GUILT, EMPATHY, AND COMPLIANCE IN A NATURALISTIC MORAL SCENARIO:
PREDICTING PROSOCIAL AND EXTERNALIZING BEHAVIOR IN 3-7-YEAR-OLD
CHILDREN
By
Caitlin J. Listro

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Psychology—Doctor of Philosophy
2017

ABSTRACT
GUILT, EMPATHY, AND COMPLIANCE IN A NATURALISTIC MORAL SCENARIO:
PREDICTING PROSOCIAL AND EXTERNALIZING BEHAVIOR IN 3-7-YEAR-OLD
CHILDREN
By
Caitlin J. Listro
Current science offers only limited answers concerning the development of empathy
disturbances. Indeed, few studies have attempted to empirically identify the developmental
trajectory of empathy to define either normal or aberrant developmental patterns. The present
study aimed to use an observational approach to assess empathy, guilt, and obedience in
children, and to validate this approach by exploring how these observational measures of child
moral behavior associated with characteristics theoretically linked to moral development and
antisocial behavior. We utilized a videotaped Picture Tearing task in which the child is
presented with a moral dilemma. Trained coders rated the tasks for several child behaviors (e.g.
guilt, gaze avoidance, defiance) using a coding scheme adapted from the Lab-TAB (Goldsmith et
al., 1993). Variations in moral behavior were investigated using person-centered (cluster
analysis) and variable-centered (factor analysis) methods, then associations between resulting
behaviors and other relevant child characteristics (temperament, externalizing behaviors) were
examined concurrently and over time. In general, results indicated that empathic verbalizations
and defiance were consistently associated with externalizing pathology. This association was
observed concurrently; empathy did not predict externalizing over time. Overall, these results
suggest that compliant without complaint is the most adaptive response at this age. Furthermore,
the Picture Tearing task does provide useful data about empathic behavior and its associations in

young children. Recommendations are made for adaptations to the task and coding scheme to
improve the measurement of moral behavior in future research.
Keywords: guilt, empathy, compliance, moral emotions, moral behavior

ACKNOWLEDGEMENTS

It is difficult to quantify the importance of other people in the process of proposing,
creating, and passing a dissertation project. They are invaluable, essential. First and foremost, I
would like to thank my adviser, Emily Durbin, without whom I might be somewhere else now, in
another life with only a master’s degree and an uncertain career path. I cannot adequately express
what her support has meant to me. I am also indebted to my committee members, Alytia
Levendosky, Brooke Ingersoll, and Jenna Neal, for their time and thoughtful consideration. I am
doubly indebted to David Clark and Allison Gornik, my wonderful labmates, who provided
emotional and statistical assistance; to Taryn Smith, who was integral in the development of the
coding scheme; and to Evan Tacey, Madeleine Lenhausen, and Amanda Kilgore, who all logged
countless hours coding videos of children being asked to do patently horrible things.
A world of thanks to my amazing family, my parents Christine and Sammy Listro and
Sarah, Debbie, and Jim Posey. You are my world. To my friends and cohort mates who sat up
long hours with me or just made me smile through the hardest parts: Katey Smagur, Britny
Hildebrant, Aurora Dixon, Amber Mandalari, David Johnson, Anna and Andrew Lennard,
Allison Gornik, Nicola Bernard, Erika Vitale, Alana Harrison, Rachael Goodman.
To my dog, Bailey, whose positivity always energizes me. And to my cat, Nemo, who
only occasionally complained while I was in the midst of dissertation chaos.

iv

TABLE OF CONTENTS

LIST OF TABLES

ix

LIST OF FIGURES

x

INTRODUCTION
1
The Development of Empathy
3
Historical Models of Empathy Development
3
Prototypical Timeline for the Development of Empathic Skills in Early Childhood
4
Empathic Skills and The Mechanisms Underlying Their Development
8
Affective Empathy: The Role of Arousal
9
Dispositional differences in arousal.
9
Environmental modulation of arousal.
11
Cognitive Empathy: Perspective Taking
12
Emotion recognition.
12
The Social Environment: Obedience and Compliance
14
A History of the Scientific Study of Obedience
16
Obedience in Children
19
Changes in Compliance Across Development
24
The Present Study
25
Aims of the Present Study
27
Overarching Aim: Validate the picture tearing coding scheme as an assessment of
child moral development characteristics.
27
Specific Subordinate Aims
28
Aim 1: Describe variations in empathic responding in a morally relevant
scenario.
28
Aim 2: Examine associations between observed moral behavior and child
characteristics theoretically linked to moral development and empathy.
29
Aim 3: Examine predictive associations between compliance, arousal, and
empathy and later externalizing behaviors.
30
METHOD
Overview of Design
Participants
Laboratory Assessment of Temperament
Exploring New Objects (Durbin, 2010; Fear, happiness).
Making a t-shirt (Durbin, 2010; engagement, happiness).
Stranger approach (Lab-TAB; Goldsmith et al., 1993; Fear).
Impossibly perfect green circles (Lab-TAB; Goldsmith et al., 1993; Anger,
sadness).
Popping bubbles (Lab-TAB; Goldsmith et al., 1993; Activity, happiness).
Diorama snakes (Fear).
Snack delay (Lab-TAB; Goldsmith et al., 1993; Effortful control).
v

31
31
32
32
33
33
33
33
33
33
34

Picture tearing (Goldsmith et al., 1995; Guilt, empathy, sadness, compliance). 34
Balloon bop (Goldsmith & Rothbart, 1996; Effortful control, happiness).
35
Transparent box (Lab-TAB; Goldsmith et al., 1993; Anger, sadness).
35
Simon says (Strommen, 1973; Effortful control).
35
Tell a story (Durbin et al., 2007; Fear).
35
Pop-up snakes (Lab-TAB; Goldsmith et al., 1993; Anticipatory positive affect,
happiness, surprise).
36
Walk-a-line slowly (Kochanska et al., 1996; Effortful control).
36
Box empty (Lab-TAB; Goldsmith et al., 1993; Anticipatory positive affect, anger,
sadness).
36
Coding of Child Temperament Laboratory Tasks
36
Observational coding of picture tearing task.
36
Observational coding of laboratory tasks.
37
Other Tasks
39
Emotion recognition training.
39
Peabody Picture Vocabulary Test (PPVT).
40
Measures
40
Child Behavior Checklist (CBCL).
40
Children’s Behavior Questionnaire (CBQ).
42
Experimenter ratings of child traits.
42
Alabama Parenting Questionnaire (APQ).
43
Data Analytic Plan
43
Aim 1: Describe variations empathic responding in a morally relevant scenario.
43
Factor analysis.
43
Cluster analysis.
44
Aim 2: Examine associations between observed moral behavior and child
characteristics theoretically linked to moral development and empathy.
45
Demographic Variables.
48
Cognitive characteristics.
50
Temperamental characteristics.
50
Parenting factors.
53
Aim 3: Examine predictive associations between compliance, arousal, and
empathy and later externalizing behaviors.
54

RESULTS
56
Recoding of Picture Tearing Variables
56
Bivariate Correlations Among Recoded Variables
57
Associations Between Task Behaviors and Demographic Variables and Environment
Characteristics
59
Age and sex.
60
Cognitive characteristics.
61
Parenting characteristics.
61
Overall.
62
Exploratory Factor Analysis
63
vi

Confirmatory Factor Analysis of Empirically Derived Factors
Confirmatory Factor Analysis of Rationally Derived Factors
Final Factor Analysis
Summary of Factor Analysis
Associations Between Picture Tearing Task Composites and Other Child and
Environment Characteristics
Demographic variables.
Cognitive characteristics.
Parenting variables.
Overall.
Derivation of Clusters
Hierarchical cluster analysis.
K-means cluster analysis.
Description of clusters in three cluster solution.
Cluster 1: Moderate compliers.
Cluster 2: Low compliers.
Cluster 3: High compliers.
External validation.
Demographics.
IQ and emotional intelligence.
Temperament traits.
Problem behaviors.
Parenting behaviors.
Overall.
Gender of Victim as a Predictor of Child Task Behaviors
Prediction of Concurrent Problem Behaviors Using Composites Derived from Factor
Analysis
Correlations between independent variables and dependent outcomes.
Picture tearing composites.
Demographic variables.
Verbal and emotional ability.
Temperament traits.
Parenting characteristics.
Overall.
Hierarchical regressions predicting problem behaviors from task behavior composites.

64
65
66
67
67
68
68
68
69
69
69
71
71
72
72
72
73
74
74
75
76
76
76
77
78
78
78
78
79
79
80
80
81

Hierarchical regressions controlling for other child and environment characteristics.
Summary of hierarchical regressions using task composites.
Prediction of Concurrent Problem Behaviors Using Individual Task Variables
Correlations between independent variables and dependent outcomes.
Hierarchical regressions predicting problem behaviors from task behavior.
Hierarchical regressions controlling for other child and environment
characteristics.
Summary of hierarchical regressions using individual task variables.
Prediction of Concurrent Temperament Traits Using Picture Tearing Variables
vii

81
82
83
83
83
84
84
85

Correlations between independent variables and dependent outcomes.
Hierarchical regressions predicting temperament traits from task behavior.
Summary of regressions predicting temperament traits from task behaviors.
Longitudinal Prediction of Problem Behaviors Using Composites Derived from Factor
Analysis
Summary of longitudinal prediction of problem behaviors.

85
87
90
92
94

DISCUSSION
95
Confirmation and Extension of Previous and Hypothesized Findings
96
Unexpected Findings
106
Conclusions and Areas for Future Study
112
Validity of the picture tearing coding system and recommendations for future use.
112
Limitations and future applications.
116
Implications for the study of moral development.
117

APPENDICES
APPENDIX A: FIGURES
APPENDIX B: TABLES
APPENDIX C: Original picture tearing coding scheme
APPENDIX D: Description of clusters in four cluster solution
APPENDIX E: Recommended final picture tearing coding scheme

126
127
137
181
185
187

REFERENCES

192

viii

LIST OF TABLES

Table 1. Intraclass correlation coefficients for video coding variables

139

Table 2. Picture Tearing codes relevant to each epoch of the task

140

Table 3. Comprehensive list of hypotheses

142

Table 4. Hypothesized results of factor analysis

148

Table 5. Hypothesized associations for variables investigated at baseline

149

Table 6. Descriptive statistics for picture tearing variables

150

Table 7. Descriptive statistics for recoded picture tearing variables

152

Table 8. Bivariate correlations among picture tearing variables

154

Table 9. Correlations between picture tearing variables and child characteristics

156

Table 10. Correlations between picture tearing variables and parenting characteristics

158

Table 11. Factor structures based on exploratory factor analysis

161

Table 12. Fit statistics and comparative analyses for confirmatory factor analyses

162

Table 13. Factor loading estimates for final five factor model

163

Table 14. Descriptive statistics for three cluster solution

164

Table 15. Correlations between task variables, child and environmental characteristics, and
problem behaviors
166
Table 16. Correlations between picture tearing variables and temperament traits

170

Table 17. Hierarchical multiple regression analyses predicting current child problem behaviors
from picture tearing task composite variables
173
Table 18. Hierarchical multiple regression analyses predicting current child problem behaviors
from single picture tearing task variables
176
Table 19. Hierarchical multiple regression analyses predicting current child temperament traits
from picture tearing task composite variables
177

ix

LIST OF FIGURES

Figure 1. Diagram of moral development and its relationship to child internal and external
characteristics and secondary outcomes
128
Figure 2. Chart of Video Progression and Relevant Codes

129

Figure 3. Hierarchical cluster analysis agglomeration coefficients plotted against stage of
analysis
130
Figure 4. Gap statistic for hierarchical cluster analysis graphed against number of clusters
131
Figure 5. Differences between cluster means for three cluster solutions

x

132

INTRODUCTION
In the wake of the Holocaust, Stanley Milgram set out to test the prevailing public notion
that the Nazi officers who perpetrated genocide were simply evil. He famously found that, when
commanded by an experimenter, the vast majority of adults in his study would obey orders to
electrically shock another person using what they believed were dangerous levels of voltage
(Milgram, 1965). As Milgram demonstrated, actual moral or immoral behavior is not solely
determined by one’s internalized morals. In understanding antisocial behavior, it is thus crucial
to understand the processes by which morality develops and factors that predict moral decisions
in various situations. It is also crucial to understand the qualities that distinguish typical moral
development from aberrant development, and subsequently the processes by which aberrant
development of factors related to moral behavior occurs. In the child literature, this problem is
especially relevant to understanding children who appear to lack moral feelings altogether: those
with Callous-Unemotional (CU) Traits.
CU traits in children encompass a constellation of features characterized by lack of guilt,
lack of empathy, callous use of others for one’s own gain, and underreactivity to threatening and
emotionally distressing stimuli (Frick et al., 2003; Frick & Morris, 2004; Frick, Ray, Thornton,
& Kahn, 2014; Frick & White, 2008). Children with CU traits are unique among antisocial
youth in that they display blunted affectivity, absent remorse, and severe impairments in
empathy. These empathy impairments are severe compared not only to healthy controls, but also
to other individuals with antisocial behavior problems (e.g. de Wied, van Boxtel, Matthys, and
Meeus, 2011). Even in children who may not reach threshold for CU traits, low empathy is
robustly and strongly associated with antisocial behavior and violence (e.g. Asendorpf &

1

Nunner-Winkler, 1992; Hastings et al., 2000; Jolliffe & Farrington, 2004; Krettenauer,
Asendorpf, & Nunner-Winkler, 2013; Lovett & Sheffield, 2007).
Despite the preponderance of research linking impaired empathy to aggression and
violence, current science offers only limited answers concerning the development of empathy
disturbances. Indeed, few studies have attempted to empirically identify the developmental
trajectory of empathy to define either normal or aberrant developmental patterns. Moral
development and decision making is also extremely complex (see Figure 1). It depends not only
on moral emotions such as empathy, but also on cognitive skills and external influences, such as
the actions of victims or authority figures. The current literature provides a limited
understanding of how these factors interrelate to produce behaviors in observable moral
situations. The present study aims to use an observational approach to assess empathy, guilt, and
obedience in children, and to validate this approach by exploring how these observational
measures of child moral behavior associated with characteristics theoretically linked to moral
development and antisocial behavior, including child temperament traits and externalizing
problems (both concurrently and prospectively) and aspects of the parent-child relationship. We
utilized a videotaped task in which the child is presented with a moral dilemma: obey the adult
experimenter and destroy someone’s “cherished” item, or preserve the item and defy the adult
authority. In this way, we hoped to clarify individual differences in empathy and obedience as
markers of moral behavior, and to test for the construct validity of these observed individual
differences by quantifying their association with factors presumed to play a causal role in
children’s moral development and with outcomes presumed to mark abnormal (versus normal)
moral development, including antisocial behavior and callous-unemotional traits.
The Development of Empathy

2

Historical Models of Empathy Development
Initial approaches to moral development delineated universal processes that were thought
to unfold along approximately the same timeline and the same stages for every child. Both
Piaget (1965) and Kohlberg (Kohlberg & Kramer, 1969) suggested that children first
demonstrate simplistic rule-responsiveness based on the dictates of authority figures and the
threat of punishment, and only later learn to think independently and relativistically about issues
of morality and to develop personal values to draw upon when engaging in moral decisionmaking. These models viewed young children as pragmatic, egocentric, pre-moral agents
responding to external strictures with relative insensibility to the reasoning behind their actions
(Zahn-Waxler, Radke-Yarrow, Wagner, & Chapman, 1992). They also identified the emergence
of empathy or guilt (critical components of moral awareness and moral behavior) as not
occurring until late childhood at the earliest. However, empirical research suggests that
primitive forms of moral emotions (i.e. empathy, guilt) can be observed in children as young as
12 months of age (Zahn-Waxler, Radke-Yarrow, Wagner, & Chapman, 1992). Indeed, children
as young as 3 months of age are capable of recognizing goal-directed activity, even in inanimate
objects (Hamlin, Wynn, & Bloom, 2010; Shimizu & Johnson, 2004), and demonstrate aversion
to objects perceived as acting antisocially, but preference towards objects perceived as acting
prosocially (Hamlin, Wynn, & Bloom, 2010). Thus, while they may not be able to articulate
moral principles in a sophisticated manner, they still demonstrate a basic awareness of ‘right’
and ‘wrong’ and the ability to evaluate personified entities based on these principles.
In contrast with early stage models of Piaget and Kohlberg, modern theories of moral
development describe a fluid trajectory that begins at birth (Hoffman, 1982; Kochanska, 1993).
Both theory and empirical work suggest that this process does not occur on a fixed timeline with

3

invariant stages, but rather, is affected by continual transactions between the child’s disposition
and his or her environment, producing individual differences in important markers of moral
behavior and reasoning. Modern empathy theory asserts that, at the most basic level, humans are
biologically prepared to develop empathy (Hoffman, 1982). Hoffman (1982) suggested that
children develop empathy first in infancy as mimicry of others’ affective expressions, and
subsequently develop the internalization of rule-oriented messages delivered by caregivers or
parents. “Internalization” is the key process in this scenario, by which the child accepts parental
strictures as their own beliefs. Eventually, the child bases their conduct on intrinsic, absorbed
messages rather than external threats of punishment (Rose, 1999). While the progression of
development described here is similar to that proposed by Piaget (1965) and Kohlberg (Kohlberg
& Kramer, 1969), it differs in that it does not view children as purely authority-driven and
insensitive to higher levels of moral reasoning. Rather, it posits that children internalize the
moral messages suggested by adult rules (e.g. do not hit your sister because it hurts her) rather
than simply following rules without awareness of the reasoning behind them (Zahn-Waxler,
Radke-Yarrow, Wagner, & Chapman, 1992).
Prototypical Timeline for the Development of Empathic Skills in Early Childhood
The hypothesis that empathy in humans is biologically predisposed is supported by a
wealth of evidence concerning the prototypical development of empathy. While research on the
empathic development in preverbal children is limited by methodological constraints, available
studies support that the development of empathic skills occurs as soon as a child is born. Even
infants 2-3 days old exhibit reflexive crying upon hearing the cries of other infants (Field,
Woodson, Greenberg, & Cohen, 1982; Sagi & Hoffman, 1976). Reflexive crying may reflect a
biological disposition for infants to respond to the distress of others. It is also unlikely that

4

infants are merely responding to a noxious stimulus (the sound of crying) with distress, given
that the infants preferentially cried in response to other infants’ crying in comparison not only to
their own pre-recorded cries, but also to other loud sounds (Field, Woodson, Greenberg, &
Cohen, 1982). These data support the contention that infants are predisposed to attend to and to
exhibit ‘empathic’ responses to the distress of others (Sagi & Hoffman, 1976).
In further support of this notion, studies of infants have found that infants displayed
negative emotional arousal in response to the negative affective states of their depressed mothers
(Cohn, Campbell, Matias, & Hopkins, 1989), to their mothers’ simulated distress, and to videos
of distressed peers (Roth-Hanania, Davidov, & Zahn-Waxler, 2011) From a developmental
standpoint, these data indicate that infants are aware of and responsive to the distress of others
much earlier than previously believed.
Expressions of empathy analogous to those found in adults are first identifiable around 18
months of age, with some expressions of mature empathy observable around 24 months
(Belacchi & Farina, 2012). During this period, children learn the function of symbols and are
increasingly capable of representational thought; they begin to understand hypotheticals and to
mentally represent complex information (e.g. Bruner, 1972; Piaget, 1932). The onset of these
skills facilitates nascent perspective taking abilities, which are predicated on the ability to
mentally represent the emotions and desires of others. Evidence for perspective taking has been
observed in the form of toddlers’ responses to others’ distress (Zahn-Waxler, et al., 1992). ZahnWaxler and colleagues (1992) demonstrated that many children in their sample as young as
twelve months exhibited attempts to comfort another person in distress; by 24 months, almost all
children exhibited comforting attempts. These responses, while relatively simplistic, suggest a

5

form of empathic role taking, in that children are recognizing the other person’s distress and are
responding as they might like to be comforted in a similar situation (Zahn-Waxler et al., 1992).
The age of approximately 24 months is a particularly important period in empathy
development. At this time, the child is developing both increased autonomy and sophisticated
differentiation of self and other (Zahn-Waxler et al., 1992), which may enable them to
distinguish between their own desires and goals and those of another person when these are in
conflict. Children at this age are also developing the emotional capacity to recognize and
experience others’ emotional states, the cognitive capacity to interpret others’ emotional states,
and the behavioral capacity to identify appropriate responses and to act (Zahn-Waxler et al.,
1992). Consistent with this notion, in the study of toddlers discussed above, increasing age was
related to a greater sophistication, variety, and appropriateness of comforting attempts (ZahnWaxler et al., 1992). Simply put, as children developed a mature theory of mind, they were able
to identify comforting behaviors tailored to the other person, presumably via rudimentary
perspective taking.
Increased autonomy and mobility also facilitate the processes of socialization and the
internalization of parental restrictions (Kochanska, 1993). For the first time, children are able to
perform behaviors that may run counter to parental desires, some of which will incur parental
prohibitions. Consequently, expressions of guilt following a transgression also emerge around
this time (e.g. Aksan & Kochanska, 2005; Kochanska & Aksan, 2006). While difficult to assess,
guilt in children has typically been operationalized via behavioral indicators, including: gaze
avoidance, apologizing, bodily tension, negative affect, and self-blame (Kochanska & Aksan,
2006). Guilt generally results from empathic awareness of another’s distress in concert with the
awareness that one has caused this distress via a transgression (Hoffman, 1975). Thus, guilt is an

6

important epiphenomenon of empathy. While measurement limitations prohibit clarity in the
question of whether children at this time feel empathy per se, the aforementioned research
suggests that at least primitive empathy characterized by an affective component (i.e.
recognizing the other’s distress, manifesting guilt) and a cognitive component (i.e. identifying
what the other might desire as comfort in the situation) can be observed in children of this age.
The next critical milestone occurs approximately between 24 and 36 months, when
children first demonstrate an ability to obey rules in the absence of a parent (Kochanska, 1993).
Prior to this milestone, children typically desist in transgressing only because a caregiver is
present to induce anxiety and guilt or to redirect the child to other activities. However, by age
two or three years, most children have internalized previous instances of parental discipline and
can mentally represent these rules in the absence of parental instruction. At this age, anxiety in
response to transgression is not only concurrent but anticipatory (i.e. in response to hypothetical
future actions). This anticipatory anxiety inhibits the enactment of future anxiety-causing
transgressions.
In the early preschool to kindergarten years (about ages 3 to 6 years), children make
significant gains in the ability to recognize different emotions and to make basic inferences about
what another person is or might be feeling (Borke, 1971, 1973). Illustratively, Borke (1971,
1973) found that children aged 3 to 6 years demonstrated high accuracy in identifying the
emotions of characters in vignettes specially matched to certain emotions. Children of age 3
were 90% successful in identifying “happy” as the emotion experienced by characters in a
“happy” story. Rates for correctly identifying “sad” were lower than for “happy” but still
significantly higher than predicted by chance, while correct recognition of “fearful” and “angry”
was low for younger children but increased across age groups. Cumulatively, children have, by

7

the school-age years, developed an ability to recognize emotions, to identify situations that might
elicit certain emotions for others, to recognize when others might require comfort.
Empathic Skills and The Mechanisms Underlying Their Development
Given the widespread use of low risk, unselected samples in studies of empathic
development, the process of socialization described above likely best represents a prototypical
process that occurs under relatively ideal circumstances. The dominant theory of development in
the field considers empathy development in relation to cognitive development and to morally
relevant environmental contexts in which children develop components of conscience, and which
may differ substantially between children (Kochanska, 1993). Contingencies in the child’s
disposition or environment can disrupt this process and result in abnormal or absent empathic
functioning. While the literature typically parses these mechanisms roughly into the categories of
affective skills (vicarious distress) and cognitive skills (perspective taking), in practice, these
skills interact reciprocally to produce mature empathic responses.
In general, empathic internalization occurs as a result of physiological arousal in response
to environmental contingencies, including the distress of others and parental prohibition
(Kochanska, 1991; Kochanska, 1993). Ideally, the child experiences enough arousal after a
transgression that the parental message regarding the transgression is salient and becomes more
easily internalized, if not always consciously remembered. Over time, avoidance learning
occurs: the child learns to associate various transgressive acts with arousal, which typically takes
the aversive forms of anxiety or discomfort (Rose, 1999). The child wishes to avoid distress, so
in the future, he or she inhibits the behaviors associated with the distress (Eisenberg, Eggum, &
Edwards, 2010). Later in the developmental process, simple anxiety-based avoidance learning is
paired with internalized and self-chosen principles that motivate moral actions.

8

Affective Empathy: The Role of Arousal
Appropriate arousal is perhaps the most fundamental precondition for the development of
vicarious distress, which is considered the most basic component of affective empathy. Multiple
discussions of empathy impairment suggest that atypical development can occur if the arousal
experienced by the child is excessively low or excessively high. These extremes may exist due
to congenital dispositions, environmental pressures, or a combination thereof.
Dispositional differences in arousal. Perhaps the best studied example of dispositional
effects on socialization is the case of the fearful versus fearless child (Kochanska, 1991, 1993).
Fearfulness is a dimension of the higher-order trait of neuroticism that inhibits the potentiation of
behaviors (Kochanska & Aksan, 2006). Fearful children avoid exploration in laboratory tasks,
while fearless children appear bold and uninhibited (Kagan, 1994; Kagan, Reznick, & Snidman,
1987). Fearful children are also especially likely to experience anxious arousal and to fear reexperiencing that discomfort and incurring consequences (Stifter, Cipriano, Conway, & Kelleher,
2009). Consequently, fearfulness has been linked to appropriately high levels of empathy and
moral conduct. Empirical evidence confirms that fearful children are less likely to cheat on tests
(Asendorpf & Nunner-Winkler, 1992), experience more empathy/guilt in response to a
transgression (Kochanska, Gross, Lin, & Nichols, 2002), and exhibit stronger internalization of
rules and standards (Kochanska, 1995; Kochanska, Coy, & Murray, 2001) than fearless children.
In fact, high levels of fearlessness have been associated with impaired acquisition of
empathic skills and higher levels of externalizing behaviors (Kochanska, 1991, 1993). This
mechanism putatively underlies the origin of poor empathy in children with CU traits, who also
demonstrate high fearlessness (e.g. Frick et al., 2014; Frick & White, 2008; Frick, Ray,
Thornton, & Kahn, 2014; Frick et al., 1999). Children high on fearlessness are less likely to

9

experience physiological arousal or empathic arousal in response to distress cues in other people,
whether they or another person are the source of the distress (Blair, 1999). If a child experiences
too little arousal in response to parental prohibitions, then the prohibition will not be salient to
them. They are less likely to attend to and internalize parental messages and are unlikely to be
deterred from transgressing, because they fear neither the typical discomfort experienced by
others nor the consequences of breaking rules. In support of this notion, high scores on measures
of fearlessness predict lower scores on measures of conscience both concurrently (e.g.
Kochanska, Gross, Lin, & Nichols, 2002) and longitudinally (Rothbart, Ahadi, Hershey, &
Fisher, 2001). In addition, both prospective and cross-sectional work supports a link between low
anxiety or physiological arousal and poor emotional empathy, few behavioral expressions of
concern towards others, and low levels of guilt following a transgression (Kochanska, 1995;
Kochanska, Gross, Lin, & Nichols, 2002; Rothbart, Ahadi, & Hershey, 1994).
While high degrees of fearlessness seem to be maladaptive, too much fearfulness can also
be detrimental for moral development in a different way. If the child experiences too much
personal distress after a transgression, they may focus more on their own distress than on the
parental message, and may consequently avoid transgressions not because they fear harming
others, but because they fear their own distress (Eisenberg et al., 2010). In this case, the distress
they feel is self-focused rather than other-focused and may encourage withdrawal instead of a
reparative or prosocial response. Over time, empathic concern may become blunted or
eliminated by the predominant feelings of personal distress.
Cumulatively, the data concerning fearfulness predict an inverted U-shaped curvilinear
association between fear and empathy, such that children with moderate levels of fearfulness

10

demonstrate the highest empathy, while children excessively high or low on fearfulness
demonstrate low or impaired empathy.
Environmental modulation of arousal. While temperamental variations in arousal are
obvious sources of variation in empathic responding, these variations are likely influenced by the
environment as well as innate dispositional differences among children. One of the most obvious
sources of environmental moderation operates via parenting. Parenting factors can positively or
adversely affect the socialization process. Parental disciplinary styles can disrupt socialization in
a manner similar to the previous example of the highly fearful child. The use of excessive or
excessively harsh punishment may cause the child to associate their discomfort with the
punishment rather than the transgression. Instead of reinforcing the desistance of transgression,
this process reinforces the avoidance of punishment. Ultimately, this process conditions the
child to feel self-focused personal distress rather than other-focused empathy (Kochanska, 1993).
Empirically, authoritarian parenting based on power and harsh discipline is correlated with
deficits in the child’s later internalization of rules (Kochanska, 1991). Indeed, for highly anxious
children, maternal power assertion predicted fewer prosocial responses to a series of vignettes
involving hypothetical transgressions (Kochanska, 1991).
Conversely, warm and responsive parenting is associated with more ingrained
internalization of rules and empathic concern (Kochanska, 2002). Attachment appears to
moderate the effects of fearlessness, such that fearless children who demonstrate secure
attachment to their caregivers exhibit less aggression and rule-breaking than children who are
insecurely attached (Kochanska, 1995). In this case, children internalize parental messages
because they are emotionally close with their parents and desire to please and to emulate them,
rather than internalizing prohibitions through traditional avoidance learning. Similarly, a

11

mutually-responsive orientation (in which parents and children have a relationship based in
shared positive affect and cooperation) has been linked to more mature expressions of
conscience and more prosocial behavior (Kochanska, 2002).
Given this evidence, ideal socializing discipline is minimally punitive and characterized
by warmth and shared positive emotion. Highly fearful children may require gentler discipline,
while insufficiently fearful children may require socialization pressure that is stronger but still
warm and fair (Dienstbier, 1984). However, exceptions to this model exist empirically.
Kochanska (1991) found that for low-fear children, maternal power assertion had few reliable
associations with measures of conscience. It is unclear whether this null finding represents a
restriction of range (in that effects can only be observed at extremes of low fear) or whether
power assertion is only problematic for fearful children who experience too much arousal, while
power assertion at levels observed in this study is insufficient to elicit even adequate levels of
arousal from low-fear children.
Cognitive Empathy: Perspective Taking
Though arousal is the key piece in the development of affective empathy, it is only one
component of a mature empathic response. Deficits in affective empathy seemingly inhibit
children from developing cognitive reactions to others’ emotions, which impairs the
development of cognitive empathy. Over time, these deficits likely operate reciprocally to
solidify a pattern of poor empathic responding.
Emotion recognition. The best-studied aspect of emotional processing with respect to
empathy is emotion recognition. Emotion recognition is a complex skill that involves an
understanding of external features (e.g. facial expressions, situational causes), mental features
(e.g. desires, motivations), and cognitive complexities of emotions (e.g. mixed emotions, mental

12

control of emotions, ability to suppress emotional expression) (Pons, Harris, & de Rosnay,
2004). Emotion recognition underlies the individual’s basic ability to feel empathic concern
(affective empathy), in that an individual must recognize another person’s distress as distress
before they can react to it. Recognizing emotions also allows an individual to develop schemas
for those situations that elicit certain emotions in other people. This schematic provides the
underpinnings for cognitive empathy (perspective taking). Inadequate emotion recognition
abilities would putatively impair a child’s ability to recognize acts that may distress another
person or those that might repair such distress.
In support of this notion, CU traits in children and adults are associated with widely
documented deficits in the recognition of and emotional responsivity to fear and potentially
sadness (e.g. Blair & Coles, 2000; Blair, Colledge, Murray, Mitchell, 2001; Dadds, Perry,
Hawes, et al., 2006; Munoz, 2009; Blair, Budhani, Colledge, Scott, 2005; Fairchild, van Goozen,
Calder, et al., 2009). These recognition deficits persist across multiple classes of stimuli,
including pictures (Kimonis, Frick, Fazekas, & Loney, 2006), words (Loney, Frick, Clements,
Ellis, & Kerlin, 2003), facial expressions (Dadds et al., 2006), and vocal tones (Blair, Budhani,
College, & Scott, 2005). On the whole, individuals with severe empathy impairments have
difficulty identifying situations in which another person might be afraid or distressed (Marsh &
Cardinale, 2012). On the contrary, children who adequately recognize the emotions of others are
likely to internalize distress as a victim’s potential response and to consider this consequence
before acting in an antisocial manner. It is likely that, as these children do not experience the
distress of others as aversive or arousing, they do not preferentially attend to fear and sadness
stimuli; consequently, they do not adequately learn to identify and distinguish these states
(Dadds et al., 2008; Dadds, Perry, et al., 2006).

13

The Social Environment: Obedience and Compliance
Understanding the development of empathy is important to the extent that empathy
predicts observable social behaviors. The degree to which empathy reliably predicts or must
predict prosocial action, however, is somewhat contested. Empirically, high levels of empathy in
children, as assessed by teacher report, have indeed been associated with rule-following behavior
(Belacchi & Farina, 2012). However, most research supports the contention that moral emotions
and moral behavior only modestly co-occur, particularly in samples of young children (Aksan &
Kochanska, 2005; Kochanska, Aksan, & Nichols, 2003; Kochanska, Forman, Aksan, & Dunbar,
2005; Kochanska, Padavich, & Koenig, 1996; Eisenberg et al., 2010; Hartshorne & May, 1928;
Zahn-Waxler, Radke-Yarrow, Wagner, & Chapman, 1992). Research has reported stronger, but
still modest, associations when situations are analyzed in the aggregate rather than singularly
(e.g. Kochanska, Coy, & Murray, 2001). More consistent is the finding that low levels of
empathic concern positively correlate with criminal offending, bullying, and aggression in
children, adolescents, and adults (e.g. Asendorpf & Nunner-Winkler, 1992; Hastings et al., 2000;
Jolliffe & Farrington, 2004; Krettenauer, Asendorpf, & Nunner-Winkler, 2013; Lovett &
Sheffield, 2007). While most of this work is correlational in nature, a small number of
prospective studies support a causal relationship. For example, Hastings and colleagues (2000)
found that children’s empathic concern at preschool age predicted lower severity and frequency
of externalizing problems at age 6-7 years and lower frequency of externalizing problems at 9-10
years.
Given these findings, recent work by Aksan and Kochanska (2005) and Kochanska and
Aksan (2004) suggests that empathy is one of two correlated, but distinct, components of a
higher-order construct of conscience. These putative components are moral emotion and moral

14

conduct (Aksan & Kochanska, 2005; Kochanska & Aksan, 2004). Moral emotion includes the
related constructs of guilt and empathy. It is the “motivational engine” that gives transgressions
importance to the self and characterizes them as negative and ego-dystonic (e.g. Kochanska &
Aksan, 2006). On the other hand, moral conduct exclusively encompasses external behavior and
the ability to abide by rules (Kochanska & Aksan, 2006). Each component can exist
independently of the other, such that an individual can feel guilt without acting to assuage the
suffering of the victim, or can act to aid the victim without feeling congruent distress.
One issue complicating the relationship between empathy and moral responding is the
fact that children exist in a social environment that continuously places demands on their
behavior. Parents, in particular, are often concerned with empathy and guilt not only because
they desire their children to be prosocial citizens, but because they believe that guilt induces
compliance with parental commands. Thus, modest correlations may occur due to situational
pressures in any given incident, not because moral emotions and moral conduct are linked
internally within the child. For example, studies like Milgram’s (1963, 1965) suggest that
individuals might choose, in a given situation, to obey authority rather than universal principles
of morality.
The role of obedience to authority has always been inextricably linked with moral
development. Indeed, explicit in Piaget’s (1965) and Kohlberg’s (Kohlberg & Kramer, 1969)
theories of moral development is the dependence of early stages of morality on rules and
authority. Even when the authority is not present to give a command, it is implicit in moral
decision making that the child is basing their decision to some extent on the proscriptions of a
hypothesized or internalized authority figure. A child’s decision to act in an immoral or
antisocial manner is a function not only of his or her empathy towards a potential victim, but also

15

of his or her conceptions of the consequences of that action. In this study, our observational
assessment of empathy is uniquely able to disentangle these reasons, given that the child must
struggle with a choice: obey the adult authority and destroy someone’s allegedly cherished item,
which could cause harm to the victim, or refuse to destroy the item to protect the victim, thereby
disobeying the adult authority.
A History of the Scientific Study of Obedience
The spirit of this dilemma precipitated arguably the most striking study of obedience to
authority: that of Stanley Milgram. Milgram’s (1963) obedience experiments definitively proved
that moral feelings and moral conduct do not universally co-occur when an authority is present
and gives commands that conflict with widely held moral principles. In a series of studies, he
invited adult males to the laboratory for an experiment purportedly about the effects of
punishment on memory (see Milgram, 1963). In each case, the subject was paired with a
confederate, and the “random” draw was rigged so that the subject would always be the teacher
and the confederate would always be the learner. The subject was presented with a board of
buttons with labels ranging from 15 volts “mild shock” to 450 volts “severe shock”; this final
button was also labelled “dangerous.” The subject was instructed to ask the learner word pairs
and to provide increasing shocks if the learner answered incorrectly and was encouraged to
continue if he or she protested.
Before conducting the study, Milgram described the experiment to his college class and
asked them to guess what percentage of subjects would continue shocking the learner through the
last level. This pre-test indicated that students expected between 0 and 3% of subjects (mean =
1.2) would deliver the full amount of shock. These predictions vastly underestimated the actual
results. Of 40 participants, none discontinued the experiment before reaching a shock of 300

16

volts. Five subjects refused to continue beyond 300 volts and 14 subjects desisted before the
final level, leaving 26 participants who obeyed the experimenter to the final shock level.
While this first experiment provided evidence that a subject would obey, it did not
empirically investigate when the subject would obey, that is, what circumstances (internal or
external) moderate compliance. In a series of adjustments to the original paradigm, Milgram
investigated a variety of factors that moderated obedience (Blass, 1991). Proximity was a
powerful factor; the more salient and visible the victim, the less likely the subject would obey
until the final shock. Thus, many subjects complied when the learner was in a different room or
provided verbal feedback, but fewer complied when the learner was in the same room and even
fewer complied when they were required to deliver the shock by physically pushing the learner’s
hand onto a shock plate (Milgram, 1965).
Status and salience of the authority figure also moderated compliance. Orders delivered
over the phone decreased compliance (Milgram, 1974). Compliance also decreased when
another “subject” (a confederate) gave the orders. This is consistent with work by Shalala (1974)
in a military setting; subjects were more likely to comply with a higher ranking officer than a
lower ranking officer. Finally, the presence of others altered compliance. Fewer participants
complied when a second teacher (a confederate) was present and desisted, and when two
experimenters were present and gave conflicting orders (stop vs. continue) (Milgram, 1965,
1974).
These early studies gave rise to the notion of “power of the situation” (Gaertner, 1976;
Ross, 1977; Zimbardo, 1974). This concept describes the finding that the demand characteristics
of certain situations constrict behavioral responses more than others—the idea of “strong” versus
“weak” situations. Strong situations tend to suppress the effects of dispositional characteristics;

17

these are situations in which no choices are given and may be imposed rather than freely chosen
(Benjamin & Simpson, 2009; Blass, 1991). Indeed, such emphasis on the power of the situation
led many researchers to conclude that research into individual differences was untenable, and
that situational factors, not putative personality traits, accounted wholly for human behavior (see
Epstein & O’Brien, 1985 for a historical review of the person-situation debate).
Yet, obedience is not wholly environmentally motivated. Despite the high rate of
obedience in Milgram’s (1965) seminal study, a number of participants did not obey the
experimenter to the conclusion of the study; additionally, rates of obedience dropped when
additional factors were introduced (Blass, 1991). In subsequent research, many personality traits
have been associated with obedience. In weaker situations, in which choices exist and behaviors
may be freely chosen, dispositional characteristics have been shown to have stronger effects on
obedience than in strong situations (Benjamin & Simpson, 2009; Snyder & Ickes, 1985). Across
studies, researchers have found that greater compliance with authority is linked to individual
differences in authoritarianism (Elms & Milgram, 1966), trust that the experimenter is benign
and would not do real harm (Mixon, 1976), lower stages of moral development according to
Kohlberg’s stages (Milgram, 1974), lower social intelligence (Burley & McGuinness, 1977), and
low suspiciousness (Holland, 1967). Dispositional aggression has been found to relate to
obedience in studies modeled after Milgram’s shock paradigm when a choice is given (e.g. the
subject may choose which level of shock to give) rather than strict limitations (e.g. subject must
give increasing levels of shock or disobey). Despite evidence supporting strong situations as
powerful determinants of behavior, a multitude of research still supports the importance of
individual differences in predicting human action.

18

Obedience in Children
The role of obedience in the moral actions of children presents an especially complex
case. Whereas adults are presumed to judge moral situations based on a variety of nuanced
factors, strict obedience to authority has historically been considered the primary motivation for
moral behavior in young children. Indeed, Piaget (1965) and Kohlberg (Kohlberg & Hersh,
1977) posited that young children regard rules as obligatory, moral decisions as functions of
reward or punishment, and adults as unilaterally legitimate authority figures to be obeyed
without question (Krahn, 1971).
Yet, research on compliance in children rejects the idea that children obey authority
figures unquestionably or consider adults to be always legitimate moral authorities. In fact, an
accumulation of evidence suggests that children view authority and their obligation to obey (or
defy) in nuanced ways, and are not simply passive recipients of adult commands (Kuczynski,
Kochanska, Radke-Yarrow, & Girnius-Brown, 1987). Nor is all compliance equal in degree.
Kochanska & Aksan (1995; see also Kochanska, 2002) distinguish between committed and
situational compliance. Committed compliance exists when the child complies wholeheartedly
without prompting or reminding, and appears to embrace the maternal (or parental) agenda.
Situational compliance involves the child following directions, but requiring assistance or
reminders to stay on task, seemingly because he or she does not fully embrace the maternal
agenda.
Kochanska & Aksan (1995) posit that these types of compliance may reflect different
motivations for complying, and thus different levels of moral internalization. Across studies,
committed compliance predicts moral internalization, both concurrently and prospectively (e.g.
Forman & Kochanska, 2001; Kochanska & Aksan, 1995; Kochanska et al., 1995; Kochanska,

19

Coy, & Murray, 2001; Kochanska, Tjebkes, & Forman, 1998). On the other hand, situational
compliance is externally regulated and not associated with internalization (Kochanska, 2002).
Qualitative research has illuminated the factors that determine whether children will
comply with an authority. Empirically, children have cited that they comply with commands in
order to avoid punishment, to obtain rewards, to please their parent, peer, or other authority, or to
serve the needs of others (Carlsmith, Lepper, & Landauer, 1974; Laupa, 1991; Lundy, Shell, &
Roth, 1985). Children also demonstrate relatively sophisticated notions of authority (Damon,
1977; Laupa & Turiel, 1986). Even children as young as four years old judge whether they
should obey a command based on the legitimacy of the authority. Laupa & Turiel (1986) and
Laupa (1991) found that younger children (age 6-8) expressed that they were more likely to obey
an adult (teacher) than a peer granted authority by the school, while older children (9-10 years
old) were equally likely to see the teacher and peer authority as legitimate.
Additionally, children determine whether they should obey based on the type of act
commanded. Importantly, the compliance literature based on observed behavior is predicated
almost exclusively on the child’s obedience to benign commands (i.e., pick up the toys, don’t
touch the toy), while the literature concerning obedience to morally relevant commands (i.e.,
steal the toy, fight with the child) is almost exclusively predicated on hypothetical vignettes.
Vignette studies have universally found that most children are likely to judge a command to
clean their room or pick up toys as legitimate and obligatory, while they view commands to steal
or fight as illegitimate and not to be obeyed, even if the command comes from a parent or
teacher (Damon, 1977; Laupa, 1991, 1994). Most children even find the commands of a
nonauthority more legitimate than the commands of an authority figure if the authority figure is
commanding a harmful act (Laupa, 1994). However, this pattern was strongest for older

20

children; some younger preschoolers still expressed that they should obey the adult authority’s
command to fight with another student because the command was given by an adult (Laupa,
1994). Given that behaving morally is socially desirable, it is possible that the demand
characteristics of vignette studies pull children to supply what they perceive as the “right” (or
morally superior) response to the adult experimenter, even if their actual behavior would differ.
A very small body of literature has investigated whether children will obey immoral
commands in real situations, even if they disavow such actions when presented with a
hypothetical. In a striking example, Shanab & Yahya (1977) replicated Milgram’s shock
paradigm in a group of children aged 6 to 15. Overall, 73% of the subjects continued to deliver
shocks until the final level, a rate strikingly similar to the results reported in Milgram (1965).
Across ages, girls were more likely to display signs of tension and were more likely to state that
they complied because they were obeying orders; boys were more likely to cite that fear of
punishment is beneficial for learning when asked why they complied.
Finally, temperamental factors also play a role in compliance. Children who demonstrate
high negative emotional reactivity exhibit more noncompliance than their peers, although this
finding could also be a function of the fact that these children also elicit greater maternal control,
which is itself related to noncompliance (Braungart-Rieker, Garwood, & Stifter, 1997). Anxiety
appears to have an interesting role in compliance. Carlsmith, Lepper, & Landauer (1974) found
that the motivation for compliance seemed to change based on anxiety. Children induced to be
anxious were more obedient to an experimenter previously experienced as negative, perhaps to
avoid disapproval. On the other hand, children not induced to feel anxious seemed to be more
reward-focused, and were more obedient to a positive experimenter. While induced anxiety is
not identical to temperamental anxiety, this study suggests that the experience of anxiety during

21

a given situation affects the child’s perception of that situation. In another study, low anger and
low social fearfulness was also related to compliance (Kochanska, 1993).
The predictive power of some temperamental factors may differ depending on the type of
task (Kochanska, 1993; Kochanska, Coy, & Murray, 2001; Lickenbrock, Braungart-Rieker,
Ekas, et al., 2013; Rothbart & Bates, 1998). Compliance has typically been measured by parent
report and by observational tasks. These tasks fall into two categories: “Do” tasks, in which the
child is asked to perform a behavior, and “Don’t” tasks, in which the child is asked not to
perform a behavior. Across the literature, a clean-up paradigm is typically used for “Do” tasks
and a forbidden toy paradigm is used for “Don’t” tasks. Incidentally, most studies have found
that “Do” tasks elicit less committed compliance and more situational compliance or
noncompliance than “Don’t” tasks (Braungart-Rieker, Garwood, & Stifter, 1997; Kochanska,
Aksan, & Koenig, 1995; Kochanska, Coy, & Murray, 2001). In fact, committed compliance
typically demonstrates weak or nonsignificant correlations across task type.
Researchers suggest that each type of task acts on different aspects of a child’s
temperament, thus contributing to the lack of consistency in compliance observed across task
types (Braungart-Rieker, Garwood, & Stifter, 1997). One important aspect is self-regulation.
Compliance can be conceptualized as an early form of self-regulation, in that it requires the child
to modulate his or her behavior in accordance with situational demands (Kochanska, Coy, &
Murray, 2001). However, effortful control, a form of self-regulation, appears to preferentially
predict compliance in “Do” tasks, with only modest associations with “Don’t” tasks (Kochanska,
Coy, & Murray, 2001; Kochanska et al., 1997). Fearfulness, on the other hand, has been found
to predict committed compliance in “Don’t” tasks but not in “Do” tasks (Kochanska, 2002;
Kochanska, Coy, & Murray, 2001). This is consistent with psychopathy research, which finds

22

that individuals high on fearlessness have difficulty suppressing prohibited acts (Kochanska,
Coy, & Murray, 2001). However, the content of both “Do” and “Don’t” tasks almost exclusively
involves benign scenarios (e.g., pick up your toys, don’t touch the forbidden toy) rather than
commands to misbehave or cause harm.
Importantly, the vast majority of research on child compliance has operationalized
compliance as obedience to a parent, usually the child’s mother. Of studies that consider
obedience to another authority, most utilize child report measures or responses to hypothetical
vignettes. The few studies that have investigated this question observationally have found
relatively high levels of obedience to non-parent authorities (Landauer, Carlsmith, & Lepper,
1970; Shanab & Yahya, 1977). In fact, Landauer, Carlsmith, & Lepper (1970) had children
receive commands from three separate adult women, one of whom was the child’s mother.
Overwhelmingly, children complied more with commands from the unrelated adult females than
with commands from their own mothers. Children’s compliance to their own mothers’ requests
is likely a function, in part, of the quality of the parent-child relationship (Abe & Izard, 1999;
Crockenberg & Litman, 1990; Kochanska & Aksan, 1995; Kuczynski & Kochanska, 1990),
while situational or temperamental factors may be more important when dealing with a novel
adult.
Finally, compliance also depends in part on the tactics used to elicit compliance.
Maternal control is a particularly robust predictor of compliance; in fact, Kuczynski &
Kochanska (1990) found that the best predictor of child compliance assessed at age 5 was the
type of maternal control assessed at toddlerhood, with power assertive tactics predicting the
poorest compliance. Negative types of control have been associated with defiance (e.g. saying
“no” with a degree of oppositionality) and passive noncompliance (e.g. ignoring the command)

23

(Abe & Izard, 1999; Crockenberg & Litman, 1990). Maternal control also preferentially predicts
situational, as opposed to committed, compliance (Braungart-Rieker, Garwood, & Stifter, 1997).
On the other hand, warm maternal guidance has been associated with assertive and compliant
behavior (Kochanska & Aksan, 1995).
Changes in Compliance Across Development
Irrespective of other factors influencing compliance, the bulk of research has found that
compliance increases as children age. Most studies report that older children demonstrate more
committed compliance and less noncompliance than younger children, both in studies comparing
cohorts and in prospective studies (Braungart-Rieker, Garwood, & Stifter, 1997; Kochanska &
Aksan, 1995; Kochanska, Aksan, & Koenig, 1995). Many studies also report higher levels of
compliance for girls than for boys (Kochanska, 2002; Kochanska, Coy, & Murray, 2001;
Kochanska, Tjebkes, & Forman, 1998). Additionally, Kochanska (2002) found specifically that
compliance in girls was related to self-reported view of a moral self and internalization, while
there was no correlation between these constructs and compliance for boys. Perhaps boys’
compliance is more related to external situational characteristics.
Forms of noncompliance also demonstrate differing frequency across development.
Passive noncompliance, considered a less mature form of noncompliance, decreases across
toddlerhood, while active opposition, including both assertiveness and oppositionality, increases
(Kuczynski, Kochanska, Radke-Yarrow, & Girnius-Brown, 1987). However, uncontrolled anger
and defiance tends to decrease by age 7 (Kuczynski, Kochanska, Radke-Yarrow, & GirniusBrown, 1987). Thus, as children age, not only are they better able to comply, but the tactics by
which they refuse to comply become more sophisticated and less emotionally dysregulated.

24

The Present Study
Despite significant advances in the field of moral and empathic development, the
understanding of moral decision making in real-life situations is limited. Historically, the moral
development literature in children has been hampered by methodological difficulties. Empathy
is typically operationalized through self-report (e.g. De Wied, van Boxtel, Matthys, Meeus,
2011) and parent or teacher report (e.g. Dadds et al., 2009). Even parents reporting on their
child’s CU traits must make inferences about their child’s internal motivations for behavior (i.e.,
callousness versus environmental provocations). Additionally, it is socially desirable to be
empathic, which suggests that parents may be motivated to rate their children as being more
empathic than they are. While parent report provides invaluable data, parents are likely unaware
of the extent of their child’s moral decision making; they can only infer their children’s moral
rationale from their child’s behaviors.
In attempt to circumvent this issue, the majority of studies of moral development have
utilized hypothetical, morally charged vignettes in order to assess the child’s own moral decision
making (e.g. Damon, 1977; Laupa, 1991, 1994). These scenarios typically involve a situation in
which the child must make a moral choice; for example, the child may be asked what he or she
would do if a peer picked a fight, and why he or she would make that decision. Vignette studies
have provided invaluable evidence about how children navigate the moral world, including how
they incorporate authority-based rules, social norms, personal empathic feelings, and broader
moral principles into their decision making. However, there is little research to determine
whether children’s answers to hypothetical scenarios correlate with their actual behavior in dayto-day life.

25

A minority of studies has empirically examined moral decision making through
performance-based laboratory tasks (Brook & Kosson, 2013; Shanab & Yahya, 1977).
Performance-based tasks may reduce some of the problems with questionnaire data, particularly
for young children. Although, as with self-report, these tasks require some inferences about
internal states, studies have found compelling predictive associations between laboratory task
behavior and aspects of moral development. For example, Kochanska, Murray, Jacques, Koenig,
& Vandegeest (1996) found that effortful control as assessed via a forbidden toy paradigm was
correlated with children’s internalization of moral principles.
Thus, laboratory tasks can circumvent the problematic lack of consistency between
questionnaire measures of a child’s past or hypothetical behavior and their actual behavior in a
morally relevant scenario. The aforementioned methodological issues suggest that a deeper
understanding of empathy, moral decision making, and its relation to externalizing problems
should include a greater emphasis on observational data, and should attempt to disentangle the
multiple factors at play.
In the present study, we used just such a laboratory observational approach to explore the
nature of children’s affective, social, and behavioral responses to a morally rich scenario,
collected during the course of a multi-task laboratory assessment of child temperament. This
study explored individual differences in these responses and their associations with child age,
sex, temperament traits, verbal intelligence, and emotion recognition abilities. Further, because
the moral task was collected in the context of a larger study assessing the development of
temperament and risk for psychopathology with multiple assessments collected over 2 years, this
study also explored concurrent and predictive associations between children’s responses to the
moral scenario and the development of externalizing psychopathology in particular.

26

Aims of the Present Study
Overarching Aim: Validate the picture tearing coding scheme as an assessment of child
moral development characteristics. For many reasons already enumerated, the observational
Picture Tearing task provides a rich and unique opportunity to assess aspects of child moral
development. However, while this task is potentially illuminating and is described as part of the
Laboratory Temperamental Assessment Battery (Goldsmith et al., 1995), no published study to
date has included this task. Moreover, the coding scheme suggested in the original Lab-TAB
manual (Goldsmith et al., 1995) has not been validated in a published paper. Our coding scheme
is based on the original scheme, but includes additional variables of interest to moral
development and increases variability by substituting Likert scales for previously dichotomous
items.
In order to support this new scheme as an effective tool for future research, we have
examined its utility in describing and predicting behaviors of interest. Specifically, we have
determined to what extent this coding scheme predicts behaviors both within the task and outside
of the lab (externalizing behaviors, in particular). We have used the results of our analyses to
make recommendations as to which variables are most valid for tapping individual differences in
moral development and which have the greatest utility for predicting outcomes related to moral
development (i.e., externalizing problems). Thus, the present study may help to encourage use of
laboratory based observational methods for assessing moral development, and illuminate the
specific codeable behaviors most valid for understanding aspects of moral development and
predicting conceptually related outcomes of applied interest. In order to achieve this overarching
goal, we divided our analyses into three specific aims that each address an issue important to
assessing the overall utility and validity of our instrument.

27

Specific Subordinate Aims
Aim 1: Describe variations in empathic responding in a morally relevant scenario.
To date, no study has simultaneously addressed the associations of empathy, guilt, and obedience
with morally relevant behavior via observational methods in youngsters. The present study
aimed to address gaps in the moral development literature by assessing the characteristics of
empathy, obedience, and noncompliance in young children in a naturalistic moral scenario: the
Picture Tearing task (Goldsmith et al., 1995). In this task, the child is asked by one adult
experimenter (with whom s/he has been interacting for over an hour) to perform an immoral act
(tearing the other experimenter’s favorite picture). The second experimenter then returns to
observe what the child has done (or not) to the allegedly cherished object. This task is
specifically designed to elicit moral emotions (e.g. empathy, guilt) and to pit higher levels of
moral reasoning (e.g. appeals to universal moral principles) against more rudimentary levels (e.g.
obedience to authority). The children in our study ranged from 3 to 7 years old; across this age
range, children should demonstrate an ability to recognize emotions, to identify situations that
might elicit certain emotions for others, to recognize when others might require comfort. Thus,
most of the children, especially in the older half of the age range, should theoretically be able to
take the perspective of the second experimenter and identify that she might feel upset if her
favorite picture is destroyed. However, we anticipated there would be wide individual
differences in all aspects of responses to this task.
Ethical concerns prohibit the use of shock paradigms as in prior obedience research
(Milgram, 1963; Shanab & Yahya, 1977). Thus, as opposed to physical harm (a shock), the
children in our study were asked to cause emotional harm. This type of immoral act is subtler,
but also more similar to the kinds of dilemmas that children typically encounter. Additionally,

28

this paradigm allowed us to assess moral emotions and moral reasoning at multiple levels. First,
we could observe the child’s actual choice to obey an authority figure and to perform or refuse
an immoral action. Second, we could observe behavioral indicators of empathy and guilt,
including verbalizations, both during the decision-making period and when the child is
confronted by the person they have wronged (for those who choose to tear the picture). These
data will be used to describe patterns of empathic responding and obedience among children,
including descriptive data on rates of compliance with the Picture Tearing Request. We
expected to find rates of obedience that are comparable with or higher than the rates reported in
Milgram (1963) and Shanab & Yahya (1977), given that children tend to appeal more to
authority (e.g., the primary experimenter in this study) when making such decisions (compared
to adults). We also examined whether the gender of the second experimenter (the “victim”)
affected children’s responses to the request.
Aim 2: Examine associations between observed moral behavior and child
characteristics theoretically linked to moral development and empathy. Not only could we
observe the child’s moral decision making in the moment, but we could also test associations
between the child’s behavior in response to the Picture Tearing task and other dimensions that
have been linked empirically to moral decision making in the literature, including normative
factors of sex, age, temperament, emotion recognition ability, and intelligence. Foremost, we
wanted to establish the construct validity of the Picture Tearing task as a measure of individual
differences in aspects of moral development. If the task produces a valid measure, then scores
on Picture Tearing composites should correlate with known correlates of moral development
according to patterns observed in previous research. We tested the relevance of empathy
measures assessed in this task for understanding abnormal development related to deficits in

29

empathy, specifically risk for CU traits and related problems (assessed both concurrently and
predictively). In this way, we clarified the temperamental characteristics associated with
variously sophisticated levels of empathic responding, and also investigated the behavioral
factors observed in this task that are predictive of later aggression and delinquency. We also
explored broader contextual contributors to moral development and empathy, specifically aspects
of the parent-child relationship. Finally, we investigated the incremental validity of each of these
domains for understanding child behaviors in response to the Picture Tearing task, including
compliance, guilt, and empathy, via hierarchical regression. Analyses are described in more
detail below.
Aim 3: Examine predictive associations between compliance, arousal, and empathy
and later externalizing behaviors. Finally, we examined whether observationally assessed
compliance, arousal, and empathy at baseline will predict outcomes at 6 months, 9 months, 12
months, 18 months, and 24 months after the initial assessment. We were especially interested in
the prediction of externalizing pathology, given its robust empirical associations with low
empathy, poorly developed conscience, and disorders of empathy (i.e., CU traits) (e.g. Asendorpf
& Nunner-Winkler, 1992; de Wied, van Boxtel, Matthys, and Meeus, 2011; Frick et al., 2003;
Frick & Morris, 2004; Frick, Ray, Thornton, & Kahn, 2014; Frick & White, 2008; Jolliffe &
Farrington, 2004). Given the accumulation of evidence, we predicted that empathy and arousal
during the Picture Tearing task would predict externalizing problems (e.g. CU traits, aggression,
rule-breaking, ODD symptoms, CD symptoms) at all follow-up time points, over and above the
variance associated with baseline levels of these behaviors. To assess discriminant validity, we
also examined the relationship between empathy and arousal and later internalizing problems.

30

METHOD
Overview of Design
A community sample of child participants (N = 271 children from 220 families) aged 3 to
7-years-old and their parents was recruited from the central Michigan area for a study of child
temperament. Participants were recruited through advertisements on the Cragslist website as
well as local postings in areas frequented by children (e.g. preschools). Children who did not
have any significant medical conditions or developmental disabilities and lived with at least one
English-speaking biological parent were eligible for participation in the study. Participating
children visited the laboratory with their mother or father for a 2-hour assessment consisting of
16 tasks designed to elicit discrete emotions and behaviors indicative of temperament traits, as
well as assessments of intelligence and emotion recognition skills. At the end of the lab visit, the
parent was given a battery of questionnaires assessing temperament traits, problem behaviors,
and parent-child relationship to complete and return by mail.
Follow-up packets of questionnaires assessing child temperament traits and problem
behaviors were mailed to participants at 6 months, 9 months, 12 months, 18 months, and 24
months after the initial visit. These packets were to be completed at receipt and returned by
mail. When the date of questionnaire completion was ambiguous (i.e. the participants mailed
back a packet several months after having received it), graduate assistants contacted the family to
obtain an accurate date. If the family failed to return three follow-up packets in a row, they were
considered uninterested in further assessment and removed from the mailing list. Participants
were compensated monetarily (gift cards) for the baseline assessment and for each returned
follow-up packet.

31

Participants
Child participants were between the ages of 3 and 7 years at baseline (49.4% girls). The
mean age of the children was 54.5 months (SD = 16.3). Data on race and ethnicity and family
income were provided by 64.6% of mothers and by 40.4% of fathers. Of those, the ethnic
composition of mothers was as follows: Caucasian/White (76.5%), Hispanic/Latino (8.4%),
African American/Black (5.0%), Asian (1.7%), Native American (1.1%), other (1.7%), and bi- or
multiracial (5.6%). The ethnic composition of fathers was as follows: Caucasian/White (73.2%),
Hispanic/Latino (6.3%), African American/Black (10.7%), Asian (0.9%), Native American
(2.7%), other (1.8%), and bi- or multiracial (4.5%). Participants were not asked about the
ethnicity of their child. In all analyses where ethnicity is used, we used mother ethnicity as a
proxy for child ethnicity, as almost all children had their mother participating, while father
participation rates were substantially lower. In the few cases where the child’s mother did not
participate, we instead used the father’s ethnicity. Yearly family income ranged from less than
$10,000 to greater than $100,000. Income distribution was as follows: under $10,000 (5.8%),
$10,000 to $20,000 (15.1%), $21,000 to $40,000 (21.5%), $41,000 to $60,000 (30.8%), $61,000
to $100,000 (22.1%), over $100,000 (4.7%).
Laboratory Assessment of Temperament
Children completed a battery of 16 laboratory tasks designed to assess differences in
three broad dimensions of temperament: positive emotionality (PE), negative emotionality (NE),
and effortful control (EC). The battery was composed of episodes from the Laboratory
Temperament Assessment Battery—Preschool Version (Lab-TAB; Goldsmith, Reilly, Lemery,
Longley, & Prescott, 1995) and from previous investigations (e.g. Durbin, 2010; Durbin et al.,
2007; Frye, Zelazo, & Palfai, 1995; Kochanska et al., 1996; Zelazo, Muller, Frye, Marcovitch,

32

2003). Tasks were administered in the same order across participants. Parents were present for
the tasks as noted. For these tasks, parents were instructed to remain affectively neutral and to
minimize their interaction with their child. Tasks are described in full below.
Exploring New Objects (Durbin, 2010; Fear, happiness). The child was left to explore
the room, which contained novel and ambiguous stimuli including: a tunnel containing a
mechanical spider, an animal crate containing toy mice, a wooden box containing sticky
“worms,” and a plastic skull hidden under a cloth. The experimenter returned after 4.5 minutes
and asked the child to interact with each object.
Making a t-shirt (Durbin, 2010; engagement, happiness). The child decorated a t-shirt
with ink and stamps. The child was allowed to take the t-shirt home at the end of the lab visit.
Stranger approach (Lab-TAB; Goldsmith et al., 1993; Fear). The child was left alone
briefly in the testing room. A male research assistant entered the room and engaged in a scripted
conversation with the child using a neutral voice. The experimenter then returned and
introduced the research assistant as a friend.
Impossibly perfect green circles (Lab-TAB; Goldsmith et al., 1993; Anger, sadness).
The experimenter repeatedly asked the child to draw green circles on a piece of paper while
mildly criticizing each circle. After 2 minutes, the experimenter positively commented on the
child’s circles.
Popping bubbles (Lab-TAB; Goldsmith et al., 1993; Activity, happiness). The
experimenter made bubbles with a bubble-shooting toy and encouraged the child to pop the
bubbles.
Diorama snakes (Fear). The experimenter showed the child a tray filled with sand and
two remote-controlled snakes and asked the child to touch the snakes. A second experimenter

33

(holding the remote controls out of sight of the child) made the snakes move. Following the
child’s response (e.g. touching the snake), the experimenter explained that the snakes are toys
and demonstrated the use of the remote controls.
Snack delay (Lab-TAB; Goldsmith et al., 1993; Effortful control). The experimenter
placed an M&M or other small snack under a clear cup and instructed the child that he/she could
eat the M&M when the experimenter rang a bell. The experimenter rang the bell at eight
predetermined time intervals (ranging from immediately to 30 seconds after M&M placement).
Picture tearing (Goldsmith et al., 1995; Guilt, empathy, sadness, compliance). The
child’s parent was not present for this task. The main experimenter told the child that she will
prepare the next game while the second experimenter shares pictures with the child. The second
experimenter showed the child a series of pictures depicting a fictional vacation home and
grandparents. The second experimenter noted that the picture of the grandparents is his/her
favorite because he/she rarely sees his/her grandparents. The second experimenter left the room.
The main experimenter returned and asked the child to tear up the second experimenter’s favorite
picture. This command was repeated several times. After the child’s response, the main
experimenter excited the room. The second experimenter then returned to the room to retriever
his/her photo album. If the child tore the picture, the second experimenter stated, “Oh no! What
happened?” and allowed time for the child to respond. The second experimenter then retrieved a
second copy of the picture and had the child help return it to the album. The main experimenter
then emphasized that she should not have asked the child to tear the picture. Intraclass
correlation coefficients are presented in Table 1. ICCs for most variables was at 0.70 or better,
though a small number of variables with very low base rates (e.g. Hunched Shoulders, Lip

34

Biting) had lower ICCs. This was taken into account when selecting variables to retain for
analysis.
Balloon bop (Goldsmith & Rothbart, 1996; Effortful control, happiness). The
experimenter and child played a game in which they took turns hitting a balloon in the air. The
child was told that he/she must keep their feet within a circle drawn on the ground. The
experimenter tempted the child to leave the circle by hitting the balloon further from the circle on
some trials.
Transparent box (Lab-TAB; Goldsmith et al., 1993; Anger, sadness). The
experimenter locked an appealing toy chosen by the child in a clear plastic box and gave the
child an incorrect set of keys. The experimenter told the child to open the box with the keys, and
then left from the room. After three minutes, the experimenter returned with the correct set of
keys and explained that she accidentally gave the child the wrong keys. The child was then
allowed to open the box and play with the toy.
Simon says (Strommen, 1973; Effortful control). The experimenter demonstrated ten
exercises (e.g., touch the floor) to the child and had them practice. The experimenter then told
the child that in the game, the child should do all of the exercises when the experimenter says,
“Simon says,” and should not do the exercises when the experimenter does not say, “Simon
says.” After two practice trials, the experimenter completed two trials each of all 10 exercises,
regardless of the child’s performance.
Tell a story (Durbin et al., 2007; Fear). The experimenter told the child that she would
like to know how good the child is at telling stories. The experimenter handed the child a picture
book without words and told the child to tell a story using the pictures. The experimenter
explained that a second experimenter is an expert at stories and will listen to the child’s story and

35

give him/her a grade. At the end of the story, the experimenter asked the second experimenter to
give the child a grade on the story.
Pop-up snakes (Lab-TAB; Goldsmith et al., 1993; Anticipatory positive affect,
happiness, surprise). The parent was not present for the first half of the task. The experimenter
showed the child what appeared to be a can of potato chips, which instead contained coiled
spring snakes. The experimenter demonstrated the trick and then encouraged the child to play
the trick on his/her parent.
Walk-a-line slowly (Kochanska et al., 1996; Effortful control). The experimenter
demonstrated how to walk a line on the floor and asked the child to walk on the line. The
experimenter then asked the child to walk the line as slowly as he/she could, then as quickly as
he/she could, then again as slowly as he/she could. The experimenter then asked the child to
walk as slowly as he/she could on a small balance beam.
Box empty (Lab-TAB; Goldsmith et al., 1993; Anticipatory positive affect, anger,
sadness). The experimenter gave the child an empty gift bag under the pretense that the bag
contains an appealing gift. The child was told to open the gift and was left alone for 2.5 minutes.
The experimenter then returned with a bag of toys and told the child that she forgot to put the
toys inside the gift bag.
Coding of Child Temperament Laboratory Tasks
Observational coding of picture tearing task. Picture tearing videos were coded with a
global system (see Appendix A). Codes were derived from the original coding system developed
for the task by Goldsmith and colleagues (2001). Items retained were: Latency to tear up picture
(in seconds), Presence of enjoyment in tearing picture, Overall compliance, and Relief in
replacement picture. The ratings for Presence of enjoyment, Overall compliance, and Relief in

36

replacement picture were expanded to four-point Likert scales in order to increase the possible
range of scores (variance) on these dimensions. Ratings for intensity of sadness, anger, empathic
statements, and visual referencing were retained, but were coded as counts (i.e., the number of
discrete instances during which these expressions occurred) rather than global dimensions. In
addition, counts were added to assess additional behaviors including: expressions of defiance,
expressions of hesitation, squirming/fidgeting, and gaze avoidance (presence of these behaviors
was coded every five seconds).
Behaviors were counted and summed separately across three epochs: during the
prompting phase (between the experimenter’s first prompt to tear the picture and the
experimenter leaving the room), during the period in which the child is left alone in the room,
and during the period after the second experimenter returned and asked what happened to the
picture. Specifically during the third epoch, counts were taken for the child’s responses to the
experimenter’s inquiry, which included: blaming someone else, blaming themselves, silence, or
lying. A global code for guilt (“How guilty did the child seem?”) was rated on a four-point
scale. For this code, coders were instructed to rate the child’s overall guilt based on a global
evaluation of their behavior, taking into account an integrated impression of facial expressions,
posture, and verbalizations.
Figure 2 and Table 2 summarize the codes relevant to each epoch.
Observational coding of laboratory tasks. All laboratory episodes (including Picture
Tearing) were coded with a global system validated in previous studies of temperament to assess
higher-and lower-order facets of Positive Emotionality, Negative Emotionality, and Effortful
Control. Of particular importance for this study, lower-order traits of fear proneness and
compliance were rated in all tasks. This enables us to explore whether individual differences in

37

children’s anxious arousal/fear or compliance with the experimenter in the Picture Tearing task
are correlated with the same traits assessed in other lab scenarios that lack the moral dilemma of
Picture Tearing and whether associations between fear and compliance in response to Picture
Tearing are associated with outcomes of interest after accounting for general trait fear proneness
and trait compliance. Coders were trained undergraduate and graduate students.
Coders recorded instances of facial, vocal, and bodily indicators of emotional states (i.e.
happiness, sadness, fear, anger, surprise). Indicators will be rated for intensity according to the
AFFEX coding system (Izard, Dougherty, & Hebree, 1983), which classifies intensity of facial
expressions in three levels: (1) ambiguous or low intensity (expressions of low intensity in one
facial region); (2) moderate intensity (expression definitely present in at least one facial region);
and (3) high intensity (expression definitely present in both facial regions, i.e. eyes and mouth).
Intensity of vocal and bodily expressions were also coded on a three-point scale (low, moderate,
high) as determined by the extent to which the vocalization or bodily movement conveyed the
emotion.
Coders also completed global ratings of the child’s behavior during each laboratory task.
We will utilize the scales indexing the child’s level of Compliance, Engagement in the task,
Initiative, Sociability, Attentional Control, and Impulsivity. Each of these behaviors was rated
on a four-point Likert scale (0 = low, 1 = moderate, 2 = moderate-to-high, 3 = high). Each
behavior code is assigned a single rating (0-3) based on the aggregate of behaviors for a given
episode. Compliance ratings are based on the extent to which the child acts in accordance with
the parent’s or experimenter’s suggestions or commands. Engagement ratings are based on the
child’s degree of attentiveness to and persistence throughout the task. Initiative ratings are based
on the child’s degree of assertiveness in interactions with the experimenter or parent. Sociability

38

ratings are based on the child’s interest in and pursuit of social interaction with the experimenter
or parent. Attentional Control ratings are based on the child’s ability to maintain attention
flexibly across the task. Impulsivity ratings are based on the child’s tendency towards impatient
and impulsive behavior, i.e. a lack of appropriate and planful behavioral control.
Codes were aggregated across episodes as appropriate, using scale construction
techniques (evaluation of internal consistency reliability, average inter-task correlations, and
interrater reliability). Aggregate codes will be combined into higher order dimensions for traits
of Positive Emotionality (PE), Negative Emotionality (NE), and Effortful Control (EC),
consistent with evidence from other studies using laboratory assessment of traits (e.g. Durbin,
Hayden, Klein, & Olino, 2007; Dyson et al., 2012; Hayden, Klein, Durbin, & Olino, 2006;
Vroman, Lo, & Durbin, 2014; Wilson & Durbin, 2012). In this system, the codes that combine
to create each higher-order code are as follows: Negative Emotionality is comprised of Fear,
Sadness, and Anger; Positive Emotionality is comprised of total Positive Emotionality,
Anticipatory Positive Emotionality, Sociability, Interest/Engagement, Initiative, and Activity;
Effortful Control is comprised of Noncompliance, Attentional Control, and Impulsivity.
Other Tasks
Emotion recognition training. Children were shown six cards with pictures of a boy or
girl demonstrating the following facial expressions: Fear, Surprise, Neutral, Sad, Happy, Angry.
The experimenter pointed to each picture in turn and asked the child to describe how the person
in the picture was feeling. Children’s responses were recorded and a total score was given for
the number of correct labels the child provided. Children were given feedback if they were
correct and were provided with the correct label if they were incorrect. The cards were then
shuffled and the experimenter asked the child to identify all six emotions again. Feedback was

39

given for any incorrect emotions. If the child correctly labeled all six emotions, the task was
discontinued. If the child labeled any emotion incorrectly, a third trial was performed, and the
final performance score (i.e., number of emotions correctly labeled on the third trial) was
recorded. Emotion recognition scores were available for 97% of children. Overall accuracy on
the first and final trial administered were recorded and were used to assess basic emotion
recognition skills. We calculated scores (total correct) separately for the first trial, which
assesses the child’s baseline knowledge, and the third trial, which assesses their knowledge after
training. We examined each of these variables separately in analyses. Deficits in emotion
recognition, especially for fear and sadness, are empirically related to deficits in empathy and
guilt (e.g. Blair & Coles, 2000; Blair, Colledge, Murray, Mitchell, 2001; Dadds, Perry, Hawes, et
al., 2006; Munoz, 2009; Blair, Budhani, Colledge, Scott, 2005; Fairchild, van Goozen, Calder, et
al., 2009).
Peabody Picture Vocabulary Test (PPVT). Children were administered the Peabody
Picture Vocabulary Test (PPVT; Dunn & Dunn, 1997) to assess global cognitive function. The
PPVT is a widely used measure of receptive vocabulary consisting of 204 items. For each item,
respondents are presented with four pictures and are asked to identify which picture best
corresponds to a given word. Respondents begin with the set of items corresponding to their
age; if the respondent makes fewer than 2 errors, that set is considered basal. Items are
administered until the respondent makes 8 or more errors in any given set. Standardized PPVT
scores were utilized in all analyses. PPVT scores were available for 99% of children.
Measures
Child Behavior Checklist (CBCL). Participating mothers and fathers completed the
Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001), which was designed to

40

measure behavior problems in children. Parents completed the CBCL at the baseline assessment
as well at 6 months, 9 months, 12 months, 18 months, and 24 months post-visit. The CBCL
consists of 145 statements about child behavior over the past six months, rated on a three-point
scale: never, sometimes true, mostly true. CBCL scores were computed for maternal and
paternal report separately. Scores were available in the following proportions across time points:
78% of mothers and 49% of fathers (baseline), 66% of mothers and 48% of fathers (6 month
follow-up), 65% of mothers and 48% of fathers (9 month follow-up), 55% of mothers and 33%
of fathers (12 month follow-up), 38% of mothers and 26% of fathers (18 month follow-up), and
28% of mothers and 22% of fathers (24 month follow-up).
We utilized the higher-order Internalizing and Externalizing scales and the lower-order
scales measuring Aggression and Rule-breaking. We also utilized the following DSM-oriented
subscales, which were derived using the items most closely linked to DSM-IV disorder criteria:
Conduct disorder and Oppositional Defiant Disorder.
We also computed and utilized the callous-unemotional subscale created by Willoughby
and colleagues (2011, 2014). As Willoughby used the preschool version of the CBCL and we
utilized the school-age version, we are able to recover four of the five items in the original scale
(“Does not seem to feel guilty after misbehaving,” “Seems unresponsive to affection,” “Shows
little affection toward people,” “Shows too little fear of getting hurt”). The item “Punishment
does not change behavior” is not included in the school-age CBCL and was excluded. Our scale
demonstrated somewhat lower Cronbach’s alphas than the original five-item scale (Motherreport: baseline α = .27, 6 months α = .46; Father-report: baseline α = .39, 6 months α = .35)
(Willoughby et al., 2011).

41

Children’s Behavior Questionnaire (CBQ). Participating mothers and fathers
completed the Children’s Behavior Questionnaire (CBQ; Rothbart, Ahadi, Hershey, & Fisher,
2001), which was designed to measure temperament in children aged 3 to 7 years. Parents
completed the CBQ at the baseline assessment as well at 6 months, 9 months, 12 months, 18
months, and 24 months post-visit. The CBQ consists of 195 items rated on a seven-point scale
from “extremely untrue” to “extremely true.” The CBQ yields three higher-order subscales
(Surgency, Negative Emotionality, and Effortful Control) and lower-order subscales measuring
more homogenous facets of temperament. For our analyses, in addition to the higher-order
subscales, we chose to examine lower-order scales based on theoretically expected associations
with guilt and empathic responding. We utilized the following subscales consistent with
empirical and theoretical associations with empathy and guilt: Anger, Attentional focusing,
Sootheability, Fear, High intensity pleasure, Impulsivity, Inhibitory control, Sadness, and
Shyness. Scores were available in the following proportions across time points: 79% of mothers
and 50% of fathers (baseline), 66% of mothers and 48% of fathers (6 month follow-up), 65% of
mothers and 48% of fathers (9 month follow-up), 55% of mothers and 33% of fathers (12 month
follow-up), 38% of mothers and 26% of fathers (18 month follow-up), and 28% of mothers and
22% of fathers (24 month follow-up).
Experimenter ratings of child traits. At the conclusion of the initial laboratory visit,
experimenters rated the child’s global behavior across the visit. Behaviors were rated on a fivepoint Likert scale. We utilized the three higher order dimensions: Effortful Control, Negative
Emotionality, and Positive Emotionality (see Vroman, Lo, & Durbin, 2014). Post-visit ratings
were available for 99.6% of children.

42

Alabama Parenting Questionnaire (APQ). Mothers and fathers completed the
Alabama Parenting Questionnaire (APQ; Frick, 1991) at the baseline assessment. The APQ is a
42-item questionnaire assessing parenting practices across five dimensions associated with risk
for conduct problems: Poor Monitoring/Supervision, Corporal Punishment, Inconsistent
Punishment, Positive Parenting, and Parental Involvement. Scores were available for 78% of
mothers and 49% of fathers.
Data Analytic Plan
Aim 1: Describe variations empathic responding in a morally relevant scenario. We
utilized two analytic strategies to describe variations in empathic responding. All participants
with Picture Tearing data were retained for factor and cluster analysis, regardless of whether
questionnaire outcome data was present. However, only those participants with CBCL, CBQ, or
APQ data were included in follow-up external validation analyses.
Factor analysis. First, we utilized a variable centered approach and conducted an
Exploratory Factor Analysis in Mplus version 7.2 (Muthen & Muthen, 1998-2012) utilizing a
principal axis approach with oblimin rotation to account for the fact that the underlying
dimensions are not orthogonal to one another. Picture Tearing task variables from all three
epochs were utilized in the EFA. Those that demonstrated no associations with other task
variables were excluded from analyses. Eigenvalues, scree plot, and parallel analysis were all
utilized to estimate the most likely number of factors. We retained the three factor structures
best supported by analyses and tested the fit of these models using Confirmatory Factor Analysis
in Mplus (Muthen & Muthen, 1998-2012). We also estimated a three-factor model in which
factors and loadings were chosen based on theoretical assumptions about the underlying structure
of task behavior (see Table 4; Hypothesis 1.1). Fit indices for each model were compared. We

43

created composite variables for further analysis based on the factor structure of the best-fitting
model.
Cluster analysis. Second, we utilized a person centered approach by performing a cluster
analysis on the Picture Tearing data. Cluster analysis included all of the coding variables from
all three epochs. To derive clusters, we used a two-step method recommended by several
researchers (e.g. Hair et al., 1994; Milligan, 1980; Steinley, 2003; Swogger, 2007). We initially
derived clusters using Ward’s hierarchical agglomerative method of cluster analysis in SPSS
(IBM version 22.0). To determine the appropriate number of clusters, we graphed the
agglomeration coefficients against the stage number and used the “elbow method” to determine
the stage at which agglomeration coefficients ceased to increase substantially from stage to stage
(e.g. Swogger, 2007).
We also computed the gap statistic, which indicates whether the pooled squared distances
of points within an increasing number of clusters are sufficiently different than would be
expected based on a random, uniform reference distribution (see Tibshirani et al., 2001). We
computed this statistic using the clusterGenomics package (Nilsen & Lingjaerde, 2013) in R (R
Core Team, 2013). Based on these analyses, we chose a number of clusters and applied this
solution to the data using the k-means cluster procedure (Tan, Steinbach, & Kumar, 2006) in
SPSS, with k set as the number of clusters derived from the hierarchical method. We validated
the resultant clusters using several methods: (1) collapsing of similar clusters that have fewer
than 5% of the sample included to avoid identification of small groups unlikely to replicate in
other samples and (2) validation of clusters by comparing them on variables not included in the
cluster analysis (i.e., parent-reported temperament and behavior problem, child sex and age,
etc.).

44

This analysis allowed us to examine characteristics that may differ between children who
present with certain patterns of behaviors. Overall, we expected that most children will comply
with the experimenter’s request (Hypothesis 1.2), but that children’s compliance may have very
different accompanying behaviors (e.g., tearing with glee versus reluctantly and with concern for
the second experimenter). We hypothesized that failure to comply could be positively associated
with assertiveness (Hypothesis 1.3) and/or extreme levels of expressed empathy (Hypothesis 1.4)
in some children. However, failure to comply may also correlate positively with defiance or
oppositionality (Hypothesis 1.5), as measured by the child’s noncompliant and defiant behaviors
in other lab tasks, ratings made by the experimenter, and parent-report of oppositional behaviors.
Compliance in tearing the picture in the absence of empathic statements could also demonstrate a
positive association with fearlessness (Hypothesis 1.6), as these children are unlikely to feel
concerned about the experimenter’s feelings, and may even enjoy tearing the picture.
Aim 2: Examine associations between observed moral behavior and child
characteristics theoretically linked to moral development and empathy. Utilizing
composites derived from factor analysis and a variable indicating cluster membership, we
examined the pattern of bivariate correlations between factors, cluster membership, and parentreported problem behaviors. We excluded variables that did not have a significant zero-order
association with the dependent variable of interest. The general strategy for regressions is
described below; in all cases, variables described in each step were only included in the
regression if they had a significant bivariate correlation with the dependent variable of interest.
In all analyses, mother-reported variables and father-reported variables were included and
analyzed separately.

45

Hierarchical regressions were run separately predicting total externalizing problems,
aggression, CU traits, Rule-breaking, ODD symptoms, CD symptoms, and total internalizing
problems. In all analyses, we first entered child sex and age. Then, we entered Picture Tearing
variables (either behavioral composites or dummy coded contrast variables describing cluster
membership) into step two. The purpose of these initial regressions was to determine if
behaviors during Picture Tearing predict externalizing and internalizing problems over and
above demographic variables that are themselves associated with variations in problem
behaviors.
The next set of regressions more stringently evaluated whether Picture Tearing behaviors
predicted problem behaviors over and above other child and environmental characteristics that
are theoretically and empirically linked to externalizing pathology (see Figure 3). We included
additional predictors only if the independent variables accounted for significant variance in the
outcome in the first set of regressions. In these additional analyses, we first entered child sex
and age; subsequent steps in the regression will include emotion recognition and verbal ability
(step 2), child temperament traits (step 3), and parenting dimensions (step 4). Finally, Picture
Tearing behaviors will be entered (step 5). The order of the sets of variables was determined by
reference to theoretical ideas regarding their casual primacy for the dependent variables. We
chose to enter emotion recognition and verbal ability next as these factors may fundamentally
impact the degree to which the child understands the moral scenario of the task. Finally, we
entered child traits, and then parenting factors; we also reversed the order of steps 3 and 4 in
subsequent analyses, as it could be argued that parenting dimensions may be more casually prior
to our dependent variables of interest than child traits.

46

Picture Tearing variables were entered

last to determine whether these task-specific behaviors are incrementally predictive beyond other
characteristics of the child and his or her environment.
Importantly, the Picture Tearing task was collected as part of battery of 16 tasks, which
were designed to measure other traits of interest in this study, including fear proneness and
effortful control, as well as compliance. Given that the other 15 tasks do not involve moral
dilemmas, coding data from these tasks were used to derive a composite score reflective of the
child’s typical levels of compliance and fear proneness. We included these lab-based trait
composites in our hierarchical regression analyses (assuming they had a significant zero-order
correlation with the dependent variable of interest) to determine whether a child’s pattern of
behavior during Picture Tearing could be accounted for wholly by that child’s typical levels of
compliance (or fear proneness), or whether compliance (or anxious arousal) in Picture Tearing is
incrementally useful in predicting other behaviors (e.g., externalizing). We expected that
compliance and fear proneness observed specifically during Picture Tearing would incrementally
predict externalizing behaviors, above and beyond typical levels of compliance and fear
proneness. Specifically, we expected that noncompliance would be positively associated with
externalizing behaviors (Hypothesis 2.1) and fear-proneness, as indexed by Anxious Arousal
during the task, would be negatively associated with externalizing behaviors (Hypothesis 2.2).
Irrespective of compliance with the request to tear the picture, we expected that empathy
would be strongly positively associated with appropriate moral socialization, operationalized as
an absence of significant externalizing problems (Hypothesis 2.3). This is consistent with
Kohlberg’s (Kohlberg & Kramer, 1969) theories of moral development, which suggest that strict
obedience to authority figures is normative for younger children. Consequently, we expected
that high anxious arousal and empathy in the Picture Tearing task would be suggestive of

47

normative moral development and thus correlate with few externalizing problems, while low
empathy and low arousal would be suggestive of aberrant moral socialization and accordingly,
predict high levels of externalizing problems (Hypothesis 2.2, 2.3).
A summary of our hypothesized associations is provided in Table 3. We discuss these
hypotheses in more detail below.
Demographic Variables. In our study, age served as a proxy for developmental period.
Across research, older children (5 to 6 years old) typically demonstrate more compliance with
the experimenter commands across a variety of tasks than do younger children (3 to 4 years old),
including fewer instances of direct defiance (Braungart-Rieker, Garwood, & Stifter, 1997;
Kochanska & Aksan, 1995; Kochanska, Aksan, & Koenig, 1995; Laupa, 1991; Laupa & Turiel,
1991). Given the demand characteristics of the situation and previous obedience research (e.g.
Shanab & Yahya, 1977), we expected that most older children will comply with the
experimenter’s request. However, older children are also more likely to demonstrate more
sophisticated perspective taking and higher levels of empathic internalization (Borke, 1971,
1973; Kochanska, 1993), and are more likely to view commands to harm others as illegitimate
(Damon, 1977; Laupa, 1991, 1994). Thus, we expected that irrespective of their choice to
comply, older children would demonstrate higher levels of empathy and anxious arousal,
including more empathic statements, than younger children (see Table 3; Hypothesis 2.5, 2.6).
Given the compliance research as a whole, we also expected that, given they display more
empathy than younger children, older children would be modestly less likely to comply with the
experimenter’s request than younger children (Hypothesis 2.4).
The potential for sex differences was more exploratory, as the literature does not support
a clear position on whether gender significantly predicts compliance or empathy in children.

48

While many studies report higher levels of compliance for girls (e.g. Kochanska, 2002;
Kochanska, Coy, & Murray, 2001; Kochanska, Tjebkes, & Forman, 1998), other research has
found no gender differences (e.g. Abe & Izzard, 1999; Braungart-Rieker, Garwood, & Stifter,
1997; Higbee, 2012). Thus, we tentatively expected that girls would exhibit greater compliance
than boys, but that this difference would be small in magnitude (Hypothesis 2.7). Similarly,
while some studies report that girls demonstrate higher levels of empathy and internalization
(e.g. Garaigordobil, 2009; Hoffman, 1977; Kochanska, DeVet, Goldman, Murray, & Putnam,
2008), other research has found that gender differences are either nonsignificant or an artifact of
assessment method (e.g. Eisenberg & Lennon, 1983).
A review by Rose and Rudolph (2006) supports the contention that sex differences may
become more apparent across development, such that girls demonstrate greater empathy than
boys in older school-age or adolescent samples, but effect sizes are small or nonexistent for
younger school-age samples. Older children, particularly those of pubertal or post-pubertal age,
may experience socializing pressures differentially by sex and may be basing their moral
decisions on different factors, whereas younger children may reason more uniformly. This
notion is supported by evidence in adolescents and adults that women are more likely to identify
“care” as a moral issue (i.e. compassion for the other, providing for others’ needs), while men are
more likely to identify “justice” as the central dilemma (i.e. fairness, reciprocal rights) (Gilligan
& Attanucci, 1988). Similarly, in their sample of 6 to 18-year-olds, Shanab & Yahya (1977)
asked children why they chose to comply with the experimenter’s request to shock a peer; girls
were more likely to state that they were obeying orders, while boys were more likely to express
that punishment is beneficial for learning. Collectively, these data suggest that, by middle
childhood, boys and girls appear to preferentially focus on different aspects of moral dilemmas

49

and employ different reasoning. Given the available evidence, we expected to find sex
differences in empathy and anxious arousal, of small magnitude (Hypothesis 2.8, 2.9).
Cognitive characteristics. We also examined the role of cognitive maturity in predicting
compliance, arousal, and empathy. While some effect of cognitive maturation will be captured
by age, even children of the same age display individual variation in cognitive capabilities. We
expected that adequately developed cognitive skills would be most strongly associated with
perspective taking, given that this empathic skill is predicated on a child’s ability to recognize
and represent the emotions of others and predict the emotions that might result from a given
situation (Pons, Harris, & de Rosnay, 2004; see Table 3). Moreover, poor emotion recognition,
especially for fear and sadness, is robustly linked to CU traits, low empathy, low anxious
arousal, and aggression (e.g. Blair & Coles, 2000; Blair, Colledge, Murray, Mitchell, 2001;
Dadds, Perry, Hawes, et al., 2006; Munoz, 2009; Blair, Budhani, Colledge, Scott, 2005;
Fairchild, van Goozen, Calder, et al., 2009). Thus, we expected that children who performed
poorly on the baseline trial of our emotion recognition task would also demonstrate low levels of
anxious arousal and empathy during the Picture Tearing task (Hypothesis 2.10, 2.11). Given the
relationship between general intelligence and the ability to mentally represent complex ideas, we
also investigated the potential correlation between perspective taking and general intelligence (as
assessed by the PPVT; Hypothesis 2.12, 2.13).
Temperamental characteristics. We expected that a number of child characteristics
would be associated with compliance, empathy, and arousal during the Picture Tearing task (see
Table 3). Broadly, we expected that compliance, empathy, and anxious arousal (fear-proneness)
in Picture Tearing would be positively related to characteristics associated with adequate
socialization (as operationalized by lower levels of externalizing problems). There is extensive

50

support linking fearlessness, low anxious arousal (e.g. Kochanska, 1991, 1993, 1995;
Kochanska, Coy, & Murray, 2001), deficient empathy, and high levels of externalizing problems
(e.g. Frick et al., 2014; Frick & White, 2008; Frick, Ray, Thornton, & Kahn, 2014; Frick et al.,
1999; Kochanska, Coy, & Murray, 2001) to poor moral internalization and poor compliance (or
oppositionality). Specifically, fearlessness has been linked to low arousal during a transgression
and low or absent guilt after the transgression (Kochanska, 1995; Kochanska, Gross, Lin, &
Nichols, 2002; Rothbart, Ahadi, & Hershey, 1994).
With respect to compliance with the request to tear the picture, we predicted that high
levels of externalizing problems (especially ODD symptoms and noncompliance) would be
positively associated with noncompliance with the experimenter’s request (Hypothesis 2.1). We
also expected that noncompliance would be positively associated with tempermental
characteristics associated with externalizing behaviors (Hypothesis 2.18), especially fearlessness
(Hypothesis 2.14), but negatively associated with sadness (Hypothesis 2.21).
With respect to moral emotions, we expected that observed anxious arousal and empathy
would be inversely associated with externalizing problems (e.g. CU traits, aggression, rulebreaking, ODD symptoms; Hypothesis 2.2, 2.3) and temperamental characteristics theoretically
and empirically associated with externalizing (e.g. anger, taking pleasure in highly stimulating
activities; Hypothesis 2.19, 2.20), especially fearlessness (Hypothesis 2.15, 2.16). Conversely,
we expected that anxious arousal and empathy/guilt would be positively associated with sadness
(Hypothesis 2.22, 2.23). Taken as a whole, these findings would be consistent with a
preponderance of findings associating moderate fearfulness and anxious arousal with appropriate
moral internalization and guilt (e.g. Kochanska, 1995; Kochanska, Coy, & Murray, 2001;
Kochanska, Gross, Lin, & Nichols, 2002; Stifter, Cipriano, Conway, & Kelleher, 2009).

51

However, extreme levels of fearfulness have also been linked to low empathy, given that
these children may be oriented more towards personal distress than the distress of others
(Eisenberg et al., 2010). Thus, we anticipated a curvilinear relationship between empathy and
fearfulness, such that children with moderate levels of fearfulness demonstrate the most
empathy, while children at either extreme demonstrate low levels of empathy (Hypothesis 2.17).
If fearlessness was related at a zero-order level to the outcome of interest, this was tested by
entering a nonlinear term for fearfulness after a linear term in a hierarchical multiple regression.
Furthermore, we expected that high fearfulness would correlate positively with statements
indicative of personal distress during the Picture Tearing task.
The expected pattern of associations for facets of effortful control was partially
exploratory. In general, research supports a positive relationship between compliance with
orders and effortful control and a negative relationship with impulsivity (Kochanska, Coy, &
Murray, 2001; Kochanska et al., 1997). This relationship is especially strong for “Do” tasks, in
which the child is being asked to perform an action. However, these findings are drawn almost
exclusively from benign orders (e.g. clean up) rather than orders to do harm (e.g. tear the
picture), and the tasks involved are typically sustained behaviors (e.g. clean up the room, carry
marbles one by one) rather than single actions (e.g. tear the picture). Results from the obedience
literature suggest that the pull to obey an authority figure will, for most children, outweigh the
pull to behave morally (e.g. Shanab & Yahya, 1977). Thus, we expected that effortful control
and its facet of inhibitory control would be positively related to compliance, while impulsivity
would inversely predict compliance (Hypothesis 2.27).
Additionally, effortful control capacities have been positively linked to high moral
internalization and a developed conscience and negatively linked to disobedience (e.g.

52

Kochanska & Knaack, 2003; Kochanska, Murray, & Coy, 1997), which suggests that children in
our study with high effortful control may also demonstrate higher levels of moral internalization,
which would suggest a reduced likelihood of committing an immoral act. Given the available
literature, we predicted that effortful control would also be positively associated with anxious
arousal, empathy, and guilt (Hypothesis 2.28, 2.29).
Parenting factors. Finally, we investigated the role of broader contextual factors in
predicting the child’s ability to comply with requests and overall moral internalization.
Specifically, we examined effects of the parent-child relationship, given that aspects of this
relationship have demonstrated numerous empirical associations with compliance and
conscience development. On the whole, the literature suggests that high levels of parental
control, particularly power assertion, harsh discipline, and corporal punishment, are inversely
related to both compliance (Abe & Izard, 1999; Braungart-Rieker, Garwood, & Stifter, 1997;
Crockenberg & Litman, 1990) and conscience (empathy, guilt) (Kochanska, 1991, 1993, 2002).
A warm parent-child relationship can even suppress the deleterious effects of fearlessness on
internalization (Kochanska, 1995), while power assertion demonstrates few associations with
conscience for highly fearless children (Kochanska, 1991). Thus, in general, we expected to find
significant negative associations between observed compliance, anxious arousal, and empathy
and parent-reported poor monitoring, corporal punishment, and inconsistent punishment (see
Table 3; Hypothesis 2.30, 2.31, 2.32). Conversely, we expected that compliance, empathy, and
guilt will be positively associated with parent-reported positive parenting and involvement
(Hypothesis 2.33, 2.34, 2.35). However, we also hypothesized that these relationships would be
moderated by child fearlessness, such that the relationship between power assertion and child
empathy would be significant only for moderately or highly fearful children

53

Aim 3: Examine predictive associations between compliance, arousal, and empathy
and later externalizing behaviors. Finally, we examined whether observationally assessed
compliance, arousal, and empathy at baseline would predict outcomes at 6 months, 9 months, 12
months, 18 months, and 24 months after the initial assessment. We were especially interested in
the prediction of externalizing pathology, given its robust empirical associations with low
empathy, poorly developed conscience, and disorders of empathy (i.e., CU traits) (e.g. Asendorpf
& Nunner-Winkler, 1992; de Wied, van Boxtel, Matthys, and Meeus, 2011; Frick et al., 2003;
Frick & Morris, 2004; Frick, Ray, Thornton, & Kahn, 2014; Frick & White, 2008; Jolliffe &
Farrington, 2004). Given the accumulation of evidence, we predicted that empathy and arousal
during the Picture Tearing task will predict externalizing problems (e.g. CU traits, aggression,
rule-breaking, ODD symptoms, CD symptoms) at all follow-up time points, over and above the
variance associated with baseline levels of these behaviors. To assess discriminant validity, we
also examined the relationship between empathy and arousal and later internalizing problems. In
all analyses, mother-reported variables and father-reported variables were included and analyzed
separately.
To investigate these questions, we analyzed the patterns of change in these variables over
time using growth curve analysis estimated in HLM (Scientific Software International Inc.). We
fit a two-factor linear growth model using multi-level modeling in which the predictive variables
of interest from Picture Tearing (dimensions derived from EFA and groups from k-means
clustering) were the primary independent variables of interest, and the dependent variables were
repeated measures of EXT in children across the multiple follow-up waves. Level-1 of the
multi-level model included the within-participant change in EXT over time, and we modeled
individual differences in the intercept of EXT (baseline values) and in the linear age slope of

54

EXT. At level-2 of the multi-level model entered between subjects predictor variables (i.e.,
behavioral dimensions and clusters from Picture Tearing; other covariates of interest identified
from hierarchical regression analyses described above, such as child age and sex). In this way,
we assessed whether baseline observed guilt, empathy, and compliance reliably relate to
individual differences in change in externalizing behaviors over a period of approximately 2
years. Depending on model fit, we also investigated the possibility that the data is best
represented by a nonlinear equation (i.e., quadratic or cubic change in externalizing problems).
Specifically, we expected that noncompliance would be positively related to later externalizing
behaviors (Hypothesis 3.1), while empathy and anxious arousal would be negatively related to
later externalizig behaviors (Hypothesis 3.2, 3.3).

55

RESULTS
Recoding of Picture Tearing Variables
Across variables, there was a relatively low endorsement of behaviors, which resulted in
skewed distributions for all of the coding variables and wide ranges of responses. Absolute
values for skewness and kurtosis ranged from 0.29 to 32.23. Only seven variables had absolute
values for both skewness and kurtosis falling below 2.00, which is generally recognized as an
acceptable cutoff, although there is disagreement in the literature as to a standard rule of thumb
(e.g. Jones, 1969; see Table 6). To decrease variable skew, we recoded each variable by coding
ranges of responses into single codes. We utilized histograms and percentiles to ensure cut
points that would provide relatively even groupings with equal percentages of total participants
in each grouping, where equal groupings were possible while still maintaining adequate
variability in the codes. However, very low base rates of some behaviors prevented us from
creating equal groups for some variables; in some cases, over half of the participants scored a 0
for a given variable, so equal groupings would have been impossible without collapsing unlike
codes. In other cases, equal groupings would have reduced variability in the final code by
obscuring differences between participants who scored at the upper range of the distribution.
Cut points differed by variable, given the incredible variation in distributions across the coding
variables. Thus, applying a standard recoding solution to each variable was unreasonable.
Rather, cut points and groupings were chosen to maintain variability while reducing skewness
and kurtosis. For example, Empathy was recoded such that responses greater than 0 but less
than or equal to 1 were coded as 1, responses greater than 1 but less than or equal to 3 were
coded as 2, and responses greater than 3 were coded as 3. In this case, over half of participants
scored a 0 for Empathy, so we were unable to create equal cut points without erasing the

56

differences between any number of statements above 0. Instead, we examined all responses from
0 to the maximum and chose relatively equal groupings within this range, so that the frequencies
of the codes above 0 were relatively equal. In this way, we were able to reduced skewness and
kurtosis while preserving potentially important distinctions between children who make only 1
empathic statement and children who make 3 or more. After recoding, almost all of the variables
had absolute values for skewness and kurtosis below 2.00, with values for only two variables
(Blaming Self and Apologizing) exceeding 3.00 (see Table 7).
Responses coded during Epoch 2 demonstrated very large skewness (with no more than
30 participants exhibiting a codeable response for any Epoch 2 variable). In addition, there were
not meaningful conceptual differences between certain behaviors (e.g. empathy) exhibited in
different epochs. Specifically, there were no theoretical reasons to assume that making empathic
statements during Epoch 1 would have different correlates than making empathic statements
during Epochs 2 or 3. Thus, variables coded across all three epochs were averaged into
composites for further analyses (Empathy, Hesitation, Defiance, Social Reference, Laughing,
Smiling). It is notable that these composites correlated between .40 and .98 with the variables
that comprised them.
Recoded and composite variables were retained for all future analyses. The variables
Squirming, Lip Biting, Crossing Arms, Hiding, and Hunched Shoulders were dropped from
factor analyses due to very low rate of occurrence across epochs, even after recoding, few to no
significant correlations with other behaviors, and/or low reliability between coders.
Bivariate Correlations Among Recoded Variables
Correlations were examined between the recoded variables (see Table 8). The magnitude
of most correlations ranged from small to moderate. The associations between variables largely

57

conformed to expectations, with some exceptions. Overall, verbalizations associated with
Empathy, Hesitation, and Defiance were associated with lower task compliance, as we had
anticipated (Hypothesis 2.36), whether noncompliance was measured by how many prompts the
experimenter gave before the child complied (Number of Prompts) or by the time elapsed
between the first prompt and the child’s compliance or experimenter’s giving up (Latency to
Tear). Only Empathy (r = -0.22) and Defiance (r = -0.60) predicted less absolute compliance
(whether the child tore the picture or not) and less complete compliance (tearing the picture a
small amount as opposed to a large amount) (r = -0.27 for Empathy, r = -0.60 for Defiance).
Compliance itself (choosing to tear the picture) was also positively associated with Social
Referencing (r = 0.19), expressions of positive affect (r = 0.18 for Laughing, r = 0.28 for
Smiling), and Enjoyment in the task (r = 0.16). Only empathic statements and nonverbal signs
of discomfort (Hunched Shoulders, Squirming, Gaze Avoidance) were associated with higher
overall guilt (respectively r = 0.17, r = 0.16, r = 0.24) (Hypothesis 2.36).
We did not have specific expectations for which variables would predict responses to the
victim, as this part of the investigation was largely exploratory. Interestingly, Silence and Lying
were both associated with more Tearing (r = 0.27 for Silence, r = 0.28 for Lying) and fewer
Prompts (r = -0.34 for Silence, r = -0.22 for Lying), while more Tearing was independently
correlated with Blaming the Other (r = 0.24), Blaming the Self (r = 0.13), and Gaze Avoidance
(r = 0.17). Overall guilt was negatively associated with Blaming the Self (r = -0.16), suggesting
that children who expressed more guilt were also less likely to take full responsibility for tearing
the picture (“I did it” as opposed to “She told me to do it”). It is possible that children who
expressed more guilt were more likely to implicate the main experimenter because they were

58

more motivated to alleviate the aversive internal state (guilt) by displacing some of the blame
from themselves.
Notably, this pattern of correlations does not meaningfully or substantially differ from the
pattern of correlations observed between the variables when separated by epoch.
Overall, children who expressed empathy and defiance were less likely to comply with
the task at all, while expressions of empathy, defiance, and hesitation were all related to slower,
less enthusiastic, less thorough compliance. Only expressions of empathy were associated with
higher guilt when confronted by the victim. Furthermore, more thorough compliance (more
tearing) was associated with silence, lying, blaming the other or self, and gaze avoidance. Guilt
was actually associated with less blaming the self. Overall, these data support that meaningful
coherence can be found within task behaviors, and that this coherence is in accordance with
theorized associations (e.g. empathy predicting more guilt). This suggests that children are
responding to the Picture Tearing task as though it were a real moral dilemma. Thus, the Picture
Tearing task appears to be a useful naturalistic scenario in eliciting moral emotions and morallyrelevant behavior from young children.
Associations Between Task Behaviors and Demographic Variables and Environment
Characteristics
While above we had examined the associations among single task variables, we were also
interested in the associations between task variables and other child and environmental
characteristics that have been empirically and theoretically linked to compliance, moral decision
making, and moral emotions. We first examined bivariate correlations between task variables
(separated by epoch, totals, and factors) as well as demographics, temperament traits, and other
variables of interest (see Table 9, Table 10).

59

Age and sex. As expected, age was significantly and positively associated with empathic
statements (epoch 1) (r = 0.14), which is consistent with previous literature suggesting that older
children demonstrate better perspective taking than younger children (Hypothesis 2.6). Older
children were more likely to express hesitation and to exhibit social referencing behavior, but
age was unrelated to defiant statements. Unexpectedly, age was not correlated with whether the
child tore the photo (absolute compliance) or with latency to tear (Hypothesis 2.4). Instead, age
was negatively associated with the number of prompts given (r = -0.12) and positively associated
with how much the photo was torn (r = 0.15), contrary to our hypothesis that older children
would, by virtue of expressing more empathy, also comply less (Hypothesis 2.4). Thus, in our
sample, older children were not necessarily more likely to comply at all, but they did comply
more quickly and more extensively than younger children. Older children also demonstrated
more discomfort when facing the victim, but did not appear guiltier overall, which only partially
supports our expectation that older children would display more anxious arousal (Hypothesis
2.5).
To further examine the effect of age, we created a dichotomous variable with children
coded as younger (3 to 4 years) or older (5 to 7 years) and conducted independent samples t tests
to examine differences between the groups. This division was chosen because it has been used
as a proxy for developmental stage in previous moral development research (e.g. BraungartRieker, Garwood, & Stifter, 1997; Kochanska & Aksan, 1995; Kochanska, Aksan, & Koenig,
1995; Laupa, 1991; Laupa & Turiel, 1991). Consistent with correlational results and our
hypotheses, older children had higher means for empathy (epoch 1; p = .03), hesitation (epoch 1;
p = .04), total social referencing (p = .03), and relief for the copy (p < .05) (Hypothesis 2.6).
Older children were also surprisingly higher in overall positive affect (p < .001) and enjoyment

60

in tearing (p < .01), and were more likely to blame the other person (p < .01) but less likely to
take the blame themselves (p = .02).
While we hypothesized that girls might show slightly more empathy and anxious arousal
and less noncompliance than boys (Hypotheses 2.7, 2.8, 2.9), findings for sex were largely
consistent with the body of research suggesting that sex differences are largely nonexistent at this
age range. Sex was significantly correlated only with Squirming (epoch 3) and Relief for the
copy. Independent sample t tests similarly found that girls demonstrated more squirming (epoch
3; p < .01) and Relief (p = .01) than boys, with no other significant differences in group means.
Cognitive characteristics. We also examined the relationships between task variables
and cognitive characteristics, specifically verbal intelligence and performance on an emotion
recognition task. Consistent with our expectations, verbal intelligence (as measured by total
PPVT score) was modestly, positively associated with total empathic statements (Hypothesis
2.13) (r = 0.16). It was not correlated with anxious arousal, however (Hypothesis 2.12).
Furthermore, we found a number of associations between task variables and emotion
recognition scores that were consistent with theoretical links between poor emotion recognition
ability and poor perspective taking. As we had expected, total empathy was modestly positively
related to total correct emotions identified (how many emotions the child correctly identified out
of the six emotions presented) (r = 0.13), and was also independently, negatively associated with
incorrectly identifying sad faces (Hypothesis 2.11) (r = -0.14). Furthermore, we observed the
same pattern of associations for Social Referencing behavior, one marker of anxious arousal
(Hypothesis 2.10). However, emotion recognition scores were not associated with overall guilt.
Parenting characteristics. We also examined the association between task variables
and parent-reported characteristics of their own parenting. Consistent with our expectations, we

61

did observe some significant associations with negative parenting characteristics and child
defiance and noncompliance and between (positive parenting and adequate socialization
behaviors. Specifically, mother-reported involvement was associated with more child
expressions of empathy in the first epoch (Hypothesis 2.35) (r = 0.14). Mother-reported poor
monitoring was associated with more child defiant statements (r = 0.14) and more
noncompliance (r = 0.19) (Hypothesis 2.30), but was also unexpectedly associated with more
social referencing (r = 0.15) (Hypothesis 2.31). Similarly, and also unexpectedly, we observed
unexpected positive associations between father-reported poor monitoring (r = 0.18) and
corporal punishment (r = 0.18) and child empathic statements (Hypothesis 2.32). Results
concerning monitoring should be interpreted cautiously, as most of the items (e.g. “Your child
stays out in the evening past the time he/she is supposed to be home”) are more applicable to
older children than to children the age of those in our sample (3 to 7 years). Though this scale is
less appropriate for younger children, we included it because poor monitoring is robustly
associated with externalizing behaviors in other research (e.g. Beyers et al., 2003), and because it
does include items potentially applicable to children the age of our sample, particularly at the
older end of the range (e.g. “You leave the house without telling your child where you are
going”). Furthermore, the scale was validated in other research using samples of children as
young as 6 years of age, and thus may have utility for the oldest children in our sample (Shelton,
Frick, & Wooton, 1996).
Overall. Overall, older children expressed more empathy and hesitation than younger
children (Hypothesis 2.5, 2.6), but despite this, were actually slightly more likely to comply with
the experimenter’s request (Hypothesis 2.4). Sex effects were limited, with girls demonstrating
slightly more discomfort in the form of squirming than boys, and seeming slightly more relieved

62

when the copy of the destroyed picture was revealed (Hypothesis 2.8). Sex differences were not
observed in empathy (Hypothesis 2.9) or noncompliance (Hypothesis 2.7).
Verbal intelligence and emotion recognition ability were both associated with more
empathic statements (Hypothesis 2.11). Accurate emotion recognition was also associated with
more anxious arousal in the form of social referencing, when the child looks at the experimenter
presumably for reassurance, but was not associated with other markers of anxious arousal
(Hypothesis 2.10).
Positive parenting for mothers was associated with more child empathic statements
(Hypothesis 2.35). Poor monitoring for mothers was related to more defiance and
noncompliance on the part of the child (Hypothesis 2.30), but also unexpectedly more empathic
statements (Hypothesis 2.32). Similarly, poor monitoring for fathers was unexpectedly related to
more guilt (including both global guilt and empathic statements; Hypothesis 2.32) on the part of
the child.
Overall, these data support that task behaviors derived from the coding scheme are
reflective of meaningful differences in child standing on other internal and external
characteristics. This suggests that the coding scheme has utility in measuring behaviors that are
of theoretical interest in themselves, and also relate to a child’s behavior outside of the artificial
laboratory setting.
Exploratory Factor Analysis
We conducted principal axis factor analyses utilizing oblimin rotation in MPlus version 7
(Muther & Muther, 1998-2011). While seven factors had eigenvalues above 1.0, evaluation of
the initial scree plot suggested that no more than four factors would be appropriate. We then
conducted a principal axis exploratory factor analysis in MPlus constraining the solutions to one,

63

two, three, four, and five factors. Chi square difference analyses indicated that fit differed
significantly between all five solutions (p < .001) and that the five factor model demonstrated the
best fit, although fit statistics were modestly below typically used cutoffs (Hooper, Coughlan, &
Mullen, 2008). We also performed a parallel analysis, which compared a scree plot generated
from the actual data to what would be expected if the data were randomly generated. The
parallel analysis suggested that only four factors were larger than could be expected by chance,
which recommended a four factor structure. Empirically derived factor structures for the
solutions indicated by the scree plot and parallel analysis are provided in Table 11. We also
retained the three factor solution, given that we had hypothesized a three factor solution. This
allowed us to compare our hypothesized number of factors with the factors suggested
empirically.
Confirmatory Factor Analysis of Empirically Derived Factors
To further examine the differences between the potential factor solutions, we conducted
confirmatory factor analyses using ML estimation. We compared fit between the three, four, and
five factor solutions (loadings are described in Table 13). Indicators that did not load above .30
on any factor (Social Referencing, Blaming Self, Apologizing, and Lying for the three, four, and
five factor models and Gaze Avoidance for the three factor model) were dropped from analyses.
Notably, these variables also demonstrated few or very modest correlations with other indicators
(see Table 8).
Investigation of the fit indices (see Table 12) suggested that fit differed substantially
between factor solutions, with the five factor solution demonstrating the best fit overall
according to standard fit indices. Notably, the fit statistics for all three were outside the bounds
of typically used cut-off scores. Given that fit indices are affected by sample size and non-

64

normality, and that our sample size was relatively small and our variables somewhat skewed, the
standard cut-offs may not be optimal in evaluating this data (Nye & Drasgow, 2011). Fit for the
three factor model was not substantially improved by modifications suggested by modification
indices. The four and five factor models both demonstrated improved fit with modifications to
the latent structure. This was contrary to our expectation of a three factor model conforming to
Noncompliance, Anxious Arousal, and Empathy (Hypothesis 1.1).
In comparing the modified four and five factor structures, the four factor structure
demonstrated the best fit, though change in fit indices was modest between the four and five
factor models. The four factor model is more parsimonious. In addition, parallel analysis
(discussed above) suggests that only the first four factors are sufficiently larger than would be
expected due to chance. However, the five factor structure is consistent with theoretical
associations. The four-factor model places both empathy-related and arousal-related variables
into a single factor, though theoretically they involve related but distinct aspects of empathic
responding. Specifically, our operationalization of empathy involves statements made by the
child, which indexes perspective taking (cognitive empathy), while arousal-related variables
should be more closely related to affective empathy (Kochanska, 1993). These two facets of
empathy are overlapping but not identical. Given that the five factor model demonstrates the
best fit theoretically, and that the increment in fit observed in the four-factor model is very
modest, we chose to retain the five-factor model for further analyses. This model is described in
greater detail below.
Confirmatory Factor Analysis of Rationally Derived Factors
We also fit a model in accordance with our hypothesized (rational) factor structure (see
Table 11) with three latent factors representing Compliance, Arousal, and Empathy. The initial

65

model did not converge. The model failed to converge even when items with low factor loadings
were dropped. In comparing the structure of this rationally derived solution to the empirically
derived solutions, it appears that our initial hypotheses for Arousal conflated a tendency to
display positive affect (Smiling, Laughing) with a tendency to display discomfort (Gaze
Avoidance, Silence, Squirming). Our hypotheses assumed that all expressions of discomfort,
whether positively or negatively valenced, were expressions of the same underlying latent
variable, anxious arousal (Hypothesis 1.1). However, empirically, expressions of positive affect
and expressions of avoidance/discomfort (Gaze Avoidance, Silence, Squirming) corresponded to
separate latent factors. While smiling and laughing in this task might still represent internal
discomfort rather than internal pleasure (factor analysis is agnostic as to this possibility), these
expressions appear to be conceptually different from the other variables measuring emotional
arousal.
Final Factor Analysis
Given the data above, we chose to apply only the empirically-derived five factor solution
in choosing composites for further analysis. The five factor model retains a Noncompliance,
Positive Affect, and Avoidance/Withdrawal factor as observed in the four-factor model, but
teases apart the Empathy/Guilt factor into aspects that can be labeled as Appeal to Authority
(Hesitation, Blame the Other) and Empathy/Guilt (Empathy, How Guilty, Relief at copy). To
create composites, we obtained standardized z-scores for each variable, then created averages in
accordance with the five-factor structure. Variables making up each factor are listed in Table 11.
Factor loadings for the final five-factor model are presented in Table 13.

66

Summary of Factor Analysis
Overall, the data did not fit well into the three factor structure of Noncompliance,
Empathy, and Anxious Arousal/Guilt that we had originally hypothesized (Hypothesis 1.1).
Noncompliance did emerge as a factor. However, Empathy and Anxious Arousal did not
separate from each other cleanly. Instead, the two major moral emotions cohered, such that
empathic statements and guilt emerged together as a Guilt factor. Expressions attributed to
Anxious Arousal actually separated into three distinct factors: a Positive Affect factor, which
appears to reflect genuine positive affect rather than nervous smiling; an Avoidance factor,
which reflects behaviors designed to withdraw from confrontation; and Appeal to Authority,
which reflects behaviors designed to elicit reassurance from the experimenter.
While the data did not factor exactly as we had expected, we did observe covariance in
behavior that made sense from a theoretical standpoint. Empathy and guilt loaded onto the same
factor; while these are theoretically distinct moral emotions, they are also highly correlated, and
it thus makes sense that they might reflect the same underlying factor. Furthermore, anxious
arousal behaviors did vary together; they were simply reflective of several distinctive types of
anxious arousal, whereas we had expected them to reflect one underlying factor. On the whole,
our coding scheme does appear to measure a variety of behaviors that conform to important
aspects of moral decision making, but the distinctions with respect to anxious arousal are more
nuanced than we had first anticipated.
Associations Between Picture Tearing Task Composites and Other Child and Environment
Characteristics
While above we had examined the associations among single task variables and other
child and environmental characteristics, we were also interested in examining how these

67

relationships applied to our factor-analysis derived composite variables. We discuss
demographic variables, cognitive characteristics, and parenting variables here; the association
between task variables and temperament traits and problem behaviors is described in depth
below.
Demographic variables. Consistent with correlational results and our hypotheses, older
children had higher means for the Guilt composite (r = 0.13) (Hypothesis 2.5, 2.6).
Cognitive characteristics. Consistent with our expectations, verbal intelligence (as
measured by total PPVT score) was modestly, positively associated with the Guilt composite (r =
0.14) (Hypothesis 2.12, 2.13).
Furthermore, emotion recognition scores were not associated with the Guilt composite
(which includes both guilt and empathic statements), contrary to hypotheses (Hypotheses 2.10,
2.11). This is also contrary to correlational results for single variables, which suggested that
accurate emotion recognition predicted more empathic statements. Given that the Guilt
composite includes both empathic statements and guilt, it appears that accurate emotion
recognition is specifically related to empathic statements, such that the children with the best
emotion recognition abilities are most likely to verbalize empathy. However, given that our
situation represents an obvious transgression, even children not skilled at emotion recognition
may understand that the task is morally wrong and feel guilty.
Parenting variables. Specifically, mother-reported involvement was associated with
higher scores on the Guilt composite (r = 0.15) (Hypothesis 2.34, 2.35). Mother-reported
appropriate punishment was associated with lower scores on the Avoidance composite (r = 0.19), suggesting that appropriate punishment is related to less withdrawal from the child in the
face of their own wrongdoing.

68

Overall. On the whole, associations for the composite variables were fewer than for the
variables analyzed separately. Age and verbal ability were associated with more Guilt
(Hypothesis 2.5, 2.6, 2.12, 2.13), while emotion recognition was not (Hypothesis 2.10, 2.11).
Maternal involvement was also associated with more Guilt (Hypothesis 2.34, 2.35), while
maternal appropriate punishment was associated with less Avoidance (Hypothesis 2.34).
Overall, these data support that task behaviors derived from the coding scheme are reflective of
meaningful differences in child standing on other internal and external characteristics. Thus,
task behaviors appear to reflect children’s actual standing on moral emotions and decision
making behaviors in relation to their peers, rather than simply reflecting artificial behaviors
exclusive to the laboratory environment.
Derivation of Clusters
To derive clusters, we used a two-step method recommended by several researchers (e.g.
Hair et al., 1994; Milligan, 1980; Steinley, 2003; Swogger, 2007). The first step involves
conducting a hierarchical cluster analysis to determine the most appropriate number of potential
clusters. As the hierarchical method is heavily influenced by initial cases assigned to each
cluster, the second step involves conducting a k-means cluster analysis utilizing the number of
clusters and centroids derived from step one.
Hierarchical cluster analysis. We initially derived clusters using Ward’s hierarchical
agglomerative method of cluster analysis in SPSS (IBM version 22.0) using the recoded picture
tearing variables. However, instead of using totals collapsed across epochs, we utilized the
variables coded separately for each epoch. We also retained the Squirming, Hunched Shoulders,
and Social Reference variables that were excluded from factor analysis.

69

To determine the appropriate number of clusters, we graphed the agglomeration
coefficients against the stage number and used the “elbow method” to determine the stage at
which agglomeration coefficients ceased to increase substantially from stage to stage (e.g.
Swogger, 2007). The number of this stage is then subtracted from the total number of stages to
determine an appropriate number of clusters. This method suggested that a four cluster solution
would be appropriate. The change in agglomeration coefficients became modest after stage 5,
showing a distinct “elbow” in the plot of coefficients (Figure 3). Subtracted from the total
number of stages (269), this suggests 4 clusters.
As a check to determining the number of clusters, we also computed the gap statistic,
which indicates whether the pooled squared distances of points within an increasing number of
clusters are sufficiently different than would be expected based on a random, uniform reference
distribution (see Tibshirani et al., 2001). We computed this statistic using the clusterGenomics
package (Nilsen & Lingjaerde, 2013) in R (R Core Team, 2013). Then, according to
recommendations from Tibshirani et al. (2001), we graphed the number of clusters against the
value of the gap statistic for each cluster (Figure 4). As the graph shows, the most wellseparated cluster solution occurs at k = 1, i.e. that the data does not contain more than one “wellseparated” cluster. However, the gap statistic rises again at k = 3, suggesting that there are at
least three less separated “sub-clusters” within the data.
On the whole, these methods suggest that there might be as many as four distinct clusters
within the dataset. However, separating the data into multiple clusters should be interpreted
cautiously, as the distance between these cluster centers is not large enough to suggest welldefined, maximally separated groups. Thus, it is possible that there are three to four subclusters

70

within one single cluster. Further analyses will evaluate the characteristics of these subclusters
and whether or not there are meaningful differences between them.
K-means cluster analysis. Next, we fit a four cluster solution to the data using the kmeans cluster procedure in SPSS, specifying k = 4, and a three cluster solution. After further
analyses, we retained the three cluster solution, described below. Descriptions of clusters in the
four factor solution are provided in Appendix B.
Description of clusters in three cluster solution. Clusters 2 and 4 significantly differed
only on Latency to tear (d = 2.98) and Relief in seeing the photo copy (d = 0.67). Notably, they
did not differ on the constructs theoretically most important to moral decision making (i.e.
Empathy, Defiance, Guilt). In addition, the fourth cluster contained only 6% of the sample. To
determine whether collapsing clusters 2 and 4 substantially changed their relationships with the
other clusters, we computed descriptives for a three cluster solution (with clusters 2 and 4
collapsed). We then conducted a one-way ANOVA using cluster membership from the new
three-cluster solution and conducted post-hoc comparisons of significant ANOVAs using
Tukey’s HSD. We evaluated the three clusters according to the characteristics on which they
significantly differed. Means and standard deviations for each cluster are presented in Table 14.
Graphs showing significant differences between cluster means are presented in Figure 5.
Clusters are described individually below. On the whole, they fell along a continuum of
behaviors, with Low Compliers demonstrating the least compliance and most empathy, High
Compliers demonstrating the most compliance and least empathy, and Moderate Compliers
falling in the middle on these markers. High Compliers expressed the most guilt of the clusters,
while Low Compliers expressed the least guilt; this lack of guilt in Low Compliers despite their
high levels of empathy may be due to the fact that Low Compliers were the most likely to refuse

71

to do the task, and thus had no reason to feel guilt afterwards, as they had not performed a
transgression. On the other hand, High Compliers were the most likely to tear the photo, and still
demonstrated substantial guilt despite their low levels of expressed empathy. There were no
differences between clusters on observer-rated sociability across laboratory tasks, suggesting that
the children’s differing patterns of behavior in the Picture Tearing task cannot simply be
accounted for by their average levels of social engagement.
Cluster 1: Moderate compliers. Members of this cluster comprised 24.9% of the sample
(n = 67). On most behaviors of interest, members of cluster 1 fell between clusters 2 and 3, and
were thus moderate in their behaviors. Cluster 1 fell between the High and Low Compliers on
expressions of defiance in response to the experimenter’s prompt, but only expressed more
empathy and hesitation than the High Compliers. With respect to compliance, cluster 1 also fell
in the middle on latency to comply, absolute compliance, and extent of compliance, as well as
enjoyment in tearing. Finally, cluster 1 also fell into the middle with respect to guilt and relief
for the copy.
Cluster 2: Low compliers. Members of this cluster comprised 16.0% of the sample (n =
43). Compared to the other clusters, cluster 2 demonstrated the most defiance. Members
showed more empathy and hesitation than the High Compliers, but did not differ from the
Moderate Compliers. With respect to compliance, cluster 2 members took the longest time to
tear, were least likely to comply, complied to a lesser extent, and showed less enjoyment while
tearing than the other clusters. Interestingly, when they did comply, they also showed the least
guilt and the least relief when the copy was presented.
Cluster 3: High compliers. Members of this cluster comprised 59.1% of the sample (n =
159). On the whole, members of cluster 3 were the most enthusiastically compliant and the least

72

empathic and defiant. Compared to the other clusters, members of cluster 3 expressed the least
empathy, hesitation, and defiance in response to the experimenter’s prompt. In keeping with
their lack of verbal dissent, they were also the most likely to comply, to the greatest extent, in the
shortest amount of time, and with the most enjoyment. Finally, they expressed the most guilt
and the most relief for the copy. Compared to the Low Compliers, they were more likely to
respond to the victim by blaming another, lying, or being silent.
External validation. After we determined the relative standing of clusters on the
variables used to derive them, we examined whether the clusters demonstrated differential
relationships with external variables on which we would expect them to show meaningful
differences. We compared each pair of clusters across the external criterion variables utilizing
analysis of variance (ANOVA) and Tukey HSD post hoc tests in SPSS (IBM version 22.0).
To determine whether collapsing clusters 2 (Low Compliers) and 4 (which was included
in the description of Low Compliers above) substantially changed their relationships with
external criterion variables, we compared results of ANOVA analyses between the four cluster
solutions and the three cluster solution (with clusters 2 and 4 collapsed). Results showed that the
three cluster solution showed the same pattern of significant external associations as the four
cluster solution (Child age, Mother-reported attention shifting, Mother-reported shyness,
Emotion recognition scores), but showed additional significant between-cluster differences that
were obscured in the four cluster solution. Given the lack of differentiation between clusters 2
and 4 on both internal and external criteria, we collapsed these clusters and treated them as one
cluster for remaining analyses. As an additional check, we computed the k-means analysis again
setting k equal to 3. In this solution, clusters 1 and 3 were identical to their counterparts in the
four-cluster solution; the only cases that changed membership were those in cluster 4, which all

73

were sorted into cluster 2. This check provided further support that clusters 2 and 4 were not
dissimilar enough to be analyzed separately. In addition, the three cluster solution is more
parsimonious, as there is no theoretical basis to assume that groups differing only on latency to
tear and relief would demonstrate substantively different relationships with important correlates.
ANOVA results for the three cluster solution are described in more detail below.
Demographics. Cluster membership was not significantly predicted by child sex or
parent ethnicity (Hypothesis 2.7, 2.8, 2.9). Cluster membership was significantly predicted by
child age (F = 5.01, p < .01). Post-hoc tests indicate that Moderate and Low Compliers did not
differ on age, and Low and High compliers did not differ on age. On average, High Compliers
were significantly older than Moderate Compliers (d = 0.47). This is consistent with our
correlational findings, given that age was positively associated with compliance and the Higher
Compliers demonstrated the most compliance of the three groups (Hypothesis 2.4). It further
supports that our hypothesis, that older children would feel more empathy and thus comply less,
was incorrect; the press to comply with authority trumped the press to act empathically.
However, the fact that High and Low Compliers did not differ on age suggests that factors other
than age were contributing to the greater noncompliance and empathy of the Low Compliers.
IQ and emotional intelligence. Cluster membership was not significantly predicted by
child verbal intelligence, as measured by the child’s baseline PPVT score (F = 826.17, p = .17),
contrary to our hypotheses that differences in empathy and anxious arousal would be associated
with differences in verbal intelligence (Hypothesis 2.12, 2.13). Cluster membership was also not
predicted by emotion recognition scores, indicating that children in different clusters did not
differ, on average, in the number of emotions they correctly identified (F = 2.80, p = .53),

74

contrary to our hypotheses that differences in empathy and anxious arousal would be associated
with differences in verbal intelligence (Hypothesis 2.10, 2.11).
Temperament traits. Cluster membership was not predicted by either mother-reported or
father-reported traits of Anger, Inhibitory Control, Approach, Smiling, Low intensity pleasure,
Impulsivity, Sootheability, Attention Focusing, Sadness, or Fearfulness (Hypothesis 2.14 through
2.29). This is surprising, given that we expected that Fearfulness and facets of effortful control
would be related to task compliance. However, clusters did differ on mother-reported
Attentional Shifting (F = 8.41, p = .02), such that the Low Compliers demonstrated, on average,
poorer ability to shift their attention according to task demands than did High Compliers (d =
0.52). Clusters differed on both mother-reported (F = 5.22, p < .01) and father-reported shyness
(F = 3.36, p = .04). According to mother-report, High Compliers were the least shy of the
sample, as compared to both Low Compliers (d = 0.50) and Moderate Compliers (d = 0.41). For
father-reported shyness, differences were no longer significant after post-hoc tests correcting for
multiple comparisons. This association was contrary to our hypotheses, as we expected that
more assertive children would be less likely to comply. Finally, clusters differed on both
mother-reported (F = 3.33, p = .04) and father-reported (F = 3.27, p = .04) High Intensity
Pleasure. High Compliers were rated as seeking more intensely stimulating, pleasurable
activities than Low Compliers by both mothers (d = 0.46) and fathers (d = 0.50). This finding
was consistent with the fact that High Compliers were rated the highest on positive affect of all
three groups. It was also inconsistent with our predictions; given that stimulation seeking is a
characteristic associated with externalizing behaviors, we expected that children who scored high
on High Intensity Pleasure would be less compliant than children who scored low. It is more
consistent with our contention that High Intensity Pleasure, as an associated feature of

75

externalizing behavior, might predict low empathy. Indeed, the High Compliers, who were rated
highest on High Intensity Pleasure, also demonstrated the lowest empathy. In this case, the
“thrill” of tearing the picture might have been more salient to these children than the opportunity
to be noncompliant.
Overall, the most compliant children were also the most stimulation seeking and
demonstrated the most effortful control (specifically ability to shift attention), while the least
compliant children demonstrated the least effortful control. The most compliant children were
also the least shy, suggesting that perhaps they were more motivated than more shy children to
acquiesce to the experimenter’s request, as a way of promoting the relationship between
themselves and the experimenter.
Problem behaviors. Cluster membership was not predicted by any of concurrent
problem behaviors measured by the CBCL, including global externalizing and internalizing and
CU traits. This result is contrary to our hypotheses that either task compliance in the absence of
empathic statements, or task noncompliance, would both be related to externalizing behaviors
(Hypothesis 2.1, 2.2, 2.3).
Parenting behaviors. Cluster membership was not predicted by any parenting behaviors
measured by the APQ, rated by either mothers or fathers. This is contrary to our hypotheses that
harsh and inconsistent parenting would be indirectly related to task compliance (Hypothesis
2.30), given that this type of parenting is empirically associated with less empathy.
Overall. Overall, the most compliant children were also the most stimulation seeking and
demonstrated the most effortful control, while the least compliant children demonstrated the least
effortful control. These differences in clusters cannot simply be accounted for by age or emotion
recognition abilities, as the two extreme groups (High and Low Compliers) did not differ on age

76

or emotion recognition. There were also no differences in externalizing problem behaviors or in
parenting behaviors theoretically linked to problem behavior. Temperament, specifically
effortful control, seems to be playing a role independent of cognitive maturity (indexed by age,
verbal ability, and emotion recognition) in predicting compliance and empathy in this task.
These analyses demonstrate that the coding scheme does allow us to observe meaningful
differences in patterns of child behavior within this task.
Gender of Victim as a Predictor of Child Task Behaviors
In order to determine whether the gender of the victim was associated with any
differences in child behavior, we first coded each video for gender of the victim (male = 0,
female = 1). The vast majority were female (n = 266, 98.5% of videos), with only four male
victims (n = 4, 1.5% of videos). We then conducted t tests comparing the means of all of our
task variables separated by epoch, task variable totals, the five empirically derived factors, and
cluster membership, to determine whether these means differed between male and female
victims. Children in the presence of female victims demonstrated more Hunched Shoulders,
Social Referencing, and Laughter in epoch 1 and more Hunched Shoulders in epoch 3. These
differences, though statistically significant, are theoretically minor. No differences were
observed on key constructs (i.e. Empathy, Guilt, any of the compliance variables) related to
moral decision making or to any of our hypotheses.
It is also important to note that the male experimenter was the same person in each of the
4 instances of a male victim, so it is unclear whether effects can be generalized to male victims
in general. In addition, given that a male victim was present in only 4 videos, and that the base
rates of each task behavior were largely positively skewed, these differences are likely due to

77

chance rather than meaningful disparities. Given these considerations, any meaningful
conclusions on the influence of victim sex at this point would be premature.
Prediction of Concurrent Problem Behaviors Using Composites Derived from Factor
Analysis
Correlations between independent variables and dependent outcomes.
Picture tearing composites. We conducted correlations between our outcome variables
(EXT, INT, CD symptoms, ODD symptoms, Rule-breaking, Aggression, and CU traits) and our
independent variables. As independent variables, we included the five composites formed by the
factor analysis (Guilt, Appeal to Authority, Noncompliance, Positive Affect,
Avoidance/Withdrawal.
We found few associations between task behaviors and problem behaviors (see Table
15). However, opposite to our expectations, we did find that the Guilt composite was positively
associated with mother-reported Aggression (r = 0.15), ODD symptoms (r = 0.17), and EXT (r =
0.14).
Demographic variables. We examined the correlations between problem behavior
outcomes and child sex and age (Table 15). With respect to problem behaviors, child sex did not
correlate with any of the outcome variables and was excluded from further analyses. This is
somewhat unexpected, given that boys typically demonstrate more externalizing behaviors than
girls (e.g. Broidy et al., 2003). However, our sample is composed of younger children, mostly of
preschool age; some studies have found that sex differences in externalizing behaviors are
smaller or nonexistent in the early preschool years (Keenan & Shaw, 1994; Rose et al., 1989).
Age demonstrated small positive correlations with CD symptoms (r = 0.17), INT (r = 0.26), and

78

Rule-breaking (r = 0.16) (mother-reported outcomes only) and was included in block 1 of these
regressions.
Verbal and emotional ability. We also examined correlations between problem
behaviors and verbal ability (measured by the PPVT overall score) and emotion recognition
ability (measured by performance on the emotion training lab task) (Table 15). PPVT score
showed a small positive correlation with mother-reported INT (r = 0.16). Of the emotion
recognition variables, total emotions correctly identified and misidentifying happy, sad, fearful,
and neutral expressions did not correlate with any outcomes. Misidentifying angry expressions
correlated positively with CU symptoms (r = 0.29), ODD symptoms (r = 0.18), and Aggression
(r = 0.20) (father-reported outcomes only).
Temperament traits. We also conducted correlations between problem behavior
variables and higher order temperament traits. We examined correlations between these traits
and problem behaviors in order to determine whether to include them as steps in hierarchical
regressions according to our planned analyses. Consistent with previous literature, higher
Effortful Control (bother father- and mother-reported) was associated with fewer externalizing
problems of any kind (mother- or father-reported Aggression, Rule-breaking, ODD, CD, EXT,
CU traits), and was not associated with internalizing problems. Mother-reported Negative
Emotionality was positively associated with mother-reported Aggression, ODD, INT, and EXT,
and father-reported Negative Emotionality was positively associated with all father-reported
problem behaviors. Finally, mother-reported Positive Emotionality was positively associated
with mother-reported Aggression, Rule-breaking, ODD, CD, and EXT, while father-reported
Positive Emotionality was positively associated with father-reported Aggression, Rule-breaking,
ODD, CD, EXT, and CU traits.

79

With respect to temperament traits coded from other laboratory tasks, we examined
correlations between fearfulness and compliance and problem behaviors. Fearfulness was not
correlated with any of the problem behaviors. Compliance was negatively associated with
father-reported Aggression, Rule-breaking, ODD, CDD, and EXT, but did not demonstrate any
correlations with mother-reported problem behavior.
Parenting characteristics. Finally, we examined the associations between parenting
characteristics rated by mothers and fathers and problem behaviors (Table 15). Father-rated Poor
Monitoring, Corporal Punishment, and Appropriate Punishment did not correlate with any of the
outcomes (with the exception of a modest positive association between Corporal Punishment and
father-rated CU traits, r = 0.20). This is somewhat surprising, given the well-validated
association between harsh and inconsistent parenting and externalizing problems. However,
consistent with hypotheses, mother- and father-reported Involvement and Positive Parenting
were associated with fewer problem behaviors overall, while mother-rated Inconsistent
Discipline, Poor Monitoring, and Corporal Punishment were associated with more problem
behaviors. Unexpectedly, mother-reported Appropriate Punishment was actually associated with
more mother-rated Aggression, ODD, and Rule-breaking, but with less father-reported total INT
and CU traits. It is possible that mothers whose children demonstrate more externalizing
behavior also view themselves as delivering more punishment in general than mothers whose
children are low on externalizing.
Overall. Overall, few associations were observed between externalizing outcomes and
Picture Tearing composities. Notably, Guilt was positively associated with aggression, ODD
symptoms, and CU traits. Age was positively associated with CD symptoms, INT, and Rulebreaking. Verbal abilities were only correlated with slightly more INT, while overall emotion

80

recognition did not predict any problem behaviors. Only misidentifying anger specifically was
correlated with more CU symptoms, ODD symptoms, and aggression.
As expected, higher effortful control was consistently associated with fewer externalizing
problems of any kind, while both higher negative emotionality and positive emotionality were
consistently associated with more externalizing problems. Laboratory-observed fearfulness was
not associated with any problem behaviors, but laboratory-observed compliance was associated
with less father-reported externalizing.
Finally, dimensions of harsh or inconsistent parenting were related to externalizing
behaviors, but only for mother-reported parenting. Positive parenting and involvement both
predicted fewer problem behaviors.
Hierarchical regressions predicting problem behaviors from task behavior
composites. According to our analytic plan, we first performed a series of hierarchical
regressions controlling only for child age in block one (where age demonstrated a significant
zero order correlation with the dependent variable of interest), with the composite task variables
in block two. For each significant regression, we calculated ΔR2 for each block to determine
whether the predictors accounted for significant variance beyond the effects of age and sex.
With respect to mother-reported child outcomes, the laboratory composite variables
significantly predicted ODD symptoms over and above child age and sex (ΔR2 = .06, p = .04,
Cohen’s f2 = .06). Only the effect for Guilt was uniquely significant (B = .12, p = .01), but in the
opposite direction expected (Hypothesis 2.3). No models predicting father-reported outcomes
were significant.
Hierarchical regressions controlling for other child and environment
characteristics. Next, we retained the significant regressions and added additional steps

81

accounting for other child and environmental characteristics. Only variables that had significant
zero order correlations with the dependent variable, as described above, were included in each
regression. Two regressions were performed for each problem behavior, one in which
temperament traits were entered in the block before parenting variables, and one with the order
reversed.
In predicting mother-reported ODD symptoms, the effect of Guilt was significant (B =
.11, p < .01), but did not account for additional variance beyond the effects of temperament and
parenting variables (ΔR2 = .04, p = .08, Cohen’s f2 = .04) (Hypothesis 2.3). The values of the
slope for Guilt and change in R2 for that block were identical whether parenting variables were
entered before or after temperament traits. Thus, regardless of whether parenting characteristics
or temperament traits are casually prior in the prediction of ODD symptoms, there is no
remaining variance in ODD that is significantly explainable by the addition of Guilt.
Summary of hierarchical regressions using task composites. Overall, there were few
significant regressions. We had hypothesized that noncompliance would predict more EXT
(Hypothesis 2.1), while anxious arousal/ guilt and empathy would predict less EXT (Hypothesis
2.2, 2.3). In fact, the Guilt composite, which contained both overall guilt and empathic
statements, predicted more ODD symptoms, but was no longer a significant predictor once
variance due to child age, temperament, cognitive characteristics, and parenting was accounted
for. Thus, task composites do not demonstrate utility in predicting concurrent externalizing
behaviors. It seems that the “slice” of behavior observable in Picture Tearing is not eliciting
either the degree or variability of moral emotions that would be indicative of variability in
externalizing problems outside of the laboratory.

82

Prediction of Concurrent Problem Behaviors Using Individual Task Variables
Correlations between independent variables and dependent outcomes. As outlined
in our aims, we also investigated the association between single variables derived from the
coding scheme and problem behavior outcomes. We utilized the variables indexing total
behavior across epochs (Empathy, Hesitation, Defiance, Social Reference, Laughing, Smiling),
the variables indexing compliance and extent of compliance (Latency to Tear, Tear or Not, How
Much Torn, Enjoyment in Tearing), and the variables indexing guilt and arousal in response to
the victim (Gaze Avoidance, Overall Guilt, Relief for Copy). Overall, these variables broadly
cover our domains of interest: compliance, empathy, and guilt/anxious arousal.
Similar to our results using factor composites, we found unexpected positive associations
between Empathy and mother-reported Aggression, ODD symptoms, and EXT (Hypothesis 2.3).
We also found unexpected positive associations between Social Referencing and motherreported Aggression, Rule-breaking, CD, and EXT (Hypothesis 2.2). Finally, we found, as
expected, a modest positive correlation between Enjoyment in Tearing the picture and motherreported CD symptoms.
Hierarchical regressions predicting problem behaviors from task behavior.
According to our analytic plan, we first performed a series of hierarchical regressions controlling
only for child sex and age in block one, with the composite task variables in block two. For each
significant regression, we calculated ΔR2 for each block to determine whether the individual task
behaviors accounted for significant variance beyond the effects of age and sex.
With respect to mother-reported outcomes, total Empathy (B = .08, p = .02) and Social
Referencing (B = .05, p = .12) predicted Aggression above and beyond sex and age (ΔR2 = .04, p
= .06, Cohen’s f2 = .04), with only the effect of Empathy being significant. Total Empathy (B =

83

.17, p < .01) also significantly predicted ODD symptoms (ΔR2 = .05, p < .01, Cohen’s f2 = .05).
Finally, the combination of Empathy (B = .05, p = .06) and Social Referencing (B = .04, p = .08)
predicted EXT over and above age and sex (ΔR2 = .03, p = .03, Cohen’s f2 = .03). All effects
were small in magnitude. No regressions were significant for father-reported outcomes. Again,
the positive association between Empathy and EXT was contrary to our hypotheses (Hypothesis
2.3).
Hierarchical regressions controlling for other child and environment
characteristics. Next, we retained the significant regressions and added additional steps
accounting for other child and environmental characteristics. Only variables that had significant
zero order correlations with the dependent variable were included in each regression. Two
regressions were performed for each outcome, one in which temperament traits were entered in
the block before parenting variables, and one with the order reversed.
With the addition of those cognitive, temperament, and parenting variables that were
significantly correlated with ODD symptoms, picture tearing task variables only remained
significant in predicting mother-reported ODD symptoms (ΔR2 = .03, p = .01), with total
Empathy as the significant predictor of more ODD symptoms (B = .14, p = .01) (Hypothesis 2.3).
Summary of hierarchical regressions using individual task variables. As with
regressions using task composites, few regressions were significant. However, notably, total
empathic statements continued to predict ODD symptoms beyond the variance explained by
cognitive characteristics, temperament, and parenting. This is contrary to our hypothesis that
empathic statements would be associated with less EXT (Hypothesis 2.3). It is also different
from results for the Guilt composite; once other variables were added, Guilt did not continue to
predict ODD symptoms. Guilt contains both overall guilt and empathic statements. These

84

results suggest that the prediction of ODD symptoms is being driven by empathic statements,
and the addition of guilt actually obscures the predictive relationship between empathic
statements and ODD symptoms. With respect to our aim of using task behaviors to predict
externalizing behaviors, it seems that only empathic statements are a useful predictor of
externalizing behavior; however, contrary to our hypotheses, they are a better index of
oppositionality than of adaptive moral understanding (and thus less externalizing).
Noncompliance and anxious arousal within the task did not emerge as useful predictors of EXT
in either direction (Hypothesis 2.1, 2.2). Thus, it would appear that while the Picture Tearing
task does allow for the elicitation of moral behavior, and variations in this moral behavior
between children do not reflect variation in externalizing behaviors outside of the laboratory.
Prediction of Concurrent Temperament Traits Using Picture Tearing Variables
Correlations between independent variables and dependent outcomes. Finally, we
decided to conduct a series of analyses examining the associations between picture tearing
variables and relevant temperament traits. These associations were not initially outlined in our
aims, as we were more interested in predicting externalizing behaviors. However, we observed
through our analyses of clusters that clusters differing on empathy and compliance also
demonstrated meaningful differences on temperament traits, particularly those related to effortful
control (e.g. inhibitory, impulsivity). Furthermore, parent-reported temperament was associated
with parent-reported behavior problems, suggesting the possibility that, at the age of children in
our sample, picture tearing behaviors are associated with traits related to risk for
psychopathology more strongly than behavior problems themselves. Thus, we wanted to assess
more stringently how behaviors during the picture tearing task relate to variations in child
temperament. Results of correlations are presented in Table 16.

85

Consistent with results from the cluster analysis, most of the significant associations were
observed with facets of effortful control. Our hypotheses for the relationship between effortful
control and compliance largely held (Hypothesis 2.27), while the results for effortful control and
empathy and guilt violated our expectations (Hypothesis 2.28, 2.29). Specifically, we
hypothesized that effortful control would be positively associated with compliance, empathy, and
guilt. Broadly, facets of effortful control were indeed modestly related to less overall
Noncompliance, less delay in complying (Latency), and greater extent of compliance (How
Much Torn), but was unexpectedly related to less Empathy and Guilt (either Overall Guilt or the
Guilt Composite). At the facet level, however, the relationships with Empathy and Guilt were
significant only for Attentional Shifting and Impulsivity, and Inhibitory Control did not relate to
Empathy, Guilt, or Noncompliance at all. Notably, these associations were all small in
magnitude and most were observed only for mother-reported temperament traits.
On the whole, the results for inhibitory control are somewhat surprising, given that
inhibitory control has empirically and theoretically been associated with better ability to inhibit
preponent responses, such as defying an adult authority. Thus, we would have expected that
children with better inhibitory control would be less noncompliant, whereas the association in
our data was null. However, it is more surprising that impulsivity was positively related to
empathy and guilt, while attentional shifting was negatively related to these constructs. Past
literature has linked poor effortful control with poorer moral internalization, which would
suggest the opposite associations.
With respect to higher order temperament variables, mother-reported Effortful Control
was associated with less Noncompliance (r = -0.18) and Defiance (r = -0.14) and father-reported
EC was associated with less Empathy (r = -0.17). Mother-reported Negative Emotionality was

86

modestly associated with more Hesitation (r = 0.14), less absolute compliance (Tear or Not) (r =
-0.19), and less extent of compliance (How Much Torn) (r = -0.16). It appears that children
higher on negative emotionality were more tentative in their response to the main experimenter’s
request, potentially because they are more motivated than children low on NE to avoid
committing an act that will “get them into trouble” and subsequently cause them to feel
additional negative emotions.
Finally, mother-reported Positive Emotionality was modestly associated with more Guilt
(composite) (r = 0.18), Authority (r = 0.14), and Relief for the copy (r = 0.17) and fatherreported PE was associated with more Social Referencing (r = 0.18). Given that positive
emotionality is generally correlated with higher sociability, it appears that children higher on
positive emotionality in our sample were accordingly more likely to seek closeness and positive
interaction with the experimenters, whether they expressed this by expressing concern for the
victim, seeking approval from the main experimenter, or sharing in the victim’s positive
emotions when the loss of the cherished photograph is repaired.
Hierarchical regressions predicting temperament traits from task behavior. We first
performed a series of hierarchical regressions controlling only for child sex and age in block one
(where age and sex were significantly correlated with the outcome in question), with the
composite task variables in block two. We also performed a series of regressions with age or sex
in block one and dummy coded contrasts comparing clusters in block two. For each significant
regression, we calculated ΔR2 for each block to determine whether the predictors accounted for
significant variance beyond the effects of age and sex.
The first set of regressions used the five empirically-derived picture tearing factors as
predictors in one block. With respect to mother-reported traits, the empirically derived factors

87

significantly predicted Attentional Shifting (ΔR2 = .09, p < .01); specifically, children who were
more likely to seek approval from the experimenter demonstrated less attentional shifting
(Authority; B = -.20, p < .05), while children who withdrew through silence or avoided gaze
(Avoidance; B = .20, p < .05) were higher on attentional shifting. Impulsivity was significantly
predicted (ΔR2 = .12, p < .01) by more Guilt (B = .22, p < .01) and more appeals to Authority (B
= .20, p = .01). On the whole, these results are consistent with the notion that children who have
poorer effortful control (here, attentional focusing and shifting) are less equipped to execute and
comply with commands issued by authorities. However, another facet of effortful control,
impulsivity, was associated not with noncompliance, but with exhibitions of empathy/guilt and
anxious arousal. It is possible that, in our task, impulsive children were more likely to comply
with the task because the thought of doing something usually forbidden was highly appealing,
without considering the consequences. Subsequently, they might be more likely to express signs
of guilt once faced with the consequences.
The regression predicting shyness was also significant (ΔR2 = .08, p < .01), with
Noncompliance (B = .49, p < .01) unexpectedly associated with more shyness. This finding may
reflect the way that shyness is measured via the Child Behavior Questionnaire. This subscale
contains items suggesting a child who is slow to engage with strangers, inhibited around
unfamiliar people, and nervous in the presence of novel adults. It is possible that the shyness
here reflects an unwillingness or reluctance on the part of the child to engage with the unfamiliar
experimenter, rather than simply the opposite of assertiveness/surgency. Thus, the shy child
might simply fail to comply because their tendency is to withdraw from the stranger and her
request, not because they are being intentionally defiant.

88

Variance in High Intensity Pleasure was also primarily driven by Noncompliance (B = .32, p = .02), such that children who complied less were rated as being more likely to seek highly
stimulating activities; this association is consistent with our hypotheses (Hypothesis 2.18), given
that both noncompliant behavior and sensation-seeking behavior are consistently empirically
associated with externalizing pathology and with each other.
Variance in Attentional Focusing (ΔR2 = .05, p = .04) was also predicted by
Noncompliance (B = -.33, p < .01), such that more noncompliant children demonstrated worse
attentional focusing, consistent with our expectations (Hypothesis 2.27). Higher-order Positive
Emotionality (ΔR2 = .06, p = .02) was predicted primarily by Guilt (B = .10, p < .01), with
children who exhibited more guilt also being rated as generally exhibiting more PE. For fatherreported temperament traits, none of the regressions predicted significant change in R2 after
accounting for age. We further examined the potential effect of cluster membership on
temperament. With respect to mother-reported temperament traits, the contrast between High
and Low Compliers (B = -.23, p = .02) predicted significant change in the variance of Attentional
Shifting (ΔR2 = .04, p = .02). The contrast between clusters High and Low Compliers (B = .40, p
< .01) significantly predicted Shyness (ΔR2 = .05, p = .01). The contrast between clusters High
and Low Compliers (B = -.19, p = .04) also predicted High Intensity Pleasure (ΔR2 = .03, p <
.05). With respect to father-reported temperament traits, the contrast between clusters High and
Low Compliers (B = .37, p = .01) predicted Shyness (ΔR2 = .05, p = .04). These results provide
further support for the conclusions suggested by the regressions utilizing factor composites as
predictors. Attentional Shifting, Shyness, and High Intensity Pleasure were all associated with
membership in the cluster whose members were most noncompliant.

89

Finally, we examined variance in child temperament measured in the other laboratory
tasks. Picture tearing composites did not predict variance in fearfulness in other tasks (ΔR2 =
.02, p = .58), contrary to our expectations (Hypothesis 2.24, 2.25, 2.26). However, they did
predict variance in compliance in other tasks above and beyond age (ΔR2 = .05, p < .05). The
effect for Noncompliance (B = -.16, p < .05) was significant, such that noncompliance in Picture
Tearing was associated with slightly less compliance in other laboratory tasks, which conforms
to our expectations (Hypothesis 2.24). These results suggest that children who were likely to be
defiant in general were, in fact, somewhat more defiant in Picture Tearing as well. However, the
very small magnitude of this effect is interesting in that it suggests that noncompliance in Picture
Tearing and in other tasks were only modestly related to one another. This further suggests that,
conceptually, noncompliance in Picture Tearing does not simply reflect a general tendency not
to comply, but is being driven by other factors.
Summary of regressions predicting temperament traits from task behaviors.
Overall, task behaviors were most useful in predicting variations in effortful control, though the
direction of the associations was opposite to what we had anticipated for empathy and anxious
arousal (Hypothesis 2.28, 2.29). Impulsivity was predicted by higher Guilt, while Attentional
Shifting was predicted by both less Appeal to Authority figures and more Avoidance/Withdrawal
behaviors. Better Attentional Focusing was predicted by less noncompliance, providing further
support for a positive relationship between task compliance and effortful control (Hypothesis
2.27). Noncompliance also predicted both more shyness and more high intensity pleasure
seeking.
With respect to emotional expression, greater Positive Emotionality was predicted by
higher Guilt. This association may be due in part to a link between extraversion / lack of shyness

90

and a tendency to verbalize empathic statements, as the Guilt composite also contains empathic
statements. There were no significant regressions for Negative Emotionality.
With respect to cluster membership, the difference between High and Low Compliers
predicted variance in Attentional Shifting, Shyness, and High Intensity Pleasure. This again
supports that ability to shift attention allowed for less noncompliance, while children high on
stimulation seeking and low on shyness were more likely to perform the task, perhaps either
because it seemed fun or because they wanted to please the experimenter.
With respect to performance in other laboratory tasks, there were no significant
regressions predicting observed fearfulness (Hypothesis 2.24, 2.25, 2.26). Noncompliance in
Picture Tearing was associated with lower compliance in other laboratory tasks (Hypothesis
2.24). However, the modest magnitude of the association suggests that noncompliance in Picture
Tearing was not wholly reflective of a general tendency not to comply, but was also being driven
by other factors unique to the task.
These data suggest that, while Picture Tearing behaviors have limited utility in predicting
variation in child externalizing behaviors, they are indeed reflective of variations in important
aspects of child temperament. In particular, variations in child noncompliance, empathy, and
anxious arousal/guilt are associated with differences in effortful control faculties, such that
children highest on effortful control were the most compliant and expressed the fewest
reservations in response to the experimenter’s request. Task behavior within Picture Tearing
seems to reflect a child’s ability to follow a request and to inhibit task-inconsistent behavior.

91

Longitudinal Prediction of Problem Behaviors Using Composites Derived from Factor
Analysis
We conducted growth curve analyses looking at effects of our composite picture tearing
variables and cluster membership on problem behavior outcomes (Aggression, Rule breaking,
CD, ODD, CU traits, EXT, and INT) over time, from baseline out to the 24-month assessment,
across all 6 waves. Multilevel models (MLM) were estimated using HLM 7.0 (Scientific
Software International Inc.). Repeated assessments (CBCL scales) were nested within
participants at level-1 of the MLM; child-level factors (i.e., picture tearing variables) were
entered at level-2 of the models. For each outcome, the level 1 model describes the linear and
quadratic effects of age on the outcome. The intercept was set at 36 months (the youngest age of
any participant) and age-related change was modeled as change in months after that. We used
MLM to test whether picture tearing task variables at level-2 predicted individual differences in
children’s initial level (intercept) and age-related change in behavior problems.
At level 1, CD symptoms increased modestly along a quadratic trajectory as children
aged (p = .04), while ODD symptoms and total EXT did not demonstrate significant age-related
change without the addition of level 2 predictors, as described below. Total INT did not
demonstrate age related change in the sample, nor did picture tearing variables significantly
affect change in INT over time.
With the addition of level 2 predictors (picture tearing variables), we observed that
individual differences in intercepts and age-related trajectories for CD, ODD, and EXT
demonstrated significant associations only with Guilt. Unexpectedly, children high on Guilt
started out at baseline with higher levels of EXT, CD symptoms, and ODD symptoms than their
low-Guilt peers (Hypothesis 3.2, 3.3). However, these children also decreased more quickly

92

over time than low-Guilt children in EXT, CD symptoms, and ODD symptoms. This decrease is
consistent with our assertion that empathy and anxious arousal/guilt would negatively predict
EXT (Hypothesis 3.2, 3.3). Specifically, at the time of laboratory assessment, children who
expressed more empathy and guilt were also higher than their peers on externalizing pathology.
However, this effect appears to be limited to concurrent associations. When considering the data
longitudinally, high levels of Guilt appear to predict a lessening of externalizing pathology and a
return to a more adaptive developmental trajectory over time.
Considering the data as a whole, this may reflect an intersection of processes, such that
children who are more impulsive and show less effortful control at baseline display more guilt
and express more empathy at the time of the laboratory task. High levels of impulsivity and low
levels of effortful control might also dispose them to be rated by their parents as displaying more
externalizing pathology, again at the time of baseline assessment. However, their high levels of
empathic concern and guilt could potentially buffer against the increase of externalizing
pathology in these children as they age. On the other hand, children who demonstrate low levels
of guilt at baseline do not demonstrate a decrease in externalizing pathology as they age. It is
possible that these children are already disposed to be on a trajectory of consistent EXT over
time, and their low levels of guilt at the baseline assessment are an early sign of this. Put another
way, their absence of guilt might represent a risk factor for stable EXT over time.
In addition, children high on Positive Affect during the picture tearing task showed a
modest nonlinear decrease of total EXT over time compared to children low on Positive Affect,
suggesting that positive affect could also act as a protective factor against the increase of
externalizing pathology as children age. Further investigation of this possibility is outside the
scope of this particular study, as positive affect is not unique to moral decision making.

93

Summary of longitudinal prediction of problem behaviors. Overall, there were few
significant regressions of longitudinal outcomes, which is consistent with concurrent results.
Guilt drove the significant analyses. Children high on Guilt started out at baseline with higher
levels of EXT, CD, and ODD than low-Guilt peers, but decreased more quickly over time on
these problem behaviors. Thus, children high on Guilt initially appeared to have more
problematic behavior, but returned to a more adaptive trajectory over time. This is partially
consistent with our assertion that anxious arousal/guilt and empathy would negatively predict
long-term EXT (Hypothesis 3.2, 3.3).
We had intended for task behaviors to be useful in predicting changes in externalizing
behavior. For the most part, behaviors during Picture Tearing were not useful in predicting
change in problem behaviors over time. However, Guilt (including empathic statements and
overall guilt) did, in fact, demonstrate utility in predicting an adaptive decrease in externalizing
behaviors for children who were initially high on both externalizing and guilt. Thus, it seems
that the Guilt composite in our task does reflect moral emotions thought to act as a protective
factor against externalizing problems. Guilt behaviors observed within Picture Tearing appear to
meaningfully distinguish children who, despite appearing to have more EXT at the time, are
more likely than their low-Guilt peers to decrease in their EXT as they age.

94

DISCUSSION
This study examined and sought to validate a coding scheme for the Picture Tearing
laboratory task, with the aim of examining the structure of empathy and compliance behaviors in
a naturalistic, morally-relevant scenario. Specifically, we aimed to refine the coding scheme for
the Picture Tearing task and create recommendations for its use in future research. In order to
examine and refine the coding scheme, we applied it to a sample of children aged 3 to 7 years
and examined its performance and utility in several areas: describing variations in moral decision
making behaviors among children aged 3 to 7 years, determining the underlying structure of
these behaviors, testing whether coherent patterns emerged in these behaviors using personcentered analysis, and determining whether behavior assessed in this task showed concurrent and
predictive validity for with child externalizing problems and temperament traits.
On the whole, we found that behaviors observed in Picture Tearing do distinguish
children on a number of external markers theoretically important to moral decision making,
including age, emotion recognition ability, and aspects of effortful control and surgency.
However, the utility in predicting variations in child EXT was somewhat different than we had
expected. Guilt and empathy (operationalized as expressed empathic statements) actually
predicted more oppositionality in children concurrently. However, these behaviors also
predicted a decrease in EXT over time. Thus, the morally-relevant behaviors children exhibit in
Picture Tearing do appear to translate outside of the laboratory.
The measurement of empathy, guilt, and compliance in vivo has significant
methodological implications for research in moral decision making, which has historically been
predicated on hypothetical vignettes or parent report of child behavior. The Picture Tearing task
provides the unique opportunity to circumvent problems of reporter bias and to clarify

95

inconsistencies in past results, which may be attributed to problems with the ecological validity
of previous paradigms. It is critical that we understand the nature of moral decision making in
children in this age range, across which empathy and moral decision making is first developing
and undergoing rapid changes. Insight into how empathy and moral decision making develop
and function in young children is a crucial foundation for future work into how these processes
can be encouraged or disrupted. While there are important limitations to the utility of our coding
scheme for the Picture Tearing task, as we will discuss below, the data on the whole suggest that
it is a useful method of examining variations in children’s moral behavior; in particular, it
provides incremental information above that which can be gleaned from self- and parent-report
methods.
Confirmation and Extension of Previous and Hypothesized Findings
In order to determine whether our coding scheme was truly measuring variations in
behaviors relevant to moral decision making, we first investigated the patterns of behaviors
observed during the task itself. On the whole, our data suggest that the coding scheme measures
behaviors that are consistent with theoretical expectations for child behavior in a morally
ambiguous scenario. Most importantly, our observational approach allowed us to actually
observe behaviors that have, heretofore, been measured only as hypotheticals.
As we had expected, most children complied with the experimenter / authority figure.
Moreover, most children complied with little to no resistance. The base rate of all behaviors was
positively skewed, with most children complying relatively quickly and without comment. This
is consistent with the age of our sample (3 to 7 years old), given that children at this age are
typically socialized to accept the commands of authority figures (Kohlberg & Kramer, 1969;
Piaget, 1965). It is also consistent with previous research using both neutral (Kochanska, Coy, &

96

Murray, 2001; Kochanska et al., 1997) and morally ambiguous (Milgram, 1963; Shanab &
Yahya, 1977) tasks, wherein both children and adults have been shown to comply with the
request of a perceived authority figure, even if the request is putatively harmful to another
person. As we had anticipated, children who demonstrated more enjoyment in the task, more
social referencing, and more positive affect were also more likely to comply and to do so quickly
and thoroughly.
Despite high rates of compliance in general, there was sufficient variation in the extent of
compliance to suggest that even children as young as three years of age do not uniformly respond
to external strictures without an awareness of the meaning of their actions. Some children waited
much longer to comply, required more prompts, and tore the picture less thoroughly than did
other children. Early moral development models (Kohlberg & Kramer, 1969; Piaget, 1965) have
suggested that children are pragmatic, premoral agents who respond to commands based simply
on the dictates of authority figures. In short, these models would suggest that children of the age
of those in our study responded to the experimenter’s request based on their relative fear of
punishment, without considering the meaning of the request in terms of their own personal
values or broader issues of morality. Moreover, both Piaget (1965) and Kohlberg (Kohlberg &
Kramer, 1969) contended that young children do not experience empathy or guilt, and might
have expected to observe signs of these moral emotions only from the oldest participants in the
sample (aged seven years) and still only rarely.
However, our data as a whole contradict this view of children as uncritically ruleresponsive and unempathic. A survey of the manner in which children responded to the
experimenter’s request verbally clearly supports that many children were not only critically
evaluating the meaning and appropriateness of the authority figure’s request, but were

97

experiencing the moral emotions that early researchers ascribed exclusively to older children.
For example, multiple children interrogated the experimenter about the request (“Are you sure?
Did she say you could do this? Why do you want me to?”). Others, even the youngest children
in the sample, revealed that they were considering the moral implications of the task and
anticipating the victim’s feelings (“What if she’s mad? She’s going to be sad. That’s not nice.”),
while some even suggested a less harmful alternative (“But it’s her favorite. How about we rip
this other picture instead.”). Some children condemned the experimenter for what they clearly
perceived as bad behavior (“You’re not supposed to do that. You’re mean. You’re being a
bully.”), and a small number of children went so far as to threaten the experimenter with
punishment (“I’m going to tell her what you said.”).
One could argue that the children were simply parroting statements taught to them by
parents, teachers, and other socializing agents. Even so, whether or not these statements are
indicative of self-chosen moral principles, they do reveal that many of the children were not
simply weighing the potential for punishment when choosing to comply or not. Instead, they
were concerned about the effect of the action on the victim and were weighing the request
against a basic set of internalized moral principles. This interpretation is consistent with more
current research suggesting that even children as young as 24 months of age demonstrate at least
primitive forms of moral emotions and possess a basic awareness of “right” and “wrong,”
regardless of whether they can articulate these principles in a sophisticated matter (Zahn-Waxler,
Radke-Yarrow, Wagner, & Chapman, 1992).
Some children did remain defiant and refuse to comply with the task. This absolute
noncompliance was, as we had expected, positively associated with statements reflecting
empathy and defiance. However, the fact that most children complied in spite of expressing

98

reservations does suggest that, even though young children experience empathy and understand
basic moral principles, they cannot necessarily translate their empathy into absolute moral
behavior (i.e. refusing to tear the picture) when faced with the strong social press to comply with
an authority figure. Yet, that does not signify that their moral sensibilities were completely
overridden by the pressure of authority. As we had anticipated, statements reflecting empathy,
defiance, and hesitation were all positively associated with noncompliance in a relative sense.
Most of these children tore the picture eventually, but, in proportion with their aforementioned
verbalizations, they delayed longer in tearing, required more prompts from the experimenter
before starting to tear, and/or tore the picture less extensively (e.g. ripping only a corner of the
picture instead of “tearing it up” completely as instructed). The latter behavior suggests a
surprisingly sophisticated sense of moral relativism; that is, if the child was going to “have to”
listen to the adult authority, then they were also going to mitigate the damage as much as
possible.
In further support of the contention that expressions of guilt after a transgression emerge
as early as 24 months of age (e.g. Aksan & Kochanska, 2005; Kochanska & Aksan, 2006), all
facets of compliance were positively associated with guilt. This effect was observed whether
guilt was operationalized as observer ratings of how guilty the child seemed or objective displays
of nonverbal discomfort (e.g. hunched shoulders, squirming, avoidance of the victim’s gaze), a
method by which guilt has been operationalized in some previous research (Kochanska & Aksan,
2006). Moreover, we observed a coherence between the moral emotions; empathic statements
and guilt were significantly and positively correlated. In fact, of the three types of verbalizations
coded, only empathic statements were associated with guilt, such that children who verbalized
more empathy towards the victim exhibited more guilt when the victim returned.

99

Empathic statements were associated not only with guilt, but also other moral behaviors
of interest. Children who demonstrated high expressed empathy for the victim were more likely
than their peers to apologize to the victim or attempt to make reparations. This finding
represents a theoretical replication of earlier findings that children as young as 24 months are
capable of taking the perspective of a person in distress and attempting to mitigate that person’s
distress (Zahn-Waxler, et al., 1992). While children in this previous study were likely to offer
comfort in forms that they themselves might desire (e.g. offering a teddy bear), the children in
our sample (who were at least 12 months older than in the previous study) demonstrated more
sophisticated perspective taking, in that the reparations they offered were more appropriate to the
victim and the situation (e.g. offering to help the victim put the photograph back together).
After observing the associations between task variables, we attempted to determine the
underlying structure of these behaviors, via both a variable centered approach (factor analysis)
and a person centered approach (cluster analysis). With respect to factor analysis of the
behaviors, the results partially confirmed our hypotheses about the underlying structure and
partially defied our expectations. As we had expected, a clear factor emerged in all solutions that
reflected noncompliance. This factor reflected defiant statements, the number of prompts given,
latency to tear, and how much the child tore the picture, which was identical to our hypothesized
results.
Additionally, we did observe a coherence between empathic statements and relief for the
copy of the photograph, which we had hypothesized would reflect the child’s ability to take the
victim’s perspective. However, empathy and guilt, the latter of which we had thought would
reflect arousal, did not separate as cleanly as anticipated. In fact, a five factor solution provided
the best fitting model, rather than the three factor solution we had hypothesized. Our

100

hypothesized arousal factor, in fact, seemed to conflate different aspects of arousal. Whereas we
had suggested that smiling and laughter might reflect internal discomfort, these variables cohered
only with each other and did not load with the other variables meant to reflect aversive arousal.
Indeed, at the level of zero order correlations, both smiling and laughter are modestly to
moderately associated with better compliance, are positively associated with other variables
reflecting positive affect (Enjoyment, Relief), and are unrelated to variables reflecting discomfort
(with the exception of a small negative correlation between laughter and gaze avoidance). Taken
as a whole, it appears that this factor reflects genuine positive emotionality rather than nervous
smiling or laughter. Finally, Guilt crossloaded onto a factor with Empathy and Relief, whereas
we had expected that Guilt would separately reflect physiological arousal while Empathy would
reflect perspective taking or affective empathy. This suggests that the behavior measured by our
overall guilt code covaries sufficiently strongly with empathy as to load onto one factor. Indeed,
it is possible that guilt as observed by our coders reflects, in part, an outward expression of
affective empathy. This is consistent with the literature’s assertion that guilt results from the
internal empathic awareness of another’s distress (e.g. Hoffman, 1975; Kochanska & Aksan,
2006). It seems likely that, at the level of coding observational data, parsing these highly
correlated phenomena into discrete, observable behaviors is difficult, if not impossible.
The other aspects of arousal were best represented by two factors. One set of variables
(Silence, Gaze Avoidance, Overall Guilt) seemed to reflect a withdrawn and avoidant response
style. That Guilt crossloads onto this factor suggests that our overall Guilt code reflects both
empathic feeling (perhaps affective empathy) and the aversive internal state we initially
expected, and which also manifests as withdrawal. The final set of codes (hesitation, blaming
the other person) seemed to reflect a tendency to appeal to authority. It is possible that both the

101

Avoidance and Authority factors reflect an internal state of aversive physiological arousal, in that
they both contain behaviors that attempt to modulate this arousal. Importantly, many of the
nonverbal discomfort behaviors (Social reference, Hunched shoulders, Squirming) and response
behaviors (Blame self, Apologize, Lying) that we expected to load onto an arousal factor did not
have substantially large loadings onto any factor and were dropped from analysis. These were
also all behaviors with very low base rates of endorsement across videos, so it is possible that
there was not enough variance in these variables for them to factor meaningfully.
Our person-centered analysis of the task behaviors was more exploratory, but yielded
patterns of behavior that cohered in interpretable ways. Results of this analyses suggest that the
Picture Tearing coding scheme yields codes that describe meaningful variations in child
behavior, and that these behaviors cohere into patterns that meaningfully characterize distinct
groups of children. Analyses suggested that a three cluster solution best fit the data. High
Compliers represented the most compliant group; these children were most likely to comply,
complied in the shortest time with the fewest number of prompts, and tore the picture to the
greatest extent. Low Compliers represented the least compliant group on all compliance
variables, with Moderate Compliers falling in the middle on most behaviors. Consistent with our
theorized associations between task variables, High Compliers also expressed less empathy and
hesitation than Low Compliers and expressed the least defiance in the sample. They also
exhibited more smiling and enjoyment in tearing and more discomfort in the form of hunched
shoulders than Low Compliers. Upon the victim’s return, cluster differences for guilt and relief
mirrored those for compliance, with children from the most compliant cluster demonstrating the
most guilt and relief and vice versa; again, Moderate Compliers fell in the middle. Similar
results were observed for squirming behaviors. There were fewer differences in other response

102

behaviors, with the least compliant children, Low Compliers, expressing less blame for the other
person, less silence, less lying, less hunched shoulders, and less smiling than the most compliant
children, the High Compliers. The Low and Moderate Compliers did not differ significantly on
these variables.
Rates of cluster membership largely conformed with our expectations for child behaviors.
High Compliers, the most compliant group, also contained the majority of participants (n = 159).
Consistent with our hypothesis, most children were willing and quick to comply, and
subsequently demonstrated guilt when faced with the victim. Low Compliers (n = 43), the least
compliant group, contained the fewest participants. On the whole, the cluster data suggest a
general coherence between task compliance and outward indications of guilt, and that children
who exhibit these behaviors tend not to express empathy or defiance. These data provide further
support for our contention that complying with the experimenter in this task is the most
normative response for children in this age group.
Furthermore, we did not find differences in cluster membership by child sex, which runs
counter to some literature suggesting that girls demonstrate more empathy than boys (e.g.
Garaigordobil, 2009; Hoffman, 1977; Kochanska, DeVet, Goldman, Murray, & Putnam, 2008)
and compliance (e.g. Kochanska, 2002; Kochanska, Coy, & Murray, 2001; Kochanska, Tjebkes,
& Forman, 1998), but is consistent with other literature suggesting that sex differences in
empathy are either spurious (e.g. Eisenberg & Lennon, 1983) or do not emerge until middle
childhood or adolescence (Rose & Rudolph, 2006), and that there are no sex differences in
compliance (e.g. Abe & Izzard, 1999; Braungart-Rieker, Garwood, & Stifter, 1997; Higbee,
2012). Consistent with cluster results, we did not find significant mean differences between boys
and girls for most of the individual task variables or factors. Girls exhibited modestly more

103

squirming in response to the victim and relief in response to the copy of the photo, but there were
no significant differences for any other variables, including empathy.
Our failure to observe notable sex differences, where other studies have, may result from
a difference in methodology. With respect to empathy, most previous work has used
hypothetical vignettes or self-report, and have found that girls are more likely to articulate
concern for the victim and to discuss themes of care and compassion (Gilligan & Attanucci,
1988). However, as multiple lines of research have found, the coherence between a person’s
hypothetical behavior and their actual moral behavior is generally low (e.g. Aksan & Kochanska,
2005; Kochanska, Aksan, & Nichols, 2003; Kochanska, Forman, Aksan, & Dunbar, 2005;
Kochanska, Padavich, & Koenig, 1996; Eisenberg et al., 2010; Hartshorne & May, 1928; ZahnWaxler, Radke-Yarrow, Wagner, & Chapman, 1992). Moreover, our measurement of empathy
is predicated on the child verbalizing empathic statements to the experimenter in what the child
believes to be an actual dilemma. As supported by the low base rate of empathic statements in
our sample, most children seem to be unlikely to express empathy aloud in defiance of the
authority figure’s request, even if they are feeling it or thinking about it internally. Had we
asked the children what they were thinking during the task, we might have observed that girls
express more empathy, as other studies have. Our study at present does not allow us to examine
this possibility.
While many of the associations between factors or cluster membership and external
predictors were unexpected, as we will discuss below, we did observe significant negative
associations between Noncompliance, Defiance, and Empathy and overall Effortful Control. In
fact, behaviors derived from the Picture Tearing coding scheme seem especially useful in
distinguishing children on the basis of their effortful control. The data for Noncompliance and

104

Defiance are consistent with findings from laboratory tasks using benign (rather than morally
aberrant) commands, which have found that better effortful control is also related to better task
compliance (Kochanska, Coy, & Murray, 2001; Kochanska et al., 1997). However, the results
for empathy are somewhat unexpected. There is some work to suggest that effortful control is
linked to high moral internalization and developed conscience (e.g. Kochanska & Knaack, 2003;
Kochanska, Murray, & Coy, 1997), which would suggest that we should have observed a
positive association between effortful control and empathy. It is possible that effortful control
could allow a child to focus on the potential consequences of their actions and to avoid
transgressions. Thus, we might have expected that effortful control was positively related to
empathy and guilt. However, again, these findings have been predicated on a small number of
studies which utilized benign commands.
Given our findings for empathy, previous findings relating EC to moral internalization
and conscience may reflect the ability of children with high effortful control to inhibit impulsive
or attractive transgressive responses and to instead choose morally-consistent acts. It may be the
case that children with high EC seem more empathic because they are better able to focus their
attention on the potential consequences of their transgressions, and to avoid transgressing.
However, it does not necessarily follow that they experience more empathy than their peers, only
that they are better able to act on their empathy appropriately. In the case of our study, effortful
control appeared to allow children to inhibit their proponent empathic response and instead
comply with the unempathic action requested by the experimenter, whereas children with poorer
effortful control seemed less able to inhibit their protestations. Given our operationalization of
empathy as verbalizations, it seems likely that whether children were experiencing empathic
thoughts about the victim, children who were more impulsive and less able to focus their

105

attention were more likely to “blurt out” these thoughts. Our coding scheme seems especially
sensitive to morally-relevant behaviors that distinguish children on the basis of their levels of
effortful control and its facets.
Our exploration of effortful control was partly exploratory, given that our primary aim
was to distinguish children on the basis of externalizing behaviors rather than effortful control. It
would be useful in future studies to further explore the utility of the Picture Tearing coding
scheme in measuring behaviors that are relevant to a child’s effortful control. Here, we found
that variations in task behavior predict variations in parent-reported EC. An more stringent test
of the notion that Picture Tearing behaviors relate to meaningful differences in EC would be to
examine whether task behaviors predict variations in EC observed in other laboratory tasks or
assessment specifically designed to measure EC or executive functioning, such as a Go/No Go
task. Furthermore, it would be useful to determine whether the association between task
variables and variation in EC is limited to concurrent associations, or whether task behaviors are
predictive of meaningful variation in EC across time.
Unexpected Findings
We had originally hypothesized that the Picture Tearing coding scheme would be useful
in identifying morally-relevant behaviors that would be predictive of variations in child
externalizing pathology. In particular, we had expected that children who demonstrated little
empathy or guilt in this task would also score higher on measures of externalizing behavior.
However, the relationship between task behaviors and externalizing differed greatly from what
we had anticipated. The most unexpected finding across analyses was the consistent association
between empathy and/or guilt and externalizing pathology. We had predicted that failure to
comply might be associated with defiance and oppositionality. This was observed in our results,

106

such that children who expressed defiance tended also to be rated by their parents as having more
symptoms of ODD and other externalizing pathology. We had assumed that, on the other hand,
empathic statements would reflect a mature level of perspective taking towards the victim and an
understanding that the experimenter’s request was morally wrong. Subsequently, we assumed
that this mature perspective taking would reflect appropriate moral socialization, as
operationalized by a lack of significant externalizing problems.
We had not anticipated, however, that empathic statements might actually reflect
defiance and boldness. In effect, empathic statements seemed to be more reflective of an
impulsive child who failed to inhibit their verbalizations than a morally mature child. Despite
the perspective taking reflected in their statements about the victim’s potential mental state,
children who scored higher on the Guilt composite (which included empathic statements and
overall guilt) were also rated by their parents as having more ODD symptoms, although this
association did not hold when parenting and temperament variables associated with ODD
symptoms were included in the regression. Furthermore, the association between empathy/guilt
and externalizing problems was observed both concurrently and longitudinally, across follow-up
periods out as far as 24 months from the laboratory visit. However, we did not find any
differences in externalizing pathology between our three clusters, even though they differed
significantly on empathy, compliance, and guilt.
In a similar vein, despite robust findings in the literature that poor emotion recognition is
associated with low empathy and low anxious arousal (e.g. Blair & Coles, 2000; Blair, Colledge,
Murray, Mitchell, 2001; Dadds, Perry, Hawes, et al., 2006; Munoz, 2009; Blair, Budhani,
Colledge, Scott, 2005; Fairchild, van Goozen, Calder, et al., 2009), we did not find significant
associations between performance on our emotion recognition task and empathy or guilt. Lack

107

of findings may be attributable to the nature of our emotion variables, which index whether a
child has correctly identified an angry, sad, surprised, happy, or fearful face, respectively. There
is only one trial per affective face, whereas previous research finding effects for emotion
recognition errors has typically employed multiple trials and thus has increased power to find
effects.
Furthermore, we did not observe the associations between behavioral responses to the
Picture Tearing task and child temperament traits that we had expected. Primarily, we
hypothesized that empathy would be associated with temperament traits that are themselves
associated with low levels of externalizing problems, and vice versa. However, as with our
findings for EXT, our findings for temperament traits suggest that empathy in this task is not
purely tapping into perspective taking or affective empathy. We had anticipated that
fearlessness, a quality associated with CU traits and thus a lack of empathy and remorse, would
be associated with less concern for the victim, lower anxious arousal or guilt, and greater
willingness to engage in, and even enjoy, committing an immoral and potentially exciting action.
This finding is, however, consistent with our surprising findings that empathy and guilt were
related to more externalizing problems. Our findings for fearlessness were thus not suggestive of
a link between low empathy and high fearlessness. In fact, the absence of empathic responses,
low anxious arousal or guilt, and enjoyment in tearing were not associated with fearlessness at
all. Furthermore, fearlessness did not distinguish clusters, even though all three clusters differed
significantly on their absolute and relative levels of compliance.
Moreover, we did not anticipate the positive relationship between empathic statements
and low levels of facets of effortful control. While previous literature does not account for the
effects of effortful control on a child’s ability to comply with harmful or immoral orders, as most

108

studies utilize benign commands, it does suggest generally that high effortful control is related
positively to both compliance (Kochanska, Coy, & Murray, 2001; Kochanska et al., 1997) and
adequate moral internalization, operationalized here as few parent-reported externalizing
problems (e.g. Kochanska & Knaack, 2003; Kochanska, Murray, & Coy, 1997). In fact, in our
sample, we either observed the opposite or found no effect. When considering our factors and
discrete task variables, there was a significant negative effect of total empathy, total guilt, and
the Guilt composite on overall effortful control. At the facet level, total empathy, total guilt, and
the Guilt composite were related to poorer Attentional Shifting and higher Impulsivity. With
respect to clusters, differences between clusters (which differed from each other primarily in
terms of compliance and guilt) were not observed for Inhibitory Control or Impulsivity.
However, the Low Compliers, who demonstrated the lowest compliance and guilt, also showed
poorer ability to shift attention according to situational demands than other children in the
sample. This finding, though unexpected, is consistent with findings for task variables. On the
whole, it suggests that variation in behaviors during the Picture Tearing task is reflectively
primarily of variation in a child’s ability to sustain and shift attention and inhibit oppositional or
defiant responses. As described above, it would be useful for further research to more
stringently test the utility of Picture Tearing behaviors in predicting variation in a child’s
effortful control faculties.
Given the unexpected nature of these findings, it is important to note that parent report of
externalizing problems may not be the cleanest or best operationalization of adequate moral
socialization. Indeed, externalizing behaviors might be motivated by a variety of factors other
than deficiencies in empathy and moral decision making abilities, such as impulsivity or negative
affectivity. Furthermore, the hypothesis that externalizing behavior reflects inadequate moral

109

socialization, and thus should be associated with low levels of empathy and guilt, is predicated
on the assumption that what parents are rating as high externalizing behavior is synonymous with
the construct of high externalizing behavior as defined in the literature. In fact, multiple studies
suggest that the norms parents consciously or unconsciously utilize when rating their child’s own
behavior problems are not necessarily consistent with child report of the same behavior, and may
be affected by parent characteristics (e.g. Kolko & Kazdin, 1993; Stranger & Lewis, 1993).
Teacher report, on the other hand, has been shown to reflect severity of a child’s externalizing
behavior in a way that is more consistent with the child’s “actual” behavior (as operationalized
as the validity of teacher ratings of EXT in predicting future behavioral problems, mental health
diagnoses, and referrals), potentially because teachers are more familiar with a wide variety of
children and are thus better able to compare a given child’s behavior to age norms (Stranger &
Lewis, 1993). Thus, while parent report provides invaluable information about a child’s
behavior, it is possible that the positive associations between EXT and guilt we observed would
be more modest or nonsignificant if other methods of assessing EXT (e.g. teacher report,
observation) were utilized. Indeed, this possibility provides additional support for the use of
Picture Tearing and our coding scheme in future research on moral development. Based on the
current literature, largely predicated on self- and parent-report, it would be expected that a
combination of empathy and noncompliance in this task was reflective of mature moral decision
making. However, our study illuminated the possibility that complying with an authority figure
in this case is actually the more adaptive choice. Our observational method provided
information incremental to what has been found in previous research.
Finally, the use of scales from the CBCL allows us to assess a wide variety of
externalizing problems and compare to the wealth of previous ASEBA-based literature, but the

110

CBCL was not intended as a comprehensive, nuanced measure of externalizing pathology.
Furthermore, it contains few items relating to moral emotions, though these are the key pieces of
externalizing pathology that we would expect to be related to unempathic behavior in the picture
tearing task. Future work would benefit from including dedicated measure of moral emotions,
such as the Inventory of Callous-Unemotional Traits (Kimonis, Frick, et al., 2008), as a more
stringent test.
Findings for age were also somewhat complex with respect to our hypotheses. While in
general we would expect older children to comply more readily with a command, we also
considered that older children might show more mature empathy and might, in our particular
experiment, comply less. Indeed, past studies of hypothetical vignettes have found that older
children, in comparison to younger children, tend to view adult commands for harm to be
illegitimate (Damon, 1977; Laupa, 1991, 1994). We did find significant correlations between
age and both empathic statements and the Guilt factor, which were further supported by
significant differences for empathy and Guilt means between younger children (participants aged
3 to 4 years old) and older children (participants aged 5 to 7 years old). Similar results were
observed for hesitation, social referencing, and relief for the copy of the photo.
However, age was also unexpectedly positively associated with task compliance. While
age was uncorrelated with absolute compliance, it did correlate with facets of compliance, such
that older children demonstrated a shorter latency to comply, tore the photo more extensively,
and exhibited more enjoyment while tearing the photo. We observed similar results when we
examined the data from a person centered perspective. Members of the most compliant cluster,
the High Compliers, were significantly older than the Moderate Compliers, who were middling
on most of behaviors. High Compliers also demonstrated the most enjoyment in tearing the

111

photo. However, they did not differ in age from the least compliant group, the Low Compliers,
nor did the Low and Moderate Compliers differ in age from each other, suggesting that our
findings for age and compliance may vary somewhat depending on a child’s other patterns of
behavior.
We hypothesized that older children would view the command as illegitimate and that
their empathy for the victim would trump the pull to comply with an authority. However, despite
expressing more empathy, older children were largely more compliant than younger children,
which is consistent with findings from observational tasks using both benign commands (e.g.
Braungart-Rieker, Garwood, & Stifter, 1997; Kochanska & Aksan, 1995; Kochanska, Aksan, &
Koenig, 1995) and more extreme harmful commands (Shanab & Yahya, 1977). In both cases,
older children were more compliant than younger children. On the whole, age appears to be an
important predictor of empathy and compliance. However, given that we controlled for age
across our analyses, it is not the case that variability in outcomes due to age is masking
variability in outcomes due to other child factors. Even controlling for age, we still observe that
empathy does not substantively predict noncompliance, and is positively and unexpectedly
associated with externalizing pathology.
Conclusions and Areas for Future Study
Validity of the picture tearing coding system and recommendations for future use.
On the whole, our data support that the Picture Tearing coding scheme has utility for describing
meaningful variations in children’s moral decision making behaviors, variations which also
reflect variability in their age, cognitive maturity, and temperament. In particular, task behaviors
relate to variations in effortful control, while their relationships to externalizing behavior are
different from what would be theoretically expected. Thus, results using the observational

112

coding scheme provide information about the potential meaning and predictive utility of a child’s
morally relevant behavior, information that is somewhat different from and incremental to the
information derived from previous questionnaire-based literature. Use of this coding scheme in
future may help to illuminate inconsistencies in the questionnaire-based literature, particularly
inconsistencies between a child’s empathy as measured by self- or parent-report and a child’s
actual moral behaviors.
The aforementioned results support the advancement of several recommendations for
researchers planning to utilize this task and coding scheme. With consideration of our results,
we have created a finalized coding scheme retaining the most useful and theoretically important
variables (see Appendix C). Variables that demonstrated few to no important associations in
either factor or cluster analysis were removed (i.e. Hiding, Lip Biting, Crossed Arms). We also
recommend that, in future, different coders be assigned to code the 1st/2nd epochs from those
assigned to code the 3rd epoch. It is possible that the correlations we found between empathic
behavior in epochs 1 and 2 and guilt in epoch 3 can be due in part to the fact that coders
observed the whole task, and were assuming that children who had expressed the most empathy
were also guiltier than their peers. Because we had the same coder code the entire task, we are
unable to examine this possibility. Having “blind” coders code the 3rd epoch in future will
ensure that codes are based purely on the child’s behavior during that epoch, and do not merely
reflect their behavior earlier in the episode.
We also have several recommendations to how the task is conducted that could
potentially increase the utility of the final codes. First, it is suggested that the length of epoch 2
be standardized. In our task, most of the epoch 2 codes were very skewed, with the majority of
children scoring a 0 for any given behavior. This might be due to the fact that the length of

113

epoch 2 was as brief as 20 seconds in some video episodes, which did not allow sufficient time
for the child to exhibit behaviors. While it is still possible that most children who not exhibit
much behavior in this epoch, given that they are alone in a room and are unlikely to speak to
themselves, standardizing the epoch at a length of one minute or more might allow additional
time for the child to exhibit bodily discomfort.
In addition, it would be useful to ask for self-report data from the child during or at the
conclusion of the task. We recommend that the child be asked questions meant to access their
thoughts about the task and their behavior: “Why did you tear the picture?” “What were you
feeling when you were asked to tear the picture?” “How did you think the owner of the picture
was going to feel when they saw it?” “How do you think another child would feel if they had
torn the picture?” These questions are similar to those used in hypothetical vignettes, and might
be helpful to clarify discrepancies between a child’s actual behavior and their self-stated moral
feelings and principles. It would also help to determine whether some children were
experiencing empathy towards the victim but did not verbalize it due to fear of being chastised or
a belief that they were supposed to do what they were told without question.
From an analytic perspective, our results also support the utility of a person-centered
strategy in examining observational data of this nature. We did find interesting and significant
associations between our single variables and composites derived from factor analysis and
constructs of interest. However, there are a number of features of the Picture Tearing design that
suggest the superiority of a person centered approach. First, demarcating behaviors into factors
in this case seems somewhat artificial. Many of the variables demonstrating bodily tension and
discomfort (e.g. Squirming, Hunched Shoulders) and apologizing to the victim did not load
strongly onto any factor, yet these behaviors are theoretically important indicators of guilt and

114

internal aversive arousal (Kochanska & Aksan, 2006). When cluster analysis was utilized, we
observed meaningful differences between bodily discomfort variables across clusters, which
were consistent in magnitude and direction with empathy and guilt observed in each cluster, such
that the cluster displaying the most guilt also displayed more bodily discomfort. This was a
difference lost in factor analysis.
Furthermore, cluster analysis allowed us to examine coherent patterns of behavior and to
speculate at the meaning of the behavior for that group. Factor analysis is predicated on the cooccurrence of variables, but is agnostic as to the potential that the co-occurrence of these
variables within one person might be attributable to different factors than the co-occurrence of
these variables within another person, and that these variables may actually demonstrate quite
different levels of covariance when other factors are considered. Cluster analysis allows us to
examine the possibility that the co-occurrence of the same variables in two different groups
might actually have different correlates (and, thus, functional significance) depending on other
characteristics of the members of those groups. Similarly, we can examine the possibility that
variables that tend to co-occur in general across the sample actually do not co-occur within
certain clusters, again due to other characteristics of the individuals in those clusters.
For instance, from a variable centered approach, we found that empathy was moderately,
positively correlated with overall guilt and relief for the copy of the photo. However, members
of the cluster with the highest mean for empathy actually demonstrated the lowest mean for guilt
and relief. Moreover, despite expressing more empathy and defiance than High Compliers (the
most compliant cluster), Low Compliers were actually shyer and less assertive, but somewhat
poorer in their ability to shift attention. The factor analysis would lead us to believe that
empathy is uniformly desirable, as it yields the adaptive response of guilt, whereas the cluster

115

analysis suggests that expressed empathy can actually reflect a subgroup of defiant, less inhibited
children. Viewing the data in this manner allows us to consider more complex interrelationships
among behaviors than is feasible from the perspective of factor analysis. It is suggested, then,
that future research examining the Picture Tearing task utilize a cluster approach in lieu of or in
addition to a factor analytic approach.
Limitations and future applications. One important limitation of the current study is
our use of a community sample. Our sample is diverse in gender, ethnicity, and income level,
which allows us to be reasonably confident that we have captured a range of behavior available
in typically developing children. However, despite varying widely on demographic
characteristics, our sample does not offer the range of behaviors that might be observed within a
high-risk or clinical sample. Given that extremes in externalizing behavior and CU traits in the
population are relatively low base rate (Canino, Polanczyk, Bauermeister, Rohde, & Frick,
2010), it is beyond the scope of a community sample to draw conclusions about children who
score at extremes for EXT, CU traits, fearlessness, or other markers of severe disturbances in
moral development. Indeed, the rates of externalizing pathology within our sample for all
varieties of EXT are positively skewed.
Within this somewhat restricted range of EXT, we found that higher levels of verbalized
empathy reflect oppositionality. The use of the Picture Tearing coding scheme in additional
samples could provide further validation of the scheme as a useful measure of variation in child
behaviors, particularly with respect to its utility for describing variations in EXT. As it stands,
the codes have little relationship with EXT apart from empathic statements, which is associated
with oppositionality. Given the robust link between low empathy and EXT in the literature, it is
important to further investigate how the Picture Tearing behaviors relate to EXT in samples

116

where there is more variability in EXT. Within a higher risk sample, where rates of CU traits,
EXT, and fearlessness are more extreme, it is possible that expressed empathy would instead
separate those children with adequate perspective taking skills from those children with severely
abnormal perspective taking. We might also expect to see higher rates of defiant and
oppositional behavior across participants during the Picture Tearing task itself. Future research
should examine these hypotheses in a high risk community or clinical sample.
However, there are also challenges to applying this paradigm to a high risk sample. In a
higher risk sample, we would, as mentioned, expect to observe higher levels of externalizing
pathology, including aggression and oppositionality, as well as lower levels of effortful control.
Given the associations between empathic statements and oppositionality in our sample, it is
possible that we would observe even stronger associations between these constructs in a high risk
sample, given that these children should be even less likely to inhibit their responses and more
likely to be defiant. Thus, it might be difficult to attribute empathic statements, and subsequence
noncompliance, to actual concern for the victim, when these behaviors could just as easily reflect
oppositionality on the part of the children. One important addition might be to more stringently
test the nature of empathic statements in this group by examining the associations between
empathic statements and purer measures of empathy and conscience, particularly the Inventory
of Callous-Unemotional Traits (Kimonis, Frick, et al., 2008). If empathic statements in this
sample are indeed indicative of concern for the victim and not purely oppositionality, then we
would expect to observe a negative association between empathic statements and ICU scores.
Implications for the study of moral development. This study sought to validate a
coding scheme designed to examine the facets of moral decision making in preschool and young
school-age children, with a specific interest in connecting facets of moral behavior to

117

externalizing pathology. Unlike most recent work in the area of moral development and
empathy, we utilized a laboratory task, which allowed us to observe a child’s actual behavior in a
naturalistic morally-relevant situation, rather than their responses to hypothetical situations.
Using a scheme to numerically code the observed behaviors, we were able to examine
contentions in the literature of what children say that they would do versus what they actually do
given the opportunity. We have found that our coding scheme provides useful information about
children’s moral decision making that confirms, clarifies, and contests long-standing findings in
the previous questionnaire-based literature. Thus, we were able to show, for example, that while
older children are less likely to say that they would comply with a request from an authority
figure if the request is perceived as being harmful (Damon, 1977; Laupa, 1991, 1994), in practice
they are more likely to comply with such a request. Furthermore, sex differences in compliance
and expressed perspective taking often found when analyzing responses to hypothetical vignettes
(e.g. Garaigordobil, 2009; Hoffman, 1977) were largely absent in our observations. Given that
previous research finding significant sex differences has been based on parent report, it is
possible that parents in these studies are more likely to view girls as empathic and compliant in
accordance with gender-based stereotypes. With our observational coding scheme, however,
expression of empathy is rated objectively as number of statements made, which is putatively
less likely to suffer gender-norm-based bias than subjective reporting of whether a child “seems”
empathic or not to the rater.
Most importantly, our results provide additional support for the longstanding contention
that the social press to comply with an authority figure can and often does supersede an
individual’s internalized moral rules to avoid harming another person. Additionally, these
findings support work by Aksan and Kochanska (2005) and Kochanska and Aksan (2004)

118

suggesting that conscience is a higher order construct with two underlying components, moral
emotions (empathy and guilt) and moral behavior. These components are correlated, but distinct,
and each can occur to the exclusion of the other. Indeed, in our sample, as with Milgram’s
(1963, 1965) seminal studies and Shanab & Yahya’s (1977) replication, the majority of children
in our study complied with the experimenter’s request to harm the victim’s cherished property.
Moreover, many of the children were compliant despite expressing concern for the victim’s
feelings or recognition that the request was immoral or illegitimate. In fact, older children, who
putatively possess a better developed understanding of appropriate behavior and the victim’s
perspective, complied with the least delay and to the greatest extent.
On the whole, our data support the notion that the role of authority in the moral
development of young children is complex, and that questionnaire-based measures may not
adequately address this complexity on their own. Piaget (1965) and Kohlberg (Kohlberg &
Kramer, 1969) seem to be accurate in their contention that an authority figure, whether present or
internalized, is crucial to early moral development. The decision to act morally (or not) is
contingent on the perceived consequences of the action. In our study, this notion manifests not
only in the fact that most children complied with the authority despite any reservations they
expressed (presumably due to fear of some imagined punishment), but also that their statements
of reservation ostensibly reflect moral strictures gleaned from the adult authorities who have
participated in their moral socialization. However, contrary to the assumptions of early theorists,
the consequences that children of this age envision appear to be not only consequences to
themselves in the form of punishment, but consequences to their victim.
Rather than passively receiving and executing adult commands, it appears that children
view authority and their obligation to obey in nuanced ways (Kuczynski, Kochanska, Radke-

119

Yarrow, & Girnius-Brown, 1987). It does not appear to be a simple matter that “I must obey
because the adult said so, and because I will be punished if I disobey.” Rather, we observed that
many children in our study seemed to evaluate the legitimacy of the authority figure critically.
This is consistent with a small body of work that has found that children, when presented with
hypothetical scenarios in which a parent or teacher commanded them to commit an immoral act
(e.g. fight with another child), stated that the command was “wrong” and that they would not
obey (Damon, 1977; Laupa, 1991, 1994). We did not find this to be the case. Rather, consistent
with the only other in vivo study to utilize harmful commands (Shanab & Yahya, 1977), we
found that most children did comply. However, we did observe that children, despite complying,
seemed to evaluate critically the legitimacy of the authority figure’s command, with the result
that the degree and nature of their compliance varied. The more children expressed reservations
about the “rightness” of the task, the longer they resisted the press to comply and the less
thoroughly they carried out the command.
The way in which children interacted with and evaluated the authority figure as part of
their decision making process appeared to interact in complex ways with age, in concert with
increasing sophistication of moral reasoning and empathic perspective taking abilities, but also
increasing efficacy of effortful control capabilities and awareness of social norms. Consistent
with this notion, older children (five to seven years old) were more likely to openly question and
disagree with the experimenter’s agenda than younger children (three to four years old), and to
express to the victim that they complied because they were commanded to do it. However,
paradoxically, they were also more likely than younger children to comply quickly and
completely. We had hypothesized that older children, with superior empathic skills, would be
less likely to comply with the picture tearing request, despite being more likely to comply with

120

adult requests in general. This proved not to be the case. Our results suggest that, even though
older children were more openly critical of the legitimacy of the command, they still tended to
accept that the experimenter’s authority to command them trumped their own preferences.
Perhaps this explains why older children did not differ from younger children in their appearance
of guilt; they might have believed that they were not personally responsible for the action, since
they had acted against their will.
While younger children were less compliant with the request than older children, this did
not appear to be a function of empathy—at least not observable empathy. Recall that younger
children articulated fewer empathic statements than older children. One could presume that
younger children experienced empathy internally, and that despite being more hesitant to express
this empathy to the experimenter, it motivated them to defy the command. Yet, this explanation
contradicts the theoretical notion that younger children do not have as sophisticated empathic
abilities as older children and are more likely to base their moral decisions on the strictures of
authority rather than abstract more principles than older children. However, empirically,
multiple studies have found that younger children are less compliant with experimenter
commands, potentially because they have poorer effortful control and are thus less able to
modulate their behavior according to task expectations (Braungart-Rieker, Garwood, & Stifter,
1997; Kochanska & Aksan, 1995; Kochanska, Aksan, & Koenig, 1995). Thus, it appears that
even though younger children are theoretically more likely to perceive authority figures as
legitimate, their ability to comply with these authority figures is largely impacted other
developmental limitations (including, potentially, aspects of effortful control).
Additionally surprising was the role that empathy and guilt played in predicting other
behaviors of interest. Within the task itself, the children who demonstrated the most guilt were

121

likely to blame the other person and, unexpectedly, less likely to blame themselves. Guilt was
also unrelated to apologizing or attempting to make reparations. It appears, then, that the
children who experienced the most aversive arousal in the presence of their victim were not
necessarily accordingly motivated to repair their relationship with the victim. It is possible that
children experiencing high levels of guilt/anxious arousal were too focused on their own
negative feelings to adequately focus on the feelings of the victim and what they might do as
recompense. Instead, they seemed to attempt to modulate their aversive arousal by placing the
blame on the authority figure who commanded them, or else withdrawing from the victim and
remaining silent.
These putative processes are consistent with the notion that children who experience too
much anxious arousal after a transgression might express low levels of guilt and empathy
because they are too focused on their personal distress to adequately respond to the emotions of
their victim (Eisenberg et al., 2010). This hypothesis in the literature is typically predicated on
the assumption that excessive personal distress is a result of extreme dispositional fearfulness,
high levels of anxiety, excessive power assertion on the part of the socialization agent (typically
operationalized as a punitive parent), or a combination thereof (Kochanska, 1991). We did not
find a link between fearfulness and empathic expression in our study. However, it is likely that
the unique qualities of the picture tearing paradigm (having an authority figure ask the child to
do something clearly harmful to another person, confronting the child with their victim) are
strong enough situational presses to elicit excessive personal distress even in children who are
not dispositionally fearful or anxious.
Not only did guilt not motivate reparative behavior during the task, but guilt and empathy
were associated with externalizing pathology rated outside of the task—in particular, with

122

oppositionality. Moreover, these constructs were associated with deficits in areas of effortful
control. These findings are unusual, given the wealth of research linking empathy and guilt to
appropriate moral socialization, and thus low levels of externalizing pathology (Kochanska,
1991, 1993), as well as adequate levels of effortful control. Instead, it seems that children in our
study who verbalized empathy were actually more oppositional and less inhibited than their
peers. It is unlikely that empathy as defined in the literature is predictive of externalizing
behaviors.
Rather, given the low-risk nature of this sample, it is probable that most children in our
sample feel some degree of affective empathy towards the victim, and are able to take the
victim’s perspective. However, it appears that less impulsive, less oppositional children were
able to override an empathic desire to refuse the experimenter in favor of the less intuitive
behavior (ripping the photo) necessitated by task demands. On the contrary, children with more
oppositional, impulsive temperaments were more likely to express that empathy in defiance of
the experimenter’s request, possibly because they were less able to inhibit the dominant response
of refusing to commit an immoral act. In addition, children with more impulsive temperaments
were possibly less likely to think through the consequences of their actions, and thus more likely
to feel badly—or to fail to modulate their observable emotional response—after confronting the
victim.
On the whole, our findings confirm that many modern notions of moral development in
young children can be observed not only via parent report and children’s assessment of
hypothetical scenarios, but also in a naturalistic, in vivo scenario in which children believe they
are facing an actual moral dilemma. In fact, using our coding scheme with the observational
Picture Tearing task appears to provide additional information about children’s moral behavior

123

than that which can be gleaned from questionnaire. It also confirmed many of the contentions
previously hypothesized by theoretical and questionnaire-based research. We confirmed, for
example, that empathy and guilt can be observed in children as young as three years of age, but
that the presence of these moral emotions does not necessarily correlate strongly, or at all, with a
child’s enacted behavior. Indeed, the press to respond to an authority figure appears, as in
previous observational studies, to trump a child’s instinctive inclination to act prosocially.
However, nor is the trade-off between empathy and compliance simple. Despite generally
complying with the authority’s immoral request, many children still evaluated the
appropriateness of the task and resisted or modulated their level of compliance in ways that
suggest a level of moral sophistication of which early theorists assumed them incapable.
Unexpectedly, we observed that children who expressed more empathy and guilt were
also rated as showing more externalizing behaviors by their parents. This would seem to
contradict the literature’s contended associations between empathy, guilt, and adequate moral
socialization, as operationalized by few externalizing problems. However, notably, the
associations between empathy and externalizing pathology became weaker as children aged,
suggesting that empathy in the Picture Tearing task is more reflective of concurrent oppositional
behavior rather than a marker for enduring patterns of externalizing pathology. On the whole, in
this age group and within the conditions of this task, compliance appears to be more adaptive
than defiance, even though compliance is synonymous with committing an immoral or harmful
act. Children who complied more extensively tended to be older, to express more empathy, and
to demonstrate better effortful control. On the whole, our results suggest that moral decision
making in preschool and school-aged children is a complex phenomenon affected by complex
interrelationships between age, temperament, situational demands, and the child’s own

124

internalized moral rules. Utilizing our coding scheme for the Picture Tearing task allowed us to
observe subtleties in children’s actual moral decision making that are not easily or not at all
accessible via questionnaire.

125

APPENDICES

126

APPENDIX A: FIGURES
Figure 1. Diagram of moral development and its relationship to child internal and external characteristics and secondary
outcomes.

Moral
behavior

Moral
Emotions

Moral Development

Guilt

Empathy
(Affective +
Cognitive)

Obedience

Child temperament + Environmental Characteristics (e.g.
Parenting)
127

Social rules

Figure 2. Chart of Video Progression and Relevant Codes

st

1
Experimenter
prompts the
child to tear

Child’s behavior
nd

when 2 returns and
asks about picture
(Squirming, Hiding,
Blame self, Blame
other, Lying, Gaze
avoidance, Overall
guilt, Apologizing)

Child’s response
to request
(Defiance,
Hesitation,
Empathy,
Squirming,
Hiding)

Tear / Not tear
(if child did not
tear, following
steps do not apply)
(Latency to tear, #
prompts given)

Child’s behavior
after tearing
(Squirming,
Hiding,
Smiling/Laughin
g)

Child’s behavior
while tearing
(Enjoyment,
Amount torn,
Smiling/Laughin
g, Squirming)

Child’s
response to
second copy of
picture (Relief
at second
picture)

128

Figure 3. Hierarchical cluster analysis agglomeration coefficients plotted against stage of analysis.
7

6

5

4

3

2

1

1
7
13
19
25
31
37
43
49
55
61
67
73
79
85
91
97
103
109
115
121
127
133
139
145
151
157
163
169
175
181
187
193
199
205
211
217
223
229
235
241
247
253
259
265

0

Stage

Coefficients

129

Figure 4. Gap statistic for hierarchical cluster analysis graphed against number of clusters.

Gap

0.45

0.4

0.35

0.3

0.25
1

2

3

4

5

6

7

8

9

130

10

Figure 5. Differences between cluster means for three cluster solutions.
Epoch 1.

131

Figure 5 (cont’d)

132

Figure 5 (cont’d)
Epoch 2.

133

Figure 5 (cont’d)
Epoch 3.

134

Figure 5 (cont’d)

135

Figure 5 (cont’d)

136

APPENDIX B: TABLES
Table 1. Intraclass correlation coefficients for video coding variables
Epoch 1
Epoch 2
Epoch 3
Variable
ICC
Variable
ICC
Variable
*
Empathy
0.89 Empathy
Blame Other
*
Hesitation
0.95 Hesitation
Blame Self
*
Defiance
0.98 Defiance
Apologize
*
Squirming
0.67 Squirming
Silence
Hunched
Hunched
Shoulders
0.23 Shoulders
0.23 Lying
*
Lip Biting
*
Lip Biting
Squirming
*
Crossed Arms
*
Crossed Arms
Hunched Shoulders
*
Hiding
*
Hiding
Lip Biting

ICC
0.78
*
0.85
0.67
0.7
0.38
0.68
0.01

Social Reference

0.57 Laughter

0.44 Crossed Arms

Laughter

0.93 Smiling

0.78 Hiding

Smiling
Number of
Prompts

0.82

Social Reference

0.66

0.94

Laughter

0.84

How Much Torn

0.99

Smiling

0.91

Enjoyment

0.86

Gaze Avoidance

0.91

How Guilty

0.92

Relief for Picture

0.75

Note: * Insufficient variance among codes to calculate ICC

137

*
0.99

Table 2. Picture Tearing codes relevant to each epoch of the task.
2. After tear, before victim
1. Request phase
3. After victim returns
returns
Latency to tear
Squirming
Squirming
Tear/No tear
Lip biting
Lip biting
Number of prompts given
Hunched shoulders
Hunched shoulders
Amount of picture torn
Leaning away/Hiding
Leaning away/Hiding
Defiance/Noncompliance
Social referencing
Social referencing
statements
Concern/empathy statements
Laughing / Smiling
Laughing / Smiling
Hesitation/questioning
Negative affect
Negative affect
statements
Squirming
Crossing arms
Silence
Lip biting
Blaming other
Hunched shoulders
Lying
Leaning away/Hiding
Gaze avoidance
Social referencing
Global guilt rating
Laughing / Smiling
Blaming self

138

Table 2 (cont’d)
Negative affect
Enjoyment in tearing
(reversed)
Crossing arms

Apologizing
Relief at second picture
Crossing arms

139

Table 3. Comprehensive list of hypotheses
Relevant analysis (if
Number
applicable)
Hypothesis
Aim 1
The task behaviors will be best represented by a three factor structure (see Table 4 for
1.1
EFA / CFA
specific hypothesized loadings)
1.2
CFA / Cluster
Most children will comply with the experimenter’s request
1.3
CFA / Cluster
Noncompliance will be positively associated with Assertiveness (parent-rated)
1.4
CFA / Cluster
Noncompliance will be positively associated with expressed Empathy in the task
Noncompliance will be positively associated with Oppositionality (parent-rated ODD
1.5
CFA / Cluster
symptoms, Noncompliance in other lab tasks)
Noncompliance in the absence of empathic statements will be negatively associated with
1.6
CFA / Cluster
Fearlessness
Aim 2
Correlations /
Noncompliance in PT will positively predict EXT, AGG, CU, Rule-breaking, ODD, and CD
2.1
Regressions
Correlations /
Anxious Arousal in PT will negatively predict EXT, AGG, CU, Rule-breaking, ODD, and
2.2
Regressions
CD

140

Table 3 (cont’d)
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11

Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions

Empathy in PT will negatively predict EXT, AGG, CU, Rule-breaking, ODD, and CD
Noncompliance will be modestly positively associated with child age
Anxious Arousal will be positively associated with child age
Empathy will be positively associated with child age
Noncompliance will be modestly negatively associated with female sex
Anxious Arousal will be modestly positively associated with female sex
Empathy will be modestly positively associated with female sex
Anxious Arousal will be positively associated with total baseline Emotion Recognition score
(more emotions correctly identified)
Empathy will be positively associated with total baseline Emotion Recognition score (more
emotions correctly identified)

141

Table 3 (cont’d)
2.12
2.13
2.14
2.15
2.16
2.17
2.18
2.19
2.20

Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions

Anxious Arousal will be positively associated with PPVT score (better verbal ability)
Empathy will be positively associated with PPVT score (better verbal ability)
Noncompliance will be positively associated with Fearlessness
Anxious Arousal will be negatively associated with Fearlessness
Empathy will be negatively associated with Fearlessness
Empathy will have a curvilinear association with Fearlessness such that extreme high and
low Fearlessness will be associated negatively with Empathy, while moderate levels will be
associated positively with Empathy
Noncompliance will be positively associated with Anger, High Intensity Pleasure,
Anxious Arousal will be negatively associated with Anger, High Intensity Pleasure,
Empathy will be negatively associated with Anger, High Intensity Pleasure,

142

Table 3 (cont’d)
2.21
2.22
2.23
2.24
2.25
2.26
2.27
2.28
2.29

Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions

Noncompliance will be negatively associated with Sadness
Anxious Arousal will be positively associated with Sadness
Empathy will be positively associated with Sadness
Noncompliance will be negatively associated with Fear-proneness in other lab tasks
Anxious Arousal will be positively associated with Fear-proneness in other lab tasks
Empathy will be positively associated with Fear-proneness in other lab tasks
Noncompliance will be negatively related to Effortful Control (high Inhibitory Control, low
Impulsivity)
Anxious Arousal will be positively related to Effortful Control (high Inhibitory Control, low
Impulsivity)
Empathy will be positively related to Effortful Control (high Inhibitory Control, low
Impulsivity)

143

Table 3 (cont’d)
2.30
2.31
2.32
2.33
2.34
2.35
2.36

Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions
Correlations /
Regressions

Noncompliance will be positively associated with Poor Monitoring, Corporal Punishment,
Inconsistent Punishment
Anxious Arousal will be negatively associated with Poor Monitoring, Corporal Punishment,
Inconsistent Punishment
Empathy will be negatively associated with Poor Monitoring, Corporal Punishment,
Inconsistent Punishment
Noncompliance will be negatively related to Positive Parenting and Involvement
Anxious Arousal will be positively related to Positive Parenting and Involvement
Empathy will be positively related to Positive Parenting and Involvement
Within the PT task, empathy, defiance, hesitation, and signs of physical discomfort will be
positively associated with Noncompliance and Guilt

Aim 3
3.1
3.2

Growth Curve
Analysis
Growth Curve
Analysis

Noncompliance in PT will positively and prospectively predict EXT, AGG, CU, Rulebreaking, ODD, and CD
Anxious Arousal in PT will negatively and prospectively predict EXT, AGG, CU, Rulebreaking, ODD, and CD

144

Table 3 (cont’d)
Growth Curve
3.3
Analysis

Empathy in PT will negatively and prospectively predict EXT, AGG, CU, Rule-breaking,
ODD, and CD

145

Table 4. Hypothesized results of factor analysis
Obedience/Compliance
Arousal
Latency to tear
Hesitation statements
Defiance/Noncompliance
Squirming
statements
Crossing arms
Lip biting
Tear/No tear
Hunched shoulders
Number of prompts given
Leaning away/Hiding
Amount of picture torn
Social referencing
Laughing / Smiling
Negative affect
Silence
Blaming other
Lying
Gaze avoidance
Global guilt rating
Blaming self
Apologizing
Hesitation/questioning
statements

Empathy
Concern/empathy statements
Enjoyment in tearing
(reversed)
Relief at second picture

146

Table 5. Hypothesized associations for variables investigated at baseline.
Construct

Positive

Negative

Compliance

Compliance in other tasks,
Fearfulness, Effortful control,
Inhibitory control

Assertiveness, High expressed
empathy, Oppositionality, Age,
Impulsivity, Harsh and inconsistent
parenting

Empathy/ Anxious Arousal

Emotion recognition, Fearfulness,
Sadness, Effortful control, PPVT

Guilt, Compliance, Fearlessness,
EXT, Anger, Harsh and inconsistent
parenting

147

Table 6. Descriptive statistics for picture tearing variables.
Epoch1
Epoch2
Skewnes Kurtosi Mea
(SD
Kurtosi Mea
Mean
(SD)
s
s
n
)
Skewness s
n
Empathy 1.15
0.13
7.49
55.50
(1.91) 2.23
5.16
0.02
Hesitation 1.35
0.12
16.25 264.00
(1.68) 2.77
13.85
0.01
Defiance 1.27
0.06
16.25 264.00
(2.47) 2.51
6.98
0.00
Squirming 0.35
(1.11) 6.08
50.90
0.07
0.36
6.32
42.76 0.59
Hunched
0.67
5.10
31.53 0.46
Shoulders 0.22
(0.59) 3.91
20.49
0.18
Lip Biting 0.07
0.14
8.48
73.75 0.09
(0.33) 6.91
56.94
0.12
Crossing
0.27
12.56 176.71 0.07
Arms 0.09
(0.40) 5.53
32.23
0.03
Hiding 0.27
0.74
4.86
31.69 0.28
(0.80) 4.15
19.81
0.23
Social
Referencin
0.00
0.12
g 1.56
(1.42) 1.89
6.17
0.00
Laughing 0.26
0.14
6.43
40.22 0.24
(0.81) 4.37
21.75
0.02
Smiling 1.90
0.57
3.74
17.18 2.20
(2.38) 1.71
2.84
0.20
Tear or Not 0.78
(0.41) -1.39
-0.08
Number of
Prompts 4.07
(2.68) 1.06
1.10
Latency to
(43.48
Tear 41.45
)
1.29
2.18
How Much
Torn 3.36
(1.47) -0.62
-1.11
Enjoyment 2.03
(0.86) 0.52
-0.29
Blame
0.95
Other
Blame Self
0.18

148

Epoch 3
(SD)

Skewness

Kurtosi
s

1.09

2.27

4.83

0.95

2.75

8.43

0.49

8.05

75.54

0.25

3.92

14.76

0.77

4.17

20.48

0.31

3.30

12.84

0.66
2.55

3.76
2.09

16.81
5.81

1.09

1.14

0.96

0.45

2.78

8.15

Table 6 (cont’d)
Apologize
Silence
Lying
Gaze
Avoidance
Overall
Guilt
Relief for
Copy

149

0.09
0.39
0.22

0.32
0.95
0.57

3.96
3.33
3.99

17.45
12.89
22.57

0.41

1.14

4.84

31.06

2.44

0.83

0.48

-0.54

2.07

0.86

0.58

-0.27

Table 7. Descriptive statistics for recoded picture tearing variables.
Epoch 1
Epoch 2
Epoch 3
Mean (SD) Skewness Kurtosis Mean (SD) Skewness Kurtosis Mean
Empathy
0.83 1.05
0.91
-0.56
0.02 0.14
7.10
48.76
Total Empathy
0.43 0.55
0.94
-0.46
Hesitation
1.23 1.12
0.40
-1.20
0.01 0.12
16.25
264.00
Total
0.63 0.57
0.57
-0.33
Hesitation
Defiance
0.72 1.05
1.10
-0.31
0.00 0.06
16.25
264.00
Total Defiance
0.37 0.56
1.38
1.37
Squirming
0.29 0.64
1.99
2.40
0.07 0.32
4.83
23.67
0.70
Hunched
0.27 0.57
2.02
2.96
0.16 0.48
3.11
8.55
0.63
Shoulders
Social
1.61 1.17
0.48
-0.57
0.25
Reference
Total Social
0.93 0.70
0.60
-0.22
Reference
Laughing
0.15 0.36
1.94
1.77
0.03 0.16
5.93
33.39
0.43
Total Laughing
0.20 0.40
2.03
3.05
Smiling
1.63 1.52
0.44
-1.27
0.19 0.49
2.56
5.73
2.02
Total Smiling
1.30 1.06
0.55
-0.68
Tear or Not
0.78 0.41
-1.39
-0.08
Number of
3.60 1.74
0.00
-1.39
Prompts
Latency to
41.45 43.48
1.29
2.18
Tear
How Much
3.30 1.46
-0.54
-1.17
Torn
Enjoyment
2.18 0.88
0.30
-0.65
Blame Other
0.95

150

(SD) Skewness Kurtosis

1.01

1.05

-0.38

1.05

1.38

0.34

0.56

2.21

3.70

0.99

2.09

2.62

1.73

0.44

-1.07

1.03

0.68

-0.79

Table 7 (cont’d)
Blame Self
Apologize
Silence
Lying
Gaze
Avoidance
Overall Guilt
Relief for
Copy

151

0.21
0.10
0.33
0.37

0.48
0.30
0.66
0.82

2.26
2.72
1.76
2.08

4.45
5.43
1.59
3.03

0.30

0.63

1.92

2.28

2.48

0.96

0.06

-0.93

1.98

0.88

0.68

-0.19

Table 8. Bivariate correlations among picture tearing variables.
1
2
3
4
5
Empathy
1.00
1
Hesitation
0.39**
1.00
2
**
Defiance
0.25
0.05
1.00
3
Social
0.07
0.12*
-0.03
1.00
4
Reference
Laughing
-0.08
-0.02
-0.13*
0.16**
1.00
5
**
Smiling
-0.04
0.06
-0.10
0.22
0.46**
6
Hunched
0.02
0.02
0.03
0.14*
0.06
7
Shoulders
Squirming
-0.04
-0.05
-0.03
0.24**
0.12
8
**
**
**
**
Tear
or
Not
-0.22
-0.04
-0.60
0.19
0.18
9
Number of
0.32**
0.35**
0.63**
-0.01
-0.16*
10
Prompts
Latency to
0.18**
0.15*
0.56**
0.00 -0.16**
11
Tear
How Much
-0.27**
-0.08 -0.60**
0.08
0.17**
12
Torn
Enjoyment
-0.13
-0.03
-0.13
0.11
0.39**
13
Blame Other
0.23**
0.20**
-0.14*
0.21**
0.04
14
Blame Self
-0.10
0.01
-0.10
-0.08
0.11
15
*
*
Apologize
0.14
0.15
-0.04
0.12
0.00
16
**
**
**
Silence -0.21
-0.26
-0.19
0.07
-0.03
17
*
**
Lying
-0.14
-0.03 -0.18
0.04
0.11
18
Gaze
-0.03
-0.05
-0.09
0.06
-0.11
19
Avoidance
Overall Guilt
0.26**
0.04
0.11
0.11
-0.05
20
Relief for
0.20**
0.10
-0.03
0.06
0.07
21
Copy

152

6

7

8

9

10

11

1.00
0.19**

1.00

0.29**
0.28**

0.14*
0.05

1.00
0.09

1.00

-0.19**

-0.02

-0.15*

-0.56**

1.00

-0.19**

-0.02

-0.15*

-0.58**

0.66**

1.00

0.24**

0.04

0.10

0.83**

-0.62**

-0.59**

0.61**
0.18**
0.11
0.07
-0.10
0.09

0.12
0.00
-0.13*
0.07
0.07
0.12

0.14*
0.13*
-0.03
0.03
0.14*
-0.01

0.16*
0.35**
0.21**
0.14*
0.26**
0.22**

-0.18**
-0.06
-0.01
0.03
-0.34**
-0.22**

-0.15*
-0.20**
-0.05
-0.03
-0.16**
-0.15*

-0.14*

0.14*

0.11

0.25**

-0.11

-0.12

-0.02

0.17*

0.16*

0.06

0.02

0.00

0.34**

0.07

0.06

-0.01

-0.12

-0.14*

Table 8 (cont’d)
12
12
13
14
15
16
17
18
19
20
21
Note:

How Much
1.00
Torn
Enjoyment
0.15*
Blame Other
0.24**
Blame Self
0.13*
Apologize
0.01
Silence
0.27**
Lying
0.28**
Gaze
0.17**
Avoidance
Overall Guilt
-0.05
Relief for
-0.02
Copy
* p < .05, ** p < .01

13

14

15

16

17

18

19

20

1.00
0.05
0.12
-0.03
-0.16*
0.06

1.00
-0.10
0.13*
-0.24**
-0.22**

1.00
0.02
-0.08
0.00

1.00
0.01
-0.03

1.00
-0.01

1.00

-0.12

-0.03

-0.07

0.03

0.37**

0.06

1.00

-0.03

0.13

-0.16*

0.00

0.08

-0.05

0.24**

1.00

0.25**

0.16*

-0.04

0.09

-0.22**

-0.03

-0.23**

0.30**

153

21

1.00

Table 9. Correlations between picture tearing variables and child characteristics.

Demographics

Epoch 1

Epoch 2

Empathy
Hesitation
Defiance
Squirming
Hunched
Shoulders
Social Ref
Laugh
Smile
Tear or Not
Latency
Number
Prompts
How Much
Torn
Enjoyment
Empathy
Defiance
Squirming
Hunched
Shoulders
Laugh

Age
0.14*
0.13*
0.00
0.02

Sex
PPVT
-0.07 0.17**
0.02
0.11
-0.02
0.10
0.07
0.00

Emotion Training Variables
Total
Happy Angry
Sad
Surprise Fear Neutral
Correct Wrong Wrong Wrong Wrong Wrong Wrong
0.14* -0.15*
-0.05 -0.13*
-0.10
-0.06
-0.13
**
0.08 -0.18
-0.12
-0.09
-0.01
0.01
-0.04
0.01
0.02
-0.03
0.03
0.02
-0.02
-0.05
-0.03
0.07
-0.03
0.01
0.00
0.03
0.04

0.11

0.02

0.03

-0.07

0.12

0.06

0.00

0.08

0.04

0.01

0.20**
0.22**
0.13
0.15
-0.11

0.08
-0.11
-0.06
-0.07
-0.03

0.07
0.07
-0.14
-0.06
0.06

0.17** -0.19**
0.16*
-0.11
-0.07
0.04
0.03
0.06
-0.04
0.03

-0.07
-0.12
-0.03
-0.01
0.04

-0.11
-0.11
0.08
0.06
0.01

-0.22**
-0.06
0.09
-0.01
0.03

-0.11
-0.13*
0.09
-0.03
0.06

-0.07
-0.17**
0.00
-0.16*
0.02

-0.12*

-0.06

-0.03

-0.05

-0.01

0.03

0.00

0.09

0.06

0.03

0.15*

0.06

-0.03

0.08

-0.01

-0.04

-0.01

-0.07

-0.07

-0.11

0.12
-0.06
-0.03
0.14*

0.00
0.14*
0.06
0.06

-0.10
0.01
-0.01

0.05
0.04
0.04
0.06

-0.06
-0.04
-0.02
-0.05

-0.06
-0.04
-0.02
-0.05

0.07
-0.05
-0.02
-0.06

-0.04
0.00
-0.03
-0.04

-0.05
0.00
-0.03
-0.03

-0.07
-0.07
-0.03
-0.02

0.14*

-0.09

0.01

0.02

-0.07

0.00

-0.03

0.00

0.01

-0.02

-0.05

-0.17**

0.00

0.00

0.07

-0.04

-0.06

0.05

-0.01

0.00

154

Table 9 (cont’d)
Smile
Blame Other
Epoch 3
Blame Self
Apologize
Silence
Lying
Squirming
Hunched
Shoulders
Social Ref
Laugh
Smile
Gaze
Avoidance
Overall Guilt
Relief for Copy
Empathy
Totals
Hesitation
Defiance
Social Ref
Laugh
Smile
Composites Guilt
Authority
Noncompliance
Positive Affect
Avoidance
Note: * p < .05, ** p < .01

0.14*
0.23**
-0.13*
0.05
0.02
0.08
0.08

0.01
-0.06
-0.07
-0.05
0.10
-0.03
0.19**

0.08
0.05
-0.12
-0.03
0.02
0.02
0.02

0.05
0.11
-0.18**
0.03
0.02
-0.02
0.05

-0.10
-0.14*
0.00
0.04
0.11
0.05
-0.02

-0.10
-0.05
0.09
-0.03
-0.06
-0.01
-0.04

0.00
0.01
0.20**
-0.06
-0.07
0.02
0.03

0.00
-0.06
0.13*
-0.06
-0.01
0.02
-0.04

0.02
-0.08
0.22**
-0.01
-0.04
0.01
-0.07

-0.06
-0.16*
0.09
0.01
0.00
0.01
-0.07

0.20**

0.05

0.02

0.04

0.05

-0.05

0.00

-0.02

-0.06

-0.05

0.13*
0.09
0.31**

-0.06
0.10
0.11

0.01
0.02
0.04

0.08
0.03
0.12

-0.08
-0.11
-0.09

-0.09
-0.02
-0.10

-0.06
0.05
-0.02

-0.05
-0.10
-0.12

-0.05
0.04
-0.08

-0.05
0.01
-0.10

0.03

-0.03

-0.02

-0.06

0.06

0.07

0.02

0.04

0.06

0.02

0.12
0.10
0.11
0.11
0.00
0.17**
0.14*
0.28**
0.13*
0.20**
-0.04
0.23**
0.08

0.10
0.19**
-0.06
0.02
-0.02
0.01
0.02
0.07
0.05
-0.03
-0.02
0.04
0.07

0.05
-0.03
0.16*
0.10
0.10
0.04
0.05
0.04
0.14*
0.09
0.05
0.01
0.02

0.05
0.02
0.02
-0.01
*
0.13
-0.15*
0.08 -0.18**
0.00
0.01
0.11 -0.19**
0.08
-0.12
0.11
-0.08
0.10
-0.10
0.11 -0.20**
-0.01
0.01
0.10
-0.11
0.01
0.09

0.00
-0.02
-0.03
-0.11
-0.01
-0.03
-0.07
-0.10
-0.02
-0.09
0.01
-0.10
0.00

-0.05
0.01
-0.14*
-0.09
0.02
-0.10
0.00
-0.01
-0.12
-0.06
0.01
0.01
-0.04

-0.10
-0.03
-0.09
-0.01
0.03
-0.12
-0.10
-0.10
-0.10
-0.04
0.03
-0.10
-0.03

-0.08
-0.01
-0.04
0.01
-0.01
-0.04
-0.01
-0.06
-0.05
-0.04
0.02
-0.04
-0.02

0.03
-0.02
-0.12
-0.03
-0.03
-0.05
-0.05
-0.12
-0.06
-0.12
-0.05
-0.11
0.01

155

Table 10. Correlations between picture tearing variables and parenting characteristics.
Mother-report
Father-report

Epoch 1 Empathy
Hesitation
Defiance
Squirming
Hunched
Shoulders
Social Ref
Laugh
Smile
Tear or
Not
Latency
Number
Prompts
How
Much
Torn
Enjoyment
Epoch 2 Empathy
Defiance
Squirming

Pos
Parent

Consist

Poor
Monitor

Corp
Punish

Approp
Punish

Pos
Parent

Involve

0.12
-0.02
-0.04
-0.07

0.14*
0.05
-0.02
-0.12

-0.08

Consist

Poor
Monitor

Corp
Punish

Approp
Punish

Involve

0.05
0.02
-0.12
0.05

0.13
0.06
0.14*
0.01

0.01
0.02
-0.06
-0.12

0.08
0.02
0.04
-0.03

-0.02
0.01
-0.02
0.05

-0.01
0.03
-0.06
-0.02

0.15
0.01
0.09
-0.02

0.19*
0.08
0.00
-0.01

0.18*
0.03
-0.08
-0.08

0.01
-0.02
0.05
0.01

-0.09

-0.01

0.07

-0.03

-0.19**

-0.12

-0.16

0.13

0.04

-0.09

-0.17

0.00
0.01
-0.11

0.08
0.03
-0.08

0.03
0.01
-0.07

0.20**
-0.10
0.07

-0.08
0.04
0.08

0.00
0.13
-0.12

0.10
0.03
-0.07

0.08
0.04
0.00

0.09
0.05
-0.04

0.01
-0.01
-0.05

-0.07
-0.07
-0.06

-0.04
0.05
-0.10

-0.04

-0.03

0.08

0.00

0.03

-0.03

0.06

0.19

-0.19

0.13

0.13

0.04

-0.06

-0.03

-0.08

0.11

-0.01

-0.03

-0.09

-0.16

0.12

-0.07

-0.11

0.00

-0.03

-0.03

-0.12

0.13

-0.05

0.04

-0.12

-0.13

0.02

-0.02

-0.08

0.07

-0.07

-0.04

0.10

0.02

0.03

-0.05

0.02

0.09

-0.11

0.04

0.11

-0.04

-0.08
0.05
0.02
-0.01

-0.10
0.03
0.02
-0.01

-0.02
-0.13
0.01
0.03

0.00
-0.06
0.02
0.05

0.13
0.00
0.00
0.11

0.01
0.03
-0.01
0.03

0.11
0.09
.c
-0.03

0.08
0.06
.c
0.03

0.18
-0.05
.c
0.03

-0.02
-0.01
.c
-0.02

-0.05
0.03
.c
0.13

0.06
0.03
.c
0.12

156

Table 10 (cont’d)
Hunched
0.03
0.04
Shoulders
Laugh
-0.17* -0.21**
Smile
-0.03
-0.01
Blame Other
-0.02
0.01
Epoch 3
Blame Self
0.01
-0.04
Apologize
0.00
0.04
Silence
-0.05
0.00
Lying
0.00
0.00
Squirming
-0.05
0.00
Hunched
0.01
0.10
Shoulders
Social Ref
-0.10
-0.06
Laugh
0.01
0.00
Smile
-0.13
-0.02
Gaze
0.01
0.02
Avoidance
Overall Guilt
0.10
0.11
Relief for Copy
0.03
0.04
Empathy
0.11
0.13
Totals
Hesitation
-0.03
0.05
Defiance
-0.06
-0.03
Social Ref
-0.06
0.02
Laugh
-0.02
-0.02
Smile
-0.12
-0.03
Guilt
0.10
0.15*
Composites

0.05

0.25**

0.01

0.03
0.04
0.04
-0.06
0.10
0.06
-0.01
0.00

-0.03
0.03
0.06
-0.08
0.06
0.05
-0.10
-0.11

0.00
0.02
-0.05
0.02
-0.08
0.00
0.03
-0.05

0.01

0.18*

0.05
0.01
0.10

-0.12

0.12

-0.05

-0.05

0.06

0.03
-0.07
0.11
-0.10
0.06
-0.04
-0.07
0.10

-0.01
-0.01
0.12
-0.09
0.12
-0.01
-0.02
0.13

0.06
-0.05
0.05
0.02
0.02
-0.11
0.08
-0.10

-0.06
0.04
0.12
-0.14
0.13
-0.02
0.04
0.03

-0.05
-0.10
0.03
-0.06
0.16
0.09
0.05
-0.01

0.03
0.00
0.04
-0.06
0.09
-0.14
-0.14
-0.06

0.07

0.01 -0.03

-0.01

-0.08

-0.03

-0.04

-0.07

0.08
-0.13
-0.01

0.14
-0.03
0.04

0.03 0.00
0.06 -0.06
0.03 -0.03

0.04
-0.04
-0.03

0.00
0.09
0.07

0.13
0.07
0.08

0.05
0.03
-0.01

0.01
-0.14
-0.18*

0.07

0.05

-0.07

-.165* -0.11

-0.06

-0.19*

0.04

0.04

-0.01

-0.09
0.04
0.03
0.02
-0.13
0.03
0.01
0.05
-0.02

0.16*
0.08
0.12
0.07
0.13*
0.15*
-0.14*
0.03
0.19**

-0.08
-0.05
0.02
0.01
-0.08
0.02
-0.03
0.05
-0.05

0.06
0.01
-0.03
0.02
-0.10
0.04
-0.03
-0.02
0.02

0.03
0.17
0.16
0.02
0.12
0.15
0.11
0.05
0.15

0.09
0.05
0.18*
0.08
0.00
0.07
0.05
0.08
0.15

0.10
-0.01
0.18*
0.03
-0.07
-0.04
-0.01
0.00
0.13

-0.13
-0.10
0.01
-0.03
0.02
-0.03
-0.11
-0.11
-0.10

157

0.05 -0.10
-0.11
0.02
0.06
0.02
-0.03
-.178*
0.02
-0.05

-0.09
-0.13
0.08
0.03
0.03
0.04
0.07
0.01
-0.04

0.14
0.02
-0.03
0.01
-0.05
0.04
-0.05
-0.04
0.04

Table 10 (cont’d)
Authority
Noncompliance
Positive Affect
Avoidance
Note: * p < .05, ** p < .01

-0.02
-0.11
-0.10
0.00

0.05
-0.06
-0.06
0.05

0.04
-0.11
0.02
0.05

0.08
0.19**
-0.04
0.11

-0.03
0.06 0.08
-0.06
0.00 -0.12
0.04
0.02 -0.02
-0.04 -0.19** -0.01

158

0.09
-0.14
-0.02
0.01

0.04
0.08
0.13
-0.14

0.12
-0.02
0.07
0.06

0.04
-0.08
-0.01
0.12

0.00
0.03
-0.09
-0.11

Table 11. Factor structures based on exploratory factor analysis.
Three Factor
Four Factor
Five Factor
Empathy Guilt
Guilt
Guilt
Hesitation Guilt
Guilt
Authority
Defiance Noncompliance Noncompliance
Noncompliance
Social Reference (dropped)
(dropped)
(dropped)
Laughing Positive Affect Positive Affect
Positive Affect
Smiling Positive Affect Positive Affect
Positive Affect
Hunched
Shoulders (dropped)
(dropped)
(dropped)
Squirming (dropped)
(dropped)
(dropped)
Number of
Prompts Noncompliance Noncompliance
Noncompliance
Latency to Tear Noncompliance Noncompliance
Noncompliance
How Much Torn Noncompliance Noncompliance
Noncompliance
Enjoyment Positive Affect Positive Affect
Positive Affect
Blame Other Guilt
Guilt
Authority
Blame Self (dropped)
(dropped)
(dropped)
Apologize (dropped)
(dropped)
(dropped)
Silence Noncompliance Avoidance
Avoidance
Lying (dropped)
(dropped)
(dropped)
Gaze Avoidance Noncompliance Avoidance
Avoidance
Overall Guilt Guilt
Guilt
Avoidance/Guilt
Guilt/Positive
Relief for Copy
Guilt
Affect
Guilt

159

Rational Structure
Empathy
Arousal
Compliance
Arousal
Arousal
Arousal
Arousal
Arousal
Compliance
Compliance
Compliance
Empathy
Arousal
Arousal
Arousal
Arousal
Arousal
Arousal
Arousal
Empathy

Table 12.
Model
1
2
3
4
5
6

Fit statistics and comparative analyses for confirmatory factor analyses.
Description
x2 (df)
prob
CFI
TLI
3 factor
<.00
0.77
0.72
310.27 (74)
4 factor
251.97 (71)
<.00
0.82
0.77
5 factor
216.87 (65)
<.00
0.85
0.79
4 factor after modifications
207.23 (69)
<.00
0.86
0.82
5 factor after modifications
200.53 (65)
<.00
0.87
0.81
No
Rationally derived 3 factor
convergence

160

RMSEA SRMR
0.11
0.11
0.1
0.1
0.09
0.09
0.09
0.09
0.09
0.09

Table 13. Factor loading estimates for final five factor model
Factor
Variable
Estimate S.E.
p-value
Authority
Hesitation
1.00
0.00
Blame Other
0.70
0.28
0.01
How Many
Prompts
1.17
0.29
<0.01
Noncompliance Defiance
1.00
0.00
Latency to tear
76.46
6.35
<0.01
How Many
Prompts
3.22
0.24
<0.01
How Much Torn
-2.60
0.21
<0.01
Positive Affect Laughter
1.00
0.00
Smiling
4.10
0.55
<0.01
Enjoyment
3.05
0.41
<0.01
Avoidance
Silence
1.00
0.00
Gaze Avoidance
0.63
0.16
<0.01
How Guilty
1.19
1.10
0.03
Guilt
Empathy
1.00
0.00
How Guilty
2.39
2.26
0.03
Relief for Picture
1.27
0.74
0.07

161

Table 14. Descriptive statistics for three cluster solution.
How
Much
Torn

Empathy

Hesitation

Squirming

Laughing

Smiling

1.04

1.51

1.16

0.40

0.30

1.21

0.12

0.76

50.83

0.66

4.72

2.84

1.40

1.16

1.17

1.11

0.74

0.58

1.04

0.33

1.02

17.19

0.48

1.20

1.61

1.19

1.23

1.60

1.84

0.47

0.47

1.19

0.12

0.35

124.00

0.30

5.56

1.60

0.63

1.25

1.26

1.07

0.77

0.77

1.10

0.32

0.69

31.64

0.46

0.63

1.07

0.95

0.62

1.00

0.22

0.19

0.20

1.31

0.18

0.92

15.16

0.96

2.54

4.01

2.14

0.89

1.00

0.61

0.53

0.49

0.95

0.38

1.03

8.44

0.19

1.29

1.04

1.01

Epoch 2
1 Mean
Std.
Dev
2 Mean
Std.
Dev
3 Mean

Empathy

Hesitation

Squirming

Hunched
Shoulders

Social
Reference

Laughing

Smiling

0.01

0.00

0.00

0.01

0.01

0.04

0.00

0.09

0.12

0.00

0.00

0.12

0.12

0.21

0.00

0.29

0.00

0.00

0.00

0.02

0.14

0.02

0.02

0.07

0.00

0.00

0.00

0.15

0.35

0.15

0.15

0.26

0.03

0.00

0.01

0.04

0.05

0.09

0.04

0.13

Defiance

162

Latency
to Tear

Number
of
Prompts

Epoch 1
1 Mean
Std.
Dev
2 Mean
Std.
Dev
3 Mean
Std.
Dev

Defiance

Social
Reference

Tear or
Not

Hunched
Shoulders

Enjoyment

Table 14 (cont’d)
Std.
0.16
Dev
Epoch 3
1 Mean
Std.
Dev
2 Mean
Std.
Dev
3 Mean
Std.
Dev

0.00

0.08

0.21

0.22

0.28

0.19

0.33
Social
Reference

Smiling

Gaze
Avoid

Overall
Guilt

Relief
for
Copy

1.55

0.24

1.64

1.37

0.67

1.61

0.58

1.41

1.20

0.19

0.21

1.05

0.12

0.72

0.58

0.80

0.50

0.77

1.25

0.45

1.22

1.01

0.91

0.79

0.26

0.58

2.42

0.36

2.36

2.14

1.08

1.14

0.58

1.12

1.77

0.67

1.05

0.98

Blame
Other

Blame
Self

Apologize

Squirming

Hunched
Shoulders

Laughing

0.88

0.22

0.07

0.22

0.28

0.55

0.43

0.22

0.18

1.02

0.49

0.26

0.57

0.65

0.93

0.86

0.55

0.53

0.19

0.09

0.12

0.09

0.09

0.30

0.88

0.55

0.29

0.45

0.43

0.37

1.08

0.20

0.11

0.43

0.47

1.04

0.45

0.31

0.72

0.93

Silence

Lying

163

Table 15. Correlations between task variables, child and environmental characteristics, and problem behaviors.

Guilt Composite
Authority
Composite
Noncompliance
Composite
Pos Affect
Composite
Avoidance
Composite
Total Empathy
Total Hesitation
Total Defiance
Total Social
Reference
Total Laughing
Total Smiling
Latency to Tear
Tear or Not
How Much
Torn
Enjoyment in
Tearing
Gaze Avoidance
Overall Guilt
Relief for Copy
Child variables

Agg
0.15*

RB
0.11

ODD
0.17*

CD
0.08

INT
0.06

EXT
0.14*

CU
0.08

Agg
0.16

RB
0.07

ODD
0.11

CD
0.10

INT
-0.05

EXT
0.13

CU
0.08

0.08

0.09

0.10

0.07

0.07

0.09

0.08

0.05

-0.06

0.03

-0.05

-0.03

0.01

0.06

-0.02

0.01

0.07

-0.04

0.05

-0.01

0.00

0.07

0.00

0.10

0.03

0.09

0.05

0.03

0.06

0.07

0.00

0.09

-0.06

0.07

0.10

0.03

0.05

0.00

0.00

-0.13

0.04

0.10

-0.03

0.00

-0.09

0.01

-0.03

-0.02

0.11

-0.08

-0.06

-0.17

-0.02

-0.16

-0.08

0.03

0.17*
0.07
-0.01

0.09
0.13
0.04

0.23**
0.10
0.06

0.06
0.09
0.00

0.11
0.04
0.09

0.15*
0.10
0.01

0.05
0.12
-0.05

0.17
0.04
0.09

0.09
-0.01
0.10

0.15
0.03
0.14

0.12
-0.02
0.11

-0.02
-0.04
0.14

0.15
0.02
0.10

0.13
0.04
-0.02

0.14*

0.16*

0.07

0.16*

0.12

0.16*

0.09

0.01

-0.07

-0.04

0.01

-0.10

-0.02

0.03

0.08
0.02
-0.08
-0.01

0.03
0.07
-0.02
-0.13

0.07
-0.04
0.00
-0.08

0.05
0.07
-0.09
-0.06

-0.06
-0.03
0.01
-0.02

0.06
0.04
-0.06
-0.07

0.07
0.12
-0.02
0.08

0.04
-0.05
-0.02
-0.10

0.08
-0.01
-0.05
-0.17

0.01
-0.07
0.04
-0.14

0.02
-0.05
-0.03
-0.15

-0.10
-0.10
0.08
-0.16

0.06
-0.03
-0.03
-0.13

0.11
0.00
-0.04
-0.06

0.04

-0.01

-0.04

0.04

0.00

0.02

0.09

-0.05

-0.07

-0.10

-0.06

-0.17

-0.06

0.06

0.07

0.14

0.00

0.16*

-0.08

0.11

0.07

0.09

0.10

0.09

0.08

-0.10

0.10

0.08

-0.04
0.10
0.08

-0.04
0.11
0.07

-0.09
0.07
0.08

0.00
0.09
0.07

-0.05
-0.03
-0.01

-0.04
0.11
0.08

0.08
0.10
0.03

0.02
0.02
0.11

-0.04
0.03
0.14

-0.08
-0.04
0.06

0.06
0.01
0.12

-0.09
-0.10
-0.01

0.00
0.02
0.13

0.03
0.09
0.01

164

Table 15 (cont’d)
Child Age
0.09
Child Sex
-0.09
PPVT Score
-0.06
Mother-report
Positive
-0.16*
Parenting
Involvement
Inconsistent
Discipline
Poor
Monitoring
Corporal
Punishment
Appropriate
Punishment
Father-report
Positive
Parenting
Involvement
Inconsistent
Discipline
Poor
Monitoring
Corporal
Punishment

0.16*
-0.07
0.01

0.02
-0.02
-0.02

0.17*
-0.13
-0.01

-0.15*

0.19**

0.21**
0.20**

0.18**

0.21**
0.21**

0.25*

0.23**

0.24**

0.20**

0.19**

0.14*

0.17*

0.28**

0.22**

0.19**

0.09

0.23**

-0.05

-0.19*

-0.01

0.16

0.11

0.03
0.08

0.26**
0.08
0.16*

0.13
-0.09
-0.04

0.09
0.00
-0.08

-0.06
-0.12
-0.06

0.03
-0.13
0.01

-0.08
-0.11
-0.03

-0.01
-0.16
-0.03

0.14
0.10
0.12

-0.03
-0.13
-0.04

-0.03
-0.04
-0.07

0.19**
0.20**

-0.17*

-0.18*

-0.18*

-0.18*

-0.15

-0.12

-0.19*

-0.14*

0.25**

0.24**

0.25**

0.23**

-0.06

0.27**

0.37**
0.34**

0.13

0.26**

0.27**

0.04

0.16

0.03

0.14

-0.05

0.09

0.10

0.15*

0.17*

0.19**

0.28**

0.05

0.06

-0.01

0.08

0.06

0.06

0.06

0.24**

0.23**

0.14*

0.27**

0.11

0.15

0.24**

0.12

0.23**

0.07

0.20*

0.09

0.23**

0.07

0.13

0.16*

0.11

-0.11

-0.10

-0.04

-0.14

-0.21*

-0.11

-0.18*

-0.08

-0.13

-0.17

-0.12

-0.20*

-0.17*

-0.12

-0.08

0.24**

-0.15

0.26**
0.30**

-0.19*

-0.07

0.32**
0.38**

-0.21*

-0.07

0.27**
0.33**

0.22*

0.06

0.09

0.15

-0.07

0.24**

0.24**

0.26**

0.21*

0.19*

0.26**

0.16

0.05

0.08

-0.01

0.07

0.04

0.12

-0.05

-0.03

-0.02

-0.07

0.11

-0.05

0.15

0.03

0.05

0.01

-0.04

0.06

0.17

0.10

0.10

0.04

0.06

-0.01

0.10

0.20*

0.30**
0.28**

-0.09
-0.02

165

-0.20*

0.22**

Table 15 (cont’d)
Appropriate
Punishment
Emotion
Recognition
# Emotions
Correct
Happy Wrong
Angry Wrong
Sadness Wrong
Surprised
Wrong
Fear Wrong
Neutral Wrong
Mother-report
Effortful
Control
Negative
Emotionality
Positive
Emotionality
Father-report
Effortful
Control
Negative
Emotionality
Positive
Emotionality
Laboratory
Tasks

0.01

0.01

0.04

-0.01

0.01

0.01

0.09

-0.07

-0.06

-0.03

-0.10

-0.03

-0.07

-0.04

0.06

0.06

-0.01

0.07

0.07

0.06

-0.02

-0.14

-0.03

-0.13

-0.08

-0.03

-0.11

-0.14

-0.02
0.01
-0.05

-0.01
0.00
-0.07

0.00
0.05
0.01

-0.02
0.01
-0.07

-0.02
-0.03
-0.05

-0.02
0.01
-0.06

0.11
0.02
0.02

0.03
0.20*
0.06

-0.04
0.05
-0.03

0.03
0.18*
0.07

-0.04
0.13
0.06

-0.02
0.08
-0.01

0.00
0.16
0.03

0.03
0.29**
-0.02

-0.09

-0.09

-0.02

-0.09

-0.03

-0.10

-0.03

0.14

0.06

0.13

0.08

0.05

0.12

0.23*

-0.01
-0.08

0.00
-0.06

0.05
-0.04

-0.03
-0.08

-0.05
-0.10

-0.01
-0.08

0.06
-0.05

0.11
0.06

0.06
0.02

0.10
0.06

0.08
0.04

0.01
0.00

0.10
0.05

0.10
0.02

-0.33*

0.32**

0.32**

0.35**

-0.06

0.35**

0.30**

0.26**

-0.21*

0.26**

-0.22*

-0.04

0.26**

0.26**

0.21**

0.09

0.24**

0.09

0.30**

0.17*

0.07

0.31**

0.11

0.30**

0.18*

0.32**

0.25**

0.17*

0.29**

0.16*

0.28**

0.17*

-0.11

0.25**

0.05

0.36**

0.27**

0.39**

0.32**

-0.06

0.35**

0.20*

0.34**

-0.20*

0.33**

-0.22*

-0.11

0.30**

-0.22*

0.35**

0.26**

0.38**

0.29**

-0.11

0.34**

0.29**

0.18*

0.10

0.21*

0.15

0.24**

0.16

0.01

0.41**

0.18*

0.40**

0.28**

0.32**

0.34**

0.25**

0.18*

0.07

0.21*

0.14

-0.08

0.15

-0.09

0.32**

0.21*

0.31**

0.25**

-0.07

0.30**

0.26**

166

Table 15 (cont’d)
Fearfulness
Compliance

-0.08

-0.07

-0.03

-0.09

-0.04

-0.08

-0.05

-0.15

-0.13

-0.14

-0.14

0.05

-0.15

-0.15

0.00
0.29*
*

Note: * p < .05, ** p < .01

167

-0.01

0.08

0.26* 0.31**

-0.01
0.27*
*

0.07
-0.07

0.00
0.30*
*

-0.05
-0.17

Table 16. Correlations between picture tearing variables and temperament traits.
Mother-reported
Inhib
Cont

Att
Shift

Shy

0.04

-0.03

-0.08

-0.10

0.10

-0.08 -0.16*

-0.13

0.07

-0.09 -0.17*

Anger

Guilt
Composite
Authority
Composite
Noncomplianc
e Composite
Pos Affect
Composite
Avoidance
Composite
Total Empathy
Total
Hesitation
Total Defiance
Total Social
Reference
Total Laughing
Total Smiling
Latency to
Tear
Tear or Not
How Much
Torn

Hi
Pleas

Sadness

Impul

Soothe

0.23**

-0.05

0.01

-0.02

0.14* 0.22**

-0.10

0.08

-0.08

0.19**

-0.13

-0.05

-0.04

0.11

0.18**

0.12

Fear

Att
Focus

EC

NE

PE

0.06

0.01

0.07

0.18**

0.05

-0.09

0.09

0.14*

0.08 -0.18*

0.09

0.01

-0.03

-0.04

0.10

-0.12

0.04

0.08

-0.04

-0.04

0.03

-0.13

0.01

-0.09

-0.02

-0.09

0.06

0.17*

0.03

-0.02

-0.01

0.10

-0.08

0.10

-0.11

0.04

-0.08

-0.08

0.08

-0.02

**

-0.10

0.05

0.18**

-0.05

0.00

-0.03

0.11

0.01

0.08

0.12

0.14*

-0.13 -0.15*

0.00

-0.01

0.12

-0.09

0.13

-0.12

0.08

-0.11

0.14*

0.09

-0.03

-0.04

-0.10

0.13

-0.10

-0.05

0.01

0.15*

-0.12

0.04 -0.14*

0.06

-0.03

0.03

-0.13

-0.05

-0.05

0.13

0.13

-0.10

-0.02

-0.03

-0.08

-0.09

-0.03

0.08

0.07
-0.10

-0.02
0.02

-0.01
0.10
-0.10
**
0.21
0.07
0.07

-0.06
-0.08

0.04
0.05

-0.10
0.01

-0.04
-0.04

0.00
0.07

0.01
-0.13

0.02
0.07

-0.02
-0.12

0.00
-0.04

-0.07

-0.04

0.08 -0.17*

0.12

-0.13

0.10

0.02

-0.10

0.06
-0.02
**
0.18
0.09

0.00

-0.02

-0.14

0.10

-0.17*

0.07 -0.19*

-0.11

-0.14*

0.17*

0.05

-0.01

-0.14*

0.09

-0.15*

0.06 -0.16*

-0.02

0.11
-0.07
-0.08

0.04

0.19

0.15*

0.22**

168

Table 16 (cont’d)
Enjoyment in
Tearing
Gaze
Avoidance
Overall Guilt
Relief for Copy

-0.08

-0.12

0.17*

-0.19*

0.08

0.16*

0.03

-0.01

-0.07

-0.23**

-0.08

-0.11

0.02

-0.01

0.02

0.11

0.03

-0.08

-0.04

0.06

-0.01

0.03

-0.03

-0.03

0.01

-0.13

0.02
0.02
0.07
-0.07
Father-reported

0.04
0.06

-0.01
-0.01

0.10
0.10

0.16*
0.14

0.02
-0.12

-0.01
0.04

-0.04
-0.10

-0.03
0.05

-0.02
-0.01

0.03
0.09

0.15
0.17*

Sadness

EC

NE

PE

Inhib
Cont

Att
Shift

Impul

Soothe

Fear

Att
Focus

Shy

0.01

-0.16

-0.01

-0.12

0.16

0.21*

-0.04

-0.04

-0.13

0.17

-0.16

-0.01

0.12

-0.01

-0.02

-0.03

-0.13

0.03

0.12

-0.07

0.01

-0.04

-0.03

-0.10

-0.08

0.03

0.07

-0.05

-0.07

0.11

-0.15

-0.01

0.05

-0.03

-0.11

0.09

-0.11

0.09

-0.05

-0.05

-0.01

0.20*

-0.08

0.12

0.03

0.00

-0.02

0.07

-0.09

0.03

-0.07

0.04

-0.05

-0.01

0.09

-0.02

0.02

0.07

-0.01

0.10

0.08

-0.02

0.05

0.00

0.02

0.02

-0.17

-0.15

-0.11

0.05

0.17

-0.05

-0.06

-0.14

0.14 -0.17*

-0.02

0.05

0.05

-0.07

0.00

-0.02

-0.03

0.08

0.01

-0.03

-0.08

0.00

-0.16

-0.03

0.02

0.06

0.02

-0.05

0.05

-0.11

-0.05

0.09

-0.01

-0.06

0.01

-0.03

0.07

-0.06

0.06

-0.16

-0.04

0.04

0.17

0.13

-0.15

0.17*

-0.09

-0.03

-0.06

0.07

0.18*

-0.01
-0.10

-0.03
0.05

0.08
0.21*

-0.07
-0.03

0.11
0.02

-0.01
-0.01

0.06
0.01

-0.06
-0.05

0.00
0.11

-0.04
-0.12

-0.02
0.09

-0.05
-0.11

0.03
-0.01

Anger

Guilt
Composite
Authority
Composite
Noncomplianc
e Composite
Pos Affect
Composite
Avoidance
Composite
Total Empathy
Total
Hesitation
Total Defiance
Total Social
Reference
Total Laughing
Total Smiling

Hi
Pleas

169

Table 16 (cont’d)
Latency to
0.08
Tear
Tear or Not
-0.03
How Much
-0.10
Torn
Enjoyment in
-0.08
Tearing
Gaze
0.08
Avoidance
Overall Guilt
-0.03
Relief for Copy
0.05
Note: * p < .05, ** p < .01

0.20* -0.18*

-0.08

0.04

-0.10

-0.12

0.17*

-0.10

0.07

-0.05

0.16

0.12

0.08

0.09

0.25*

-0.11

0.18

-0.03

0.13

-0.15

.020*

0.20*

-0.05

0.07

0.06

-0.12

0.03

-0.08

0.12

0.20*

-0.11

0.15

0.13

0.00

0.05

-0.02

-0.01

-0.01

0.01

0.11

-0.03

0.04

0.02

-0.12

-0.04

-0.01

0.20*

0.10

0.04

0.05

0.13

-0.07

-0.03
-0.17

0.08
0.19

-0.04
-0.02

0.16
0.10

0.16
0.12

-0.03
0.01

0.04
-0.11

-0.04
-0.12

0.11
0.11

-0.06
-0.10

0.03
-0.03

0.16
0.10

-0.02

-0.08

0.08

0.10

-0.02

-0.02

0.11

-0.04

170

Table 17. Hierarchical multiple regression analyses predicting current child problem behaviors from picture tearing task
composite variables
Effects of predictors

Effects of predictors
ΔR2 for block
ΔF

Mother
reported
AGG
Step 1

B

ΔR2

p

.037
Guilt
Authority
Noncomp
Pos Affect
Avoidance

RB

t

0.06
0.00
-0.02
0.02
-0.03

2.40
0.15
-0.41
0.68
-0.92

0.02

1.94

.178

B

Guilt
Authority
Noncomp
Pos Affect
Avoidance
.027

5.70

.018

.012

0.49

.783

t

ΔR2

p

Step 1

.017
.883
.683
.495
.359

Step 1
Age

ΔF

Father
reported
AGG

p
1.54

ΔR2 for block

0.04
0.00
0.00
0.00
-0.04

1.38
-0.10
-0.09
0.10
-1.22

p

.023

0.58

.713

.015

0.38

.865

.042

1.10

.366

.172
.920
.926
.921
.227

RB

.054

Step 2

Step 1

Guilt
Authority
Noncomp
Pos Affect

0.02
0.01
0.00
0.01

1.22
0.28
0.13
0.58

.224
.780
.894
.565

Guilt
Authority
Noncomp
Pos Affect

0.02
-0.02
-0.01
0.01

0.91
-0.83
-0.42
0.31

.363
.409
.679
.757

Avoidance

-0.01

-0.26

.798

Avoidance

-0.02

-0.97

.332

ODD

ODD
.055

Step 1

2.33

.044

Step 1

Guilt
Authority

0.10
0.01

2.75
0.26

.006
.794

Guilt
Authority

0.06
0.00

1.19
0.02

.236
.983

Noncomp
Pos Affect
Avoidance

0.02
0.00
-0.07

0.38
-0.01
-1.68

.707
.993
.095

Noncomp
Pos Affect
Avoidance

0.01
-0.02
-0.12

0.12
-0.41
-2.09

.906
.683
.038

171

Table 17 (cont’d)
CD

Step 1
Age

0.02

2.53

Pos Affect
Avoidance
INT

0.02
0.00
-0.01
0.01

1.05
0.04
-0.33
0.74

.295
.965
.745
.457

-0.01

-0.26

.794

Step 1
Age

0.03

3.83

Authority
Noncomp
Pos Affect
Avoidance

.012

.010

0.41

.845

CD
Step 1
Guilt
Authority
Noncomp
Pos Affect
Avoidance

.067

14.68

.000

.025

1.08

.373

0.02
-0.02
-0.01
0.00

0.90
-0.85
-0.20
0.01

.368
.399
.840
.993

-0.01

-0.56

.578

.010

0.26

.936

.049

1.29

.273

.020

0.50

.775

.011

0.28

.923

INT

.000

Step 2
Guilt

6.42

.012

Step 2
Guilt
Authority
Noncomp

.030

0.01
0.00

0.59
0.02

.556
.987

0.00
-0.03
-0.02

0.18
-1.91
-1.07

.861
.057
.288

Step 1
Guilt
Authority
Noncomp
Pos Affect
Avoidance

EXT

-0.01
0.00

-0.33
0.02

.745
.984

0.01
-0.03
-0.04

0.20
-1.56
-1.71

.845
.120
.090

EXT
Age
.032

Step 1
Guilt
Authority
Noncomp
Pos Affect
Avoidance

0.04
0.01
-0.01
0.02
-0.01

2.08
0.38
-0.21
0.88
-0.58

1.35

.245

Step 1

.039
.702
.835
.381
.560

Guilt
Authority
Noncomp
Pos Affect
Avoidance

CU

0.03
-0.01
-0.01
0.00
-0.03

1.28
-0.40
-0.23
0.19
-1.20

.203
.691
.820
.849
.233

CU
.034

Step 1
Guilt
Authority
Noncomp

0.01
0.02
0.02

0.50
0.74
0.54

1.43

.216

.618
.462
.587

Step 1
Guilt
Authority
Noncomp

172

0.00
0.03
0.00

-0.10
0.90
-0.07

.922
.371
.947

Table 17 (cont’d)
Pos Affect
Avoidance

0.04

1.71

.089

0.04

1.67

.097

Pos Affect
Avoidance

173

0.01

0.40

.690

0.01

0.38

.705

Table 18. Hierarchical multiple regression analyses predicting current child problem behaviors from single picture tearing
task variables
ΔR2 for block

Effects of predictors
B
AGG

0.08
0.05

2.27
1.58

0.02

1.30

ODD

0.17

3.35

0.03

1.88

Social Ref

1.79

.033

3.56

.030

.017

3.64

.058

.052

11.21

.001

.021

3.53

.062

.034

3.71

.026

.001
.062

Step 1
0.04

.016

.058

Step 1
Enjoyment

EXT

1.91

Step 1
Empathy

CD

0.04

.039

.198

Step 2
Social Ref

p
4.20

.024
.115

Step 1
Age

ΔF

ΔR2

p

Step 1
Empathy
Social Ref

RB

t

.076

174

Table 19. Hierarchical multiple regression analyses predicting current child temperament traits from picture tearing task
composite variables
ΔR2 for
block

Effects of
predictors

ΔR2 for
block

Effects of
predictors

Lab task
variables
B
Fear

t

p

Step 1
Age

-0.63

-1.38

ΔR2
.01
2

Noncomp
Pos Affect
Avoidanc
e

B
.168

1.23

1.58

.115

-1.09

-1.37

.172

-0.01

-0.01

.994

0.25

0.29

.769

-0.61

-0.69

.489

0.77

p

0.15

5.16

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e

0.04

0.78

-0.03

-0.68

-0.16

-1.98

-0.02

-0.32

0.11

1.81

ΔR2
.11
8

F
26.6
4

.05
0

2.31

p
.000

.00
0

Step
2

.576

Mother
reported
Anger

Comp

t

Step
1
Age

.02
4

Authority

p

.168

Step 2
Guilt

F
1.92

.046

.43
5
.50
0
.04
9
.75
0
.07
1

Father
reported
.03
6

Step 1
Age

-0.13

-2.60

Noncomp

Anger

Step
1
Age

.02
5

Authority

.024

.010

Step 2
Guilt

3.80

0.03

0.34

.731

0.17

1.78

.077

0.03

0.23

.819

1.08

.374

-2.58

Authority
Noncomp

-0.01

-0.13

0.13

1.10

0.01

0.07

.015

.01
3

.875

.01
1

Step
2
Guilt

175

-0.16

.06
4

.89
7
.27
3
.94
6

Table 19 (cont’d)
Pos Affect
Avoidanc
e
Inhib

-0.02

-0.23

.815

-0.07

-0.76

.446
.05
7

Step 1

Contr

Age

0.13

3.13

Authority
Noncomp
Pos Affect
Avoidanc
e

.003

0.00

-0.06

.954

-0.12

-1.48

.139

-0.14

-1.11

.270

-0.13

-1.65

.100

0.00

0.03

.976

Inhib

1.37

-0.43

-0.07

-0.55

.66
9
.58
6

Age

0.15

2.75

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Att Shift

-0.17

-1.80

-0.07

-0.66

-0.01

-0.08

-0.09

-0.85

0.05

0.42

.09
5

.002

.04
1

.323

.05
0

.228

.00
7

Step
2

.236

Step 1

-0.05

Step
1

Contr
.03
1

Guilt

6.16

.002

Step 2

Att Shift

Pos Affect
Avoidanc
e

.07
4
.51
2
.93
7
.39
5
.67
8

Step
1
Age

.08
5

Step 2
Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Shyness

Step 1

-0.09

-0.96

.339

-0.20

-1.97

.051

-0.22

-1.51

.133

0.12

1.24

.215

0.20

1.96

.051

Step
2

.003

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Shyness

Step
1
Age

176

-0.05

-0.40

-0.14

-1.15

0.02

0.10

0.26

2.03

0.19

1.36

.69
1
.25
3
.92
2
.04
4
.17
6

Table 19 (cont’d)
.07
9

Step 2
Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Hi Pleas

-0.18

-1.64

.102

-0.17

-1.38

.168

0.49

2.71

.007

-0.18

-1.59

.114

0.09

0.76

.446

Step
2

.003

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e

Step 1

Hi Pleas

Age
.05
4

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Impuls

0.15

1.86

.064

0.12

1.35

.178

-0.32

-2.40

.017

0.02

0.20

.844

-0.07

-0.77

.443

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Impuls

0.23

1.06

-0.12

-0.78

0.03

0.18

0.10

1.62

0.18

1.77

-0.04

-0.37

-0.28

-1.75

0.10

0.87

-0.07

-0.55

.10
2

.001

.05
1

.205

.07
8

.057

.10
7

.07
9
.70
9
.08
2
.38
7
.58
7

Age
.11
6

Step 2

Authority

-1.29

Step
1

Age

Guilt

-0.19

.236

.19
9
.20
0
.29
1
.43
9
.86
0

Step
2

.035

Step 1

-1.29

Step
1

Age
Step 2

-0.18

.05
1

0.22

3.01

.003

0.20

2.58

.011

.000

Step
2
Guilt
Authority

177

0.21

2.38

0.11

1.18

.01
9
.24
1

Table 19 (cont’d)
Noncomp
Pos Affect
Avoidanc
e
Soothe

-0.14

-1.21

.229

0.13

1.74

.083

-0.01

-0.08

.940

Noncomp
Pos Affect
Avoidanc
e

Step 1

Soothe

Step 2

Noncomp
Pos Affect
Avoidanc
e
Fear

-0.05

-0.80

.426

-0.04

-0.59

.555

-0.04

-0.41

.686

-0.01

-0.18

.857

0.10

1.39

.165

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Fear

Step 2

Pos Affect
Avoidanc
e
Att
Focus

Step 1

0.27

-0.02

-0.22

-0.09

-1.03

0.16

1.21

0.03

0.37

0.02

0.15

.02
0

.774

.02
5

.679

.02
4

.211

.82
7
.30
5
.22
7
.71
4
.88
0

Age
.02
4

Noncomp

0.03

Step
1

Age

Authority

0.72

Step
2

.532

Step 1

Guilt

0.07

.99
3
.47
5
.78
9

Age
.02
0

Authority

-0.01

Step
1

Age

Guilt

0.00

-0.05

-0.53

.598

0.10

0.96

.336

0.17

1.17

.245

-0.07

-0.78

.438

-0.08

-0.83

.407

Step
2

.399

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
.05
3

.004

Att
Focus

178

Step
1

-0.14

-1.18

0.02

0.19

-0.05

-0.26

-0.05

-0.38

0.19

1.37

.24
2
.84
8
.79
6
.70
7
.17
3

Table 19 (cont’d)
Age

0.10

2.47

.014
.05
4

Step 2
Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Sad

Age

-0.02

-0.25

.803

-0.09

-1.06

.290

-0.33

-2.71

.007

-0.09

-1.19

.234

0.04

0.53

.599

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
Sad

Step 2

Noncomp
Pos Affect
Avoidanc
e
EC

0.08

1.12

.263

0.04

0.46

.648

0.02

0.19

.848

-0.16

-2.07

.039

-0.14

-1.70

.090

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
EC

-0.01

-0.13

-0.13

-0.92

0.01

0.14

0.13

1.21

.403

.04
5

.328

.05
4

.208

.13
7
.89
3
.36
1
.88
8
.22
9

0.18

2.13

-0.07

-0.80

0.07

0.53

-0.04

-0.49

-0.07

-0.75

.03
5
.42
3
.59
6
.62
5
.45
3

Step
1

Age

Age
.04
2

Step 2

Authority

-1.50

Step
2

.107

Step 1

Guilt

-0.14

.03
9

Age
.04
3

Authority

.60
7

Step
1

Age

Guilt

0.52

Step
2

.039

Step 1

0.03

0.04

0.74

.463

-0.09

-1.51

.131

.103

Step
2
Guilt
Authority

179

-0.10

-1.61

-0.08

-1.27

.11
0
.20
6

Table 19 (cont’d)
Noncomp
Pos Affect
Avoidanc
e
NE

-0.21

-2.49

.014

-0.05

-0.92

.356

-0.02

-0.41

.685

Noncomp
Pos Affect
Avoidanc
e

Step 1

NE

Step 2

Noncomp
Pos Affect
Avoidanc
e
PE

0.03

0.68

.495

0.07

1.36

.176

0.03

0.48

.630

-0.06

-1.33

.186

-0.05

-0.95

.342

Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e
PE

Step 2

Pos Affect
Avoidanc
e

0.87

-0.02

-0.33

-0.02

-0.43

0.06

0.70

-0.02

-0.27

0.01

0.10

.00
8

.956

.01
7

.797

.74
2
.67
0
.48
4
.78
6
.91
8

Age
.06
4

Noncomp

0.06

Step
1

Age

Authority

-0.15

Step
2

.236

Step 1

Guilt

-0.01

.55
2
.88
0
.38
7

Age
.03
2

Authority

-0.60

Step
1

Age

Guilt

-0.06

0.10

2.76

.006

0.05

1.42

.157

-0.04

-0.74

.462

0.00

0.07

.943

-0.05

-1.30

.196

.017

Step
2
Guilt
Authority
Noncomp
Pos Affect
Avoidanc
e

180

0.06

1.31

0.00

-0.02

-0.03

-0.47

0.03

0.51

-0.02

-0.29

.19
4
.98
3
.63
9
.61
2
.77
5

Appendix C. Original picture tearing coding scheme.
ID: _________
Coder: _________
Date: __________

Durbin Lab Guilt Coding Scheme

Time - first prompt given: ___________
Time - child rips the picture / experimenter gives up: _____________

Code timestamps for verbalizations: (examples below)
Concern: It’s his favorite, What if she’s mad, She’ll be sad / Hesitation: Are you sure? Can I? /
Defiance: No, I can’t, I’m not going to
Concern/empathy

Hesitation

Defiance/Noncompliance

Bodily reactions (AFTER PROMPT but BEFORE ALBUM OWNER RETURNS)
Squirming

Hunched shoulders

Lip biting

Crossing arms

Leaning

Social reference

Laughing

Smiling

away/Hiding

181

Did the child tear the picture when prompted? How many prompts had to be given?
-

No

-

Yes

-

How many prompts: _______

How much did the child tear the picture?
1. Not at all (refused to tear)
2. A small amount (the corner, a small rip)
3. A moderate amount (Torn in some places, around edges)
4. A large amount (Torn into several pieces, in half)
5. Most of the picture (torn into pieces, can’t see the picture at all)

How much enjoyment did the child exhibit in tearing the picture?
1. None
2. A little
3. A moderate amount
4. A lot

182

AFTER ALBUM OWNER RETURNS
How the child responded when the experimenter came back?
Blame other

Blame self

Apologize

Silence (every five

Lying

seconds)

Bodily reactions
Squirming

Hunched shoulders

Lip biting

Crossing arms

Leaning

Social reference

Laughing

Smiling

away/Hiding

183

Gaze avoidance (code every five seconds of averted gaze)

How guilty does the child seem?
1. Not at all
2. A little
3. Moderate
4. A lot

How much relief does the child show at the second picture?
1. None
2. A little
3. Moderate amount
4. A lot

Other comments:

184

Appendix D. Description of clusters in four cluster solution.
Description of clusters in four cluster solution. In order to determine the similarities
and differences between clusters, we conducted a one-way ANOVA using cluster membership
and the variables used to create the clusters as dependent variables. We also conducted post-hoc
comparisons of significant ANOVAs using Tukey’s HSD.

Average characteristics of the four

clusters and characteristics on which the clusters differed are described below.
Cluster 1. Members of this cluster comprised 24.9% of the sample (n = 67). Compared
to other clusters, children in cluster 1 took more time to comply and tore the picture more
completely than children in cluster 3, but took less time and tore less than children in clusters 2
and 4. During the first epoch (between the first prompt and the main experimenter leaving), they
showed more empathy and hesitation than children in cluster 3 but did not differ from clusters 2
and 4 on these variables. They displayed less defiance than cluster 4 and more defiance than
cluster 3. In response to the victim, they showed less physical discomfort (i.e. squirming,
hunched shoulders) than cluster 3 and more than clusters 2 and 4. They also displayed more
guilt and less relief than cluster 3, but less guilt and more relief than clusters 2 and 4.
Cluster 2. Members of this cluster comprised 9.3% of the sample (n = 25). Children in
cluster 2 differed from children in cluster 4 only on Latency to tear (d = 2.98) and Relief in
seeing the photo copy (d = 0.67). Compared to the other clusters, they showed more empathy,
hesitation, and defiance than cluster 3 and tore the picture less completely than clusters 1 and 3.
In response to the victim, they showed less physical discomfort (i.e. squirming, hunched
shoulders) and less guilt than clusters 1 and 3. They were also less likely to lie or be silent than
cluster 3.
Cluster 3. Members of this cluster comprised 59.1% of the sample (n = 159). Compared
to other clusters, children in this cluster waited the shortest amount of time before tearing the
185

picture and were the most likely to tear the picture. During the first epoch, they displayed less
empathy and hesitation than clusters 1 and 2 and less defiance than all the other clusters. They
also tore the picture more completely than the other clusters and demonstrated the most
enjoyment while tearing. Notably, they also displayed the most guilt upon the victim’s return
and the most relief on receiving the copy of the picture. Also in response to the victim, they
were the most likely to be silent and to appear visibly uncomfortable (i.e. squirming, hunched
shoulders).
Cluster 4. Members of this cluster comprised 6.7% of the sample (n = 18). Children in
cluster 4 differed from children in cluster 2 only on Latency to tear (d = 2.98) and Relief in
seeing the photo copy (d = 0.67). Compared to the other clusters, they showed more empathy,
hesitation, and defiance than cluster 3 and tore the picture less completely than clusters 1 and 3.
In response to the victim, they showed less physical discomfort (i.e. squirming, hunched
shoulders) and less guilt than clusters 1 and 3. They were also less likely to lie or be silent than
cluster 3.

186

Appendix E. Recommended final picture tearing coding scheme retaining most empirically
useful variables.

ID: _________
Coder: _________
Date: __________

Durbin Lab Guilt Coding Scheme

Time - first prompt given: ___________
Time - child rips the picture / experimenter gives up: _____________
Code timestamps for verbalizations: (examples below)
Concern: It’s his favorite, What if she’s mad, She’ll be sad / Hesitation: Are you sure? Can I? /
Defiance: No, I can’t, I’m not going to
Concern/empathy

Hesitation

Defiance/Noncompliance

Bodily reactions (AFTER PROMPT but BEFORE ALBUM OWNER RETURNS)
Social reference

Laughing

Smiling

187

Did the child tear the picture when prompted? How many prompts had to be given?
-

No

-

Yes

-

How many prompts: _______

How much did the child tear the picture?
6. Not at all (refused to tear)
7. A small amount (the corner, a small rip)
8. A moderate amount (Torn in some places, around edges)
9. A large amount (Torn into several pieces, in half)
10. Most of the picture (torn into pieces, can’t see the picture at all)

How much enjoyment did the child exhibit in tearing the picture?
5. None
6. A little
7. A moderate amount
8. A lot

AFTER EXPERIMENTER LEAVES, BEFORE EXPERIMENTER COMES BACK
When child is alone in room
Code timestamps for verbalizations: (examples below)
Concern: It’s his favorite, What if she’s mad, She’ll be sad / Hesitation: Are you sure? Can I? /
Defiance: No, I can’t, I’m not going to
Concern/empathy

Hesitation

Defiance/Noncompliance

188

Bodily reactions (AFTER PROMPT but BEFORE ALBUM OWNER RETURNS)
Social reference

Laughing

Smiling

189

AFTER ALBUM OWNER RETURNS
How the child responded when the experimenter came back?
Blame other

Blame self

Apologize

Silence (every five
seconds)

Bodily reactions
Social reference

Laughing

Smiling

190

Lying

Gaze avoidance (code every five seconds of averted gaze)

How guilty does the child seem?
5. Not at all
6. A little
7. Moderate
8. A lot

How much relief does the child show at the second picture?
5. None
6. A little
7. Moderate amount
8. A lot

Other comments:

191

REFERENCES

192

REFERENCES

Abe, J., & Izard, C. (1999). Compliance, Noncompliance Strategies, and the Correlates of
Compliance in 5-year-old Japanese and American Children. Social Development. Retrieved
from http://onlinelibrary.wiley.com/doi/10.1111/1467-9507.00077/full
Achenbach, T., & Rescorla, L. (2001). ASEBA school-age forms & profiles. Retrieved from
http://aseba.com/ordering/ASEBA Reliability and Validity-School Age.pdf
Aksan, N., & Kochanska, G. (2005). Conscience in Childhood: Old Questions, New Answers.
Developmental Psychology, 41(3), 506–516. Retrieved from
http://eric.ed.gov/?id=EJ684980
Asendorpf, J. B., & Nunner-Winkler, G. (1992). Children’s Moral Motive Strength and
Temperamental Inhibition Reduce Their Immoral Behavior in Real Moral Conflicts. Child
Development, 63(5), 1223–1235. http://doi.org/10.1111/j.1467-8624.1992.tb01691.x
Belacchi, C., & Farina, E. (2012). Feeling and Thinking of Others: Affective and Cognitive
Empathy and Emotion Comprehension in Prosocial/Hostile Preschoolers. Aggressive
Behavior, 38(2), 150–165. http://doi.org/10.1002/ab.21415
Beyers, J. M., Bates, J. E., Pettit, G. S., & Dodge, K. A. (2003). Neighborhood structure,
parenting processes, and the development of youths' externalizing behaviors: A multilevel
analysis. American journal of community psychology, 31(1-2), 35-53.
Blair, R. J., Colledge, E., Murray, L., & Mitchell, D. G. (2001). A selective impairment in the
processing of sad and fearful expressions in children with psychopathic tendencies. Journal
of Abnormal Child Psychology, 29, 491–498. http://doi.org/10.1023/A:1012225108281
Blair, R. J. R., Budhani, S., Colledge, E., & Scott, S. (2005). Deafness to fear in boys with
psychopathic tendencies. Journal of Child Psychology and Psychiatry, and Allied
Disciplines, 46(3), 327–36. http://doi.org/10.1111/j.1469-7610.2004.00356.x
Blair, R. J. R., & Coles, M. (2000). Expression recognition and behavioural problems in early
adolescence. Cognitive Development, 15(4), 421–434. http://doi.org/10.1016/S08852014(01)00039-9
Blass, T. (1991). Understanding behavior in the Milgram obedience experiment: The role of
personality, situations, and their interactions. Journal of Personality and Social Psychology.
Retrieved from http://psycnet.apa.org/journals/psp/60/3/398/
Borke, H. (1971). Interpersonal Perception of Young Children: Egocentrism or Empathy?.
Developmental Psychology. Retrieved from http://eric.ed.gov/?id=EJ043606
193

Borke, H. (1973). The Development of Empathy in Chinese and American Children between
Three and Six Years of Age: A Cross-Culture Study. Developmental Psychology. Retrieved
from http://eric.ed.gov/?id=EJ081919
Braungart-Rieker, J., Garwood, M., & Stifter, C. (1997). Compliance and noncompliance: The
roles of maternal control and child temperament. Journal of Applied …. Retrieved from
http://www.sciencedirect.com/science/article/pii/S0193397397800081
Broidy, L. M., Nagin, D. S., Tremblay, R. E., Bates, J. E., Brame, B., Dodge, K. A., ... & Lynam,
D. R. (2003). Developmental trajectories of childhood disruptive behaviors and adolescent
delinquency: a six-site, cross-national study. Developmental psychology, 39(2), 222.
Brook, M., & Kosson, D. S. (2013). Impaired cognitive empathy in criminal psychopathy:
Evidence from a laboratory measure of empathic accuracy.
Bruner, J. S. (1972). The Uses of Immaturity. New York University Education Quarterly, 27(8),
687–708. http://doi.org/10.1037/h0033144
Burley, P., & GUINNESS, J. M. (1977). Effects of social intelligence on the Milgram paradigm.
Psychological Reports. Retrieved from
http://www.amsciepub.com/doi/pdf/10.2466/pr0.1977.40.3.767
Carlsmith, J., Lepper, M., & Landauer, T. (1974). Children’s obedience to adult requests:
Interactive effects of anxiety arousal and apparent punitiveness of the adult. Journal of
Personality and …. Retrieved from http://psycnet.apa.org/journals/psp/30/6/822/
Cohn, J. F., & Campbell, susan b.; Matias, reinaldo; Hopkins, J. (1989). Face-to-Face
Interactions of Postpartum Depressed and Nondepressed Mother-Infant Pairs at Two
Months. Developmental Psychology, 26(1), 15–23. http://doi.org/10.1037/00121649.26.1.15
Crockenberg, S., & Litman, C. (1990). Autonomy as competence in 2-year-olds: Maternal
correlates of child defiance, compliance, and self-assertion. Developmental Psychology.
Retrieved from http://psycnet.apa.org/journals/dev/26/6/961/
Dadds, M. R., Hawes, D. J., Frost, A. D. J., Vassallo, S., Bunn, P., Hunter, K., & Merz, S.
(2009). Learning to “talk the talk”: the relationship of psychopathic traits to deficits in
empathy across childhood. Journal of Child Psychology and Psychiatry, 50(5), 599–606.
http://doi.org/10.1111/j.1469-7610.2008.02058.x
Dadds, M. R., Perry, Y., Hawes, D. J., Merz, S., Riddell, A. C., Haines, D. J., …
Abeygunawardane, A. I. (2006). Attention to the eyes and fear-recognition deficits in child
psychopathy. The British Journal of Psychiatry : The Journal of Mental Science, 189, 280–
1. http://doi.org/10.1192/bjp.bp.105.018150

194

Damon, W. (1977). Measurement and social development. The Counseling Psychologist.
Retrieved from http://psycnet.apa.org/psycinfo/1978-31504-001
De Gelder, B. de, Snyder, J., Greve, D., & Hadjikhani, N. K. (2004). Fear fosters flight: A
mechanism for fear contagion when perceiving emotion expressed by a whole body.
Proceedings of the National Academy of Sciences of the United States of America (PNAS),
101(47), 16701–16706. Retrieved from
http://www.narcis.nl/publication/RecordID/oai:wo.uvt.nl:141110
De Wied, M., Van Boxtel, A., Matthys, W., & Meeus, W. (2012). Verbal, facial and autonomic
responses to empathy-eliciting film clips by disruptive male adolescents with high versus
low callous-unemotional traits. Journal of Abnormal Child Psychology, 40, 211–223.
http://doi.org/10.1007/s10802-011-9557-8
Dienstbier, R. a. (1984). The role of emotion in moral socialization. Psychology, Paper 114,
484–514. http://doi.org/10.1016/j.tics.2008.09.006
Dunn, L. (1997). Examiner’s manual for the PPVT-III peabody picture vocabulary test: Form
IIIA and Form IIIB. Retrieved from
https://scholar.google.com/scholar?q=Dunn+%26+Dunn%2C+1997+ppvt&btnG=&hl=en&
as_sdt=0%2C23#0
Durbin, C. (2010). Validity of young children’s self-reports of their emotion in response to
structured laboratory tasks. Emotion. Retrieved from
http://psycnet.apa.org/journals/emo/10/4/519/
Durbin, C., Hayden, E., Klein, D., & Olino, T. (2007). Stability of laboratory-assessed
temperamental emotionality traits from ages 3 to 7. Emotion. Retrieved from
http://psycnet.apa.org/journals/emo/7/2/388/
Dyson, M.W., Olino, T.M., Durbin, C.E., Goldsmith, H.H., & Klein, D.N. (2012). The structure
of temperament in preschoolers: A two-stage factor analytic approach. Emotion, 12(1), 4457.
Eisenberg, N., Eggum, N. D., & Edwards, A. (2010). Empathy-related responding and moral
development.
Eisenberg, N., & Lennon, R. (1983). Sex differences in empathy and related capacities.
Psychological Bulletin. Retrieved from http://psycnet.apa.org/journals/bul/94/1/100/
Eisenberg, N., & Lundy, T. (1985). Children’s justifications for their adult and peer-directed
compliant (prosocial and nonprosocial) behaviors. Developmental Psychology. Retrieved
from http://psycnet.apa.org/journals/dev/21/2/325/

195

Eisenberg, N., & Shell, R. (1987). Prosocial development in middle childhood: A longitudinal
study. Developmental …, 2(5), 712–718. Retrieved from
http://psycnet.apa.org/journals/dev/23/5/712/
Elms, A., & Milgram, S. (1966). PERSONALITY CHARACTERISTICS ASSOCIATED WITH
OBEDIENCE AND DEFIANCE TOWARD AUTHORITATIVE COMMAND. Journal of
Experimental Research in …. Retrieved from http://psycnet.apa.org/psycinfo/1967-04552001
Epstein, S., & O’Brien, E. (1985). The person–situation debate in historical and current
perspective. Psychological Bulletin. Retrieved from
http://psycnet.apa.org/journals/bul/98/3/513/
Fairchild, G., Van Goozen, S. H. M., Calder, A. J., Stollery, S. J., & Goodyer, I. M. (2009).
Deficits in facial expression recognition in male adolescents with early-onset or
adolescence-onset conduct disorder. Journal of Child Psychology and Psychiatry, and
Allied Disciplines, 50(5), 627–36. http://doi.org/10.1111/j.1469-7610.2008.02020.x
Field, T. M., Woodson, R., Greenberg, R., & Cohen, D. (1982). Discrimination and imitation of
facial expression by neonates. Science (New York, N.Y.), 218(4568), 179–81. Retrieved
from http://www.ncbi.nlm.nih.gov/pubmed/7123230
Forman, D., & Kochanska, G. (2001). Viewing imitation as child responsiveness: a link between
teaching and discipline domains of socialization. Developmental Psychology. Retrieved
from http://psycnet.apa.org/journals/dev/37/2/198/
Fowler, P. J., Tompsett, C. J., Braciszewski, J. M., Jacques-Tiura, A. J., & Baltes, B. B. (2009).
Community violence: a meta-analysis on the effect of exposure and mental health outcomes
of children and adolescents. Development and Psychopathology, 21(1), 227–59.
http://doi.org/10.1017/S0954579409000145
Frick, P. J., & Ellis, M. (1999). Callous-Unemotional Traits and Subtypes of Conduct Disorder.
Clinical Child and Family Psychology Review, 2(3), 149–168.
http://doi.org/10.1023/A:1021803005547
Frick, P. J., & Morris, A. S. (2004). Temperament and developmental pathways to conduct
problems. Journal of Clinical Child and Adolescent Psychology : The Official Journal for
the Society of Clinical Child and Adolescent Psychology, American Psychological
Association, Division 53, 33(1), 54–68. http://doi.org/10.1207/S15374424JCCP3301_6
Frick, P. J., Ray, J. V, Thornton, L. C., & Kahn, R. E. (2014). Can callous-unemotional traits
enhance the understanding, diagnosis, and treatment of serious conduct problems in
children and adolescents? A comprehensive review. Psychological Bulletin, 140(1), 1–57.
http://doi.org/10.1037/a0033076

196

Frick, P., & Kimonis, ER, Dandreaux, DM, Farrell, J. (2003). The 4 year stability of
psychopathic traits in non‐referred youth. Behavioral Sciences & …, 21(6), 713–736.
Retrieved from http://onlinelibrary.wiley.com/doi/10.1002/bsl.568/full
Frick, P., & White, S. (2008). Research review: The importance of callous‐unemotional traits for
developmental models of aggressive and antisocial behavior. Journal of Child Psychology
and Psychiatry. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.14697610.2007.01862.x/full
Frye, D., Zelazo, P., & Palfai, T. (1995). Theory of mind and rule-based reasoning. Cognitive
Development. Retrieved from
http://www.sciencedirect.com/science/article/pii/0885201495900241
Gaertner, S. L. (1976). Situational determinants of hurting and helping behavior.Social
psychology: An introduction, 111-142.
Garaigordobil, M. (2009). A comparative analysis of empathy in childhood and adolescence:
Gender differences and associated socio-emotional variables. International Journal of
Psychology and Psychological …. Retrieved from
http://www.sc.ehu.es/ptwgalam/art_completo/2009/IJPPT Empatia 2009.pdf
Gilligan, C., & Attanucci, J. (1988). Two moral orientations: Gender differences and
similarities. Merrill-Palmer Quarterly (1982-), 223-237.
Goldsmith, H., Reilly, J., & Lemery, K. (1995). Preschool Laboratory Temperament Assessment
Battery. … Instrument, University of …. Retrieved from
https://scholar.google.com/scholar?q=Goldsmith%2C+Reilly%2C+Lemery%2C+Longley%
2C+%26+Prescott%2C+1995&btnG=&hl=en&as_sdt=0%2C23#0
Goldsmith, H., Reilly, J., Lemery, K., & Prescott, A. (2001) The Laboratory Temperament
Assessment Battery: Middle Childhood Version. Unpublished assessment manual.
Goldsmith, H., & Rothbart, M. (1993). The laboratory temperament assessment battery (LABTAB). University of Wisconsin. Retrieved from
https://scholar.google.com/scholar?q=goldsmith+1993+labtab&btnG=&hl=en&as_sdt=0%2C23#0
Goldsmith, H., & Rothbart, M. (1996). The laboratory temperament assessment battery.
Madison: University of Wisconsin, …. Retrieved from
http://www.waisman.wisc.edu/twinresearch/researchers/LabTAB- Middle Childhood.pdf
Hamlin, J. K., Wynn, K., & Bloom, P. (2010). Three‐month‐olds show a negativity bias in their
social evaluations. Developmental Science. Retrieved from
http://onlinelibrary.wiley.com/doi/10.1111/j.1467-7687.2010.00951.x/full

197

Hartshorne, H., & May, M. (1928). 1930. Studies in the Nature of Character. Retrieved from
https://scholar.google.com/scholar?q=Hartshorne+%26+May%2C+1928&btnG=&hl=en&a
s_sdt=0%2C23#0
Hastings, P. D., Zahn-Waxler, C., Robinson, J., Usher, B., & Bridges, D. (2000). The
development of concern for others in children with behavior problems. Developmental
Psychology, 36(5), 531–46. Retrieved from
http://www.safetylit.org/citations/index.php?fuseaction=citations.viewdetails&citationIds[]
=citjournalarticle_270616_38
Hayden, E. P., Klein, D. N., Durbin, C. E., & Olino, T. M. (2006). Positive emotionality at age 3
predicts cognitive styles in 7-year-old children.Development and Psychopathology, 18(02),
409-423.
Higbee, K. (1979). Factors affecting obedience in preschool children. The Journal of Genetic
Psychology. Retrieved from
http://www.tandfonline.com/doi/pdf/10.1080/00221325.1979.10534059
Hoffman, M. (1977). Sex differences in empathy and related behaviors. Psychological Bulletin.
Retrieved from http://psycnet.apa.org/journals/bul/84/4/712/
Hoffman, M. L. (1975). Moral internalization, parental power, and the nature of parent-child
interaction. Developmental Psychology, 11(2), 228–239. http://doi.org/10.1037/h0076463
Hoffman, M. L. (1982). Development of prosocial motivation: Empathy and guilt. In The
development of prosocial behavior (pp. 218–231).
Holland, R. (1967). Representation of dielectric, elastic, and piezoelectric losses by complex
coefficients. Sonics and Ultrasonics, IEEE Transactions on. Retrieved from
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1538402
Izard, C., Dougherty, L., & Hembree, E. (1983). A system for identifying affect expressions by
holistic judgments (AFFEX). Retrieved from
https://scholar.google.com/scholar?q=Izard%2C+Dougherty%2C+%26+Hebree%2C+1983
&btnG=&hl=en&as_sdt=0%2C23#0
Jolliffe, D., & Farrington, D. P. (2004). Empathy and offending: A systematic review and metaanalysis. Aggression and Violent Behavior. http://doi.org/10.1016/j.avb.2003.03.001
Jr, L. B., & Simpson, J. (2009). The power of the situation: The impact of Milgram’s obedience
studies on personality and social psychology. American Psychologist. Retrieved from
http://psycnet.apa.org/journals/amp/64/1/12/
Kagan, J., Reznick, J., & Snidman, N. (1987). The physiology and psychology of behavioral
inhibition in children. Child Development. Retrieved from
http://www.jstor.org/stable/1130685
198

Kagan, J., Snidman, N., Arcus, D., & Reznick, J. (1994). Galen’s prophecy: Temperament in
human nature. Retrieved from http://psycnet.apa.org/psycinfo/1994-97715-000
Keenan, K., & Shaw, D. S. (1994). The development of aggression in toddlers: A study of lowincome families. Journal of abnormal child psychology, 22(1), 53-77.
Kimonis, E. R., Frick, P. J., Fazekas, H., & Loney, B. R. (2006). Psychopathy, aggression, and
the processing of emotional stimuli in non-referred girls and boys. Behavioral Sciences &
the Law, 24(1), 21–37. http://doi.org/10.1002/bsl.668
Kimonis, E. R., Frick, P. J., Skeem, J. L., Marsee, M. A., Cruise, K., Munoz, L. C., ... & Morris,
A. S. (2008). Assessing callous–unemotional traits in adolescent offenders: Validation of
the Inventory of Callous–Unemotional Traits. International journal of law and
psychiatry, 31(3), 241-252.
Kochanska, G. (1991). Socialization and Temperament in the Development of Guilt and
Conscience. Child Development, 62(6), 1379–1392. http://doi.org/10.1111/j.14678624.1991.tb01612.x
Kochanska, G. (1993). Toward a synthesis of parental socialization and child temperament in
early development of conscience. Child Development. Retrieved from
http://onlinelibrary.wiley.com/doi/10.1111/j.1467-8624.1993.tb02913.x/abstract
Kochanska, G. (1993). Toward a Synthesis of Parental Socialization and Child Temperament in
Early Development of Conscience. Child Development, 64(2), 325–347.
http://doi.org/10.1111/j.1467-8624.1993.tb02913.x
Kochanska, G. (2002). Mutually Responsive Orientation Between Mothers and Their Young
Children: A Context for the Early Development of Conscience. Current Directions in
Psychological Science, 11(6), 191–195. http://doi.org/10.1111/1467-8721.00198
Kochanska, G., & Aksan, N. (2004). Conscience in Childhood: Past, Present, and Future.
Merrill-Palmer Quarterly, 50(3), 299–310. http://doi.org/10.1353/mpq.2004.0020
Kochanska, G., & Aksan, N. (2006). Children’s conscience and self-regulation. Journal of
Personality, 74(6), 1587–617. http://doi.org/10.1111/j.1467-6494.2006.00421.x
Kochanska, G., Aksan, N., & Koenig, A. (1995). A longitudinal study of the roots of
preschoolers’ conscience: Committed compliance and emerging internalization. Child
Development. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.14678624.1995.tb00963.x/abstract
Kochanska, G., Aksan, N., & Nichols, K. (2003). Maternal power assertion in discipline and
moral discourse contexts: commonalities, differences, and implications for children’s moral
conduct and cognition. Developmental Psychology. Retrieved from
http://psycnet.apa.org/journals/dev/39/6/949/
199

Kochanska, G., Coy, K. C., & Murray, K. T. (2001). The Development of Self-Regulation in the
First Four Years of Life. Child Development, 72(4), 1091–1111.
http://doi.org/10.1111/1467-8624.00336
Kochanska, G., DeVet, K., Goldman, M., Murray, K., & Putnam, S. P. (1994). Maternal Reports
of Conscience Development and Temperament in Young Children. Child Development,
65(3), 852–868. http://doi.org/10.1111/j.1467-8624.1994.tb00788.x
Kochanska, G., & Forman, D. (2005). Pathways to conscience: Early mother–child mutually
responsive orientation and children’s moral emotion, conduct, and cognition. Journal of
Child …. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.14697610.2004.00348.x/full
Kochanska, G., Gross, J. N., Lin, M.-H., & Nichols, K. E. (2002). Guilt in Young Children:
Development, Determinants, and Relations with a Broader System of Standards. Child
Development, 73(2), 461–482. http://doi.org/10.1111/1467-8624.00418
Kochanska, G., & Knaack, A. (2003). Effortful control as a personality characteristic of young
children: Antecedents, correlates, and consequences. Journal of Personality. Retrieved from
http://onlinelibrary.wiley.com/doi/10.1111/1467-6494.7106008/full
Kochanska, G., Murray, K., & Coy, K. (1997). Inhibitory control as a contributor to conscience
in childhood: From toddler to early school age. Child Development. Retrieved from
http://onlinelibrary.wiley.com/doi/10.1111/j.1467-8624.1997.tb01939.x/full
Kochanska, G., Murray, K., Jacques, T. Y., Koenig, A. L., & Vandegeest, K. A. (1996).
Inhibitory control in young children and its role in emerging internalization. Child
development, 67(2), 490-507.
Kochanska, G., Padavich, D. L., & Koenig, A. L. (1996). Children’s Narratives about
Hypothetical Moral Dilemmas and Objective Measures of Their Conscience: Mutual
Relations and Socialization Antecedents. Child Development, 67(4), 1420–1436.
http://doi.org/10.1111/j.1467-8624.1996.tb01805.x
Kochanska, G., Tjebkes, J., & Fortnan, D. (1998). Children’s emerging regulation of conduct:
Restraint, compliance, and internalization from infancy to the second year. Child
Development. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.14678624.1998.tb06218.x/abstract
Kohlberg, L., & Hersh, R. (1977). Moral development: A review of the theory. Theory into
Practice. Retrieved from
http://www.tandfonline.com/doi/pdf/10.1080/00405847709542675
Kohlberg, L., & Kramer, R. (1969). Continuities and discontinuities in childhood and adult
moral development. Human Development. Retrieved from
http://www.karger.com/Article/Abstract/270857
200

Kolko, D. J., & Kazdin, A. E. (1993). Emotional/behavioral problems in clinic and nonclinic
children: Correspondence among child, parent and teacher reports. Journal of Child
Psychology and Psychiatry, 34(6), 991-1006.
Krahn, J. (1971). A comparison of Kohlberg’s and Piaget's type I morality. Religious Education.
Retrieved from http://www.tandfonline.com/doi/pdf/10.1080/0034408710660509
Krettenauer, T., Asendorpf, J. B., & Nunner-Winkler, G. (2013). Moral emotion attributions and
personality traits as long-term predictors of antisocial conduct in early adulthood: Findings
from a 20-year longitudinal study. International Journal of Behavioral Development, 37,
192–201. http://doi.org/10.1177/0165025412472409
Kuczynski, L. (1987). A developmental interpretation of young children’s noncompliance.
Developmental …. Retrieved from http://psycnet.apa.org/journals/dev/23/6/799/
Kuczynski, L., & Kochanska, G. (1990). Development of children’s noncompliance strategies
from toddlerhood to age 5. Developmental Psychology. Retrieved from
http://psycnet.apa.org/journals/dev/26/3/398/
Kuczynski, L., Kochanska, G., Radke-Yarrow, M., & Girnius-Brown, O. (1987). A
developmental interpretation of young children's noncompliance.Developmental
Psychology, 23(6), 799.
Landauer, T., Carlsmith, J., & Lepper, M. (1970). Experimental analysis of the factors
determining obedience of four-year-old children to adult females. Child Development.
Retrieved from http://www.jstor.org/stable/1127210
Laupa, M. (1991). Children’s reasoning about three authority attributes: Adult status, knowledge,
and social position. Developmental Psychology. Retrieved from
http://psycnet.apa.org/journals/dev/27/2/321/
Laupa, M. (1994). “Who’s in charge?” Preschool children’s concepts of authority. Early
Childhood Research Quarterly. Retrieved from
http://www.sciencedirect.com/science/article/pii/0885200694900264
Laupa, M., & Turiel, E. (1986). Children’s conceptions of adult and peer authority. Child
Development. Retrieved from http://www.jstor.org/stable/1130596
Lickenbrock, D. (2013). Early temperament and attachment security with mothers and fathers as
predictors of toddler compliance and noncompliance. Infant and Child …. Retrieved from
http://onlinelibrary.wiley.com/doi/10.1002/icd.1808/full
Loney, B. R., Frick, P. J., Clements, C. B., Ellis, M. L., & Kerlin, K. (2003). Callousunemotional traits, impulsivity, and emotional processing in adolescents with antisocial
behavior problems. Journal of Clinical Child and Adolescent Psychology : The Official
Journal for the Society of Clinical Child and Adolescent Psychology, American
201

Psychological Association, Division 53, 32(1), 66–80.
http://doi.org/10.1207/S15374424JCCP3201_07
Lovett, B. J., & Sheffield, R. A. (2007). Affective empathy deficits in aggressive children and
adolescents: a critical review. Clinical Psychology Review, 27(1), 1–13.
http://doi.org/10.1016/j.cpr.2006.03.003
Eisenberg, N., Lundy, T., Shell, R., & Roth, K. (1985). Children's justifications for their adult
and peer-directed compliant (prosocial and nonprosocial) behaviors. Developmental
Psychology, 21(2), 325.
Marsh, A. A., & Cardinale, E. M. (2014). When psychopathy impairs moral judgments: neural
responses during judgments about causing fear. Social Cognitive and Affective
Neuroscience, 9(1), 3–11. http://doi.org/10.1093/scan/nss097
Milgram, S. (1961). Nationality and conformity. Scientific American. Retrieved from
http://psycnet.apa.org/psycinfo/1962-06297-001
Milgram, S. (1963). Behavioral study of obedience. The Journal of Abnormal and Social
Psychology. Retrieved from http://psycnet.apa.org/journals/abn/67/4/371/
Milgram, S. (1965). Some conditions of obedience and disobedience to authority. Human
Relations. Retrieved from http://psyc604.stasson.org/Milgram2.pdf
Milgram, S. (1974). The experience of living in cities. Crowding and Behavior. Retrieved from
https://books.google.com/books?hl=en&lr=&id=XA8Ho_aITVoC&oi=fnd&pg=PA41&dq=
milgram+1974&ots=JwPHwRr9jF&sig=rzyyuqi7vGmYPvZDAXtrMqxRuD4
Mixon, D. (1976). Studying feignable behavior. Representative Research in Social Psychology.
Retrieved from http://psycnet.apa.org/psycinfo/1977-29451-001
Muñoz, L. C. (2009). Callous-unemotional traits are related to combined deficits in recognizing
afraid faces and body poses. Journal of the American Academy of Child and Adolescent
Psychiatry, 48(5), 554–62. http://doi.org/10.1097/CHI.0b013e31819c2419
Piaget, J. (1932). The moral development of the child. Kegan Paul, London. Retrieved from
https://scholar.google.com/scholar?q=piaget+1932&btnG=&hl=en&as_sdt=0%2C23#1
Piaget, J. (1965). The moral judgment of the child (1932). New York: The Free. Retrieved from
http://scholar.google.com/scholar?q=piaget+1965+moral+judgement+of+the+child&btnG=
&hl=en&as_sdt=0,23#4
Pons, F., Harris, P. L., & de Rosnay, M. (2004). Emotion comprehension between 3 and 11
years: Developmental periods and hierarchical organization. European Journal of
Developmental Psychology, 1(2), 127–152. http://doi.org/10.1080/17405620344000022

202

Rose, S. R. (1999). Towards the Development of an Internalized Conscience. Journal of Human
Behavior in the Social Environment, 2(3), 15–28. http://doi.org/10.1300/J137v02n03_02
Rose, S. L., Rose, S. A., & Feldman, J. F. (1989). Stability of behavior problems in very young
children. Development and Psychopathology, 1(01), 5-19.
Rose, A. J., & Rudolph, K. D. (2006). A review of sex differences in peer relationship processes:
potential trade-offs for the emotional and behavioral development of girls and
boys. Psychological bulletin, 132(1), 98.
Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution
process. Advances in experimental social psychology, 10, 173-220.
Rothbart, M., Ahadi, S., & Hershey, K. (1994). Temperament and social behavior in childhood.
Merrill-Palmer Quarterly (1982-). Retrieved from http://www.jstor.org/stable/23087906
Rothbart, M., & Bates, J. (1998). Temperament. Hoboken. Retrieved from
https://scholar.google.com/scholar?q=Rothbart+%26+Bates%2C+1998&hl=en&as_sdt=0%
2C23&as_ylo=1998&as_yhi=1998#0
Rothbart, M. K., Ahadi, S. A., Hershey, K. L., & Fisher, P. (2001). Investigations of
Temperament at Three to Seven Years: The Children’s Behavior Questionnaire. Child
Development, 72(5), 1394–1408. http://doi.org/10.1111/1467-8624.00355
Roth-Hanania, R., Davidov, M., & Zahn-Waxler, C. (2011). Empathy development from 8 to 16
months: early signs of concern for others. Infant Behavior & Development, 34(3), 447–58.
http://doi.org/10.1016/j.infbeh.2011.04.007
Sagi, A., & Hoffman, M. (1976). Empathic distress in the newborn. Developmental Psychology,
12(2), 175–176. http://doi.org/10.1037/0012-1649.12.2.175
Shalala, S. R. (1975). A study of various communication settings which produce obedience by
subordinates to unlawful superior orders (Doctoral dissertation, ProQuest Information &
Learning).
Shanab, M., & Yahya, K. (1977). A behavioral study of obedience in children. Journal of
Personality and Social …. Retrieved from http://psycnet.apa.org/journals/psp/35/7/530/
Shelton, K. K., Frick, P. J., & Wootton, J. (1996). Assessment of parenting practices in families
of elementary school-age children. Journal of clinical child psychology, 25(3), 317-329.
Shimizu, Y. A., & Johnson, S. C. (2004). Infants’ attribution of a goal to a morphologically
unfamiliar agent. Developmental Science, 7(4), 425–430. http://doi.org/10.1111/j.14677687.2004.00362.x

203

Snyder, M., & Ickes, W. (1985). Personality and social behavior. Handbook of Social
Psychology. Retrieved from
https://scholar.google.com/scholar?q=Snyder+%26+Ickes%2C+1985&btnG=&hl=en&as_s
dt=0%2C23#0
Stanger, C., & Lewis, M. (1993). Agreement among parents, teachers, and children on
internalizing and externalizing behavior problems. Journal of Clinical Child
Psychology, 22(1), 107-116.
Stifter, C. A., Cipriano, E., Conway, A., & Kelleher, R. (2009). Temperament and the
Development of Conscience: The Moderating Role of Effortful Control. Social
Development (Oxford, England), 18(2), 353–374. http://doi.org/10.1111/j.14679507.2008.00491.x
Strommen, E. (1973). Verbal Self-Regulation in a Children’s Game: Impulsive Errors on“ Simon
Says.” Child Development. Retrieved from http://www.jstor.org/stable/1127737
Tan, P., Steinbach, M., & Kumar, V. (2006) Cluster Analysis: Basic Concepts and Algorithms.
Retrieved from http://www-users.cs.umn.edu/~kumar/dmbook/ch8.pdf
Tibshirani, R., Walther, G., & Hastie, T. (2001). Estimating the number of clusters in a data set
via the gap statistic. Journal of the Royal Statistical Society: Series B (Statistical
Methodology), 63(2), 411-423.
Vroman, L., Lo, S., & Durbin, C. (2014). Structure and convergent validity of children’s
temperament traits as assessed by experimenter ratings of child behavior. Journal of
Research in Personality. Retrieved from
http://www.sciencedirect.com/science/article/pii/S0092656614000531
Willoughby, M. T., Mills-Koonce, W. R., Gottfredson, N. C., & Wagner, N. (2014). Measuring
Callous Unemotional Behaviors in Early Childhood: Factor Structure and the Prediction of
Stable Aggression in Middle Childhood. Journal of Psychopathology and Behavioral
Assessment, 36(1), 30–42. http://doi.org/10.1007/s10862-013-9379-9
Willoughby, M. T., Waschbusch, D. A., Moore, G. A., & Propper, C. B. (2011). Using the
ASEBA to Screen for Callous Unemotional Traits in Early Childhood: Factor Structure,
Temporal Stability, and Utility. Journal of Psychopathology and Behavioral Assessment,
33(1), 19–30. http://doi.org/10.1007/s10862-010-9195-4
Wilson, S., & Durbin, C.E. (2012). Dyadic parent-child interaction during early childhood:
Contributions of parental and child personality traits. Journal of Personality, 80(5), 13131338.
Zahn-Waxler, C., Radke-Yarrow, M., Wagner, E., & Chapman, M. (1992). Development of
concern for others. Developmental Psychology, 28(1), 126–136. Retrieved from
http://cat.inist.fr/?aModele=afficheN&cpsidt=5053281
204

Zelazo, P., Müller, U., & Frye, D. (2003). The development of executive function in early
childhood. Monographs of the …. Retrieved from http://www.jstor.org/stable/1166202
Zimbardo, P. G. (1974). On" Obedience to authority.".

205