
Furthermore, the happy victimizer pattern was less frequent mean on sympathy) showed an increase in the happy mor-
among females than males, OR = 0.40, p \ .001, and was alist pattern from T1 to T3. As a result, the difference
associated negatively with cognitive skills, OR = 0.93, between the two groups at T1 had disappeared at T3 (see
p \ .01, and educational level, OR = 0.80, p \ .001. Fig. 2). Differences in the happy moralist pattern that were
related to self-reported sympathy at the age of 15 years thus
The Unhappy Victimizer Pattern by time, Story diminished over time. Furthermore, the happy moralist
Context, Sympathy, and Personality Characteristics pattern occurred more frequently in females than males,
OR = 1.56, p \ .001, and was associated positively with
The unhappy victimizer pattern decreased from T1 to T3, educational level, OR = 1.08, p \ .05.
OR = 0.77, p \ .001 (see Table 5). In addition, this pattern
occurred more frequently in the bike story than in the purse
story, OR = 1.46, p \ .001. These main effects, however, Discussion
were qualified by a significant two-way interaction between
story context and time, OR = 0.64, p \ .001, indicating that Adolescents’ emotions in the context of moral decision-
the pattern occurred more frequently in the bike story than in making have been associated repeatedly with actual behavior.
the purse story at T1 (p \ .001). There was no story effect on However, little systematic information on developmental
this pattern at T2 and T3. The unhappy victimizer pattern change regarding these emotion expectancies has been
was predicted negatively by agreeableness, OR = 0.82, available so far. The present study investigated adolescents’
p \ .001, and conscientiousness, OR = 0.84, p \ .001. anticipated moral emotions in the context of moral decision-
There was a significant interaction effect between self- making by integrating three different analytical perspectives
reported sympathy and time on the unhappy victimizer on moral emotion expectancies: We examined normative
pattern, OR = 1.34, p \ .001, indicating that participants age-graded change and non-normative change related to
with high levels of sympathy at T1 (i.e., one point higher preexisting individual differences in dispositional sympathy,
than the grand mean on sympathy) showed a decrease in the agreeableness and conscientiousness. Finally, we analyzed
unhappy victimizer pattern from T1 to T3, whereas task-related effects. Whereas previous research on anticipated

123
123
Table 5 Parameter estimates (SE) of independent variables on patterns of moral decision making and emotion attribution: two-level HLM Bernoulli model analyses
Parameter Patterns of moral decision-making and emotion attribution
Happy victimizer Unhappy victimizer Happy moralist
b (SE) t OR (CI) b (SE) t OR (CI) b (SE) t OR (CI)

Time level
Time 0.04 (0.04) 0.84 1.03 (0.93–1.14) -0.27 (0.04) -5.96*** 0.77 (0.67–0.81) 0.16 (0.04) 4.36*** 1.17 (1.09–1.24)
Story level
Story context 0.64 (0.08) 8.51*** 1.90 (1.65–2.37) 0.38 (0.08) 4.83*** 1.46 (1.32–1.86) -0.96 (0.07) -13.98*** 0.38 (0.35–0.45)
Child level
Sympathy (SR) -0.35 (0.06) -6.36*** 0.70 (0.64–0.79) -0.06 (0.06) -1.05 0.94 (0.84–1.07) 0.21 (0.05) 4.51*** 1.24 (1.10–1.32)
Sympathy (PR) -0.21 (0.05) -4.67*** 0.81 (0.74–0.89) -0.05 (0.05) -1.11 0.95 (0.86–1.04) 0.12 (0.04) 3.11** 1.13 (1.04–1.21)
Agreeableness -0.09 (0.05) -1.75 0.91 (0.82–1.01) -0.20 (0.05) -4.07*** 0.82 (0.74–0.90) 0.11 (0.04) 2.59* 1.12 (1.09–1.28)
Conscientiousness -0.01 (0.04) -0.08 0.99 (0.90–1.07) -0.18 (0.04) -4.73*** 0.84 (0.77–0.90) 0.13 (0.04) 3.53*** 1.14 (1.10–1.26)
Cross-level interactions
Story context 9 time -0.40 (0.07) -5.61*** 0.67 (0.58–0.77) -0.45 (0.09) -5.31*** 0.64 (0.53–0.75) 0.45 (0.07) 6.55*** 1.56 (1.36–1.73)
Sympathy (SR) 9 time 0.16 (0.10) 1.59 1.08 (0.97–1.19) 0.29 (0.06) 4.77*** 1.34 (1.19–1.49) -0.11 (0.04) -2.58* 0.89 (0.82–0.97)
Sympathy (PR) 9 time -0.08 (0.07) -2.00* 0.92 (0.85–1.00) 0.06 (0.08) 0.77 1.07 (0.95–1.14) -0.08 (0.07) -1.23 0.92 (0.90–1.08)
Agreeableness 9 time -0.05 (0.09) -0.58 0.95 (0.85–1.02) 0.03 (0.05) 0.58 1.03 (0.94–1.15) -0.13 (0.07) -1.81 0.88 (0.84–1.01)
Conscientiousness 9 time 0.01 (0.04) 0.21 1.01 (0.94–1.08) 0.07 (0.04) 1.71 1.07 (1.00–1.16) -0.04 (0.03) -1.36 0.96 (0.90–1.02)
Control variables
Sex -0.92 (0.08) -11.43*** 0.40 (0.34–0.47) 0.01 (0.08) 0.15 1.01 (0.85–1.20) 0.44 (0.07) 6.11*** 1.36 (1.19–1.55)
Cognitive skills -0.08 (0.03) -2.69** 0.93 (0.88–0.98) -0.05 (0.03) -1.76 0.95 (0.89–1.01) 0.04 (0.03) 1.54 1.04 (1.00–1.10)
Educational attainment -0.22 (0.04) -4.99*** 0.80 (0.73–0.87) -0.03 (0.04) -0.81 0.97 (0.91–1.08) 0.08 (0.04) 2.06* 1.08 (0.98–1.13)
SES (ISEI) 0.01 (0.00) 1.35 1.00 (0.99–1.01) 0.01 (0.00) 1.39 1.01 (0.99–1.01) 0.00 (0.01) -1.15 1.00 (0.99–1.00)
OR odds ratio, CI 95 % confidence interval, SR self report, PR primary caregiver report
* p \ .05; ** p \ .01; *** p \ .001
J Youth Adolescence
J Youth Adolescence

alone. This finding underscores the importance of studying

moral emotions in combination with decision-making.
It should be noted that self-conflicting decisions resulting
in negative feelings within the person (either unhappy
victimizer or unhappy moralist) tended to decrease over
time. Thus, the moral decisions that adolescents make and
the emotions that they expect following these decisions
tend to converge over time. This might explain why regret
after a moral decision (‘‘unhappy moralist’’) was found to
be relatively rare in late adolescence and early adulthood.
Fig. 1 Two-way-interaction of sympathy x time on probability of While previous research on anticipated moral emotions
happy victimizer pattern has addressed adequately negative emotions following a
moral transgression, it has investigated rarely positive
emotions for acting morally (for exceptions, see Malti and
Keller 2010; Krettenauer and Johnston 2011). In the pres-
ent study, longitudinal change in positive emotions fol-
lowing a moral decision mirrored the happy victimizer
response pattern. Hence, development in the anticipation of
positive emotions was situation-specific. The anticipation
of positive emotions increased for the bike story over time
whereas the purse scenario evidenced a higher frequency of
positive emotions at the onset of the study that did not
change over time. Miller et al. (2011) have argued that
satisfaction following a moral decision is an important
Fig. 2 Two-way-interaction of self-reported sympathy x time on moral emotion that stems from a higher level of internal-
probability of happy moralist pattern ization or integration of moral responsibilities. Thus, the
increase in positive anticipated emotions following a pro-
moral emotions in adolescence and early adulthood has been social decision found in the present study might be
cross-sectional or based on rather small and selective sam- reflective of adolescents’ and young adults’ moral identity
ples, the current study utilized data from a 6-year, represen- development and corresponding higher levels of internal
tative longitudinal study of Swiss adolescents. The results of moral motivation (Krettenauer 2011).
this study boast greater generalizability than most previous Systematic associations between moral emotions and
findings on the development of moral emotions in adoles- personality characteristics also were found. Measures of
cence and beyond. agreeableness and conscientiousness at the age of 15 years
Normative developmental change in anticipated moral were associated inversely with unhappy victimizer respon-
emotions was contingent on the content of stories. Two ses over the 6-year time interval and positively associated
scenarios were used. One story described a situation in with positive anticipated emotions following a moral deci-
which a decision had to be reached between keeping a sion. This finding is consistent with results reported by
promise versus making a higher profit from selling a bike Krettenauer et al. (2013a), who found that personality traits
to a new customer. The second story depicted a situation were associated with the development of moral emotion
where the protagonist needed to decide whether or not to anticipations. The present study goes beyond these findings
return a lost purse to its rightful owner. While the bike as it demonstrates that the predictive effect of personality
situation evidenced a decrease in the happy victimizer traits is independent of socio-demographic characteristics
pattern over time, a similar age trend was not found for the (SES, level of education and gender) and independent of
purse scenario. For the bike scenario, the frequency of dispositional sympathy.
happy victimizer responses was significantly higher at the Research has demonstrated that sympathy is related to
onset of the study (i.e., at age 15 years) and thus reflected a multiple aspects of moral functioning (Eisenberg et al.
more tempting situation for teenagers at this age. 2010). The present study adds another piece of evidence to
The decreasing happy victimizer response pattern in the this growing literature as it shows that sympathy (self- and
bike story was paralleled by a decrease of unhappy feelings other-reported) is related systematically to the development
following the decision to break a promise. Taken together, of moral emotion expectancies in adolescence and early
both trends suggest an age-related change in adolescents’ adulthood. Lack of sympathy at the age of 15 years (self-
decision-making as opposed to a change in moral emotions and parent-report) was associated with more frequent happy

123
 J Youth Adolescence

victimizer responses over the 6-year time interval. More- Consequently, the extent to which findings from the present
over, adolescents who scored lower on self-reported sym- study reflect moral socialization is unknown. For instance,
pathy at the age of 15 years evidenced a slight increase in the base rate of positive moral choices in the purse story at
happy victimizer responses, whereas the opposite was found the onset of the study turned out to be higher than in other
for adolescents who were high on sympathy. Conversely, countries (cf. Saelen and Markovits 2008). This might
positive emotions following a moral decision were associ- reflect a particular emphasis that Swiss parents attach to
ated with higher levels of sympathy over the 6-year time respecting others’ property. In a similar vein, it is important
interval for both self- and parent-reports. For self-reported to keep in mind that the present study is confined to one
sympathy, this effect diminished in early adulthood (i.e., at particular cohort from an individualistic society. Evidence
21 years). The significant interactions between time and from Krettenauer and Jia (2013) and Malti and Keller (2010)
sympathy as predictors of moral emotions suggest that the highlights cultural differences in moral emotion attributions
influence of empathy related responding on moral devel- and, in doing so, underscores the importance of extending
opment is not limited to childhood. the current study to different cultural contexts.
Research indicates that the personality traits of agree-
ableness and conscientiousness, together with individual
differences in empathy related responding, comprise Conclusions
important dimensions of moral and prosocial personality
development (cf. Eisenberg et al. 2002; Lapsley and Hill The current study offers valuable insight into the devel-
2009). The present study shows that these dimensions are opment of moral emotions in adolescence and early
related systematically to anticipated moral emotions. If adulthood. It demonstrates that research on anticipated
increases in the happy moralist response pattern over moral emotions can be extended meaningfully into ado-
adolescence reflect increases in internal moral motivation lescence and beyond. Although happy victimizer responses
as part of adolescents’ moral identity development, the in early adulthood are rare, they are by no means the
current findings suggest that moral identity development exception. More importantly, the most frequent emotional
and personality are not antagonistic. Instead, they system- outcomes that young adults anticipate from moral decisions
atically intersect in adolescents’ and young adults’ moral are positive. Positive emotions generally serve the function
development (cf. Hill and Roberts 2010). It is important to of maintaining and broadening established action tenden-
note, however, that the present study did not directly cies whereas negative emotions press for behavioral
include a measure of adolescents’ moral identity develop- change (Fredrickson, 1998; Baumeister et al. 2007). Thus,
ment. As a consequence, the developmental mechanisms the present findings suggest that the influence of moral
that lead to changes in adolescents’ emotions and decision- emotions on one’s moral life may shift across the lifespan
making remain speculative. from primarily corrective to sustentative. This life span
Besides limitations that apply to longitudinal studies in view of moral emotions and their effect on individual
general (e.g., sample attrition, potential effects of repeated morality clearly warrants further research.
testing), several limitations of the current study need to be
noted. First, due to the sheer vastness of the present study, Acknowledgments The Swiss National Science Foundation funded
the research reported in this paper. The authors would like to express their
the interview measure of moral emotions and decision- sincere thanks to the children and parents who participated in the study.
making was restricted to two scenarios. Even though these Moreover, the authors are grateful to all the interviewers and under-
two scenarios have been validated in previous studies (e.g., graduate students who helped with the collection and coding of data.
Krettenauer and Eichler 2006; Nunner-Winkler et al. 2007),
Author contributions TM and MB participated in the design of the
and piloted prior to the main study, the situational variability study and coordinated data collection. TK was responsible for man-
of anticipated moral emotions was rather limited and factors uscript preparation and data interpretation. TC assisted in running
that account for this variability were not systematically statistical analyses and drafting the manuscript. All authors read and
investigated in the present study. As a consequence, it approved the final version of the manuscript.
remains unclear which story characteristics produced the
differences in response patterns between scenarios. Second,
the study included only multiple informants at the age of References
15 years. As data on personality characteristics were not
available for the later points in time, it was impossible to Arsenio, W. (in press). Moral emotion attributions and aggression. In
study reciprocal developmental relationships between per- M. Killen, & J. Smetana (Eds.), Handbook of moral development
(2nd ed.). New York: Taylor & Francis.
sonality measures and moral emotions. Third, although Arsenio, W. F., & Kramer, R. (1992). Victimizers and their victims:
parents were involved in data collection, parenting data in Children’s conceptions of the mixed emotional consequences of
relation to prosocial or antisocial behavior was not available. moral transgressions. Child Development, 63, 915–927.

123
J Youth Adolescence

Arsenio, W. F., & Lemerise, E. A. (2004). Aggression and moral Hill, P. L., & Roberts, B. W. (2010). Propositions for the study of
development: Integrating social information processing and moral personality development. Current Directions in Psycho-
moral domain models. Child Development, 75, 987–1002. logical Science, 19, 380–383.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Hoffman, M. L. (2000). Empathy and moral development. Cambridge:
Mechanisms of moral disengagement and the exercise of moral Cambridge University Press.
agency. Journal of Personality and Social Psychology, 71, 364–374. Hox, J. J., & Roberts, J. K. (2011). Handbook of advanced multilevel
Baumeister, R. F., Vohs, K. D., DeWall, C. N., & Zhang, L. (2007). analysis. New York: Taylor and Francis.
How emotion shapes behavior: Feedback, anticipation, and Jeynes, W. (2010). A call for character education and prayer in the
reflection, rather than direct causation. Personality and Social schools. Westport, CT: Praeger.
Psychology Bulletin, 11, 167–203. Krettenauer, T. (2011). The dual moral self: Moral centrality and
Blasi, A. (1999). Emotions and moral motivation. Journal for the intrinsic moral motivation. The Journal of Genetic Psychology,
Theory of Social Behavior, 29, 1–19. 172, 309–328. doi:10.1080/00221325.2010.538451.
Curran, P. J., Bauer, D. J., & Willoughby, M. T. (2006). Testing and Krettenauer, T., Asendorpf, B., & Nunner-Winkler, G. (2013a). Moral
probing interactions in hierarchical linear growth models. In C. emotion attributions and personality traits as long-term predic-
S. Bergeman & S. M. Boker (Eds.), The Notre Dame series tors of antisocial conduct in early adulthood: Findings from a
on quantitative methodology. Methodological issues in aging 20 year longitudinal study. International Journal of Behavioral
research (Vol. 1, pp. 99–129). Mahwah, NJ: Lawrence Erlbaum Development, 27, 192–201. doi:10.1177/0165025412472409.
Associates. Krettenauer, T., Campbell, S., & Hertz, S. (2013b). Moral emotions
Dunfield, K., Kuhlmeier, V. A., O’Connell, L., & Kelley, E. (2011). and the development of the moral self in childhood. European
Examining the diversity of prosocial behavior: Helping, sharing, Journal of Developmental Psychology, 10, 159–173. doi:10.
and comforting in infancy. Infancy, 16, 227–247. 1080/17405629.2012.762750.
Durkheim, E. (1893/1997). The division of labor in society. New Krettenauer, T., & Eichler, D. (2006). Adolescents’ self-attributed
York: Free Press. moral emotions following a moral transgression: relations with
Eisenberg, N. (2000). Emotion, regulation, and moral development. delinquency, confidence in moral judgment, and age. British
Annual Review of Psychology, 51, 665–697. Journal of Developmental Psychology, 24, 489–506.
Eisenberg, N., Carlo, G., Murphy, B., & Van Court, P. (1995). Krettenauer, T., & Jia, F. (2013). Investigating the actor effect in
Prosocial development in late adolescence: A longitudinal study. moral emotion expectancies: A comparison of Canadian and
Child Development, 66, 1179–1197. Chinese adolescents. British Journal of Developmental Psychol-
Eisenberg, N., Eggum, N. D., & Edwards, A. (2010). Empathy-related ogy. Advance online publication. doi:10.1111/bjdp.12012.
responding and moral development. In W. Arsenio & E. Krettenauer, T., & Johnston, M. (2011). Positively versus negatively
A. Lemerise (Eds.), Emotions, aggression and morality in charged moral emotion expectancies in adolescence: The role of
children: Bridging development and psychopathology (pp. situational context and the developing moral self. British Journal
115–135). Washington: American Psychological Association. of Developmental Psychology, 29, 475–488. doi:10.1348/
Eisenberg, N., Fabes, R. A., & Spinrad, T. L. (2006). Prosocial 026151010X508083.
behavior. In N. Eisenberg (Ed.), Handbook of child psychology. Krettenauer, T., Malti, T., & Sokol, B. (2008). The development of
Social, emotional, and personality development (6th ed., Vol. 3, moral emotion expectancies and the happy victimizer phenom-
pp. 646–718). New York: Wiley. enon: A critical review of theory and application. European
Eisenberg, N., Guthrie, I. K., Cumberland, A., Murphy, B. C., Journal of Developmental Science, 2, 221–235.
Shepard, S. A., Zhou, Q., et al. (2002). Prosocial development in Lapsley, D. K., & Hill, P. L. (2009). The development of moral
early adulthood: A longitudinal study. Journal of Personality personality. In D. Narvaez & D. K. Lapsley (Eds.), Personality,
and Social Psychology, 82, 993–1006. identity, and character (pp. 185–213). Cambridge: Cambridge
Eisenberg, N., Guthrie, I. K., Murphy, B. C., Shepard, S. A., University Press.
Cumberland, A., & Carlo, G. (1999). Consistency and develop- Little, R. J. A. (1988). A test of missing completely at random for
ment of prosocial dispositions: A longitudinal study. Child multivariate data with missing values. Journal of the American
Development, 70, 1360–1372. Statistical Association, 83(404), 1198–1202.
Eisenberg, N., Zhou, Q., & Koller, S. (2001). Brazilian adolescents’ Lourenço, O. (1997). Children’s attributions of moral emotions to
prosocial moral judgment and behavior: Relations to sympathy, victimizers: Some data, doubts and suggestions. British Journal
perspective taking, gender-role orientation, and demographic of Developmental Psychology, 15, 425–438.
characteristics. Child Development, 72, 518–534. Malti, T., & Buchmann, M. (2010). Socialization and individual
Fredrickson, B. (1998). What good are positive emotions? Review of antecedents of adolescents’ and young adults’ moral motivation.
General Psychology, 2, 300–319. Journal of Youth and Adolescence, 39, 138–149.
Ganzeboom, H. B. G., Degraaf, P. M., Treiman, D. J., & Deleeuw, J. Malti, T., & Keller, M. (2010). The development of moral emotions in
(1992). A standard international socio-economic index of a cultural context. In W. F. Arsenio & E. A. Lemerise (Eds.),
occupational status. Social Science Research, 21, 1–56. doi:10. Emotions, aggression, and morality in children: Bridging
1016/0049-089X(92)90017-B. development and psychopathology (pp. 177–198). Washington:
Gibbs, J. C., Basinger, K. S., Grime, R. L., & Snarey, J. R. (2007). American Psychological Association.
Moral judgment development across cultures: Revisiting Kohl- Malti, T., Keller, M., & Buchmann, M. (2012). Do moral choices
berg’s universality claims. Developmental Review, 27, 443–500. make us feel good? The development of adolescent’s emotions
Graziano, W. G., & Tobin, R. (2009). Agreeableness. In M. R. Leary following moral decision-making. Journal of Research on
& R. H. Hoyle (Eds.), Handbook of individual differences in Adolescence, 23, 389–397. doi:10.1111/jora.12005.
social behavior (pp. 46–61). New York: Guilford. Malti, T., & Krettenauer, T. (2013). The relation of moral emotion
Hardy, S. A., & Carlo, G. (2011). Moral identity: What is it, how does attributions to pro- and antisocial behavior: A meta-analysis. Child
it develop, and is it linked to moral action? Child Development Development, 84, 397–412. doi:10.1111/j.1467-8624.2012.01851.x.
Perspectives, 5, 212–218. Malti, T., & Ongley, S. (in press). The development of moral emotions and
Harris, P. L. (1989). Children and emotion: The development of moral reasoning. In M. Killen, & J. Smetana (Eds.), Handbook of
psychological understanding. Cambridge, MA: Blackwell. moral development. (2nd ed.). New York: Taylor & Francis.

123
 J Youth Adolescence

McCrae, R. R., & Costa, P. T. (2004). A contemplated revision of the Tracy, J., & Robins, R. W. (2007). The self in self-conscious
NEO five-factor inventory. Personality and Individual Differ- emotions. In J. Tracy, R. W. Robins, & J. P. Tangney (Eds.), The
ences, 36, 587–596. self-conscious emotions: Theory and research (pp. 3–20). New
Miller, J. G., Chakravarthy, S., & Rekha, D. (2008). The moral York: Guilford.
emotions of guilt and satisfaction: A cross-cultural perspective. Warneken, F., & Tomasello, M. (2009). The roots of human
European Journal of Developmental Science, 2, 236–250. altruism. British Journal of Psychology, 100, 455–471. doi:10.
Miller, J. G., Rekha, D., & Chakravarthy, S. (2011). Culture and the 1348/000712608X379061.
role of choice in agency. Journal of Personality and Social Weiß, R. (1998). CFT-20. Grundintelligenztest (culture fair intelli-
Psychology, 101, 46–61. doi:10.1037/a0023330. gence test). Goettingen: Hogrefe.
Nunner-Winkler, G. (2007). Development of moral motivation from Wintre, M., & Vallance, D. D. (1994). A developmental sequence in
childhood to early adulthood. Journal of Moral Education, 36, the comprehension of emotions: Intensity, multiple emotions,
399–414. and valence. Developmental Psychology, 30, 509–514.
Nunner-Winkler, G. (2009). Moral motivation from childhood to Zhou, Q., Valiente, C., & Eisenberg, N. (2003). Empathy and its
early adulthood. In W. Schneider & M. Bullock (Eds.), Human measurement. In S. J. Lopez & C. R. Snyder (Eds.), Positive
development from early childhood to early adulthood (pp. psychological assessment: A handbook of models and measures
91–118). New York: Psychology Press. (pp. 269–284). Washington: American Psychological Association.
Nunner-Winkler, G., Meyer-Nikele, M., & Wohlrab, D. (2007).
Gender differences in moral motivation. Merrill-Palmer Quar-
terly, 53, 26–52. Author Biographies
Paciello, M., Fida, R., Tramontano, C., Lupinetti, C., & Carprara, G. V.
(2008). Stability and change of moral disengagement and its
impact on aggression and violence in late adolescence. Child Tobias Krettenauer is Professor in the Department of Psychology at
Development, 79, 1288–1309. Wilfrid Laurier University. He received his doctorate at the Free
Piaget, J. (1932/1965). The moral judgment of of the child. New York: University of Berlin, Germany. His research focuses on the devel-
Free Press. opment of morality, self and emotions in children and adolescents.
Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006). Computational
tools for probing interactions in multiple linear regression, Tyler Colasante is a graduate student in the Department of
multilevel modeling, and latent curve analysis. Journal of Psychology at the University of Toronto. He received his undergrad-
Educational and Behavioral Statistics, 31, 437–448. uate training in Psychology at the University of Toronto. His research
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R., & Du interests include moral emotions, socio-emotional development,
Toit, M. (2011). HLM 7: Hierarchical linear and nonlinear antisocial behavior, emotion regulation strategies, and psychophys-
modeling. Lincolnwood, IL: Scientific Software International. iological techniques.
Saelen, C., & Markovits, H. (2008). Adolescents’ emotion attributions
and expectations of beahvior in situations involving moral Marlis Buchmann is Professor of Sociology and Director of the
conflict. Journal of Experimental Child Psychology, 100, 53–76. Jacobs Center for Productive Youth Development at the University of
Schneider, W., & Bullock, M. (Eds.). (2009). Human development Zurich. She received her doctorate in Sociology from the University
from early childhood to early adulthood. New York: Psychology of Zurich. Her research interests include social inequality and the life
Press. course, with a special emphasis on the life stage youth, school-to-and
Stets, J. E., & Carter, M. J. (2012). A theory of the self for the work transitions, and the transition to adulthood, socialization and
sociology of morality. American Sociological Review, 77, competence development.
120–140.
Svetlova, M., Nichols, S. R., & Brownell, C. A. (2010). Toddlers’ Tina Malti is an Associate Professor in the Department of
prosocial behavior: From instrumental to empathic to altruistic Psychology at the University of Toronto. She received her doctorate
helping. Child Development, 81, 1814–1827. in Developmental Science from the Max Planck Institute for Human
Swiss Federal Office of Statistics. (2012). Bildungsstatistik 2011 Development and Free University of Berlin. Her research interests
(Education statistics 2011). Neuchâtel: Swiss Federal Office of include moral emotions, social-emotional development, aggression
Statistics. and prosocial behavior, and interventions to promote mental health,
Tangney, J. P., Stuewik, J., & Mashek, D. J. (2007). What’s moral especially in school and out-of-school time settings.
about self-conscious emotions? In J. Tracy, R. W. Robins, &
J. P. Tangney (Eds.), The self-conscious emotions: Theory and
research (pp. 21–37). New York: Guilford.

123