TEMPERAMENT IN CHINESE CHILDREN:
A COMPARISON OF GENDER AND SELF/PARENTAL RATINGS
A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE
UNIVERSITY OF HAWAI'IIN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
IN
PSYCHOLOGY
DECEMBER 2004
BY
Chuan Chang
Dissertation Committee:
Anthony J. Marsella, Chairperson
Elaine Hatfield
Clifford O'Donnell
Ashley E. Maynard
Majid Tehranian
ACKNOWLEDGEMENTS
I would like to thank my academic advisor, Professor Emeritus Anthony J.
Marsella, for his guidance, wisdom, and patience throughout my work on this
study. Most of all, I would like to thank him for his unwavering support during the
toughest time of my study in the US. Without him, I would not have come this far
in my academic pursuits and personal growth.
I am also grateful to the principals and teachers at the elementary schools
I visited in Beijing, including Yihai Elementary, Zuo Jia Zhuang Elementary, and
Bai Jia Zhuang Elementary and its affiliated south site.
They, along with my
former colleagues-professors Meiling Zhang and Gang Zheng at the Institute of
Psychology, Chinese Academy of Sciences, and chief editor Dagang Song of the
China Children's Press and Publication Group in Beijing, China-made it
possible for me to collect a huge amount of data from both students and their
parents in a short period of time.
During the course of working on my
dissertation, I have been blessed with many good friends who kept me motivated
and focused. I especially want to thank Sally Nhomi, who had the patience to
read the first drafts of my chapters and gave me helpful feedback.
Finally, lowe greatly to my family, who has supported me from the very
beginning.
I am especially indebted to my mother for bringing me up and
believing in me. Though my father passed away before I left for America, his
spirit has always been a part of what I do and he would have been prouder than
anyone else about my accomplishment in completing this doctoral dissertation.
111
ABSTRACT
The present study explores the temperamental characteristics of Chinese
youth, ages nine to twelve, in Beijing, People's Republic of China (PRC), using
the revised Early Adolescent Temperament Questionnaire (EATQ-R, Ellis &
Rothbart, 2001). An exploratory factor analysis was conducted on the complete
responses from 687 children and 428 mothers. The factorial structure based on
Chinese children's self-reports included seven first-order and three second-order
factors, differing from Ellis and Rothbart's structure (2001).
Significant gender
differences were found in both the children's and mothers' reports. An analysis
of reports by mothers indicated an interaction between gender and only-child
status
affecting
the
temperamental
dimension
of
shyness.
Mothers
acknowledged the importance of temperament in the formation of their respective
children's personalities, but rated it less important than non-individualistic factors
such as parental discipline and school education.
IV
TABLE OF CONTENTS
Acknowledgements
,
Abstract
iii
iv
List of Tables
viii
List of Figures
x
Chapter 1: Introduction
1
History of Temperament Studies
4
Definitions of Temperament
9
Thomas and Chess and NYLS Model
9
Buss and Plomin and EAS
11
Rothbart and Developmental Theory of Temperament
13
Goldsmith and Campos Emotion-Oriented Theory
15
Kagan's Psychobiological Theory
16
Measurement of Temperament
18
Cross-Cultural Considerations
20
Cross-Cultural Research of Temperament..
23
Current Status
23
Chinese Children and Temperament: Substantial Findings
27
Gender Differences in Temperament
Gender and Cultural Studies
37
40
Objectives of Current Study
43
Chapter 2: Method
45
Participants
45
Measures
46
EATQ-R
46
Family Information Questionnaire
47
Parent Attitude Questionnaire
48
48
Procedures
v
Chapter 3: Results
52
EATQ-R Self-Report
52
Descriptive Statistics of the EATQ-R Items
52
Descriptive Statistics and Reliability of EATQ-R Self-Report Scales
56
EFA on Chinese Children's Self-Reports
59
First-Order EFA
59
Second-Order EFA
66
Multivariate Analysis
68
Summary of Results from Children's Self-Reports
70
EATQ-R Parent-Report
71
Demographic Information
71
Descriptive Statistics of the EATQ-R Items
72
Descriptive Statistics and Reliability of EATQ-R Parent-Report Scales
75
EFA on Chinese Mothers' Reports
76
First-Order EFA
76
Second-Order EFA
82
Multivariate Analysis
84
Interaction between Gender and Only-Child Status
85
Gender Differences
86
Differences between Only Children and Those with Siblings
86
Parental Attitudes on Temperament
86
Summary of Results from Mothers' Reports
88
Comparison of Self- and Mother-Reports
Chapter 4: Discussion
89
91
Validity Limitations of the EATQ-R in PRC Sample
General Patterns of Temperamental Characteristics in PRC Sample
91
102
Gender and Only-Child Status Variations
107
Agreement between Chinese Children and Mothers
109
VI
Chapter 5: Conclusion and Future Research
112
Appendix A: Early Adolescents Temperament Questionnaire-Revised
116
Appendix B: Early Adolescents Temperament Questionnaire-Revised
Parent Report Form
121
References
125
VB
LIST OF TABLES
1.
Ten Highest Means of EATQ-R Scores among Chinese Children
53
2.
Ten Lowest Means of EATQ-R Scores among Chinese Children
54
3.
Ten EATQ-R Self-Report Items with the Largest Standard Deviations .... 55
4.
Descriptive Statistics of EATQ-R Self-Report Scales
56
5.
Coefficient Alpha for EATQ-R Self-Report Scales by
PRC and US Samples
57
Factor Pattern Loadings (Standardized Regression Coefficients) of
EATQ-R Self-Report Items in the PRC Children's Sample
60
Items Loading on Factors 1 to 7 According to the
Rotated Factor Pattern Matrix in the PRC Children's Sample
62
Factor Correlations among the New Chinese EATQ-R
Self-Report Scales
64
Descriptive Statistics and Coefficient Alpha of the New Chinese
EATQ-R Self-Report Scales Overall and by Gender
65
Parameter Estimates of the Second-Order
Three-Correlated Factor Model in the PRC Children's Sample
66
11.
Demographic Information of the PRC Mothers' Sample
71
12.
Ten Highest Means of EATQ-R Scores among Chinese Mothers
72
13.
Ten Lowest Means of EATQ-R Scores among Chinese Mothers
73
14.
Ten EATQ-R Parent-Report Items with the
Largest Standard Deviations
74
Descriptive Statistics and Coefficient Alpha of EATQ-R
Parent-Report Scales
75
Factor Pattern Loadings (Standardized Regression Coefficients) of
EATQ-R Parent-Report Items in the PRC Mothers' Sample
77
6.
7.
8.
9.
10.
15.
16.
Vlll
17.
18.
19.
20.
21.
Items Loading on Factors 1 to 6 According to the
Rotated Factor Pattern Matrix in the PRC Mothers' Sample
79
Factor Correlations among the New Chinese
EATQ-R Parent-Report Scales
81
Descriptive Statistics and Coefficient Alpha of the New EATQ-R
Parent-Report Scales Overall and by Children's Gender.
82
Parameter Estimates of the Second-Order Three-Correlated
Factor Model in the PRC Mothers' Sample
83
Descriptive Statistics on Shyness for the 4 Groups, Categorized by
Gender and Only-Child Status
86
22.
Chinese Mothers' Views of Factors Shaping Children's Personalities ..... 87
23.
Convergence between Chinese Children's and Mothers on Shared
Temperamental Dimensions
IX
89
LIST OF FIGURES
Figure
Page
1.
The First-Order and Second-Order EATQ-R Factor Structure in the PRC
Children's Sample
68
2.
The First-Order and Second-Order EATQ-R Factor Structure in the PRC
Mothers' Sample
84
3.
Comparison of the EATQ-R Factor Structures in the US and PRC,
Based on Self-Reports
x
102
CHAPTER 1
INTRODUCTION
Over the last two decades, there has been a dramatic increase in the
study of temperament during infancy and childhood. This renaissance is due to
an increasing interest in the biological bases of human individuality and in the
functional significance of individual variation in human development across life
(Talwar, Nitz, Lerner, & Lerner, 1991). Temperament is commonly viewed as
early appearing, relatively stable, biologically based individual differences in
behavior tendencies (Bates,
1989).
Temperamental
characteristics are
considered the raw materials of individual variability in later personality (Rothbart
& Ahadi, 1994).
The importance of temperament lies in its impact on person-social context
relations.
Far from being a tabula rasa, infants come into this world with
distinctive temperamental characteristics and are capable of engaging in a full
range of activities from the earliest moments of birth. The interactions between a
child's temperament and his or her social context are dynamic, as both the child
and its surrounding environment have to adjust and adapt to each other. While a
child's temperamental characteristics can moderate the context through "niche
picking," the surrounding world also places demands on the growing child. Thus,
temperamental individuality gains significance for its adaptive functioning through
its link with social context (Talwar, Nitz, Lerner, & Lerner, 1991).
1
Among the major issues in contemporary temperament research are the
definition and measurement of temperament, biological bases of temperament,
patterns of temperamental stability and change, and functional significance of
temperament.
Despite the large body of evidence accumulated over the last
several decades in the field, many questions still remain. As a matter of fact,
diverse views abound in the field of temperament. At present, temperament is
considered to follow a developmental course, both in terms of its expressions
and the number of temperamental characteristics (Rothbart & Bates, 1998).
Since the stability of temperamental characteristics during the infants and
children stage is usually low (the correlation is in the range of r
= 0.30 - 0.40),
more recent studies have begun to look at the relations between temperament
and contextual factors (Wachs & Kohnstamm, 2001).
Both biological (e.g.,
physical ecology, diet) and psychosocial conditions (e.g., the family, culture)
have been related to the development of temperament. Studies of infants and
children from different cultures have revealed cross-cultural variation in patterns
of temperament,
and
have
suggested
that
culture
may
interact with
temperamental characteristics to modify their expressions and affect individual
developmental outcomes.
The present investigation is a study of Chinese adolescents' temperament
in the People's Republic of China (PRC), with a particular
emphasi~
on (a) the
pattern of temperamental characteristics in Chinese adolescents, (b) gender
variations in temperamental characteristics, and (c) the agreement between
parental reports of their children's temperament and self-reports by the
2
adolescents themselves. In addition, this study offers an opportunity to further
examine the psychometric properties of a temperamental instrument in a PRC
sample of children and parents,
The use of a PRC sample is a strength of the proposed research. Firstly,
the Chinese adolescents sample comprises of "only" children due to the
governmental one-child-per-family policy implemented nationwide in PRC since
1979. The researcher is particularly interested in the gender-related variations in
temperamental characteristics of the "only" children, as parental perceptions,
beliefs, and expectations for the male and female children may undergo changes
because of the implementation of the policy. Secondly, Chinese adolescents are
under tremendous pressure from families and peers to succeed academically.
Children are typically expected to concentrate on their school performance, and
the goal of parenting is to make sure children study hard, do well in school, and
obey the rules (Stevenson, Lee, Chen, Stigler, Hsu, & Kitamura, 1990). Thus,
Chinese parents may tend to socialize those temperamental characteristics that
correlate with school performance. Thirdly, Chinese culture traditionally values
the concept of "training" or "educating," and this is reflected in parental beliefs
and value systems, as well as in child-rearing practices (Chao, 1994). Typically,
parents emphasize the importance of effort in motivating their children for
academic achievement (Ho, 1986; Stevenson, et aI., 1990). Such a deep-rooted
belief in "training" may diminish the significance of temperamental individuality in
Chinese culture.
Traditionally, the idea of individual variation is prevalent in
Chinese culture, especially in Chinese medicine, but it is mainly used to explain
3
variation in the nature of physical body/health and vulnerability to certain illness.
Fourthly, the relationship between Chinese adolescents and their parents is
viewed closer than that in the US. because Chinese culture values and
encourages a strong parent-children relationship even when children become
grown and mature adults.
What follows is a brief review oLthe history of temperament research, a
discussion of definitional and measurement issues in the field of temperament
research, and a review of cross-cultural studies on temperament.
History of Temperament Studies
Historically, the concept of temperament dates back 2000 years to the
humoral theory of ancient Greek and Roman physicians. The Greek physician
Hippocrates (about 400 B.C.) attributed individual differences in behavior to the
imbalance of four bodily fluids or humors - black bile, yellow bile, blood, and
phlegm. The Roman physician Galen further refined the theory and proposed
four well-known temperamental types: melancholic, choleric, sanguine, and
phlegmatic. Each was thought to be the result of a predominance of one of the
bodily humors. Without a detailed understanding of genetics or physiology, the
ancient humoral theory of temperament assumed an association between an
inner balance of bodily fluids and the observed variations in behavior and
emotionality.
A similar belief also existed in ancient Chinese civilization. For example,
the ancient Chinese believed that human nature was determined by a balance
4
between the forces of "Yin" and "Yang" ch'i, and this inner system of ch'j was
subject to the influences of the five fundamental elements of the nature-water,
fire, wood, earth, and air (Kagan, 1994).
But according to Kagan (1994), the
concept of temperament as consisting of permanent mood and behavioral styles
is unique to the Western culture.
The Ancient Chinese idea of yin and yang
balance is used more generally, from explaining the operation of the universe,
seasons, elements of nature to the bodily system, health, and human emotion.
All in all, the idea that individual characteristics in behavior and emotion are
rooted within the human body greatly shaped the current understanding of
temperament.
Contemporary research continues to search for the biological
bases of temperament, specifically the individual differences in aspects of central
nervous system structures and neurochemistry that may effect temperament.
At the beginning of the 20 th century, a typology of temperament based on
constitutional types was popular.
Kretschmer believed that particular human
qualities correlated with the physical makeup through the mediation of the
humoral system, and thus suggested an association between mental disorders
and body types. Sheldon further developed the classification of the possible link
between personality and body type, and proposed three physique types and their
corresponding temperaments in his famous book "Personality and Physique"
(1940, as cited in Kagan, 1998). Presumably, the tall, thin ectomorph was an
introvert, the chubby endomorph an extrovert, and the athletically built
mesomorph was assertive.
This constitutional typology of temperament didn't
grow into a major research movement because the idea was too close to Hitler's
5
Aryan types and it encouraged the development of racist attitudes.
psychodynamic
theory
still
dominant
and
behaviorism
rising
With
quickly,
temperament research went into decline until its revival in the 1960s and early
1970s.
The study of temperament in childhood was influenced by another line of
research in the 1920s and 1930s-normative studies that intended to establish
the normal sequences of motor and mental development in children.
Even
though temperament was not the focus of study in these studies, researchers
such
as
Gesell
and
Shirley
noticed
striking
individual
differences
in
temperamental characteristics among the children they studied. Gesell observed
that "certain fundamental traits of individuality, whatever their origin, exist early,
persist late and assert themselves under varying environmental conditions." He
also noted the critical importance of environment in temperament development,
"environment retains a critical role even though heredity sets metes and bounds."
(1928, as cited in Rothbart & Bates, 1998, p.106). Temperament research since
then has sought to examine, quantify, and describe these differences.
The modern history of temperament research began with the revolutionary
New York Longitudinal Study (NYLS) conducted by two pediatricians - Alexander
Thomas and Stella Chess - and their colleagues in 1956 (reported in Buss &
Plomin, 1984). Thomas and Chess initially interviewed parents of 138 infants
aged 2 to 6 months, and followed 132 individuals from early infancy into early
adulthood (Thomas & Chess, 1977). The original purpose of the project was to
study the role of temperament in predicting future psychiatric problems of
6
children, but it gradually shifted to exploring the role of interactions between a
child's
individual
characteristics
and
his/her environment (such
as the
expectations of others, demands and opportunities of environment) in realizing
various developmental outcomes. Key findings were reported at different stages
of the project (Chess & Thomas, 1984; Thomas, Chess, Birch, Hertzig, & Korn,
1963; Thomas, Chess, & Birch, 1968; Thomas & Chess, 1977).
In their first volume on the NYLS, Thomas et al. (1963) proposed nine
temperamental dimensions together with three temperamental constellations
from their rich interview data with parents. The nine dimensions were (a) activity
level; (b) rhythmicity or regularity of biologic functions; (c) approach / withdrawal,
referring to positive or negative responses to a new situation, person, or
environmental demand; (d) ease or difficulty of adaptability to the requirement for
change in an established behavior pattern; (e) threshold of responsiveness; (f)
quality of mood, rated as the preponderance of positive versus negative mood
expression; (g) intensity of mood expression, regardless whether it is positive or
negative; (h) ease or difficulty of distractibility from an ongoing activity by an
extraneous stimulus; and (i) length of attention span and degree of persistence
with a difficult task (this is a double category).
The three temperamental constellations were identified based on the
overall pattern of temperamental traits on the nine dimensions, and they were
easy child, difficult child, and slow-to-warm-up child. The easy child is identified
by a positive mood, biological regularity, high approach, low withdrawal,and high
adaptability. The difficult child is identified by frequent negative mood, biological
7
irregularity, low approach, many withdrawal responses, and poor adaptability.
The slow-to-warm-up child exhibits a negative mood, but at a lower intensity than
the difficult child. These children are slower to adjust to new situations, but with
repeated exposure can adapt. They are often so-called shy or inhibited children.
Standardized questionnaire instruments were soon developed using the
NYLS model to assess the nine temperamental dimensions and the easy,
difficult, and slow-to-warm-up temperaments.
These instruments have been
widely used in research, clinical, and school settings.
Another important
contribution of the NYLS is its interactional approach, which is best embodied by
the formulation of the goodness-of-fit concept (Chess & Thomas, 1991). This
formulation stems from the observation that normal or pathological development
does not depend on temperament alone.
According to Chess and Thomas,
goodness-of-fit (or poorness-of-fit) refers to the match (or mismatch) between the
child's temperament and the opportunities, demands and expectations of the
social environment of the child. Goodness-of-fit emphasizes the multiple, varied,
and unpredictable types of interactional patterns that may lead to normal or
pathological development.
Much research on temperament during recent years has been based on
the NYLS model of Thomas, Chess, and their colleagues.
Outside the U.S.,
Pavlov's typology of nervous systems and theories proposed by Hans Eysenck
and Jeffery Gray have also shaped modern views on temperament, especially
the link between physiological processes and behavioral features (Strelau, 1998).
8
Definitions of Temperament
Since the seminal NYLS project by Thomas, Chess and their colleagues,
several research groups have developed definitions of temperament.
In a
roundtable format, theorists of four major perspectives on temperament
presented their divergent viewpoints on temperament's definition and dimensions
(Goldsmith, Buss, Plomin, Rothbart, Thomas, Chess, Hinde, & McCall, 1987).
Many of these temperament researchers
temperament
that
corresponded
with
also developed
their
definitions.
measures of
Five
major
conceptualizations will be reviewed, and they include the NYLS model of Thomas
and Chess, the EAS theory of Buss and Plomin, the developmental model of
temperament by Rothbart and Derryberry, the emotion-oriented theory of
Goldsmith and Campos, and Kagan's psychobiological theory of temperament.
Thomas and Chess and NYSL Model
Since Thomas and Chess's NYLS project has been summarized earlier,
the emphasis here will be on their definition of temperament, goodness-of-fit
concept, and instruments that assess the NYLS dimensions. Thomas and Chess
used reports from parents to develop their conceptualizations of temperament, a
purely descriptive approach as they claimed (Thomas & Chess,
1977).
Temperament was defined as the behavioral style or the "how" of behavior, as
contrasted with the abilities or "what" of behavior, and the motivations or "why" of
behavior.
Based on this definition, Thomas and Chess operationalized
temperament as the nine-dimension construct described earlier.
9
Based on Thomas and Chess's theory, Carey, McDevitt and their
colleagues developed five instruments for use with children from infancy through
age 11. They include the Early Infancy Temperament Questionnaire for 1- to 4month-old infants (EITQ; Medoff-Coper, Carey, & McDevitt, 1993, as cited in
Strelau, 1998), Revised Infant Temperament Questionnaire for 4- to 8-month-old
infants (RITQ; Carey & McDevitt, 1978), Toddler Temperament Scale for children
ages 1 to 3 (TTS; Fullard, McDevitt, & Carey, 1984, as cited in Strelau, 1998),
Behavior Style Questionnaire for children ages 3 to 7 (BSQ; McDevitt & Carey,
1978; as cited in Strelau,
1998), and Middle Childhood Temperament
Questionnaire for 8 to 11-year-old children (MCTQ, Hagvik, McDevitt, & Carey,
1982; as cited in Strelau, 1998). Among the five instruments, the RITQ is widely
used, especially in cross-cultural situations. The main purpose of these studies
is to investigate whether infants from different nations demonstrate different
levels of easiness / difficulty as compared to the US norms.
Despite the
widespread use of their instruments in temperament research, Thomas and
Chess's nine temperamental dimensions were found to be difficult to replicate
and were viewed less psychometrically meaningful and independent (Bates,
1987).
Another important component of Thomas and Chess's theory of
temperament is the concept "goodness (or poorness) of fit." The concept refers
to the match or mismatch between an individual's temperamental characteristics
and expectations, demands, and opportunities of the individual's environment. If
a person's characteristics of individuality match the demands of a particular
10
social context, then positive interactions and adjustment are expected.
In
contrast, negative adjustment is expected to occur when there is a poorness of fit
between the two.
In other words, a temperamental trait becomes a factor in
pathological behavior, not by itself alone, but when it is combined with a
poorness-of-fit. As Chess and Thomas (1991) pointed out, the development of
goodness-of-fit is subject to influences of the socioeconomic and cultural factors,
because the contextual demands will vary from culture to culture, and from one
socioeconomic class to another. The goodness of fit model has become a useful
framework for the analysis of cultural variations in temperament development.
Buss and Plomin and EAS
Buss and Plomin (1984), from personality psychology perspective, defined
temperament as inherited personality traits that appear early in life. In this view,
temperament is a subclass of personality traits that meet two critical criteria (a)
inheritability and (b) appearance within the first year of life.
Buss and Plomin
identified three personality traits that meet these criteria, and they are
emotionality, activity, and sociability (1984), from which the acronym EAS theory
are derived.
Impulsivity was dropped from the original temperament list
proposed by Buss and Plomin due to the lack of strong evidence from behavioral
genetic research on this trait.
Emotionality is defined as primordial distress, which is assumed to
differentiate into fear (which emerges at 3 months of age) and then anger (at six
months of age). The authors didn't include positive emotions in their emotionality
temperament because (a) the level of autonomic arousal typical for positive
11
emotions is below that of distress, fear, and anger emotions; and (b) there is no
sufficient evidence for the inheritance of positive emotions.
Activity is the
expenditure of physical energy, and best measured by the tempo and vigor of
physical movements. Sociability is the preference for being with others rather
than being alone.
The combination of sociability and activity leads to
extroversion, and emotionality is central to neuroticism.
Buss and Plomin (1989) distinguished three levels of arousal - behavioral
arousal, autonomic arousal, and brain arousal - in correspondence to the three
temperamental traits.
Specifically, individual differences in behavioral arousal
underlie Activity, and autonomic arousal of the sympathetic system underscores
Emotionality.
Buss and Plomin (1984) did not directly mention the biological
correlate of sociability, but suggested that brain arousal might be a part of the
physiological correlate of Sociability.
Buss and Plomin emphasized "genetic inheritance" in temperament, and
identified Activity, Emotionality, and Sociability as the essential foundations of
individuality.
But recent behavioral genetic studies consistently point to the
importance of environmental influences including both non-shared and shared
experiences on temperamental characteristics. It has been suggested that the
terms "biological," "genetic," and "inheritable" do not always means the same
thing.
Buss and Plomin (1984) developed inventories that measure the
Emotionality, Activity, and Sociability (EAS) traits in children (EAS-Temperament
Survey for Children) and adults including adolescents (EAS-Temperament
12
Survey).
Though the instrument for adults includes five scales to assess
distress, fear, anger, activity, and sociability, the instrument for children has only
four scales - measuring distress, activity, shyness, and sociability.
Rothbart and Developmental Theory of Temperament
Rothbart and Derryberry (1981) defined temperament as "constitutionally
based individual differences in reactivity and self-regulation, with constitutional
referring to the person's relatively enduring biological makeup influenced over
time by heredity, maturation, and experiences" (p. 37).
Reactivity refers to
emotional, motor, and attentional reactivity, as reflected in the arousability of
somatic, autonomic, endocrine, and cognitive reactions to internal and external
stimuli. "Self-regulation refers to the processes that modulate (facilitate or inhibit)
reactivity, and those processes including attention, approach, withdrawal, attack,
inhibitory control, and self-soothing" (Rothbart, 1989, p. 59).
So the key
components of temperament are arousal (reactivity), affect (emotionality), and
self-regulation (Derryberry & Rothbart, 1988).
As a child psychologist, Rothbart brought a developmental perspective to
her conceptualization of temperament-that is, she proposes that temperament
itself develops.
Specifically, temperamental traits related to reactivity usually
appear in the first year of life.
As the child develops, such traits become
increasingly influenced and controlled by the other temperamental process-selfregulation or effortful control, which only develops later in life with advances in
cortical functioning (Rothbart, Ahadi, & Evans, 2000; Rothbart & Bates, 1998).
Clearly, development and temperament are inextricably intertwined in Rothbart's
13
theory.
Different from Thomas and Chess, Rothbart prefers not to use value-
laden names for labeling temperament patterns, as whether a temperamental
characteristic is difficult or easy is relative to contexts. For example, a persistent
infant might be difficult to console or distract, and therefore be considered difficult
by parents. But a persistent child might be able to remain focused on a task,
which is a good quality for learning. Therefore, Rothbart (1989), in agreement
with Goldsmith, uses the value-free temporal (latency and duration) and intensive
parameters (peak magnitude and shape) to describe behavior.
Guided by such a view, Rothbart and her colleagues developed a series of
questionnaires to measure the conventional reactivity-related temperamental
dimensions as well as dimensions that is related to effortful control from infants to
adults, such as negative affect (fear, anger/frustration), positive affect (smiling,
pleasure), activity level, and attention/effortful control.
The questionnaires
include the Infant Behavior Questionnaire (IBQ; Rothbart, 1981; Garstein,
Putnam, Becken-Jones, & Rothbart, 2002), Toddler Behavior Assessment
Questionnaire (Goldsmith, 1996), Children's Behavior Questionnaire (CBQ;
Rothbart, Ahadi, Hershey, & Fisher, 2001), Early Adolescent Temperament
Questionnaire-Revised
(EATQ-R;
Ellis
&
Rothbart,
2001),
and
Adult
Temperament Questionnaire (ATQ; Rothbart, Ahadi, & Evans, 2000). All of the
questionnaires have demonstrated good psychometric properties. In addition to
the questionnaires, Goldsmith and Rothbart (as cited in Strelau, 1998) developed
standardized laboratory procedures Laboratory Temperament Assessment
Battery (LAB-TAB) to measure behavioral aspects of temperament during
14
infancy. LAB-TAB is available in a locomotor version for 12- to 18-month-olds,
and a prelocomotor version for 6-month-olds.
It assesses the following
dimensions: Activity level, fearfulness, anger proneness, interest/persistence,
and joy/pleasure (Goldsmith & Rothbart, 1991).
Goldsmith and Campos Emotion-Oriented Theory of Temperament
Goldsmith (as cited in Strelau, 1998) specifically points out that their
theory is limited to infants and early childhood.
temperament is defined as
individual differences
In this approach, infant
in
the
probability of
experiencing and expressing primary emotions and arousal (Goldsmith et aI.,
1987). Though both Goldsmith and Campos have conducted extensive empirical
work on genetic and physiological determinants of temperament, they chose to
confine their definition to the behavioral level because they think temperament is
most meaningful in social context, and the stimuli that evoke temperamental
responses are part of the context. Goldsmith and Campos (Goldsmith, et aI.,
1987) proposed a series of inclusion and exclusion criteria for temperament. The
inclusion criteria are that temperament is emotional in nature, it is about
individual differences, it refers to behavioral tendency rather than actual
occurrences of behavior, and it is indexed by the expressive aspect of emotion.
Temperament doesn't include cognitive or perceptual factors.
According to the definition, the primary emotions became the content
dimensions of temperament. The primary emotions adopted by Goldsmith and
Campos include anger, sadness, fear, joy and pleasure, disgust, interest
(corresponds to temperamental persistence), and surprise (Goldsmith, et aI.,
15
1987).
Goldsmith also added motor activity level because it may reflect
emotional arousal that is not differentiated into any of the primary emotions.
Goldsmith (Goldsmith, et aI., 1987) summarized how their emotion-based
approach differs from other temperament approaches. Although the behaviors
Goldsmith and Campos studied overlap substantially with Rothbart's, Rothbart
includes much broader dimensions of temperament (at least in theory) than just
the primary emotions and activity. Unlike Thomas and Chess who emphasize
that temperament is about behavioral style, not about the content or motivation of
behavior,
Goldsmith
and
Campos assume that emotion
has important
motivational properties. Unlike Buss and Plomin, Goldsmith and Campos do not
view heritability as an inclusion criterion of temperament. This is based on their
skepticism that human behavioral characteristics can be nearly divided into
heritable and inheritable.
Kagan's Psychobiological Theory
The roundtable discussion omitted an important figure in temperament
research and that is Jerome Kagan from Harvard University. What makes Kagan
and his research different from other theorists is that he (a) emphasizes the use
of both biological markers and behavioral markers to identify a temperamental
category, (b) criticizes the sole use of the questionnaire method in temperament
research by advocating the use of standardized laboratory behavior observation
along with physiological measures in temperament studies, (c) argues for the
validity of temperamental categories instead of temperamental dimensions.
-Consistent with the above views, Kagan (1994) considers temperament as:
16
an inherited profile of behavior, affect, and physiology that is best
discovered by observing directly the young child's psychological and
biological reactions to specific incentives and charting how these initial
biases lead to distinctly different envelopes of behavior and mood that are
moderately stable over later childhood and adolescence. (p. 50).
Kagan and his colleagues have focused on two basic temperamental
types (categories) in research-inhibited and uninhibited children.
They have
identified correlations, in some degrees, between certain physiological process
(highly or lowly reactive limbic system area) and actual shy or bold behaviors.
For example, a high and stable heart rate (low vagal tone) is often related to an
inhibited and fearful behavioral style (Kagan, 1998). The challenge is that the
link between particular physiological measures and behavioral expressions is
complex-it is neither consistent nor linear.
For example, some children may
have the biological markers indicating a highly reactive limbic system, but do not
necessarily display inhibited behaviors when encountering strangers, or vice
versa.
More studies need to be done in search of both better physiological
indicators and potential processes that may modify behavioral expressions.
Kagan and colleagues also found
interactions between a child's
temperamental characteristics and biological sex.
About 15% of low reactive
girls who were uninhibited at nine and fourteen months became very fearful at 21
months; but very few low reactive boys shifted from being uninhibited to inhibited.
As Kagan (1994) suggested, "parents unconsciously treat sons and daughters in
different ways and produce the larger number of older fearful girls" (p. 263).
17
Though Kagan emphasizes incorporating potential underlying physiology into a
temperament profile, he does not believe that temperament phenomena can be
reducible to physiological processes.
In summary, despite all the differences, the definitions of temperament
proposed by the theorists agree on the following points: Temperament is a rubric
term of a group of related traits and not a trait itself. Temperament is largely
manifest and meaningful in the context of social interactions (Goldsmith et aI.,
1987).
A temperamental dimension or quality is seen as (a) varying among
individuals, (b) relatively stable over time and situations, (c) under some genetic
influences, and (d) appearing early in life. Theorists' disagreements center on
the characteristics nominated as primary. A general definition given by Bates
(1989) is that temperament "consists of biologically rooted individual differences
in behavior tendencies that are present early in life and are relatively stable
across various kinds of situations and over the course of time" (p. 4).
Measurement of Temperament
Studies of temperament have employed a variety of measurement tools,
including questionnaires completed by parents, other caregivers, and the
subjects themselves. Other tools have included home observations, laboratory
methods, and physiological measures.
Home observation is often used with
other types of methods for validation and rarely used alone.
measures are less common because of the high cost involved.
18
Physiological
A major source of information on temperamental characteristics in children
comes from parental questionnaires or interviews.
But as the area of
temperament becomes more sophisticated, it becomes more sensitive to the
limitations of parent report. Obviously, one concern is the subjectivity of parental
reports.
Parents' own personalities, experiences, expectations and ideals of
child are involved in their perceptions of their children's temperaments. Mothers
who described their children on the RITQ as difficult were themselves more
anxious, suspicious, and impulsive than mothers who described their children as
easy (Kagan, 1998).
This issue may become more complicated when involving
parents from different ethnic and cultural backgrounds.
Another often-criticized problem with parent report is the low interrater
agreement between parent and observers (Hubert, Wachs, Perters-Martin, &
Gandour, 1982; Siabach, et aI., 1991). As demonstrated by a number of studies,
the correlations between the parents' ratings and the trained observers'
evaluations of the same characteristics are usually low. Parents and teachers
similarly tend to disagree on the presence of behavior problems in preschool
children; the parent-teacher agreement was only r
review of nine studies (Strelau, 1998).
= .38 on average based on a
Regarding the interparent agreement,
numerous studies have shown that correlations between mother and father
temperament ratings are only moderate, at best around r
= .40
on average
(Martin & Halverson, 1991; Strelau, 1998). As Siabach et al. (1991) said, "both
for interparent agreement and interobserver agreement, the moderate and highly
variable relations appear to be the norm" (p. 212).
19
However, such norms are
obtained in studies conducted primarily in Western cultures, especially those in
America.
One recent study conducted in Mainland China demonstrated very
high rates of agreement between mothers and fathers' ratings of their children's
temperamental traits (Zeng, 1999).
Zeng considered this finding unique in
Mainland China because of the increased sharing of childcare and domestic
responsibilities between the wife and husband due to cultural differences and the
"one child" policy. More studies using people from Eastern cultures may help to
clarify the issue, given the distinct impact Eastern cultural values may have on
family and interpersonal relationships.
Although Kagan (1994, 1998) argued that parental reports are not
appropriate for the study of temperament, the fact that parents spend a
significant amount of time with their children and can usually observe behaviors
of their children in a wide range of situations may, in fact, make them a good
source of information. In addition, a number of currently available temperament
questionnaires have demonstrated adequate psychometric properties (Goldsmith
& Rothbart, 1991; Siabach, Morrow, & Wachs, 1991).
Given the above reasons
and the great convenience, the questionnaire approach has become the most
frequently used method in measuring temperament.
Cross-Cultural Considerations
Cross-cultural
studies
have
typically
used
translated
versions
questionnaires to compare children from different countries on
dimensions of temperament.
of
various
Between-group differences in scores are then
20
examined to determine whether the samples differ significantly.
There are
several challenges for researchers who use questionnaire measures developed
in another culture and language. Linguistic equivalence, conceptual equivalence,
scale equivalence, and normative equivalence must occur to ensure cultural
equivalence of measurement (Marsella, Dubanoski, Hamada, & Morse, 2000;
Marsella & Leong, 1995).
Linguistic equivalence refers to precise translation of the instrument, and it
is the first step towards equal assessment of temperament cross culturally. But
correctly translated items may not be applicable or meaningful to respondents in
another culture. For example, Super and Harkness (1986) reported that one item
on their translated questionnaire asked about the infant's ability to remain
occupied when left alone for 30 minutes, most of the Kigsigis mothers living in
rural Africa said their infants could do so. But the authors had observed that their
infants were never left alone for this situation.
When they asked for an
explanation, the mothers said that they thought the question meant alone with
the infant's brothers and sisters, since leaving an infant isolated for this duration
of time was not a practice in their childcare so that the other meaning of infants
being alone by themselves did not make any sense to these mothers.
Conceptual equivalence refers to the similarity in the nature and meaning
of a concept. The same construct may carry different meanings across cultures.
For example, the NYLS temperamental dimension Rhythmicity may be very
important to the New York middle class Caucasian parents, and an infant who
does not conform to schedules of sleeping or eating is considered difficult. But
21
mothers from Chinese culture do not consider irrhythmicity during infancy as
problematic.
Instead, they consider their infants' biological irregularity normal
and temporary. So irrhythmicity carries different meanings for different cultural
groups.
Similarly, the construct Sociability may have different emphasis in
different cultures (Ahadi, Rothbart, & Ye, 1993). A sociable person in American
culture suggests an individual who goes to party frequently, is the center of
attention at gatherings, and is lively, active, and talkative in general.
But in
Chinese culture, a sociable person may be someone who is sensitive to others'
need, considerate, and cooperative. A difference in score on a temperament
scale or factor cannot be interpreted literally, specific cultural realities and values
must be taken into account (Yang, 1986).
Scale equivalence refers to the cultural comparability of the scales that are
used in the assessment instrument.
Many non-Western cultural groups are
unfamiliar with Likert scale, and researchers should not assume that people from
other cultures share similar understandings of rating scales.
Kohnstamm (1989)
found that the Dutch-speaking mothers from Belgium were significantly more
likely than American mothers to use only the scale points that were verbally
labeled (the midpoint and the two endpoints of the 7-point scale). This tendency
led to much higher variances on questionnaire items for the Dutch sample
compared to the U.S. sample.
According to Kohnstamm (1989), cultural
differences in the amount of exposure to rating scales may contribute to the
difference between the two samples.
22
In addition to the linguistic, conceptual, and scale equivalence, normative
equivalence requires that norms be available for the group being studied.
Interpretations of scale scores are likely to be questionable unless culturally
appropriate norms are utilized.
Response bias is another possible confound to be considered when using
questionnaires in cross-cultural studies.
Compared to Chinese and Japanese
mothers, American mothers tend to rate their children more highly on intellectual,
motivational,
and
academic factors
(Stevenson,
Lee,
&
Stigler,
1987).
Kohnstamm (1989) found similar tendencies in a sample of American mothers,
who were more likely than Dutch mothers to view their children more favorably
(rating them as easier than average children).
Kohnstamm (1989) interpreted
such a result in relation to cultural values, pointing out that "knowing how to sell
things, including oneself, is a major value in America" (p. 491).
Cross-Cultural Research of Temperament
Current Status
The view that temperament consists of constitutional characteristics that
appear early in life indicates temperament may have strong biological roots.
Earlier studies on temperament across cultures aimed to illustrate inherent
between-group differences in temperament due to contributions of different
genetic pools.
Most of these studies used intranational ethnic groups for
comparison and interpreted ethnic differences in temperament as primarily
biologically based.
But other lines of cross-cultural research suggest a more
23
complex and interesting picture of environmental factors. Studies conducted on
ethnically similar samples raised in different nations have shown significant
variations in temperament traits between groups.
For example, The Australia
Study (Prior, Sanson, & Oberklaid, 1989) has compared children with parents
who have immigrated to Australia to ethnically similar children who live in the
country of their parents' birth. These results offer an informative look into relative
influence of culture on the biologically based construct of temperament.
The goodness of fit concept proposed by Thomas and Chess motivated
cross-cultural studies to further examine and understand the relation between
temperamental individuality and the social context. A good example was a study
by DeVrise (as cited in DeVrise, 1994) who found that infants with difficult
temperament in an African village were able to survive a severe famine, whereas
infants with easy temperament in the same village died. The "difficult" infants
successfully got attention and better care from their parents by being "difficult."
Studies like this suggested the necessity and importance of understanding
temperamental characteristics in relation to the contextual demands.
It is a big leap from studying temperament in isolation (i.e., temperament
considered almost solely as a biological phenomenon) to studying temperament
in context. According to Super and Harkness (1986), the developing person's
context or developmental niche is influenced by (a) the physical and social
setting; (b) culturally regulated customs in the areas of childcare, socialization;
and (c) the psychology of caregivers. This psychology is termed ethnotheory,
and refers to caregivers' preferences, aversions, beliefs, or expectations
24
regarding the meaning of particular behaviors.
Temperament can be best
understood as resulting from an interaction between individual characteristics
and the influence of the developmental niche. Super and Harkness suggest that
comparative studies on temperament incorporate the understanding of these
organized environments so that inquires into group differences could be more
meaningful.
This change of direction has greatly stimulated and enriched cross-cultural
research on temperament.
Studies have been conducted to investigate
developmental outcomes of particular temperamental characteristics in different
cultures.
For example, the quality "shy" or "inhibited" often has negative
meanings in American society, so that shy people often encounter difficulties in
both career development and personal life.
Specifically, shy Americans,
especially males, were found to be less successful in their careers, more delayed
in getting married, and having a first child than non-shy males (Capsi, Elder,
Bem, 1988, as cited in Kerr, 2001).
In contrast, in Sweden where the shy
personality is more acceptable than in the U.S., shy males had no difference in
their career advancement when compared with the
non~shy
males but they
averaged being three years later than non-shy males in getting married and
having the first child (Kerr, Lambert, & Bem, 1996, as cited in Kerr, 2001).
Similarly, Chen (2000) argued that the shy characteristics are favored in Chinese
culture traditionally, thus shy children may thrive better than the bold children. In
a recent study conducted in PRC, two different forms of shyness were identified
in a group of children from Shanghai: Executive shyness and anxious shyness
25
(Xu, Farver, Chang, Zhang, Yu, & Cai, 2004). Executive shyness represents a
positive form of shyness that is highly valued in Chinese culture and often refers
to a behavioral pattern of modesty, discretion, and politeness in social
interaction. In Xu et ai's study, executive shyness was found to be associated
with positive social outcomes including high social preference among peers and
low levels of self-reported loneliness and social anxiety.
However, anxious
shyness predicted negative outcomes in the Chinese children, which is
consistent with most Western findings according to Xu et al.
There have been few studies investigating how culture (or developmental
niche) enhances some temperamental characteristics and mutes others.
A comparative study (Ahadi, Rothbart, & Ye, 1993)-conducted using a
Mainland Chinese and American sample of 6- to 7-year-olds with the Children's
Behavior Questionnaire (CBQ)-provided some information to the question.
Factor analyses of the two datasets found the same three-factor structure but
different patterns of factor loadings across the two cultures.
For example, the
factor Effortful Control and the scale Approach were uncorrelated in the US
sample, whereas in the PRC sample, the two factors were negatively correlated.
Ahadi and Rothbart (1994) suggested that in Chinese culture, one might expect
an individual's self-regulative capacities to work to suppress approach, impulsive,
and high activity level behaviors because Chinese culture does not value those
behavioral tendencies.
Thus, Effortful Control might have superordinate
regulative functions over other temperament systems.
Contextual demands
including particular cultural values may work though Effortful Control to modulate
26
the expressions of temperamental tendencies (Ahadi & Rothbart, 1994; Posner &
Rothbart, 2000).
In summary, in the last decade cross-cultural studies on temperament
began to go beyond simple comparisons of the similarities and differences to
seriously consider the role of culture in the development of temperament.
Chinese Children and Temperament: Substantive Findings
The following is a brief review of substantive findings from cross-cultural
studies on temperament.
Since the current investigation focuses on Chinese
adolescents, the review will be composed mainly of studies involving Asian
samples.
Freedman and
Freedman (1969) compared 24 newborn Chinese
American and 24 newborn European American babies aging from five to 75
hours who were similar in weight, mother's age, length of labor, and use of drugs
during labor. They administrated several Brazelton measures to the newborns,
and reported that the Chinese American infants were calmer, less emotionally
labile, more easily habituated, and more easily yonsoled when distressed than
were European American babies. Specifically, Chinese American babies were
less likely to protest a light cloth draped over the face, or being positioned with
their faces down on bedding, habituated more quickly to a light shone in the
eyes, and tended to stop crying sooner than the European-American newborns
after being picked up. Similarly, five-month-old Japanese infants took twice as
long as European American infants to show distress when their arms were
27
restrained (as cited in Kagan, 1998). Results of the Freedman study described a
temperament profile of a calm, steady, more easily soothed Chinese infant. The
Freedmans' study is important because the observation of neonatal differences
suggests strong biological foundations of the early expression of temperament.
Similar findings have been reported in other studies.
Kagan, Kearsley,
and Zelazo (1978) found that Chinese American infants living in Boston were
less active, vocal, less smiling, and more inhibited in response to visual and
auditory stimuli during the first year of life than were European American infants
living in the same neighborhoods. Lewis, Ramsay, and Kawakami (as cited in
Kagan, 1998) reported that Japanese infants were less likely to show intense
distress to inoculations.
Kagan, Arcus, Snidman, Wang, Hendler, and Greene (1994) compared
infants from Beijing, China, Dublin, Ireland, and Boston, United States on
behaviors of motor activity, cry, fret, vocalization, and smiling, using a
standardized 40-minute laboratory battery of visual, auditory, and olfactory
stimulations. Analyses of videotapes of infants from the three sites found that 4month-old Chinese infants born and living in Beijing had much lower levels of
motor activity, irritability (fret and cry), and vocalization than did Caucasian
infants (from both Boston and Dublin). Smiling was the only behavior for which
the Chinese infants did not differ from other groups.
Significantly more
Caucasian than Chinese infants were high reactive, and significantly more
Chinese than Caucasian infants were low reactive. The most important message
of this study was that the typical Chinese infant was far less reactive than the
28
average low reactive American infant. So far, Chinese infants are portrayed as
low reactive or less easily aroused, indexed by their lower levels of motor activity,
irritability, and vocalization.
Hsu, Soong, Stigler, Hong, and Liang (1981) administrated Carey's
Revised Infant Temperament Questionnaire (RITQ) to the mothers of 349 four- to
eight-month-old infants in Taipei, Taiwan. The authors used back translation and
pilot testing to ensure some cultural equality between the original RITQ and its
Chinese version. The findings were compared with Carey and McDevitt's (1978)
standardization sample from the United States. The distribution of the Chinese
sample among the basic categories of temperament (such as the easy, slow-towarm-up, and difficult temperament) closely resembled that of the American
sample. But significant differences between the two samples were found in all
nine temperamental dimensions except for persistence.
Specifically, Chinese
infants were less active, adaptable, distractible (soothable), less likely to
approach the new, and less rhythmic in biological functions; they were also more
intense and negative in mood than American infants. In addition, Chinese infants
had lower threshold for responsiveness.
In contrast to Carey's American standardization sample where no sex
differences were observed, Hsu, et al. (1981) also found significant sex
differences in their Chinese sample on the dimension of approach, with males
being more willing to approach the new than females. In conclusion, the authors
suggested that response biases, racial differences or a combination of these two
factors might contribute to the observed cultural differences in their study.
29
Though no specific types of responses biases were discussed in the study, the
authors (1981) pointed out that "it is a possibility that mother's perception of
temperamental characteristics may prove to be more significant than the actual
characteristics themselves" (p. 1340).
Overall, when compared to the US sample, Chinese infants in Hsu et al.'s
study had lower thresholds for responsiveness, were more intense and negative
in emotion, and less distractible (or soothable), active, approaching of new
stimuli, and adaptable.
These results contradict the findings of Freedman and Freedman (1969),
who suggested a temperament profile of a calm, steady, and more easily
soothed Chinese American infant. Several differences in the two studies might
contribute to the inconsistent results here; such differences include those
concerning methods (Freedmans' observation versus Hsu et al.'s questionnaire
methods), age-related factors (newborns versus four- to eight-month-olds), and
potential response biases. Further study on samples of older age is necessary to
clarify the issue.
The Australian Temperament Project (ATP) represents the most extensive
study of temperament across cultures. It was initiated in 1983 with a sample of
2443 four- to eight-month-old infants using the RITQ to assess the Thomas and
Chess's nine dimensions of temperament and the, easy, difficult, and slow-towarm-up categories (Prior, Sanson, & Oberklaid, 1989).
Infants who had
Australian-born parents (about 75% of the sample) were compared to infants with
at least one parent who was born in another country. In addition to Australia (N
30
=
1593), the ethnic cultural groups represented in the sample were as follows: UK,
Ireland, or New Zealand (combined for analyses because of perceived similarity
in cultural background and origins, N =279), Italy (N
=124), Northern or Western
Europe (N = 110), Asia, excluding the Indian subcontinent (N =48), Greece (N =
46), Yugoslavia (N = 27), Indian subcontinent (N = 25), Middle East, excluding
Lebanon (N = 22), Africa (N = 16), Lebanon (N = 16), South or Central America
(N =14), and North America (N =11). The number from each cultural group was
representative of 1981 census figures but the reason why the ATP researchers
recruited samples of children with foreign-born parents in proportion to census
data was not clear. Seeking such representativeness led to very small sample
sizes for some of the ethnic groups, which raises questions about the reliability of
the results.
When compared to infants with parents born in Australia and other
Western nations, infants with parents born in Asia, Lebanon, Italy, Greece, and
Yugoslavia consistently scored higher (more negatively) on the dimensions of
Approach, Adaptability, and Distractibility. They were rated less approaching,
adaptable, and distractible (soothable). These five groups of infants were also
more likely to be classified in the Difficult category due to lower levels of reported
approach, adaptability, and rhythmicity and higher levels of negative mood. Prior
et al. (1989) rejected several hypotheses for the differences emerged, including
the possibility of language difficulties in immigrant parents, confounding ethnic
status with SES, mental health problems of immigrant parents, and sample
selection bias. However it was still extremely difficult to attribute the between-
31
group differences seen in the ATP data to genetic influences as such influences
As Prior et al. (1989) suggested, the
entangled with the cultural influences.
middle class North American conceptualization of what constitutes "difficult
temperament" might have less relevance in other cultures.
Windle, Iwawaki, and Lerner (1988) examined temperament in a sample
of older children from the United States and Japan.
They used the Revised
Dimensions of Temperament Survey (DOTS-R; Windle & Lerner, 1986) with a
sample of 234 Japanese preschoolers and 114 American preschoolers from
middle-class families.
The DOTS-R is a parent report instrument constructed
according to the NYLS model that measures Activity Level-General, Activity
Level-Sleep, Rhythmicity-Sleep, Rhythmicity-Eating, Rhythmicity-Daily Habits,
Approach-Withdrawal, Flexibility-Rigidity, Quality of Mood, and Task Orientation
(distractibility). The Japanese preschoolers were rated by their parents as more
withdrawn, rigid, and negative in mood with less predictable, more active sleep
compared to the American preschoolers. The ratings of Japanese children as
lower in approach and adaptability and more negative in mood are consistent
with the findings of Hsu et al. on Chinese babies.
In a parallel study, Windle, Iwawaki, and Lerner (1987) also examined
temperament in a sample of early- and late-adolescents from Japan and America
using self-report data of the DOTS-R.
Consistent with the data on Japanese
preschoolers, both the early and late Japanese adolescents rated themselves
significantly less approaching, and less positive in mood than their American
counterparts.
The late Japanese adolescents also rated themselves as less
32
flexible in adjusting to changes in the environment than their American
counterparts. Despite the results, Windle, lwawaki, and Lerner (1987; 1988) did
not offer explanations for the differences that emerged.
Another study comparing the temperament of young children from Eastern
cultures to a U.S. sample was done in Malaysia.
Banks (1989) administered
translated versions of the Infant Temperament Questionnaire (ITQ, former
version of RITQ) to parents and grandparents of 23 Malay infants and toddlers
under the age of 2. When compared to norms derived from the infants in Carey
and McDevitt's sample (1978), these Malay children demonstrated significantly
higher (or negative) scores on the measures of Regularity, Adaptability,
Approach, and Threshold. These results indicated that Malay infants were less
regular, adaptable, approaching, and lower in thresholds of responsiveness when
compared to the American norms. Banks (1989) discussed these differences in
relation to cultural influences including the customs in Malay childcare.
For
example, the Malay culture expects low sensory thresholds because the Malay
children are taught to be sensitive to sounds, lights, heat, cold, dirt, pain, and
odors. Additionally, Malaysian mothers did not consider their children's rejection
of new foods as less adaptable and problematic, instead they considered food
rejection decisive. These mothers found it ridiculous to force their children to eat
more when they already had enough or to eat the food they did not like. But
such items were used to assess the Adaptability dimension. "Thus, the relatively
low adaptability scores in this group [Malay children] may not imply the level of
difficulty that they would among Americans" (Banks, 1989, p. 395).
33
A majority of cross-cultural studies on temperament used the NYLS
conceptualization of temperament and questionnaires (ITa, RITa, DOTS-R) that
yielded nine dimension scores and the easy, slow-to-warm-up, and difficult
temperament categories.
As demonstrated by the studies reviewed so far,
concepts such as difficult temperament are especially dependent on the context
of cultural expectations.
Ahadi, Rothbart, and Ye (1993) conducted another study on cross-cultural
similarities and differences in temperament using Chinese and American
samples. They administrated csa to the parents of 468 Chinese children from
Shanghai, PRC and 156 American children aged 6 to 7. The csa is based on
Rothbart's theoretical conceptualization of temperament as constitutionally based
individual differences in reactivity and self-regulation.
It contains 15 scales to
assess Activity Level, Anger/Frustration, Approach, Attentional Focusing,
Discomfort, Falling Reactivity and Soothability, Fear, High Intensity Pleasure,
Impulsivity, Inhibitory Control, Low Intensity Pleasure, Perceptual Sensitivity,
Sadness, Shyness, and Smiling & Laughter. The major findings included:
1. Data on both samples yielded three similar temperament factors"Surgency/Extroversion,"
"Negative
Affect,"
and
"Effortful
Control,"
suggesting the structure of temperament remained same across both
cultures.
Specifically, "Surgency/Extroversion" was a factor defined by
high scores on the csa scales of Activity Level, Approach, High Intensity
Pleasure, and Impulsivity, and lower scores on Shyness.
"Negative
Affect" was made up by higher scores on Discomfort, Fear, Anger,
34
Sadness, and Shyness, and lower scores on Soothability.
The factor
"EffortfuI Control" was composed of higher scores on Inhibitory Control,
Attentional Focusing, Low Intensity Pleasure, and Perceptual Sensitivity.
The scale Smiling and Laughter loaded differently in the two samples, it
loaded on the "Surgency/Extroversion" factor in the Chinese sample, but
the "Effortful Control" factor in the American sample.
2. Significant differences in temperament between the Chinese and
American sample were obtained on 13 of the 15 scales measured. The
two exceptions were Attentional Focusing and Soothability where no
differences were found.
Specifically, when compared to the American
children, the Chinese children scored lower on the scales contributing to
the "Surgency/Extroversion" (which included the Activity Level, Approach,
High Intensity Pleasure, and Smiling & Laughter) and "Effortful Control"
(which included the Inhibitory Control, Low Intensity Pleasure, and
Perceptual Sensitivity), but scored higher on the scales contributing to the
"Negative Affect" (which included the Anger, Discomfort, Fear, Sadness,
and Shyness).
3. Gender differences in the Chinese sample were in the opposite direction
of those secured in the U.S. sample (see the section on gender
differences in temperament for details).
Ahadi, Rothbart, and Ye's study had some limitations. Firstly, the sample
sizes (156 American subjects vs. 468 Chinese subjects) were not ideal for factor
analysis, given the numbers of scales and items the
35
csa
contains (15 scales
and 195 items).
Secondly, significant differences in a wide range of
temperamental dimension were obtained by using adjusted scores. Since the
overall mean scores averaged across all 15 scales were above the midpoint of 4
for the American sample (M
= 4.61),
and below the midpoint for the Chinese
sample (M = 3.65), Ahadi, Rothbart, and Ye subtracted the overall mean score of
the participants' country from each individual scale scores to avoid possible
response bias between the two cultures.
American and Chinese participants on all
The adjusted mean scores of the
csa scales were then compared.
As a
result, Chinese children were different from American sample on more scales
than using the unadjusted scores.
Taking the "Negative Affect" factor as an
example, the Chinese children scored higher on only Fear and Shyness scales
than their American peers, but when the adjusted scores were used, the Chinese
children also scored higher on Anger, Discomfort, Impulsivity, and Sadness than
the American children in addition to Fear and Shyness. These higher scores of
Chinese children suggested that they were not only more fearful and shy, but
also angrier and sadder, as well as more impulsive than the American children ..
There is a possibility that using the adjusted mean scores for comparison might
inflate the differences on some
csa scales between the two samples, especially
on the scales contributing to the "Negative Affect" factor. This is because the
response bias-higher ratings from American mothers versus lower ratings from
Chinese mothers-tends to be more pronounced on the socially desirable scales
than the undesirable ones.
36
In addition to such limitations, some findings of the study were "atypical."
For example, the Chinese children in Ahadi, Rothbart, and Ye's study showed
higher threshold of responsiveness through their lower scores on the Perceptual
Sensitivity and Low Intensity Pleasure scales, which contradicted the previous
findings indicating lowered sensory thresholds of Asian infants. Another example
would be the results about the temperamental gender differences in Chinese
children.
More studies using the same Rothbart's conceptualization of
temperament should be conducted to further investigate these atypical findings.
There are limited cross-cultural studies on temperament using adult and
adolescent samples.
Most of the studies at this age groups used Eysenck
Personality Questionnaire. Barrett and Eysenck (1984) reported data comparing
a Chinese sample and British sample. The Chinese subjects scored lower on
Extraversion and higher on Psychoticism than the British subjects. With respect
to Neuroticism, men in the UK sample had the lowest scores, followed by
Chinese men and then Chinese women, with women in the UK scoring highest.
Results of virtually all the cross-cultural research have posited the need to
closely
examine
childrearing
practices,
parental
perceptions
of
child
temperament, parental tendencies to report favorably about their children, and
the appropriateness of individual items and tasks that comprise questionnaires
method.
Gender Differences in Temperament
Kohnstamm (1989) pointed out that the sex distinction creates two natural
groups in any cultures. He distinguished sex differences from gender differences
37
by emphasizing that the use of "gender differences" implies the culturally or
socially construct on sex.
Furthermore, "to choose either word seems to
indicated a preference for one explanation: the observed differences in behavior
are inborn or learned ... I prefer "sex" over "gender" because I want to keep the
possibility open that the behavioral differences are, at least partly, caused or
facilitated by constitutional factors" (Kohnstamm, 1989, p.494).
Despite the biological and psychosocial salience, gender differences have
received little attention in temperament studies. Several previous reviews of the
literature on a gender difference in temperament appear as valid today as then
(Kohnstamm, 1989; Buss and Plomin, 1975; Maccoby and Jacklin, 1974).
A
common finding given by these reviews is the interaction between age and
gender differences.
In other words, whether there is a gender difference
depends on age.
Maccoby and Jacklin concluded that boys were more active than girls in
general, however, no gender differences in activity level was reported in children
less than 12 months.
Alternatively, Buss and Plomin (1975) concluded that
before the preschool years, there were no consistent gender differences in
activity, fear, anger, and sociability. But starting from the preschool year, girls
tended to become more fearful and sociable than boys, and boys consistently
more active and aggressive than girls.
In adolescence and adulthood, the
gender differences became consistent and significant.
Accordingly, Buss and
Plomin also reasoned that gender differences in temperament were results of
socialization due to culturally shaped gender roles. Up to recently, Buss (1991)
38
still maintained his views regarding the gender differences in temperament, as he
saw no reasons to change the views.
Support also comes from other empirical evidence that was not included in
the early overviews of Buss and Plomin, and of Maccoby and Jacklin. Rothbart
(1986) found no temperamental sex differences using either parental report (the
Infant Behavior Questionnaire) or home observations in a study of 46 infants at 3,
6, and 9 months of age. She concludes: "The differences do not seem to be built
on differences observable in the first year, at least as we have been able to
measure them" (Rothbart, 1986, p. 364). However gender differences became
observable when children get older.
Using the Child Behavior Questionnaire
(CBQ), Goldsmith and colleagues (Goldsmith, Buss, & Lemery, 1997) noted boys
were rated higher than girls on Activity Level and High Intensity Pleasure
whereas girls were rated higher on Inhibitory Control and Perceptual Sensitivity.
The Australian Temperament Project followed up 2443 4- to 8-month-old
infants from 1983 to 1986 for 4 years (Prior, Sanson, Oberklaid, 1989). Parents
of these infants filled out an Australian revision of the RITQ in the first year.
Each year after, an identical procedure was followed, with two thirds of the
original sample being surveyed using age appropriate instruments. The Toddler
Temperament Scale (TTS) was used in the second and third year (1984 and
1985), and the Child Temperament Questionnaire (CTQ) was used in the fourth
year (1986) along with the parent and teacher's version of the Preschool
Behavior Questionnaire (PBQ; Behar & Stringfield, 1974, as cited in Prior,
Sanson, & Oberklaid, 1989).
39
The overall trend of gender differences reported by the 4-year ATP project
was that gender differences in temperament were minimal in the first year of life
but emerged quite strongly in the second, third, and fourth years.
In the first
year, gender differences were observed only on the dimensions of Approach and
Irritability, with males rated as more approaching and irritable. There were no
gender differences in the distribution of easy, slow-to-warm-up and difficult
temperament at this stage. In the second year, gender differences were found
on Approach, Cooperation/Manageability, Rhythmicity and Intensity, with boys on
the negative ends of these scales except for Approach.
In the third year, the
gender differences emerged on Approach, Cooperation/Manageability, and
Activity, with boys more approaching and active but less cooperative.
In the
fourth year when children were 3- to 4-years old, boys were significantly more
likely to be on the difficult end of the easy, slow-to-warm-up, and difficult
temperament distribution; they were less persistent and more inflexible. Prior et
al. (1989) noted that they didn't find high scores on the early activity dimension
for boys, for which gender difference has often been reported.
However,
according to the review by Kohnstamm (1989), the Activity Level is the one
dime'nsion of temperament for which gender differences have been consistently
obtained across age, contexts, and cultures. Though the gender difference in
activity is age-dependent, boys tend to be more active than girls since infancy.
Gender and Cultural Studies
Very few cross-cultural studies focused on the gender differences in
temperament across cultures, and findings about temperamental gender
40
differences were often by-products.
Kagan et al.'s (1994) study on the three
groups (European American Caucasians, Irish Caucasians, and Chinese) of
infants found gender differences on vocalization and smiling only in the Chinese
sample, where the Chinese male infants vocalized and smiled more than the
Chinese female infants. Ahadi, Rothbart, and Ye's (1993) cross-cultural study on
Chinese and American children temperament using the
csa revealed interesting
gender differences across the two cultures.
Specifically, in the American sample, the boys were rated higher than the
girls on Activity Level, and lower on Inhibitory Control, Smiling, and Perceptual
Sensitivity.
Such gender differences were consistent with both the previous
findings and the gender stereotypes. However the direction was reversed in the
Chinese sample, with Chinese girls scoring higher on the Surgency scales of
Activity Level, High Intensity Pleasure, and Impulsivity, lower on the Effortful
Control scales of Inhibitory Control, Low Intensity Pleasure, and Perceptual
Sensitivity, and lower on the Negative Affect scales of Discomfort and Sadness.
As seen in the American sample, Chinese girls scored higher on the
Smiling/Laughter scale than Chinese boys. Such findings are puzzling given the
fact that the Chinese society is stricter on gender roles than the US.
Ahadi,
Rothbart, and Ye (1993) did not offer any explanations but suspected that
Chinese boys might receive more severe socialization than girls. They note: "It is
difficult to know how to interpret these findings, and it would be important to
attempt their replication." (p.373).
41
Other recent studies involving Chinese children failed to replicate the
gender differences found in Ahadi, Rothbart, and Ye's study.
Zeng (1999)
reported gender differences on only one temperamental dimension in a study of
Chinese parenting styles and child temperament. The 5-year-old Chinese boys
(N
= 89) from
same city.
Beijing were rated more active than the girls (N
= 101) from
the
No significant gender differences were obtained on Emotionality,
Sociability, and Attentional Focusing. But because Zeng's study used a "tailored"
child temperament questionnaire that was composed of Buss and Plomin's EAS
Child
Temperament Scale
(1984)
measuring Activity,
Emotionality,
and
Sociability and the Attentional Focusing scale from Rothbart's CBa, its findings
can not be directly compared with Ahadi, Rothbart, and Ye's findings on the
gender differences in their PRC sample.
Lee & Chen (1998) reported a wide range of gender differences using the
Chinese adolescents from the high schools in Taipei, Taiwan. Six hundred and
twenty two male and 543 female children completed the self-rating form of a
temperament instrument composed of some scales of the DOTS-R.
The
Chinese male students scored higher on the Activity Level, Task Orientation,
Intensity of Reaction, and Tactile Threshold, but lower on Adaptability than their
female counterparts, indicating they were more active, persistent on tasks,
intense in emotion, but less sensitive and adaptable than Chinese female
students from Taiwan.
Gender difference in children may affect parents' perceptions of children's
temperament regardless of the actual temperament.
42
It would be helpful to
explore the relationship between parents' perceptions of their children's
temperament and the gender of the children.
Objectives of Current Study
The proposed study will investigate temperamental characteristics of early
adolescents aged from 9 to 12 years from Beijing, Peoples' Republic of China
using the self- and parent-report forms of the Revised Early Adolescent
Temperament Questionnaire (EATQ-R; Ellis & Rothbart, 2001).
The three
primary aims of the study are:
1. To validate the structure or the internal organization of temperament
proposed and measured by the EATQ-R with a PRC sample of both
children and parents. According to the most recent poster presentation
(Ellis & Rothbart, 2001), four factors were extracted from the exploratory
factor analyses of the temperament scores of 177 adolescents: Surgency,
Effortful Control, Negative Affect, and Affiliativeness. The proposed study
will explore whether these factors can be replicated with the self-report
data of a large Chinese sample.
2. To examine basic patterns of temperamental characteristics in early
Chinese
adolescents,
including
further
investigating
the
gender
differences reported by Ahadi, Rothbart and Ye (1993) in a PRC sample of
6- to 7-year-old children using the CBQ. The current study will use early
adolescents to examine the gender differences in temperament as the
literature suggests that temperamental gender variations tend to become
43
more significant and stable during adolescence. In addition to the parent
report, self-reports will also be obtained to provide a perspective from the
adolescents themselves. The self-report data will provide very important
information to further clarify the patterns of the gender variations in
Chinese children reported by Ahadi et aI., whose findings were based on
parents' reports alone. Other than gender differences, potential effects of
only-child status on temperament will also be addressed.
3. To examine the agreement between Chinese adolescents' self-ratings and
their mothers' ratings. Due to the relatively close relationship between the
Chinese children and their parents (especially mothers), the agreement
between the adolescents' self-perceptions and the perceptions of their
mothers is expected to be higher than that is usually reported in the US.
44
CHAPTER 2
METHOD
In this chapter the methods of developing and adapting the measures and
the data collection procedures used are described.
The section will include
descriptions of participants, measures, and procedures.
Participants
Participants for the present study were recruited from 1225 children
attending 4th to 6th grade at four elementary schools in Beijing, PRC.
Excluding
subjects who did not complete the questionnaire and who did not meet the age
criterion of 9- to 12-years old, the final student sample included 1104 children
subjects.
The children were on average 10.49 years old (SO = .83) and
comprised of 589 (53.4%) boys and 515 (46.6%) girls.
About 14% of the
children were from non-only child families.
In addition to the student sample, there was a 68% response rate from the
mothers or primary caregivers of the students. Among the 805 responses, about
73% (589) were from mothers, 19% (152) were from fathers, 6.5% (52) were
from grandparents, and 1.5% (12) from other relatives or nannies.
Considering that the low inter-rater convergence is commonly reported in
studies of temperament and personality using a questionnaire method, the
present study will use one consistent source of parent report, namely the
mothers' ratings, for factor analysis and subsequent statistical analyses.
45
Measures
The instruments used for the present study included:
(1) Revised Early Adolescent Temperament Questionnaire (EATQ-R).
The EATQ-R has parent report and self-report versions.
The parent
version was completed by mothers or primary caregivers if mothers were
not available, and the self-report version was completed by the youths
themselves.
(2) Family Information Questionnaire, completed by mothers or primary
caregivers if mothers were not available.
(3) Parental Attitudes Questionnaire, completed by mothers or primary
caregivers if mothers were not available.
Revised Early Adolescent TemperamentQuestionnaire (EATQ-R)
The self-report form of the EATQ-R (Ellis & Rothbart, 2001) is a 103-item
questionnaire that provides scores on eleven temperamental scales (Activation
Control, Affiliation, Activity Level, Attention, Fear, Frustration, High Intensity
Pleasure, Inhibitory Control, Pleasure Sensitivity, Perceptual Sensitivity, and
Shyness) and two behavioral scales (Aggression and Depressive Mood). The 11
subscales of the EATQ-R can be classified using four higher level factors,
Surgency, Effortful Control, Negative Affect, and Affiliativeness (Ellis & Rothbart,
2001). Scale names, descriptions, and sample items are presented in Appendix
A.
Adolescents were instructed to read each item and decide whether the
statement is a true or untrue description of themselves. Ratings were made on a
46
=almost always untrue; 2 = usually untrue; 3 = sometimes true
untrue; 4 = usually true; 5 = almost always true. The internal
5-point scale: 1
sometimes
consistency estimates for EATQ-R ranges from 0.64 to 0.81, with a mean
reliability estimate of 0.73 across the 13 scales.
The EATQ-R parent report form is a 62-item questionnaire that provides
scores on eight temperamental scales (Activation Control, Affiliation, Attention,
Fear, Frustration, High Intensity Pleasure, Inhibitory Control, and Shyness) and
two behavioral scales (Aggression and Depressive Mood).
descriptions, and sample items are presented in Appendix B.
Scale names,
Parents were
instructed to read each item and decide whether the statement is a true or untrue
description of their child. The same 5-point rating scale of the self-report form
was used in the parent report form. The internal consistency estimates for the
parent report form of the EATQ-R ranges from 0.65 to 0.86, with a mean
reliability estimate of .73 across the 10 scales. Convergence with self-report was
low to moderate.
Family Information Questionnaire
The family information questionnaire obtained background information of
both children and parents, including child's age, birthplace, only-child status,
ethnicity, sleeping arrangement, and the amount of time the child spent on
various tutoring and interest classes.
Information about parents included age,
birthplace, ethnicity, years living in Beijing, education level, type of work, and
estimated family income. Information on who the primary caregiver is and who
filled out the parent version of the EATQ-R was also obtained. Such information
47
helps to identify the parent report completed by the primary caregivers of the
children from the parent report by non-primary caregivers, and hence ensures
the validity of the information used in the study.
Parent Attitude Questionnaire
The parent attitude questionnaire was designed to survey the Chinese
parents' views on temperament. Parents were asked to rate how important each
of the following factors would be in terms of their children's personality
development:
children's
own
temperament,
family
environment,
parental
disciplines, school education, traditional cultural values, and values of modern
society, mother's and father's hereditary traits, children's own life experiences,
and peer influence. Each factor was rated using a five-point Likert scale, where
1 indicated the response of "not important" and 5 indicated the response of
"extremely important." The intention for developing and using the questionnaire
is to understand the functional significance of temperament in Chinese culture,
as perceived by Chinese parents.
For simplicity, the family information questionnaire, EATQ-R parent report
form, and parent attitude questionnaire were arranged into three sections that fit
two double-sided sheets.
Procedures
The EATQ-R was selected as the temperament measure for the study.
Both the parent and youth versions of the questionnaire were translated into
Chinese using back translation procedures. A pilot test was conducted to gather
48
input as to item, response format, and instruction clarity, as well as completion
time.
Translated questionnaires were shared with principals, some parents
known to the researcher, and editors in Beijing. This led to further adjustments.
Some questionnaire items make straightforward statements, which sometimes
sound blunt to Chinese.
For example, an item in Fear scale states, "I worry
about my parents dying or leaving me" in the youth version, and "(my child) is
afraid of the idea of me dying or leaving her/him" in the parent version, the term
"dying," especially "me dying" had to be changed because mentioning somebody
dying when he or she is still alive is considered inauspicious in Chinese culture.
After the pilot testing, the word "dying" was replaced by "never seeing my parents
again" in the youth version and "never seeing me again" in the parent version to
avoid offending parents.
Grammar was also changed to emphasize the
imaginative nature of the question.
Another problem challenging translation was that some items described
American life that is unfamiliar to the regular Chinese parents. For example, one
item in the Fear scale stated that "(my child) doesn't enjoy playing softball or
baseball because s/he is afraid of baiL"
Since softball and baseball are not
common sports in China, parents had a hard time understanding why the child
can be afraid of the ball. One parent asked if it was meant to say the child was
afraid of physical tiredness. So the item was changed to "my child doesn't enjoy
competitive and dangerous sports, such as basebalL"
49
Similarly, "I enjoy
exchanging hugs with the people I like" also had to be changed to "I enjoy being
physically close with the people I like."
The most difficult problem with the Chinese translations was that most
questionnaire items lacked contextual information or at least the hidden
contextual information was unavoidably lost in some translations because the
Chinese audience does not share the information as Americans would.
For
example, item 32 of the Activation Control scale is typical of this problem. The
item stated "If someone asks me to do something, I do it right away, even if I
don't want to." Both children and parents asked about who this someone was,
and what kind of thing was asked to be done. The items with the similar problem
remained in the questionnaire and it would be interesting to see how much
validity these items will hold for Chinese in the context of present Chinese
society.
Three elementary school principals were contacted by the researcher
regarding their schools' participation in the present study. After careful reviews
of the questionnaires, the principals agreed to have their students participate in
the study. Teachers and students were then informed about the study. A time
was scheduled for researcher to administer the questionnaire at each school. To
ensure uniformity and consistency, the instruction was given to the students via
video. The homeroom teachers were present when students were completing
their questionnaires.
The researcher and two assistants were available to
answer any questions students might have during the process.
50
Packets about the study were distributed to the parents of the school
children through the children.
Each packet contained a letter to the parent
explaining the nature of the study and a copy of the three-section questionnaire
mentioned previously. Parents, especially mothers, were asked to complete the
questionnaires and return them to their children's homeroom teachers in two
days.
51
CHAPTER 3
RESULTS
In this chapter, the results of the study will be presented in three sections:
(1) the analysis of EATQ-R self-reports, (2) the analysis of EATQ-R mothers'
reports, and (3) a comparison of the self- and mothers' reports. Results of the
first two sections will include descriptive statistics of the individual EATQ-R
scales, scale reliability test, exploratory factor analysis at the item and scale
levels, and effects of gender differences and family structure (only children vs.
those with siblings) on temperamental ratings. The third section will focus on the
agreement between children's self-ratings and mothers' ratings. The chapter will
conclude with a summative discussion of significant results for further discussion
in Chapter Four.
EATQ-R Self-Report
The EATQ-R self-report included 11 temperamental scales and 86 items.
Of the entire children's sample, six hundred and eighty seven subjects had
complete responses on 86 items without any missing values. This sample of 687
children was used in the preliminary descriptive analyses, scale reliability test,
exploratory factor analysis, and multivariate general linear model test.
Descriptive statistics of the EATQ-R items
Descriptive statistics of Chinese children's responses to the EATQ-R
items are obtained and summarized in Tables 1 to 3, with Table 1 and Table 2
52
listing the ten highest and lowest means of the EATQ-R scores from the Chinese
children's sample, respectively.
Table 3 lists the ten EATQ-R items with the
largest deviations.
Table 1
Ten Highest Means of EATQ-R Scores among Chinese Children
Item #
Scale / Item content
M(SD)
Affiliation
38
I would like to be able to spend time with a good friend every day.
4.38 (1.09)
75
It is important for me to have close relationships with other people.
4.26 (1.16)
51
I will do most anything to help someone I care about.
4.24 (1.05)
Activation Control
R 82
I put off working on projects until right before they are due.
4.32 (1.04)
R 24
I do something fun for a while before starting my homework, even
when I'm not supposed to.
4.30 (1.14)
63
I finish my homework before the due date.
4.29 (1.11)
65
I tend to be on time for school and appointments.
4.26 (1.10)
66
If I have a hard assignment to do, I get started right away.
4.21 (1.17)
Activity
100
I prefer outdoor activities to those indoors.
4.25 (1.14)
Inhibitory control
71
4.20 (1.08)
It is easy for me to keep a secret.
Note. N
= 687.
R
= Reverse-coded
items.
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
Chinese children were found to have particularly high scores on the items
on the Activation Control and Affiliation scales. Given their means and standard
53
deviations, 68% of the sample was estimated to score above the midpoint of 3 in
their ratings on these items.
Table 2
Ten Lowest Means of EATQ-R Scores among Chinese Children
Item #
Scale / Item content
M(SD)
Frustration
R 40
I am a patient person.
2.48 (1.27)
91
It really annoys me to wait in long lines.
2.46 (1.47)
98
I get very frustrated when I make a mistake in my school work.
2.30 (1.40)
79
I get irritated when I have to stop doing something that I am enjoying.
2.15 (1.22)
Shyness
R 87
76
9
43
6
I am not shy
2.41(1.40)
I am shy
2.12 (1.30)
I feel shy with the kids of the opposite sex.
2.08 (1.33)
If I am asked to deliver a message to an adult, I feel uncomfortable
about going up to them.
2.07 (1.30)
I feel shy about meeting new people.
1.72 (1.20)
Fear
54
Note. N
I get frightened riding with a person who likes to speed.
= 687.
R
= Reverse-coded
items.
2.20 (1.42)
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
The children were found to score Iowan the items on the Shyness and
Frustration scales. The average means for these items were below the midpoint
of 3.
The low ratings suggested that Chinese children tend not to consider
themselves frustrated or shy.
54
Most of the items with large standard deviations were from the Fear,
Pleasure Sensitivity, and High Intensity Pleasure scales, which indicated that
Chinese children varied from one another in their opinions of these items more
than they did on other scales. The possible reasons why these scales elicited
stronger responses than others will be discussed in the next chapter.
Table 3
Ten EATQ-R Self-Report Items with the Largest Standard Deviations
Scale litem content
Item #
M
SO
Fear
85
I worry about my parents dying or leaving me.
3.26
1.61
93
I feel scared when I enter a darkened room at home
3.01
1.60
I worry about getting into trouble.
3.49
1.50
I am nervous of some of the kids at school who push people
into lockers and throw your books around.
3.37
1.49
3
77
Pleasure Sensitivity
48
I like the crunching sound of autumn leaves.
3.01
1.58
70
I enjoy listening to the birds sing
3.53
1.52
High Intensity Pleasure
17
I wouldn't be afraid of skateboard or ride a bike really fast
down a steep hill.
3.09
1.55
61
I find the idea of driving a race car exciting.
3.28
1.51
3.45
1.50
3.48
1.48
Inhibitory Control
R 22
It's hard for me not to open presents before I'm supposed to.
Shyness
R 52
Note.
N
I can generally think of something to say, even with
strangers.
= 687.
R
= Reverse-coded
items.
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
55
Descriptive Statistics and Reliability of EATQ-R Self-Report Scales
Adjusted mean scale scores were obtained by averaging the raw scores
for the items defining the corresponding scales.
Descriptive statistics of the
EATQ-R self-report scales are presented in Table 4.
Table 4
Descriptive Statistics of EATQ-R Self-Report Scales among PRC Children
PRC
EATQ-R self-report
Subscale
(N
nj
=687, Mage =10.49 years)
M
SO
Skewness
Kurtosis
Activation Control
8
4.01
.63
-.84
1.02
Affiliation
8
3.76
.67
-.62
.63
Attention
7
3.77
.66
-.43
.55
Activity Level
6
3.74
.77
-.54
.18
Fear
6
3.20
.78
-.16
-.43
Frustration
9
2.66
.80
.22
-.31
Inhibitory Control
11
3.79
.62
-.33
.01
Pleasure Sensitivity
7
3.54
.86
-.48
-.10
High Intensity Pleasure
11
3.48
.61
-.08
-.16
Perceptual Sensitivity
6
3.45
.72
-.15
-.37
Shyness
7
2.51
.64
.28
-.09
Note. nj
=number of items in a subscale.
The children had the highest mean scores on the Activation Control,
Inhibitory Control, and Attention scales. In contrast, they had the lowest scores
on the Shyness and Frustration scales.
The score pattern displayed on the
individual scale was consistent with the score patterns displayed on the individual
56
items, as shown in Tables 1 and 2. The skewness and kurtosis for most scales
were high, indicating skewed distributions of the scores on such scales. Scores
on the Activation Control scale clustered more at the higher end of the
distribution as indicated by its high negative skewness and positive kurtosis
values. A similar score pattern was also found on the Affiliation, Attention, and
Activity Level scales. In contrast, Frustration and Shyness displayed an inverse
pattern in which the scores were skewed toward the lower end of the distribution
with fewer clusters, as indicated by the scales' positive skewness and negative
kurtosis values.
Scale reliability for the EATQ-R self-report was assessed by calculating
Cronhach's coefficient alpha.
Reliability estimates for 11 EATQ-R scales are
presented in Table 5.
Table 5
Coefficient Alpha for the EATQ-R Self-Report Scales by PRC and US sample
=
*American sample: N =
172, Mage 13.78 yr.
PRC sample: N 687,
Mage 10.49 yr.
Activation control
.73
.64
>.09 US
Affiliation
.76
.64
>.12 US
Attention
.65
.58
>.07 US
Activity Level
.65
.65
=PRC
Fear
.64
.48
>.16 US
Frustration
.67
.76
>.09 PRC
Inhibitory Control
.77
.67
>.10US
Pleasure Sensitivity
.77
.69
>.08 US
High Intensity Pleasure
.77
.56
>.21 US
Perceptual Sensitivity
.77
.47
>.30 US
Shyness
.80
.47
>.33 US
EATQ-R scales
=
Note. * Ellis & Rothbart (2001)
57
=
Alpha
direction
The internal consistency for 11 subscales ranged from .47 to .76, with a
mean of .60. The current study had lower alpha values on nine of the eleven
scales than Ellis and Rothbart's study using the EATQ-R self-report (2001). The
Shyness, Perceptual Sensitivity, Fear, High Intensity Pleasure, and Attention
scales yielded particularly low levels of internal consistency when administrated
to the Chinese children. In contrast, the PRC children's sample demonstrated
higher or equal alpha values in the Frustration and Activity Level in comparison
with the American sample.
Several reasons may contribute to the overall lower scale reliability of the
EATQ-R in the current study, and they will be discussed in the next chapter.
Scale reliability can often be improved by deleting items with poor item-total
correlations.
However even after deleting the items with low item-total
correlations from their individual scales, the coefficient alpha for the Fear,
Perceptual Sensitivity, Attention, and High Intensity Pleasure scales remained
poor.
Given that at least four EATQ-R subscales demonstrated inadequate
levels of reliability, it would not be appropriate to treat the subscale scores as
observed variables in factor analysis and subsequent analyses. When several
scales on a questionnaire display poor reliability, it is advised that a principal
component analysis or an exploratory factor analysis should be performed on
responses to all items on the questionnaire (Hatcher, 1994). Following Hatcher's
advice, an exploratory factor analysis (EFA) was therefore performed using all 86
individual items as variables, to determine which items tended to group together
58
empirically.
Estimated factor scores based on the factorial structure of the
Chinese sample were then generated and used in second-order EFA and
multivariate analyses.
EFA on Chinese Children's Self-Reports
First-order EFA.
Responses to 86 EATQ-R self-report items were
subjected to an exploratory factor analysis using squared multiple correlations as
prior communality estimates.
The principal axis factor method was used to
extract the factors, and this was followed by a promax (oblique) rotation. Twentysix factors displayed eigenvalues greater than 1, but the results of a scree test
suggested 7 meaningful factors, so only these 7 factors were retained for
rotation. Combined, factors 1 to 7 accounted for 48% of the total variances.
Factor pattern loadings (standardized regression coefficients), Cronbach's
alphas, and eigenvalues are presented in Table 6. In interpreting the rotated
factor pattern, an item was interpreted to load on a given factor if the factor
loading was .30 or greater, and all other factors were less than .30. Using these
criteria, items that loaded equally low across all the factors and/or cross-loaded
on multiple factors were eliminated.
Stevens (2002) recommended against
blindly using the rule of interpreting factors with loadings greater than .30, and
urged that sample sizes be taken into account. For example, the critical value for
a simple correlation at a
= .210.
=.01
(two-tailed test) for a sample size of 600 is 2(.105)
Thus, loadings greater than .210 in absolute value would be declared
statistically significant with the present PRC sample. This further assured the
use of the cut-off value of .30 in interpreting factors in the current EFA.
59
60
= Attention; FE = Fear; FR =
Frustration; HIP =High Intensity Pleasure; IC =Inhibitory Control; PE =Perceptual Sensitivity; PL
Note. N
= 687.
AC
= Activation
Control; AF
=Pleasure Sensitivity; SH =Shyness.
= Affiliation;
AT
Loadings greater than .30 are presented in bold.
Four original Activation Control items, one Attention item, and one
Inhibitory Control item loaded on the first factor, which was labeled the
"Activation Control" factor.
Five items from the original Frustration scale were
found to load on the second factor, which was subsequently labeled the
"Frustration" factor. Four original Pleasure Sensitivity items loaded on the third
factor, which was labeled the "Pleasure Sensitivity" factor.
Five High Intensity
Pleasure items and one Fear item loaded on the fourth factor, which was
subsequently labeled the "High Intensity Pleasure" factor. The fifth factor
consisted of four original Affiliation items, and was subsequently labeled the
"Affiliation" factor. Four original Shyness items loaded on the sixth factor which
was subsequently labeled the "Shyness" factor.
Two original Perceptual
Sensitivity items and one Attention item grouped together to form the seventh
factor, which was subsequently labeled the "Perceptual Sensitivity" factor.
A
Fear factor was also generated but only two items loaded on the factor. Since it
is commonly accepted that a factor should have at least three items to be
considered a factor, the Fear factor was eliminated. Items that loaded on factors
one to seven are listed in Table 7.
61
Table 7
Items Loading on Factors 1 to 7 according to the Rotated Factor Pattern Matrix in
the PRC Children's Sample
Items loading on Factor 1 Activation Control
7. AC (R)
I have a hard time finishing things on time.
82. AC (R)
I put off working on projects until right before they're due.
24. AC (R)
I do something fun for a while before starting my homework, even when I'm not
supposed to.
I tend to get in the middle of one thing, then go off and do something else.
86. AT (R)
46.IC (R)
63. AC
The more I try to stop myself from doing something I shouldn't, the more likely I
am to do it.
I finish my homework before the due date.
Items loading on Factor 2 Frustration
42. FR
It bothers me when I try to make a phone call and the line is bUsy.
73. FR
It bothers me when people are slow about getting ready for something.
79. FR
I get irritated when I have to stop doing something that I am enjoying.
91. FR
It really annoys me to wait in long lines.
101. FR
It frustrates me if people interrupt me when I'm talking.
Items loading on Factor 3 Pleasure Sensitivity
70. PL
80. PL
50. PL
28. PL
I enjoy listening to the birds sing.
I like to look at the pattern of clouds in the sky.
I like to feel a warm breeze blowing on my face.
I like the sound of words.
Items loading on Factor 4 High Intensity Pleasure
95. HIP (R)
I wouldn't want to go on the frightening rides at the fair.
53. HIP
I would not be afraid to try a risky sport, like deep-sea diving.
14. HIP (R)
Skiing fast down a steep slope sounds scary to me.
99. HIP
I wouldn't be afraid to try something like mountain climbing.
17. HIP
I wouldn't be afraid to skateboard or ride a bike really fast down a steep hill.
I get frightened riding with a person who likes to speed.
54. FE (R)
Items loading on Factor 5 Affiliation
75.
88.
51.
38.
AF
It is important to me to have close relationships with other people.
AF
I am quite a warm and friendly person.
AF
I will do most anything to help someone I care about.
AF
I would like to be able to spend time with a good friend every day.
62
Items loading on Factor 7 Shyness
76. SH
I am shy.
87. SH (R)
I am not shy.
9. SH
I feel shy with kids of the opposite sex.
6. SH
I feel shy about meeting new people.
Items loading on Factor 6 Perceptual Sensitivity
11. PE
31. PE
I notice even little changes taking place around me, like lights getting brighter in a
room.
I tend to notice little changes that other people do not notice.
67. AT
I am good at keeping track of several different things that are happening around
me.
Note. AC = Activation Control; AF = Affiliation; AT = Attention; FE = Fear; FR = Frustration; HIP =
High Intensity Pleasure; IC = Inhibitory Control; PE = Perceptual Sensitivity; PL = Pleasure
Sensitivity; SH
=Shyness. R =reverse coding.
When compared to the original EATQ-R self-report, the newly generated
Chinese version had fewer numbers of items (33 instead of 86 items) and scales
(7 instead of 11 scales). The Activation Control, Attention, and Inhibitory Control
scales were not extracted as separate scales in the Chinese children's factor
structure. Instead, the items from these three scales correlated with one another
strongly to form one factor. The Activity Level and Fear scales were not in the
PRC structure. Items of the Activity Level scale were found to load on more than
one factor, which led to its communality value going over 1 and its elimination
from the EFA.
Personal communications with Ellis confirmed that the Activity
Level scale was also dropped from Ellis and Rothbart's EFA factorial structure
due to its multiple cross-loadings on other factors.
The Fear factor was
eliminated because it was too weak to be retained as a subscale with the current
Chinese children sample.
63
Among the 7 factors that were extracted, the Affiliation, Shyness, and
Perceptual Sensitivity factors were not as strong as the Activation Control,
Frustration, Pleasure Sensitivity, and High Intensity Pleasure factors, as
indicated by their moderate internal consistency levels.
The Shyness scale
included four items but it is obvious that the two items that correlated highly in the
scale comprised one item worded in reverse ways, and the other two shyness
items did not load high on the scale. The Perceptual Sensitivity scale contained
only three items with moderate loadings.
The seven factors were expected to be correlated and the correlations
among the factors are presented in Table 8.
Table 8
Factor Correlations among the New Chinese EATQ-R Self-Report Scales (N
=
687)
Factor
F1. Activation Control
F2. Frustration
F1
F2
F3
F4
F5
F6
F7
-.510
.319
.144
.383
-.297
.201
-.266
-.104
-.033
.176
.009
.024
.492
-.092
.376
.247
-.380
.289
-.245
.482
F3. Pleasure
Sensitivity
F4. High Intensity
Pleasure
F5. Affiliation
F6. Shyness
-.157
F7. Perceptual
Sensitivity
The seven factors extracted from the item-level or first-order EFA became
the new PRC EATQ-R self-report subscales.
64
Descriptive statistics and
coefficient alpha reliability estimates for the scales are displayed in Table 9.
Reliability estimates of the 7 new subscales ranged from .59 to .75, with a mean
of .66.
When compared to the EATQ-R scale reliability reported earlier (see
Table 5), the coefficient alpha improved substantially for 5 of the 7 scales. They
were also comparable to those reported in Ellis and Rothbart's study (2001),
despite the reduction in the number of items within each scale.
In general,
except for Perceptual Sensitivity, these scales demonstrated acceptable internal
consistency and are therefore better suited for use in subsequent analyses than
are the original EATQ-R scales.
Table 9
Descriptive Statistics and Coefficient Alpha of the New Chinese EATQ-R SelfReport Scales Overall and by Gender
Total (N
=687)
Boys (n
=367)
Girls (n
=320)
Subscale
nj
M(SD)
Alpha
M(SD)
Alpha
M(SD)
Alpha
Frustration
5
2.71 (0.94)
.71
2.74 (0.97)
.71
2.67 (0.89)
.72
Activation Control
High Intensity
Pleasure
6
4.22 (0.73)
.75
4.15 (0.79)
.74
4.26 (0.75)
.75
6
3.61 (0.87)
.65
3.67 (0.87)
.64
3.50 (0.87)
.68
Pleasure Sensitivity
4
3.52 (1.07)
.70
3.33 (1.11)
.71
3.73 (0.97)
.67
Affiliation
4
4.22 (0.76)
.62
4.13 (0.81)
.63
4.32 (0.68)
.58
Perceptual sensitivity
3
3.51 (0.96)
.59
3.46 (1.00)
.58
3.56 (0.90)
.58
Shyness
4
2.08 (0.89)
.62
2.06 (0.89)
.55
2.11 (0.91)
.71
Note. nj
=number of items in a subscale.
Adjusted mean scale scores were obtained by averaging the raw scores
for the items loaded on the corresponding scales in the EFA. Chinese children
65
rated themselves high on the Activation Control and Affiliation scales, and low on
the Frustration and Shyness scales. Results regarding gender differences on the
scales will be reported in the multivariate analysis section of this chapter.
Second-order EFA.
Estimated factor scores for the seven new factors
were obtained via the first-order EFA and were used in a second-order
exploratory factor analysis, where three correlated second-order factors were
extracted. The overall factor pattern loadings are presented in Table 10.
Table 10
Parameter Estimates of the Second-Order Three-Correlated Factor Model in the
PRC Children Sample
Standardized regression coefficients
First-order factors
Affiliativeness
Frustration/Control
Surgency
Factor 5: Affiliation
.861
.066
.095
Factor 3: Pleasure Sensitivity
.750
-.208
-.261
Factor 7: Perceptual Sensitivity
.686
.159
.174
Factor 2: Frustration
.149
.966
-.040
Factor 1: Activation Control
.265
-.591
.109
Factor 4: High Intensity Pleasure
-.008
.018
.827
Factor 6: Shyness
-.020
.193
-.544
Affiliation, Perceptual Sensitivity, and Pleasure Sensitivity factors were
found to load on the first second-order or general factor, which was subsequently
labeled the "Affiliativeness" factor. The second general factor was defined by a
positive loading for Frustration and a negative loading for Activation Control, and
was subsequently labeled "Frustration/ControL"
66
High Intensity Pleasure and
Shyness were negatively correlated on the third general factor with a positive
loading for High Intensity Pleasure and a negative loading for Shyness, which
was subsequently labeled the "Surgency" factor.
Combined, second-order
factors 1 to 3 accounted for 78% of the total variances.
Figure 1 presents the factorial structure of the new Chinese EATQ-R selfreport scales in the PRC children's sample. The three circles represent the three
general temperamental factors of Affiliativeness,
Frustration/Control,
and
Surgency, which were extracted in the preceding analyses. Scales loaded on
each of the three factors are shown within the circles.
Affiliativeness is
negatively associated with Frustration/Control, suggesting that low levels of
frustration and high levels of self-control help children become more affiliated
with their peers or vice versa. Affiliativeness is also positively associated with
Surgency, suggesting that children with low levels of shyness and high levels of
adventurousness are more likely to make friends or become popular with peers.
Surgency and Frustration/Control are negatively correlated at such a low level
that this relationship remains unclear.
67
I .~90
GF1: Affiliativeness
GF3: Surgency
Affiliation (.861)
Pleasure Sensitivity
(.750)
Perceptual Sensitivity
(.686)
High Intensity Pleasure
(.827)
Shyness (-.544)
GF2:
Frustration/Control
Frustration (.966)
Activation Control
(-.591 )
Figure 1. The First-Order and Second-Order EATQ-R Factor Structure in the
PRC Children's Sample. GF1
General factor 3.
=General factor 1; GF2 =General factor 2;GF3 =
Numbers in parentheses are standardized regression
coefficients of the first-order factors (scales). Numbers in boxes are correlations
among the general factors.
Multivariate Analysis
To examine whether children's temperament was related to gender and
their status as only-child or child with siblings, multivariate analysis procedure
was used to examine the effects of these two variables on the seven dimensions
of children's temperament: Frustration, Activation Control, Affiliation, Pleasure
Sensitivity, High Intensity Pleasure, Perceptual Sensitivity, and Shyness.
One
advantage of using multivariate analysis of variance (MANOVA) was that it
enabled the alpha level to remain the same while testing multiple dependent
variables.
68
The two independent variables, gender and only child status, are both
categorical variables and each has two levels.
Therefore, a two by two
unbalanced factorial MANOVA was conducted. The initial multivariate effects for
gender, only-child status, and the interaction between the two were not
=.295, P =.956;
F (7,484) = 1.110, P =
statistically significant: Gender by Only-Child status, F (7, 484)
Gender, F (7,484)
= .704, P = .668; Only-Child status,
.356, indicating that there were no significant differences among the four groups
of children categorized by gender and only child status on the seven
temperamental subscales.
Though the results of the two-way MANOVA test
were not significant, it should be noted that this study's group sample was
unbalanced, as there was a much larger subgroup of only children and a very
small subgroup of children with siblings.
Since the non-only children group was very small in comparison to the
only-children group (52 vs. 442) in the 2-way MANOVA test, the only-child
variable was dropped from the model to further examine the effect of gender on
children's temperament.
A one-way MANOVA was re-run with gender as the
independent variable and 7 temperamental scales as the dependent variables.
The overall multivariate test was statistically significant at the .05 level, F (7, 679)
= 5.128, P < .05.
Subsequent univariate test indicated statistically significant
=6.796, P =.009, Affiliation,
F (1, 685) = 10.173, P = .001, and Pleasure Sensitivity, F(1, 685) =24.903, P =
gender differences on Activation Control, F (1, 685)
.000.
Girls scored higher than boys on all three areas, suggesting that they
might be more self-controlled and more able to enjoy low-intensity pleasures than
69
boys. Girls might also care more about relationships than boys. This will be
discussed in a greater detail in the next chapter.
Summary of Results from Children's Self-reports
As may be expected of a research instrument devised in one culture and
used in another, internal consistencies of the original EATQ-R scales were lower
for the present Chinese sample than they were for the American sample. Among
the 11 scales, the Fear, Shyness, Perceptual Sensitivity factors showed
particularly low internal consistency.
An exploratory factor analysis yielded a
factorial structure that included seven correlated first-order factors (equivalent to
the scales) and three correlated second-order factors of Affiliativeness,
Frustration/Control,
and
Surgency.
Affiliation,
Pleasure
Sensitivity,
and
Perceptual Sensitivity were found to load on the Affiliativeness factor; high
Frustration
and
low
Activation
Control
were
found
to
load
on
the
Frustration/Control factor; high High Intensity Pleasure and low Shyness loaded
on the Surgency factor.
Boys and girls differed on the Activation Control,
Affiliation, and Pleasure Sensitivity areas of temperament, with girls scoring
higher than boys. The only children and non-only children did not seem to differ
in any of the seven temperamental dimensions.
EATQ-R Parent-report
In contrast to the EATQ-R self report, the parent report has a smaller
number of temperamental scales and items (9 scales and 50 items). Out of 805
responses from the children's caretakers, 589 (73%) were mothers, and only 428
70
of the mothers had no missing values on the EATQ-R parent form items. This
sample of 428 mothers was used in the preliminary descriptive analyses, scale
reliability test, first-order and second-order exploratory factor analyses.
Demographic Information
The average age of mothers was 37 years (SO
= 3.35),
and the average
length of residence in Beijing was 28 years. Education levels for these mothers
were varied, ranging from elementary school to graduate school, with the modal
level of education as high school (including trade school) completion.
The
average family income was between 3,000 to 5,000 Chinese Yuan (about $350 $600 US) per month, with about 33% of the families reporting more than 5,000
Chinese Yuan in monthly income and 38% reporting less than 3,000 Chinese
Yuan. The demographic information is listed in Table 11.
Table 11
Demographic Information of the PRe Mothers' Sample
Demographic category
n
M
so
Min
Max
Mothers
Age (in years)
417
37.31
3.35
28
52
Level of education
416
3.71
1.10
2
6
Years in Beijing
407
28.21
13.80
1
50
Age (in years)
408
39.31
3.75
29
62
Level of education
407
3.89
1.12
2
6
Years in Beijing
398
29.23
14.11
1
50
Monthly income (Chinese Yuan)
393
4.08
1.25
1
6
Fathers
Family
71
Descriptive Statistics of the EATQ-R Items
Descriptive statistics of Chinese mothers' responses to the EATQ-R items
were obtained and are summarized in Tables 12 to 14, with Table 12 and Table
13 listing the ten highest and lowest means of the EATQ-R scores from the
Chinese mothers' sample, respectively. Table 14 lists the ten EATQ-R items with
the largest deviations.
Table 12
Ten Highest Means of EATQ-R Scores among Chinese Mothers
Item #
Item content
M(SD)
Activation control
36
Usually finishes her/his homework before it is due.
4.44 (0.72)
38
Usually gets started right away on difficult assignments.
3.96 (1.03)
Affiliation
51
If having a problem with someone, usually tries to deal with it right
away.
4.42 (0.80)
43
Wants to have close relationship with other people.
4.18 (0.82)
18
Would like to be able to spend time with a good friend every day.
4.13 (0.98)
24
Enjoys exchanging hug with people s/he likes.
4.07 (0.90)
High Intensity Pleasure
37
4
29
Likes it when something exciting and different happens at school.
4.24 (0.85)
Thinks traveling to Africa or India would be exciting and fun.
4.06(1.13)
Expresses a desire to travel to exotic places when s/he hears
about them.
3.84 (1.15)
Attention
60
Note. N
Pays close attention when someone tells her/him how to do
something.
= 428.
R
= Reverse-coded
items.
3.91 (0.81)
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
72
The item with the highest mean score was an Activation Control item (item
36) "usually finishes her/his homework before it's due." Both the high score and
the low variability of the item may be explained by the Chinese traditions of
upholding obedience and high academic expectations at schools.
Finishing
homework on time is considered a basic requirement for being a student, not "a
good student," but just "a student." The high scores on the Affiliation items reflect
the characteristics typically valued by both Chinese teachers and families.
Table 13
Ten Lowest Means of EATQ-R Scores among Chinese Mothers
Item #
Item content
Means (SO)
Shyness
44
Is shy
2.71 (1.22)
50 R
Is not shy
2.69 (1.22)
62
Feels shy about meeting new people
2.39 (1.13)
54
Likes meeting new people.
1.90 (0.95)
Frustration
57
Hates it when people don't agree with her/him.
2.69 (1.01)
31
Gets irritated when I will not take her/him someplace s/he
wants to go.
2.68(1.10)
58
Gets very frustrated when s/he makes a mistake in her/his
school work.
2.41 (1.08)
Fear
61
2.32(1.21)
Is nervous being home alone.
Inhibitory Control
8R
Opens presents before s/he is supposed to.
2.24 (1.18)
6R
Has a hard time waiting her/his turn to speak when excited
1.81 (0.94)
Note. N
= 428.
R
= Reverse-coded
items.
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
73
Low scores were mainly obtained on items of the Shyness and Frustration
scales, and this pattern was consistent with the Chinese children's data earlier.
Low scores on Frustration were expected because character education in China
emphasizes the cultivation of abilities in children to manage frustration. But low
scores on Shyness were not expected, and this will be discussed in the next
chapter.
Table 14
Ten EATQ-R Parent-Report Items with the Largest Standard Deviations
Item content
Item #
M
SD
High Intensity Pleasure
34
Would like driving a racing car
2.79
1.47
56 R
Wouldn't want to go on the frightening rides at the fair.
3.18
1.37
Would be frightened by the thought of skiing fast down a steep
slope.
3.03
1.33
Wouldn't be afraid to try a risky sport like deep sea diving.
3.12
1.24
9R
28
Shyness
27 R
Can generally think of something to say, even with strangers.
3.14
1.32
44
Is shy
2.72
1.23
Fear
55
Feels scared when entering a darkened room at night.
3.43
1.31
48
Is afraid of the idea of never seeing me again.
3.65
1.24
53
Doesn't enjoy playing softball or baseball because s/he is
afraid of the ball.
2.83
1.24
3.26
1.24
Activation Control
14 R
Note. N
Usually does something fun for a while before starting her/his
homework, even though s/he is not supposed to.
= 428.
R
= Reverse-coded
items.
After reverse coding, high (low) mean scores
indicated high (low) scores on the individual scales rather than the specific items.
74
Large deviation items were mostly found in the High Intensity Pleasure
and Fear scales, and are consistent with the variability of such traits which are
usually observed among children.
Descriptive Statistics and Reliability of EATQ-R Parent-Report Scales
Adjusted mean scale scores were obtained by averaging the raw scores
for the items defining corresponding scales. The mothers' highest scores were in
the Affiliation and Activation Control scales, and their lowest scores were in the
Shyness and Frustration scales. The score patterns observed were consistent
with those displayed on the individual items.
Table 15
Descriptive Statistics and Coefficient Alpha for EATQ-R Parent-report Scales
*US sample (N = 62)
PRC sample (N = 428)
Alpha direction
EATQ-R scales
Alpha
Alpha
M(SD)
Activation control
.66
.75
3.63 (0.65)
>.08 PRC
Affiliation
.82
.65
4.01 (0.57)
>.18 US
Attention
.65
.65
3.52 (0.60)
>.01 US
Fear
.69
.50
3.11 (0.66)
>.21 US
Frustration
.74
.70
2.80 (0.67)
>.04 US
Inhibitory Control
.86
.35
2.85 (0.56)
>.51 US
Shyness
.72
.65
2.58 (0.76)
>.06 US
High Intensity Pleasure
.70
.61
3.51 (0.60)
>.10 US
Note. * Ellis & Rothbart (2001).
The internal consistency for 8 EATQ-R parent subscales ranged from .35
to .75, with a mean of .61, for the Chinese mothers' sample (See Table 15).
75
When compared to the coefficient alpha obtained in Ellis and Rothbart's study, all
of the EATQ-R parent-report scales showed lower internal consistency for the
PRC sample than for the American sample except for the Activation Control
scale (Ellis & Rothbart, 2001). The Inhibitory Control and Fear scales yielded
particularly low levels of internal consistency for the Chinese mothers, with
coefficient alpha of .35 and .50, respectively.
EFA on Chinese Mothers' Reports
First-order EFA. A total of 428 mothers' responses to 50 EATQ-R parentreport items were subjected to an exploratory factor analysis using squared
multiple correlations as prior communality estimates. The principal axis factor
method was used to extract the factors, and this was followed by a promax
(oblique) rotation. Fifteen factors had eigenvalues greater than 1 and could have
been extracted, but the result of a scree test suggested 6 meaningful factors, so
only these factors were retained for rotation. Combined, factors 1 to 6 accounted
for 50% of the total variances.
Questionnaire items and corresponding factor pattern loadings (in bold)
are presented in Table 16 along with the factors' (or subscales') eigenvalues and
coefficient alpha. For a sample size of 400, the critical value for a statistically
significant loading at a
= .01
(two-tailed) was .26, slightly lower than the .30 cut-
off value. Adopting 1.301 as the cut-off factor loading value, items that loaded
equally low across all the factors and/or cross-loaded on multiple factors were
eliminated.
76
Table 16
Factor Pattern Loadings (Standardized Regression Coefficients) of EATQ-R
Parent-Report Items in the PRC Mothers Sample
Standardized regression coefficients
EATQ-R
parent-report
items
AC3
AC46
AT35
AC14
AT15
F1
F2
F4
F3
Activation
Outgoing ness Affiliation Frustration
Control
.739
.024
-.109
.024
.690
-.041
-.001
-.026
.680
-.147
.042
-.029
.652
.067
-.155
-.051
.591
.026
.123
.065
F5
Fearlessness
F6
Shyness
-.016
.044
-.003
-.020
.012
-.013
.067
.040
.063
.006
.022
AT49
IC59
.581
.576
-.058
.042
-.063
.075
.064
.095
AC36
.558
.131
IC23
.521
-.182
.001
-.116
.021
.054
.031
-.036
-.012
-.148
-.108
HIP37
HIP16
.130
.015
.671
.105
.035
-.006
-.079
.029
-.144
-.071
.033
.018
AF24
.031
HIP4
IC6
-.029
-.007
.398
.215
.063
.033
-.088
-.386
-.042
HIP40
-.079
.363
-.027
-.018
-.158
-.225
AF43
.046
-.015
-.054
.103
.793
-.049
-.042
-.069
-.008
-.618
.525
.050
-.058
.055
-.025
-.051
-.100
.512
-.086
.084
.025
-.060
-.014
-.051
-.073
-.035
.676
.066
.019
-.016
.077
.051
.808
.666
.408
SH54
AF51
-.019
AF18
-.043
FR31
FR45
.037
.038
-.120
-.033
-.051
-.092
-.062
.046
.167
-.093
.518
.349
.116
.034
.016
.025
.043
.078
-.138
FR57
FR21
HIP56
HIP9
.056
-.029
-.046
.132
.768
.685
.067
.019
.086
.055
.063
-.028
.070
.044
.167
FE53
-.124
.147
.025
.034
FE55
-.085
.010
.107
.490
-.445
.137
-.004
-.435
HIP28
SH44
-.011
-.030
.147
.047
.054
.023
.043
-.025
-.057
SH62
SH50
-.012
-.009
.027
-.110
.009
.001
.056
-.065
-.018
-.017
.559
5.05
3.67
.66
2.40
1.50
1.41
1.18
.70
.71
.61
.69
Eigenvalue
Cronbach's a
.85
77
.395
.109
.593
Similar to the results of the children's EFA, the original scales of Activation
Control, Attention, and Inhibitory Control could not be extracted as individual
factors.
Instead, items from the three scales (four Activation Control, three
Attention, and two Inhibitory Control items) were found to group together and
load on the first factor, which was subsequently named the "Activation Control"
factor. The second factor was primarily defined by positive loadings for four High
Intensity Pleasure items along with one Affiliation and Inhibitory Control item, and
was subsequently named "Outgoingness." Three original Affiliation items and
one Shyness item were found to load on the third factor, which was subsequently
named the "Affiliation" factor. Four original Frustration items loaded on the fourth
factor, which was subsequently labeled the "Frustration" factor. The fifth factor
was primarily defined by positive loadings for three original High Intensity
Pleasure items and negative loadings for two Fear items.
Because the three
High Intensity Pleasure items contained words like "not be afraid of' or "not be
frightened," the factor seemed to be about fearlessness. Hence, the fifth factor
was subsequently named the "Fearlessness" factor.
Three original Shyness
items loaded on the sixth factor which was subsequently labeled the "Shyness"
factor. Items loading on the new factors one to six are listed in Table 17.
78
Table 17
Items Loading on Factors 1 to 6 according to the Rotated Factor Pattern Matrix in
the PRe Mothers Sample
Items loading on Factor 1 Activation Control
35. AT (R)
Has a difficult time tuning out background noise and concentrating when trying
to study.
46. AC (R)
Usually puts off working on a project until it is due.
3. AC (R)
Has a hard time finishing things on time.
14. AC (R)
Usually does something fun for a while before starting her/his homework, even
though s/he is not supposed to.
49. AT (R)
Is often in the middle of doing one thing and then goes off to do something else
without finishing it.
23. IC (R)
Is more likely to do something s/he shouldn't do the more s/he tries to stop
her/himself.
15. AT
Finds it easy to really concentrate on a problem.
59.IC
Is usually able to stick with his/her plans and goals.
36. AC
Usually finishes her/his homework before it's due.
Items loading on Factor 2 Outgoingness (split from original High Intensity Pleasure)
37. HIP
Likes it when something exciting and different happens at school.
16. HIP
Thinks it would be exciting to move to a new city.
24. AF
Enjoys being physically close with people s/he likes.
40. HIP
Is energized by being in large crowds of people.
6.IC
Has a hard time waiting his/her turn to speak when excited.
4. HIP
Thinks traveling to Africa or India would be exciting and fun.
Items loading on Factor 3 Affiliation
43. AF
Wants to have close relationships with other people.
54. SH
Likes meeting new people.
51. AF
Is quite a warm and friendly person.
18. AF
Would like to be able to spend time with a good friend every day.
79
Items loading on Factor 4 Frustration
31. FR
Gets irritated when I will not take her/him someplace s/he wants to go.
45. FR
Gets irritated when s/he has to stop doing something s/he is enjoying.
57. FR
Hates it when people don't agree with him/her.
21. FR
Gets very irritated when someone criticizes her/him.
Items loading on Factor 5 Fearlessness (split from original High Intensity Pleasure)
56. HIP (R)
9. HIP (R)
Wouldn't want to go on the frightening rides at the fair.
Would be frightened by the thought of skiing fast down a steep slope.
53. FE (R)
Doesn't enjoy playing softball or baseball because s/he is afraid of the ball.
55. FE (R)
Feels scared when entering a darkened room at night.
28. HIP
Wouldn't be afraid to try a risky sport like deep sea diving.
Items loading on Factor 6 Shyness
44. SH
Is shy.
62. SH
Feels shy about meeting new people.
50. SH (R)
Is not shy.
Except for the Activation Control, Inhibitory Control, and Attention scales,
which were combined into one Activation Control scale, most of the original
EATQ-R parental scales were retained in the new Chinese EFA structure after
eliminating items that either loaded poorly or cross-loaded on more than one
factor. As with the Chinese children's first-order EFA, the Fear factor failed to be
extracted due to too few items within the factor.
But uniquely in the PRC
mothers' sample, the original High Intensity Pleasure scale split into two factors Factor 2 and 5, one was named Outgoingness (Factor 2), while the other was
labeled Fearlessness. Correlations among the six factors are presented in Table
18.
80
Table 18
Factor Correlations among the New Chinese EATQ-R Parent-Report Scales
Factor
F1
F1. Activation Control
F2
F3
F4
F5
F6
-.133
.099
-.494
.162
-.099
.522
.290
.033
-.160
-.077
.177
-.274
-.293
.278
F2. Outgoingness
F3. Affiliation
F4. Frustration
F5. Fearlessness
-.385
F6. Shyness
The six factors extracted from the first-order EFA became the new
Chinese EATQ-R parent-report scales.
Adjusted mean scale scores were
obtained by averaging the raw scores for the items defining corresponding
scales.
Descriptive statistics and coefficient alpha reliability estimates for the
scales are displayed in Table 19. Reliability estimates of the 6 new subscales
ranged from .61 to .85, with a mean of .70. They were higher than the original
EATQ-R scale reliability reported earlier for Chinese mothers (see Table 15), and
comparable to the coefficient alpha reported in Ellis and Rothbart's study (2001)
despite the reduction in the number of items within each scale.
Chinese mothers rated their children high on the Affiliation, Outgoingness,
and Activation Control scales, and low on the Frustration and Shyness scales.
This pattern was basically consistent with that of the data collected from the
Chinese children's sample except for a higher rating on a new scale labeled
Outgoingness.
Results regarding gender differences on these scales will be
reported in the multivariate analysis section of this chapter.
8!
Table 19
Descriptive Statistics and Coefficient Alpha of the New EATQ-R Parent-Report
Scales Overall and by Children's Gender
Subscale
nj
Total
428)
(N
=
Mothers with boys
Mothers with girls
(n=212)
(n=216)
M(SD)
a
M(SD)
a
M(SD)
a
9
3.52 (.71)
.85
3.34 (.73)
.85
3.69 (.64)
.83
Outgoing ness
6
3.97 (.63)
.66
3.88 (.63)
.62
4.07 (.62)
.69
Affiliation
4
4.21 (.64)
.70
4.17(.67)
.70
4.24 (.62)
.68
Frustration
4
2.89 (.77)
.71
2.97 (.76)
.69
2.82 (.77)
.72
Fearlessness
5
3.03 (.82)
.61
3.03 (.82)
.62
3.04 (.82)
.61
Shyness
3
2.61 (.94)
.69
2.65 (.93)
.64
2.56 (.95)
.74
Activation
Control
Note. nj =number of items in each scale.
Second-order EFA. Estimated factor scores for the six new factors were
obtained via the first-order EFA and then used in a second-order exploratory
factor analysis, where three correlated second-order factors were extracted. The
Outgoing ness and Affiliation scales were found to load on the first second-order
factor, which might correspond to a general factor that can be labeled
"Affiliativeness." The second general factor was defined by a positive loading for
Frustration and negative loading for Activation Control, and was subsequently
labeled "Frustration/ControL" The third general factor was defined by a positive
loading for Shyness and a negative loading for Fearlessness, and was
subsequently labeled the "Shyness" factor. Combined, second-order factors 1 to
3 accounted for 78% of the total variances. Factor pattern loadings for the three
general factors are presented in Table 20.
82
Table 20
Parameter Estimates of the Second-Order Three-correlated Factor Model in the
PRC Mothers Sample
Standardized regression coefficients
New EATQ-R parent-report
scales
Affiliativeness
Frustration/Control
Shyness
F2. Outgoing ness
.888
.161
.015
F3. Affiliation
.764
-.233
-.058
F4. Frustration
.077
.839
.141
F1. Activation Control
.112
-.736
.137
F6. Shyness
-.026
-.130
.895
F5. Fearlessness
.015
-.110
-.554
Figure 2 presents the factorial structure of the new Chinese EATQ-R
parent-report scales in the PRC sample. The three circles represent the three
general temperamental factors
of Affiliativeness,
Frustration/Control,
and
Shyness, which were extracted from the preceding analyses. Scales loaded on
each of the three factors are indicated within the circles. The general factor of
Affiliativeness is negatively associated with Shyness, indicating that children who
are outgoing and enjoy being with friends are less likely to be shy or vice versa.
Frustration/Control is found to be positively associated with Shyness, indicating
that children with high level of shyness are also likely to have high level of
frustration or vice versa.
correlated
at such
Affiliativeness and Frustration/Control are positively
a low level that this
83
relationship
remains
unclear.
I
-.265
GF1: Affiliativeness
GF3: Shyness
Outgoing ness
(.888)
Affiliation (.764)
Shyness (.895)
Fearlessness (-.554)
GF2:
Frustration/Control
Frustration (.839)
Activation Control
(-736)
Figure 2. The First-order and Second-order EATQ-R Factor Structure in the PRC
Mothers' Sample. GF1
General factor 3.
= General
factor 1; GF2
= General
factor 2; GF3
=
Numbers in parentheses are standardized regression
coefficients of the first-order factors (scales). Numbers in boxes are correlations
among the general factors.
Multivariate analysis
To examine whether parents' perception of their children's temperament
was related to children's gender and their only-child status, a two-way factorial
MANOVA was used to examine the effects of two independent variables of
gender and only-child status on the six dimensions of children's temperament.
The initial multivariate test found significant effects of gender and the interaction.
These results are detailed in the following order: effects of gender and onlychild/non-only child interaction, effects of gender difference, and effects of
differences between only-children and those with siblings.
84
Interaction between gender and only-child status.
statistically significant at the a level of .05, F (6, 576)
Multivariate test was
=4.132, P =.000, indicating
that the effect of the only child status on temperament is different for boys and
girls.
Subsequent univariate analysis revealed that this interaction effect was
significant on Shyness at the .05 level, F (1, 581)
= 14.329, P = .000.
When
holding children's gender constant, boys without siblings were reported by
mothers to be less shy than boys with siblings. However, girls without siblings
were more shy than girls with siblings. When holding children's only-child status
constant, boys without siblings did not differ from girls without siblings on
Shyness. But boys with siblings were reported by mothers to be more shy than
girls with siblings.
Means and standard deviations of Shyness scores are
presented in Table 21 for the four groups of children: only-child boys, only-child
girls, boys with siblings, and girls with siblings. Among the four groups, the boys
with siblings scored the highest on Shyness, followed by only children, and then
the girls with siblings.
This interesting pattern of finding will be discussed in
greater details in the next chapter.
It should be noted that the four groups
categorized by gender and only-child status were unbalanced, and this result
should be taken with caution because only children greatly outnumbered children
with siblings in the sample tested.
85
Table 21
Descriptive Statistics on Shyness for the 4 Groups of Children Categorized by
Gender and Only-Child Status
Child gender
M
SO
n
Only
2.60
.91
251
Non-only
3.00
.75
40
Only
2.66
.94
250
Non-only
2.23
.97
44
Only child or not
Boys
Girls
Gender differences.
The initial multivariate test was significant at .05
level, indicating that boys differed from girls on certain dimensions of
temperament, F (6, 576)
= 5.706,
P
= .000.
Subsequent univariate analysis
indicated statistically significant gender differences in Activation Control, F (1,
581)
=8.517, P = .004 and Outgoingness, F (1, 581) =9.508, P = .002 at the a
level of .05, with boys being rated lower on both these scales than girls.
Differences between only children and those with siblings.
Multivariate
test did not show any differences between only children and those with siblings
on six dimensions of temperament, F (6, 576)
=2.026, P =.060.
Again, it should
be noted that the only children group and non-only children group were
unbalanced in the current study's sample.
Parental Attitudes on Temperament and Other Factors
Following the EATQ-R parent report, parents answered a questionnaire
regarding the importance of each of the following factors on the development of
86
their children's future personalities. Means and standard deviations of Chinese
mothers' ratings are presented in Table 22.
Table 22
Mothers' Views on the Importance of Factors Shaping Children's Personalities
N
M
SD
Skewness
Temperament
571
3.86
.86
-.476
-.043
Family environment
578
4.54
.63
-1.599
4.167
Parental discipline
579
4.63
.57
-1.555
3.253
School education
579
4.72
.54
-2.163
6.332
Traditional culture & values
578
3.79
.81
-.290
-.179
Values of modern society
573
3.83
.86
-.526
.334
Father's hereditary traits
579
3.48
.94
-.172
-.436
Mother's hereditary traits
576
3.50
.94
-.182
-.452
Children's experiences
577
4.15
.77
-.631
-.055
Peers influences
577
3.80
.91
-.459
-.029
Factors influencing personality
Kurtosis
According to Chinese mothers, school education, parental discipline,
family environment, and children's own experiences were among the most
influential factors shaping children's future personalities.
Mean scores on the
importance of family environment, parents' discipline, and school education were
not only high but also skewed toward the higher end of the distribution with more
clusters, indicating that Chinese mothers' opinions on these factors were quite
strong and similar.
In contrast to high scores on environment, discipline and
education, father's and mother's hereditary traits scored the lowest as an
influence in future children's personalities.
87
Children's own temperament was
rated by mothers as somewhat important, followed by the influences of the
values of modern society, peers, and traditional cultural values.
Summary of Results from Mothers' Reports
Consistent with children's self-rating patterns, mothers rated their children
high on Affiliation and Activation Control and low on Frustration and Shyness.
The exploratory factor analysis of mothers' responses yielded a factorial structure
that included six correlated first-order factors: Activation Control, Frustration,
Outgoingness, Affiliation, Fearlessness, and Shyness.
The second-order EFA
based on the 6 estimated factor scores generated three higher-order factors:
Surgency, Frustration/Control, and Shyness. Outgoingness and Affiliation loaded
positively on the Surgency factor. Frustration and Activation Control loaded on
the Frustration/Control factor, with positive loading for Frustration and negative
loading for Activation Control. Fearlessness and Shyness correlated negatively
on the Shyness factor.
Boys were rated by their mothers to be lower than girls on Activation
Control and Outgoingness. Boys with siblings were rated by their mothers to be
more shy than girls with siblings, but boys and girls did not differ on Shyness if
they were both only children.
Chinese parents consider education and training more important in
shaping their children's personalities than children's individual temperament and
hereditary parental traits.
88
Comparison of the Self- and Mothers' Reports
Though the numbers of factors (scales) extracted from children's and
mothers' data were different, the following scales were shared across both
samples: Activation Control, Frustration, Affiliation, Shyness, and High Intensity
Pleasure. Convergence between self- and parent-reports was assessed using a
series of Pearson's correlations. Agreements between self- and mother-report
scores in general and by gender are shown in Table 23.
Table 23
Convergence between Chinese Children and Mothers on Shared Temperamental
Dimensions
Activation
Control
1High Intensity Pleasure
Frustration
Affiliation
Shyness
Fearlessness Outgoing ness
Boys and
mothers
(n
.37**
267)
=
(n
.23**
260)
=
(n
.17**
265)
=
(n
.22**
270)
=
(n
.29**
260)
=
(n
.093
260)
Girls and
mothers
(n
.38**
264)
(n
.17**
270)
=
(n
.17**
270)
=
(n
.21**
270)
=
(n
.38**
258)
(n
.056
258)
Children and
mothers
.39**
(n=531)
(n
.20**
530)
(n
.18**
535)
(n
.21**
540)
.34**
(n=518)
=
=
=
=
=
=
=
.061
(n=518)
Note. 1Children's High Intensity Pleasure scores were correlated with parents' Fearlessness and
Outgoingness scores because the latter two scales split from the original High Intensity Pleasure
scale in Chinese mothers' EFA.
** Correlation is significant at the .01 level (2-tailed).
As shown in Table 23, even though the Fearlessness and Outgoingness
scales split from the original EATQ-R parent-report High Intensity Pleasure scale,
89
Fearlessness demonstrated much higher correlations with children's High
Intensity Pleasure scale than Outgoingness did.
Excluding the pair of
Outgoingness and High Intensity Pleasure, convergence for boys and their
mothers ranged from r
= .17
to r
= .37
and agreement was significant for all
scales. Convergence for girls and their mothers ranged from r
= .17 to r = .38
and was statistically significant for all scales. With gender combined, agreement
was statistically significant for all scales, ranging from r =.18 to r =.39.
When comparing Chinese children's factorial structure to their mothers'
factorial structure, the following common components of children's temperament
could be seen in both structures and they were Activation Control, Frustration,
Affiliation, High Intensity Pleasure, and Shyness. Both children and their mothers
demonstrated similar rating patterns on the above dimensions, with high scores
on Activation Control and Affiliation and low scores on Frustration and Shyness.
Gender differences were found in both children's and mothers' reports.
Girls were considered by themselves and mothers to be more self-controlled than
boys. However, while girls rated themselves higher than boys on Affiliation and
Pleasure Sensitivity scales, the mothers' ratings did not indicate such
differences. Mothers also considered girls to be higher than boys on
Outgoingness, a new scale emerged only in the Chinese mothers' factor
structure. An interaction effect between gender and only-child status was found
in mothers' reports but not in the children's self-rating data.
This chapter presented preliminary results only; further discussions of the
implications and limitations of these results will be given in the next chapter.
90
CHAPTER 4
DISCUSSION
This study has been led by two major concerns related to Chinese
children's temperament: First, whether or not the EATQ-R, a temperament
instrument created for use in America, can retain its original construct validity
when used on a Chinese sample in PRC.
To address this question, the
instrument's psychometric properties and its factorial structure were analyzed
using both Chinese children's self-reports and their mothers' reports. Second,
the study explored general temperamental patterns of early Chinese adolescents
living in Beijing, including effects of gender differences, variations in only-child
status, and areas of child-parent agreement and disagreement.
The first question will be discussed with regard to why the scale
reliabilities are generally lower in the current study than in Ellis and Rothbart's
study, and why the factorial structures differed in the two studies. The second
question will be discussed in terms of how Chinese culture influences the
temperamental patterns and gender differences displayed in the PRC sample.
Limitations and problems with regard to each of these questions are also
included in the discussion.
Validity Limitations of the EATQ-R in PRC Sample
Results of the current study suggest that EATQ-R has limited reliability
and validity when it is imported and applied to a PRC sample of children and
91
mothers. This is indicated by the lower scale reliability and different first-order
and second-order factorial structures obtained in the PRC sample as compared
to the US sample (Ellis & Rothbart, 2001).
As may be expected of a research instrument devised in one culture and
used in another, scale reliability was lower for Chinese children and mothers on
most of the EATQ-R self-report and parent-report scales than it was for the
corresponding American
samples.
Among
them,
the
Fear,
Perceptual
Sensitivity, Shyness for the children, and the Fear and Inhibitory Control scales
for the parents had particularly low internal consistency with coefficient alpha
levels lower than .50 (see Table 5 in Chapter 3).
Several factors may have contributed to the less satisfactory scale
reliability values found in the current Chinese samples. First, the generally lower
scale reliabilities might be partly due to the limited applicability of some items to
the Chinese culture and lives of Chinese children.
For example, one of the
Perceptual Sensitivity items, "I don't really notice the color of people's eyes," had
little relevance to most Chinese children because almost all Chinese people have
"black" (brown) eyes.
Items like this did not tap into Chinese children's
experiences of perceptual sensitivity and affected the internal consistency of the
scale.
This item demonstrated minimum correlation (r = .06) with the other
Perceptual Sensitivity items, and removing it helped the coefficient alpha improve
slightly from .47 to .52.
Similarly, item 54 in the Fear scale, "I get scared riding with someone who
likes to speed," does not apply to the reality of life in Beijing. In Beijing, no speed
92
limits are observed on roads, which are usually congested with cars, buses,
bicycles, and pedestrians. Only newcomers to Beijing are likely to be
unaccustomed to the lack of enforcement of traffic regulations and rules,
including speeding in this city.
Also, for most Chinese people, cars are
considered as a symbol of wealth and status more than a necessity. Learners
are required to go through 3-month training at driving schools to be eligible for
getting driver licenses. Therefore, there is a sense of pride for someone to be
able to own and drive a car in Beijing. Likewise, the pride felt by a child about
riding in a car may overshadow any fears about the dangers of speeding. As
shown in Table 3, Chinese children had one of the ten lowest scores on this item,
indicating a low level of fear experienced by Chinese children when riding in
speeding cars. Within the context of living in Beijing, the low score on this item is
likely to indicate that Chinese children enjoy riding cars, without much reference
to the level of fear associated with speeding.
Not surprisingly, this item
demonstrated a poor item-total correlation with the other items in the Fear scale,
suggesting it was not measuring the same construct of Fear that was measured
by the other items in the same scale.
Examples like the above could be found in almost every scale, some were
obvious while others involved subtle differences between culturally specific
realities
and
values.
Given
the
differences
in
social,
economic,
and
environmental conditions in the two cultures, such applicability issues were no
surprise and aligned with the literature reviewed in Chapter 1.
93
A second factor behind the generally lower scale reliabilities could be the
construct inequality. In the Activation Control scale, the item "If my friends are
mad at me I try to stay away from them" stood out with a low item-total
correlation (r
= .11) with
the other Activation Control items.
Since Activation
Control is defined as the capacity to perform an action when there is a strong
tendency to avoid it in the EATQ-R, a high score on this item will indicate a low
level of Activation Control in a child because s/he chooses to avoid the situation.
But to Chinese children and parents, the item is likely to have been interpreted as
being about how to handle conflicts in interrelationships rather than about
individual differences in activation control. This was indicated by the tendency of
the item grouping together with some Affiliation items in the PRC mothers'
sample in the item-level EFA. In comparison to other Activation Control items,
for Chinese children and mothers, the situation posed in this item had little to do
with a child's activation control and more to do with enacting strategies to resolve
the matter. A similar problem emerged in the High Intensity Pleasure scale and
led to the splitting of the original scale into two separate scales of Outgoingness
and Fearlessness for the Chinese mothers. Even though High Intensity Pleasure
is defined as the pleasure derived from activities involving high intensity or
novelty in the EATQ-R, Chinese mothers seem to differentiate the items
concerning novelty, such as travel, from the items concerning risky sports. This
indicates that these items failed to measure the same construct.
Translations of the questions from English to Chinese itself may also have
contributed to the overall lower scale reliabilities in the Chinese sample.
94
As
discussed in Chapter 2, the most common problem with the translated version is
the lack or confusing forms of contextual information for Chinese audiences. For
example, the Fear item "I worry about getting into trouble" for children and "My
child worries about getting into trouble" for parents could be open to different
interpretations. The word "trouble" in Chinese could refer to situations ranging
from getting scolded for doing bad things to being burdened with extra
responsibilities or hassles - something not necessarily bad.
This item also
demonstrated low item-total correlation with other items in the Fear scale.
Finally, age and sample size differences between the PRC and US
samples may also have led to disparities in scale reliabilities between the two
samples. The average age of respondents in Ellis and Rothbart's sample in their
original EATQ-R study was 13.78 years old, 3 years older than the average age
of Chinese children in the current study's sample (Mage
= 10.49
years).
The
younger Chinese children may have had greater difficulty understanding some of
the concepts and terms in the questionnaire. For example, several children in
the fourth grade asked the researcher what "opposite sex" means in the item "I
feel shy with kids of the opposite sex" during the data collection processes. In a
study using a younger sample of American students (Mage
= 12.31
years), Ellis
(2002) reported lower coefficient alpha levels on all of the EATQ-R scales when
compared to the original EATQ-R study (Ellis & Rothbart, 2001).
In the later
study, coefficient alphas for self-reported EATQ-R scales ranged from .55 to .78,
comparable to what this current study found: .47 to .76.
In general, some EATQ-R scales in particular, the Fear, Perceptual
95
Sensitivity, Shyness, Attention, and High Intensity Pleasure scales for the
children and the Fear, Inhibitory Control, and High Intensity Pleasure scales for
the parents, displayed less satisfactory reliability in assessing Chinese children.
This should be noted if one is interested in assessing any of these constructs
with a Chinese sample that is comparable to that of the current study.
One of the main goals of the current study is to provide evidence of the
reliability and validity of the EATQ-R dimensions when the measure is applied to
a PRC sample.
Results of the first-order factor analysis of Chinese children's
responses suggest that some EATQ-R dimensions were not as meaningful for
Chinese children as for their American counterparts.
Differences from the
American sample were seen in that fear could not be measured reliably for
Chinese children; the three "Effortful Control" scales, Activation Control,
Attention, and Inhibitory Control, collapsed into one scale.
Even though other
EATQ-R scales were replicated in the current PRC sample of children, there
were substantial changes on the items comprising these scales, including item
shifting and deletion of poor items.
Dimensions derived from the first-order factor analysis of Chinese
mothers' reports bear more similarity to the results of Chinese children than to
the dimensions in the original EATQ-R parent measure. Once again, validity of
the Fear scale was challenged among Chinese mothers; the Activation Control,
Inhibitory Control, and Attention scales also collapsed into one scale, providing
further evidence that the constructs underlying these scales were difficult to keep
distinct among both Chinese children and mothers. But High Intensity Pleasure
96
split into two scales for Chinese mothers only, suggesting that Chinese mothers
tend to distinguish children's desires for exploration from their risk-taking
behaviors, especially those concerning physically dangerous and risky behaviors,
such as riding a bike down a steep slope. Such fine discriminations were likely to
have been beyond the grasp of Chinese children.
To the children, items
concerning travel simply did not fit in with the other High Intensity Pleasure items,
indicated by the poor item-total correlations.
The Fear scale appeared to be a particularly problematic scale with low
internal consistency, poor item-total correlations, and poor inter-item correlations.
There are 6 items in this scale: Two about fears of separation from family (items
57 and 85), one about fear of riding in speeding cars (item 54), one about fear of
bullies at school (item 77), one about fear of the dark (item 93), and another
about "getting into trouble" (item 3). These items are quite heterogeneous in
terms of the objects of fear, and tend to elicit inconsistent responses. In addition,
responses to some of the items are context-dependent, such as, item 85, which
states "I worry about my parent(s) leaving me or I will never see them again."
For a child whose parents are sick or are divorced, this might be something he or
she often worries about. But for a child with a loving family and healthy parents,
this question is less likely to be of concern. Items in the Fear scale also have
problems in several aspects that have been discussed earlier: applicability,
construct inequality, and translation. All these factors may explain why a fear
scale was not generated from the item-level EFA based on both Chinese
children's and mothers' responses. Fear is a measurable dimension of
97
temperament, but the items developed to assess it, as in the EATQ-R Fear scale,
are not adequate for Chinese early adolescents.
Activation Control, Attention, and Inhibitory Control were not extracted as
separate meaningful factors in the current study. Instead, they overlapped with
one another and collapsed into one factor. The items in the Activation Control,
Attention, and Inhibitory Control scales have much in common; they are about
making efforts, learning to become self-disciplined, following through with plans,
paying attention, and suppressing impulsive behaviors.
Chinese parents and
teachers might consider these temperamental traits as one category - good
study behavior and habits. For example, Chinese mothers put the Attention item
"(My child) has a hard time tuning out background noise and concentrating when
trying to study" along with the Inhibitory Control item "(My child) is more likely to
do something he or she shouldn't do the more he or she tries to stop herself or
himself' into the same category with the Activation Control item "(My child) has
as a hard time finishing things on time."
Despite the theoretical distinctions
assumed by the developers of the questionnaire among the three constructs,
Chinese children and parents found it difficult to discriminate among them
because the items measuring these constructs overlapped in both meaning and
purposes, at least from the perspective of Chinese parents and children.
It should be noted that both the EATQ-R scales (Ellis & Rothbart, 2001)
and its previous version, the EATQ scales (Capaldi & Rothbart, 1992), were not
examined with item-level exploratory factor analysis, hence the distinction among
the three constructs lacked psychometric support. Thus the Activation Control,
98
Attention, and Inhibitory Control scales remain, at best, as only theoretically
distinct constructs. It also does not exclude the possibility for items from one of
the three scales to cross-load on other scales and or for a complete collapse of
the three scales, as found in this study's PRC sample.
If there is a lack of
evidence to support the actual existence of the three dimensions in American
sample, then it is premature to accept the existence of differences on these
dimensions between the American and PRC samples.
Exploratory factor
analyses conducted at the EATQ-R item level with a larger American sample will
help to clarify the disparity found between this study and the previous EATQ-R
study (Ellis & Rothbart, 2001).
The higher-order factors identified in the current study are also different
from those identified in the factor analysis of the EATQ-R scales in the US
sample (Ellis & Rothbart, 2001) despite some consistencies in the factor
solutions in both studies. The structure of temperament in the US sample (based
on self-report data) is presented in Figure 3 alongside the PRC structure derived
from the Chinese children's self-reports.
As in the US structure, Affiliation,
Pleasure Sensitivity, and Perceptual Sensitivity all loaded on the same factor,
Affiliativeness, for Chinese children; the broad factor, Surgency, was also defined
similarly across the two structures except for the absence of the Fear scale in the
PRC structure. However, the broad factor, Frustration/Control, in the PRC
structure showed major differences from the broad factor, Effortful Control, in the
US structure. Besides the Activation Control component (which is a combination
of the original Activation Control, Attention, and Inhibitory Control scales),
99
Frustration also loaded on this factor for Chinese children, whereas in the US
structure, it formed a separate negative temperamental factor "Negative
Affectivity" along with an "Irritability" scale 1 . Consistent with Chinese children, a
similar Frustration/Control factor emerged in the factorial structure of Chinese
mothers.
The differences on the ways Frustration loaded in the PRC and US
structures are important because it suggests that Frustration is defined or
experienced differently by Chinese and American adolescents. Frustration is an
individual characteristic in that it is a negative affect related to the interruptions of
ongoing tasks or the blockings of goals. In American society where pursuits of
individual interests are respected and encouraged, expressions of frustration
may be more expected and acceptable when things do not go according to
individual desires. In contrast, in China both teachers and parents teach children
to regulate emotions of frustration from an early age.
In fact, the ability to
regulate frustration is often called the "anti-frustration ability" by Chinese,
expressing a belief that frustration may be controlled and overcome.
To understand why Chinese children and mothers associate frustration
with effortful control requires examining the socio-cultural context of PRC.
1 Just recently, I received Ellis's reply to my inquiry about this factor.
Her reply was that the
Irritability scale was from EATQ, a previous version of EATQ-R, and it was dropped from EATQ-R
due to the overlap between Irritability and the newly constructed Frustration scale. In the final
structure, the Negative Affectivity factor includes the Frustration scale and two other socialbehavioral scales - Aggression and Depressed Mood. Ellis advised me to rerun the EFA with
Aggression and Depressed Mood included for the Chinese children and see if a similar negative
affectivity factor would be generated. Due to the time constraint, I skipped the item-level EFA and
reran a second-order EFA using EATQ-R scales. Interestingly, the Negative Affectivity factor still
did not emerge for Chinese children. Activation Control, Frustration, Aggression, and Depressive
100
Chinese culture has traditionally valued balance and harmony, as reflected in its
emphasis on interdependence and relationships between all things.
Chinese
social structure is hierarchical with strictly defined roles and prescriptions for
behavior to ensure the stability of the big system.
In this system, a person
cannot be and has never been identified as a separate entity without placed
within a network of relations. Accordingly, individual goals and pursuits are often
discouraged unless they can be subsumed under family or group goals.
The
dense population and tight political control in PRC, more generally, further deny
an individual's space, both physically and psychologically. All these lead to an
emphasis on obedience, conformity, and respect for authority in PRC's school
education. The social situations and everyday practices require Chinese children
to control/regulate their behavioral and emotional expressions according to the
demands of the situations they participate in. Usually they must control feelings
of frustration, aggression, and motivations to engage in disruptive behavior when
their goals are blocked and desires are unsatisfied. Though frustration may be
perceived as an individual emotion attribute in the US, it is perceived in China to
be a character weakness that one must work on.
Frustration and Activation
Control are likely to have loaded on the same factor in the PRC structure
because they are two aspects of what effortful control means to Chinese children
and their mothers. In the US structure, the two dimensions are not subsumed
under one factor, despite a moderate negative association between the factors
defined by these dimensions.
Mood all loaded on the same factor, with a positive loading for Activation Control, and negative
101
Effortful Control
Attention
Activation Control
Inhibitory Control
.03
High Intensity Pleasure
Low Levels of Shyness
Low Levels of Fear
Affillativeness
Surgency
Affiliation (.861)
High Intensity
Pleasure (.827)
Shyness (-.544)
Pleasure sensitivity
(.750)
Perceptual sensitivity
(.686)
Irritability
Frustration
Affiliation
Perceptual Sensitivity
Pleasure Sensitivity
Frustration (.966)
Activation control
(-.591)
EATA-R factor structure and
Correlations between factors (from Ellis & Rothbart,
2001)
New Chinese Factorial Structure according to Chinese Children
Figure 3. Comparison of the EATQ-R Factor Structures in the US and PRC,
Based on Self-reports
In summary, though there are some consistencies between the two
factorial structures derived for the US and the PRC samples, there are important
.J
differences in terms of the factors extracted and correlations among the factors.
In the PRC structure, Frustration was found to load with Activation Control on the
same factor for both Chinese children and mothers. This pattern did not change
even after the Aggression and Depressive Mood scales were included in the
EFA. This seems to suggest that negative emotions, especially the ones that
have disruptive effects, are strongly associated with behavioral control in
Chinese children. More analyses are needed to examine the overall fit of the
American model in PRC sample.
General Patterns of Temperamental Characteristics in PRC sample
Even though the original EATQ-R scales showed less satisfactory scale
reliability when applied to Chinese children and mothers, the newly constructed
loadings for Frustration, Aggression, and Depressive Mood.
102
Chinese
scales
based
on
the
first-order
demonstrated adequate internal consistency.
exploratory
factor
analyses
This allowed examining the
temperamental characteristics in Chinese children. In the current study, Chinese
children rated themselves high on Activation Control and Affiliation, but low on
Frustration and Shyness. Chinese mothers rated their children in similar ways.
As discussed earlier, Chinese culture, to a large extent, encourages its
children to become controlled and group-oriented individuals with high levels of
tolerance to cope with adversity. Thus, the high scores on Activation Control,
Affiliation, and low scores on Frustration might reflect such cultural influences as
much as the temperamental traits of individual children in China. For example, in
Beijing and most cities in China, first grade students (ages 6- to 7-years-old)
have full days at school, with four 45-minute class sessions each morning and
afternoon. Such long hours demand children to learn how to control themselves
in order to behave properly in class from an early·age. Teachers not only teach
children subject matter, they also teach what are called "good study behaviors
and habits," ranging from sitting with good posture, waiting their turns to speak,
and forming the habit of finishing all homework before playing.
There are few comparable studies examining the Activation Control
dimension of temperament in Chinese children.
But in a study using parent-
report CBa, Ahadi, Rothbart, and Ye (1993) reported low scores on scales
contributing to Effortful Control (Inhibitory Control, Low Intensity Pleasure, and
Perceptual Sensitivity) in a PRC sample of 6- to 7-year-old children from
Shanghai.
The finding was interpreted by the authors as resulting from a
103
loosened need for self-regulation in Chinese children in a relatively restraint and
rigid Chinese social structure.
The mixed results on effortful control in PRC
children may have something to do with the following differences in the two
studies.
First, Ahadi, Rothbart, and Ye used Chinese mothers' reports only,
while the current study used both children's and mothers'. In the current study,
Chinese mothers' mean score on Activation Control was lower than Chinese
children's (3.52 vs. 4.22), suggesting that mothers are likely to hold a higher
standard than children when assessing behavioral control. Though the Activation
Control scale had the highest mean scale score according to Chinese children, it
was the third highest score among Chinese mothers, after Affiliation and
Outgoingness. If behavioral control is particularly valued in the Chinese culture,
the socialization process associated with behavioral control tends to be also
severe. This makes it plausible for Chinese mothers to use a higher standard in
behavioral control assessment and give relatively lower scores on the dimension.
Furthermore, as discussed in Chapter 1, American parents seem to have a
response bias of over-evaluating their children on desirable traits.
When
combined with a tendency in Chinese parents to use lower ratings, this response
bias may have driven up the rating differences on scales concerning behavioral
control between the American and Chinese mothers. The second possible factor
could be the content differences between the Activation Control scale in EATQ-R
and the Inhibitory Control scale in CSQ. Most items in the newly constructed
Activation Control scale are about "good study behaviors and habits," and
Chinese children and mothers seem to be able to relate to these easily and
104
consistently. Items in CSQ's Inhibitory Control scale do not focus so much on
study behavior, and Chinese mothers may have reacted to these items less
consistently.
Affiliation is one of the newly added scales during the revision of the
EATQ-R to better understand the growing desire of adolescents for warmth and
closeness with others, especially among females.
In a society in which
interpersonal relatedness is emphasized and encouraged, a high level of
affiliation is expected.
A low level of Frustration was also expected in Chinese children and
confirmed in the current study's results.
However, Ahadi, Rothbart, and Ye
(1993) reported a higher level of negative affect including Frustration in Chinese
children. As discussed in Chapter 1, results regarding frustration or irritability, a
closely related temperamental dimension to frustration, were mixed in Chinese
infants.
The results of observation-based studies tended to indicate a less
reactive, less irritable, and more easily soothed Chinese infant profile; while
results of questionnaire-based studies tended to indicate that Chinese infants
had a lower level of sensory thresholds, higher level of irritability (and other
negative affect), and were less easily soothed. Unfortunately, few studies have
focused on examining Chinese children's behavioral expressions of frustration,
which would be more informative for the present study.
The low level of Shyness displayed in the present PRC sample was
unexpected, because most studies using observation method and questionnaire
method tend to report a higher level of shyness in Asian children in comparison
105
to American children.
As reviewed in Chapter 1, Asian infants and children
tended to be more behaviorally inhibited and less approaching. Ahadi, Rothbart,
and Ye (1993) found 6- to 7-year-old Chinese children from Shanghai, PRC,
scored the highest on the Shyness scale among 15 parent-reported CSQ scales.
Sut in the current study, both children and their mothers gave one of the lowest
ratings to the dimension of Shyness, along with Frustration.
Though shyness itself is defined as behavioral inhibition in social
contexts, the two Shyness items that were highly influential in defining the
Shyness scale were general statements: "I'm shy" and "I am not shy." Items that
make general statements without specifying contexts have been shown to be
less effective than the items that are context-based. This is especially true in
current study because Chinese children are likely to evaluate themselves against
different contexts. They may consider themselves shy outside home, yet not shy
at home where they are allowed to speak louder and act more freely. There is
even the possibility of two groups of shy children, one group of those who are
consistently shy, both inside and outside, and a second group of those who may
act differently depending on social contexts.
Shyness is also a multi-faceted
concept in Chinese culture (Xu et aI., 2004).
In a follow-up study, Xu further
identified "stranger shyness," a third form of shyness along with executive and
anxious shyness (personal communication). This stranger shyness resembles a
biological form of shyness and is related to the shyness construct measured by
the EATQ-R. Chinese children in Xu's follow-up study did not display a high level
of stranger shyness, as measured by the EATQ-R Shyness scale and by
106
biological indicators such as heart rates and vagal tone. Such findings may help
to explain why Chinese children generally considered themselves to be less shy
on the EATQ-R shyness scale.
Gender and Only-Child Status Variations
An area of special interest in the current study was to examine gender
differences of temperament in Chinese children, especially in the light of Ahadi,
Rothbart, and Ye's findings of "unconventional" gender differences. Specifically,
Chinese girls were reported to be higher than boys on the Surgency scales of
Activity Level, High Intensity Pleasure, and Impulsivity, lower on the Effortful
Control scales of Inhibitory Control, Low Intensity Pleasure (corresponding to
EATQ-R Pleasure Sensitivity), and Perceptual Sensitivity, and lower on the
Negative Affect scales of Discomfort and Sadness.
In the current study, the
above patterns were not replicated in the Chinese children's self-reports.
Compared to boys, Chinese girls in the current study considered themselves to
have better behavioral control; they valued their friendships more and found low
intensity pleasure more enjoyable than boys did.
These were upheld by the
higher scores reported by girls on the Activation Control, Affiliation, and Pleasure
Sensitivity scales.
Similar gender differences on Affiliation and Pleasure
Sensitivity were also reported in an age-comparable American sample (Ellis,
2002) using the EATQ-R. Gender differences found in the current older PRC
sample were "conventional" in that they were consistent with previous findings
and gender stereotypes in Chinese culture. In China, the strength among girls is
107
expected to be seen in their being more obedient, self-disciplined, caring, and
feminine than boys.
Chinese mothers agreed that girls had better behavioral control than boys,
but they did not endorse the other gender differences reported by Chinese
children on Affiliation and Pleasure Sensitivity. Instead, these mothers rated girls
significantly higher on the Outgoingness scale.
The Outgoingness scale was
split from the original High Intensity Pleasure scale and contained elements of
curiosity, adventurousness, and novelty. This scale grouped with Affiliation in the
second-order EFA instead of Shyness and Fearlessness, and gender differences
found on this dimension bears certain similarity to girls' higher Affiliation score
reported by Chinese children.
Chinese mothers' reports offered an unexpected profile of Shyness among
four groups of Chinese children - boys with siblings, boys without siblings, girls
without siblings, and girls with siblings.
The scores on Shyness followed an
interesting pattern: boys with siblings scored highest (the most shy), followed by
boys and girls without siblings who scored equivalently (somewhat shy), and girls
with siblings scored the lowest (the least shy). This pattern seems to indicate
that shyness has more to do with family dynamics and implementation of onlychild policy than with biological sex differences on temperament.
Since boys have traditionally been valued more highly than girls in
Chinese culture, it is possible that families having more than one child tend to
have first born girls and later born boys-a "big sister and baby brother" family
structure. These boys are often overly protected and cared for because of their
108
special status in the families; in contrast, girls are more likely to help with taking
care of younger siblings and running errands for parents. The family dynamics
resulting from such traditional culture values, the implementation of only-child
policy, and birth order may contribute to the unique variation of gender and onlychild-status in shyness in the current study.
It should be noted that the only-
children and non-only children groups were very unbalanced in the current study,
and reflects how China's national one-child policy is enforced well in urban cities,
but less so in rural places (Falbo, Poston, & Feng, 1996).
For example, in a
recent study conducted in Gansu (Liu, 2004), a northwestern province in PRC,
70% of the families were reported to have more than one child. The non-only
children in the current study were mainly from families of workers who migrated
to Beijing from less economically developed rural provinces. It is possible that
the families of these children retain more traditional gender values than the more
long-established families in Beijing.
There is a need for further study on the
effects of gender and only-child status on temperamental dimensions with a more
balanced sample to clarify the preliminary findings in the current study.
Agreement between Chinese Children and Mothers
It was expected that the agreement between Chinese children and
mothers might be higher in this PRC sample than is reported in studies involving
American samples because a study conducted in PRC found a high level of
agreement between mothers and fathers on their rating of their children's
temperaments (Zheng, 1999).
Zheng suspects that the high convergence
between Chinese mothers and fathers results from the increased sharing of
109
childcare caused by the implementation of China's one-child policy, and close
relationship in Chinese families. A higher level of convergence between Chinese
children's and their mothers' temperamental ratings was expected on such
beliefs that children and their parents in Asian cultures have closer relationships
than children and their parents in the U.S.
However the results regarding mother-child agreement obtained in the
current study did not support such predication.
Agreements for comparable
scales ranged from .18 to .39 in the current study. Mother and children agreed
more on Activation Control, but less on Affiliation.
This is consistent with the
higher agreement observed on scales measuring externalizing behaviors and
lower agreement on scales measuring internalizing behaviors. In contrast to the
observable behaviors in the Activation Control scale, Affiliation contained items
about affiliative feelings and desires, which were less observable to mothers.
Agreement between girls and mothers was somewhat equal to agreement
between boys and mothers, except on the High Intensity Pleasure scale. Girls
and mothers showed higher agreement than boys and mothers on this scale. In
China, high intensity activities are culturally more associated with boys than with
girls, and mothers may notice girls more than boys when they engage in such
activities.
Another important result is that Chinese mothers reported placing more
importance on the role of parental discipline and school education in their
children's personality development than on the individual temperaments of their
children. This attitude is consistent with the temperamental patterns displayed by
110
Chinese children: high scores on Activation Control and Affiliation, and low
scores on Frustration. "No rules, no shape" is a popular saying among parents in
China, which reflects the importance placed on training and education in Chinese
culture.
If temperament is considered primarily biological in American culture,
Chinese parents and teachers tend to consider it changeable. For example, a
child's tendency to get frustrated easily is often viewed to be changeable given
the right instructions and training. While all cultures tend to encourage culturally
desirable
temperamental
traits
and
discourage
undesirable
ones,
the
socialization process might be particularly explicit in China because of such a
strong belief in the importance of cultivation. In China, the process of education
is believed to encompass not just the teaching of academic subjects, but also the
cultivation of culturally desirable behavioral traits, especially during the formative
years of childhood and adolescence.
111
CHAPTER FIVE
CONCLUSIONS AND FUTURE RESEARCH
There has been a dramatic increase in the numbers of studies on the
temperament of American children during the last two decades, but very few
studies have examined the temperament of early adolescents in the People's
Republic of China (PRC). This study was among the first to: (1) examine the
reliability and validity of the EATQ-R dimensions when the measure is applied to
a PRC sample; and (2) explore basic patterns of temperamental characteristics
in early Chinese adolescents ages 9-12, including the effects of gender and
only-child status on Chinese adolescents' temperament.
The results indicated that the transfer of a Western scale, its scale items
and factorial structure is limited in validity and reliability when used on a nonWestern sample.
This emphasized the importance of validating a measure
before applying it to a particular population.
It also suggested that the
measurement of temperament in Chinese children should begin with items
generated and rooted in China.
This study also furthered understanding about Chinese culture by
exploring the views of parents. It was found that Chinese mothers believe that
non-individual characteristics such as school education, family, and discipline
have more influence on their children's social development than individual
temperament does. Such beliefs tend to be shared among Chinese children,
teachers, and parents, and are likely to suppress culturally undesirable
112
temperamental traits, such as the tendency to get frustrated easily, and to
enhance the desirable ones, such as behavioral control.
Finally, this study's
findings suggested that China's one child policy may affect the temperament of
Chinese children because only children of both genders shared certain traits
which were not shared across non-only boys and non-only girls.
While these preliminary findings are hoped to increase understanding
about the temperament of Chinese youth with greater cultural sensitivity, several
limitations remain. Although back translation procedures were used to maximize
the linguistic equivalence of the translated versiol) of the EATQ-R items, the
underlying meanings and assumptions shared by members of one cultural
community could not be conveyed through literal translations to members of
another cultural community. Furthermore, as explained in Chapter four, Chinese
children and parents found many items in the EATQ-R lacking in 'contextual
information which was crucial for enabling them to make accurate judgments.
Such problems may have reflected inherent limitations of any study
involving questionnaires, as the use of scaled statements devoid of contexts
could have the effect of simplifying the complexities of human behavior and
perceptions (Marsella, Purcell, & Carr, 2004). Accordingly, the limitations of the
EATQ-R necessitate supplementary qualitative approaches to more fully assess
the experiences and views of Chinese children and parents. Marsella, Purcell,
and Carr (2004) point out that qualitative methods, such as narrative accounts
and conversational interviews, can be more sensitive than a quantitative
113
approach regarding "the context, meaning, and origins of knowledge" which
influence the perspectives of research participants (p. 9).
Thus, future research involving such qualitative methods would be useful
for developing a temperamental scale from within the context of Chinese culture
and society.
Specifically, additional research is needed on socio-Iinguistic
aspects of Chinese views regarding children's temperament.
For example,
researchers must take into account larger behavioral patterns of modesty,
discretion, and politeness which are associated with shyness in Chinese society
(Xu, et. aI., 2004). Such an approach would allow researchers to develop scales
that better incorporate the indigenous experiences and perceptions of Chinese
people.
Improved socio-linguistic understanding of specific temperament
constructs could also guide important cross-cultural adjustments to existing
psychological instruments originating in North America.
Along these lines, qualitative research is needed to clarify the relationship
between the constructs of effortful control and negative affectivity (or frustration)
among Chinese children. While the two constructs were extracted as distinct
broad factors in Ellis and Rothbart's American sample, this study suggested that
in a Chinese sample, they constituted only one broad factor. There is a need to
better understand what "effortful control" or "negative affectivity" might mean with
reference to concepts of self-regulation, discipline, obedience, and maturity in
China.
Further research, including both quantitative and qualitative forms, is
needed to examine the direct and mediating roles of only-child status on
114
temperament. A more rigorous study on the effects of only-child status requires
a larger sample with more balanced proportions of non-only and only children.
This would suggest the need to draw the sample from a more diverse range of
provinces, including both rural and urban areas. The influences of the One Child
policy on parental expectations of their children and resulting effects on children's
social behavior and attitudes are also worthy of closer attention.
Continued studies on how cultural values and social expectations help
shape the temperament of Chinese children are particularly relevant considering
the fast-paced modernization underway in China today.
115
APPENDIX A
Early Adolescent Temperament Questionnaire-Revised
Scale Definitions - Long Form
Temperament Scales
Activation Control: The capacity to perform an action when there is a strong
tendency to avoid it.
Affiliation: The desire for warmth and closeness with others, independent of
shyness or extraversion.
Activity Level: Participation in activities requiring high levels of physical activity.
Attention: The capacity to focus attention as well as to shift attention when
desired.
Fear: Unpleasant affect related to anticipation of distress.
Frustration: Negative affect related to interruption of ongoing tasks or goal
blocking.
High Intensity Pleasure/Surgency: The pleasure derived from activities
involving high intensity or novelty.
Inhibitory Control: The capacity to plan, and to suppress inappropriate
responses.
Pleasure Sensitivity: Amount of pleasure related to activities or stimuli
involving low intensity, rate, complexity, novelty, and incongruity.
Perceptual Sensitivity: Detection or perceptual awareness of slight, lowintensity stimulation in the environment.
Shyness: Behavioral inhibition to novelty and challenge, especially social.
Behavioral Scales
Aggression: Hostile and aggressive actions, including person- and objectdirected physical violence, direct and indirect verbal aggression, and hostile
reactivity.
Depressive Mood: Unpleasant affect and lowered mood, loss of enjoyment and
interest in activities.
116
EATQ-R Scale Assignments
Composition of scales and scale alphas based on item analysis, EATQR initial study, N=177 (92 females, 85 males). Age=10.59-15.99 years, mean
age=13.78.
Scales are scored such that a high score on a scale indicates that the
individual is high in that attribute. Reversed scored items indicated by "R".
Activation Control, N=8, Alpha=.73
7)
24)
32)
55)
63)
65)
66)
82)
23)
38)
47)
51)
75)
78)
88)
94)
5)
10)
18)
33)
37)
59)
68)
72)
74)
84)
96)
R I have a hard time finishing things on time.
R I do something fun for a while before starting my homework, even when
I'm not supposed to.
When someone asks me to do something, I do it right away, even if I
don't want to.
R If my friends are mad at me, I try to stay away from them.
I finish my homework before the due date.
I tend to be on time for school and appointments.
If I have a hard assignment to do, I get started right away.
R I put off working on projects until right before they're due.
Affiliation, N=8, Alpha=.76
I want to be able to share my private thoughts with someone else.
I would like to be able to spend time with a good friend every day.
I enjoy exchanging hugs with people I like.
I will do most anything to help someone I care about.
It is important to me to have close relationships with other people.
I like to look at other people's photographs.
I am quite a warm and friendly person.
I like to listen to other people talk about themselves.
Aggression, N=11, Alpha=.81
If I'm mad at somebody, I tend to say things that I know will hurt their
feelings.
When I am angry, I throw or break things.
·If I get really mad at someone, I might hit them.
When I compete in games or sports, I really try to crush my opponents.
I tend to be rude to people I don't like.
When I am mad, I slam doors.
I tend to talk about other people behind their back.
My friends and I make fun of how other people look.
R I don't criticize other people.
When I'm really mad at a friend, I tend to explode at them.
I pick on people for no real reason.
117
Activity Level, N=6, Alpha=.65
I would rather playa sport than watch TV.
When I do things, I do them with a lot of physical energy.
I like to be physically active whenever I have the chance (sports,
dancing, etc.).
I have the energy for hard physical work, like digging in the yard or
chopping wood.
Long winter weekends make me want to get out of the house and do
something physical.
I prefer outdoor activities to those indoors.
1)
16)
36)
64)
83)
100)
25)
44)
.56)
R
R
62)
R
67)
86)
97)
13)
15)
26)
49)
60)
90)
3)
54)
57)
77)
85)
93)
12)
R
Attention, N=7, Alpha=.65
It is easy for me to really concentrate on homework problems.
When interrupted or distracted, I forget what I was about to say.
I find it hard to shift gears when I go from one class to another at
school.
When trying to study, I have difficulty tuning out background noise and
concentrating.
I am good at keeping track of several different things that are happening
around me.
I tend to get in the middle of one thing, then go off and do something
else.
I pay close attention when someone tells me how to do something.
Depressive Mood, N=6, Alpha=.69
My friends seem to enjoy themselves more than I do.
It often takes very little to make me feel like crying.
R I feel pretty happy most of the day.
I get sad more than other people realize.
I get sad when a lot of things are going wrong.
I feel sad even when I should be enjoying myself, like at Christmas or
on a trip.
Fear, N=6, Alpha=.64
I worry about getting into trouble.
I get frightened riding with a person who likes to speed.
I worry about my family when I'm not with them.
I am nervous of some of the kids at school who push people into
lockers and throw your books around.
I worry about my parent(s) dying or leaving me.
I feel scared when I enter a darkened room at home.
Inhibitory Control, N=11, Alpha=.77
R When I'm excited, it's hard for me to wait my turn to talk.
118
19)
20)
21 )
R
22)
39)
R
R
45)
46)
R
R
71)
81 )
89)
When someone tells me to stop doing something, it is easy for me to
stop.
I could easily change a bad habit if I wanted to.
I tend to say the first thing that comes to my mind, without stopping to
think about it.
It's hard for me not to open presents before I'm supposed to.
When I'm having a really good time, I have a hard time leaving to go
home when I'm supposed to.
I blurt out answers in class before the teacher calls on me.
The more I try to stop myself from doing something I shouldn't, the
more likely I am to do it.
It's easy for me to keep a secret.
I am good at self-discipline.
I can stick with my plans and goals.
Frustration, N=9, Alpha=.67
40) R I am a patient person.
42)
It bothers me when I try to make a phone call and the line is busy.
58)
I get very upset if I want to do something and my parents won't let me
73)
It bothers me when people are slow about getting ready for something.
79)
I get irritated when I have to stop doing something that I am enjoying.
91)
It really annoys me to wait in long lines.
98)
I get very frustrated when I make a mistake in my school work.
101)
It frustrates me if people interrupt me when I'm talking.
102)
I get upset if I'm not able to do a task really well.
28)
30)
34)
48)
50)
70)
80)
2)
8)
14)
17)
27)
29)
53)
Pleasure Sensitivity, N=7, Alpha=.77
I like the sound of words.
I like to look at trees and walk amongst them.
I like the feel of hot water running over me, like in the shower.
I like the crunching sound of autumn leaves.
I like to feel a warm breeze blowing on my face.
I enjoy listening to the birds sing.
I like to look at the pattern of clouds in the sky.
High Intensity Pleasure/Surgency, N=12, Alpha=.77
I enjoy going places where there are big crowds and lots of excitement.
R I wouldn't like living in a really big city, even if it was safe.
R Skiing fast down a steep slope sounds scary to me.
I wouldn't be afraid to skateboard or ride a bike really fast down a steep
hill.
When people tell me about trips to exotic places, it makes me really
want to go there.
I think it would be exciting to move to a new city.
I would not be afraid to try a risky sport, like deep-sea diving.
119
61)
I find the idea of driving a race car exciting.
95) R I wouldn't want to go on the frightening rides at the fair.
99)
I wouldn't be afraid to try something like mountain climbing.
103)
I prefer friends who are exciting and unpredictable
4)
11)
31)
35)
41)
92)
6)
9)
43)
52)
69)
76)
87)
Perceptual Sensitivity, N=6, Alpha=.77
I notice when other people are coughing during movies or shows.
I notice even little changes taking place around me, like lights getting
brighter in a room.
I tend to notice little changes that other people do not notice.
I am very aware of noises.
I can tell if another person is angry by their expression.
R I don't really notice the color of people's eyes.
Shyness, N=7, Alpha=.80
I feel shy about meeting new people.
I feel shy with kids of the opposite sex.
If I am asked to deliver a message to an adult, I feel uncomfortable
about going up to them.
R I can generally think of something to say, even with strangers.
It is a lot easier for me to talk to people I know than to strangers
I am shy.
R I am not shy.
120
Appendix B
Early Adolescent Temperament Questionnaire-Revised
Scale Definitions -Parent Report Form
Temperament Scales
Activation Control: The capacity to perform an action when there is a strong
tendency to avoid it.
Affiliation: The desire for warmth and closeness with others, independent of
shyness or extraversion.
Attention: The capacity to focus attention as well as to shift attention when
desired.
Fear: Unpleasant affect related to anticipation of distress.
Frustration: Negative affect related to interruption of ongoing tasks or goal
blocking.
High Intensity Pleasure/Surgency: The pleasure derived from activities
involving high intensity or novelty.
Inhibitory Control: The capacity to plan, and to suppress inappropriate
responses.
Shyness: Behavioral inhibition to novelty and challenge, especially social.
Behavioral Scales
Aggression: Hostile and aggressive actions, including person- and objectdirected physical violence, direct and indirect verbal aggression, and hostile
reactivity.
Depressive Mood: Unpleasant affect and lowered mood, loss of enjoyment and
interest in activities.
EATQ-R Parent Scale Assignments
Composition of scales and scale alphas based on item analysis, EATQR initial study, N=69 parent respondents
Scales are scored such that a high score on a scale indicates that the
individual is high in that attribute. Reversed scored items indicated by "R".
121
Activation Control, N=7, Alpha=.66, correlation w/self-report=.27,
p<.05
3) R
5)
14) R
17)
36)
38)
46) R
Has a hard time finishing things on time.
If having a problem with someone, usually tries to deal with it right
away.
Usually does something fun for awhile before starting her/his
homework, even though s/he is not supposed to.
When asked to do something, does it right away, even if s/he doesn't
want to.
Usually finishes her/his homework before it's due.
Usually gets started right away on difficult assignments.
Usually puts off working on a project until it is due.
Affiliation, N=6, Alpha=.82, correlation w/self report=.35, p<.01
12)
13)
18)
24)
43)
51)
Likes taking care of other people.
Likes to be able to share his/her private thoughts with someone else.
Would like to be able to spend time with a good friend every day.
Enjoys exchanging hugs with people s/he likes.
Wants to have close relationships with other people.
Is quite a warm and friendly person.
Aggression, N=6, Alpha=.71, correlation w/self report=.46, p<.001
2)
11)
19)
32)
42) R
25)
41)
When angry at someone, says thing s/he knows will hurt that person's
feelings.
If very angry, might hit someone.
Tends to be rude to people s/he doesn't like.
Slams doors when angry.
Doesn't criticize others.
Tends to try to blame mistakes on someone else.
Makes fun of how other people look.
Attention, N=6, Alpha=.65, correlation w/self report=.28, p<.05
15)
22) R
35) R
39)
49) R
60)
Finds it easy to really concentrate on a problem.
When interrupted or distracted, forgets what s/he was about to say.
Has a difficult time tuning out background noise and concentrating
when trying to study.
Is good at keeping track of several different things that are happening
around her/him.
Is often in the middle of doing one thing and then goes off to do
something else without finishing it.
Pays close attention when someone tells her/him how to do something.
122
7)
10)
26)
33)
52)
Depressive Mood, N=5, Alpha=.76, correlation w/self report=.42,
p<.001
Often does not seem to enjoy things as much as his/her friends.
Feels like crying over very little on some days.
Is sad more often than other people realize.
Is hardly ever sad, even when lots of things are going wrong.
Sometimes seems sad even when s/he should be enjoying her/himself
like at Christmas, or on a trip.
55)
61)
Fear, N=6, Alpha=.69, correlation w/self report=.40, p<.001
Worries about getting into trouble.
Worries about our family when s/he is not with us.
Is afraid of the idea of me dying or leaving her/him.
Doesn't enjoy playing softball or baseball because s/heis afraid of the
ball.
Feels scared when entering a darkened room at night.
Is nervous being home alone.
20)
21)
31)
45)
57)
58)
Frustration, N=6, Alpha=.74, correlation w/self-report=.74, p<.001
Is annoyed by little things other kids do.
Gets very irritated when someone criticizes her/him.
Gets irritated when I will not take her/him someplace s/he wants to go.
Gets irritated when s/he has to stop doing something s/he is enjoying.
Hates it when people don't agree with him/her.
Gets very frustrated when s/he makes a mistake in her/his school work.
1)
30)
48)
53)
47)
59)
Inhibitory Control, N=5, Alpha=.86, correlation w/self report=n.s.
(females=.33, p=.052, males=n.s.)
Has a hard time waiting his/her turn to speak when excited.
Opens presents before s/he is supposed to.
Is more likely to do something s/he shouldn't do the more s/he tries to
stop her/himself.
Is able to stop him/herself from laughing at inappropriate times.
Is usually able to stick with his/her plans and goals.
27) R
44)
50) R
54) R
Shyness, N=5, Alpha=.72, correlation w/self report=n.s.
(females=.31, p=.07, males=n.s.)
Can generally think of something to say, even with strangers.
Is shy.
Is not shy.
Likes meeting new people.
6) R
8) R
23) R
123
62)
4)
g) R
16)
28)
29)
34)
37)
40)
56) R
Feels shy about meeting new people.
Surgency, N=9, Alpha=.70, correlation w/self report=.29, p<.05
Thinks traveling to Africa or India would be exciting and fun.
Would be frightened by the thought of skiing fast down a steep slope.
Thinks it would be exciting to move to a new city.
Wouldn't be afraid to try a risky sport like deep sea diving.
Expresses a desire to travel to exotic places when s/he hears about
them.
Would like driving a racing car.
Likes it when something exciting and different happens at school.
Is energized by being in large crowds of people.
Wouldn't want to go on the frightening rides at the fair.
124
REFERENCES
Ahadi, S. A, & Rothbart, M. K. (1994).
Temperament, development, and Big
Five. In C. F. Harverson Jr, G. A Kohnstamm, & R. P. Martin (Eds.), The
developing structure of temperament and personality from infancy to
adulthood (pp.189-207). Hillsdale, NJ: Erlbaum.
Ahadi, S. A, Rothbart, M. K., & Ye, R. (1993). Children's temperament in the
U.S. and China: Similarity and differences.
European Journal of
Personality, 7, 359-377.
Banks, E. (1989).
Temperament and individuality: A study of Malay children.
American Journal of Orthopsychiatry, 59, 390-397.
Barrett, P. & Eysenck, S. B. (1984).
The assessment of personality factors
across 25 countries. Personality and Individual Differences, 5, 615-632.
Bates, J. E. (1987). Temperament in infancy. In J. D. Osofsky (Ed.), Handbook
ofinfant development (2 nd ed., pp. 1101-1149). New York: Wiley.
Bates, J. E. (1989).
Concepts and measurements of temperament.
In G. A
Kohnstamm, J. E. Bates, & M. K. Rothbart (Eds.), Temperament in
childhood (pp. 3-26). New York: Wiley.
Buss, A H. (1991).
The EAS theory of temperament.
In J. Strelau & A
Angleitner (Eds.), Explorations in temperament: International perspectives
on theory and measurement (pp. 43-60). New York: Plenum Press.
Buss, A H., & Plomin, R. (1975).
A temperamental theory of personality
development. New York: Wiley.
125
Buss, A. H., & Plomin, R. (1984).
Temperament: Early developing personality
traits. Hillsdale, NJ: Erlbaum.
Capaldi, D. M., & Rothbart, M. K. (1992). Development and validation of an
adolescent temperament measure.
Journal of Early Adolescence, 12,
153-173.
Chao, R. K. (1994). Beyond parental control and authoritarian parenting style:
Understanding Chinese parenting through the cultural notion of training.
Child Development, 65, 1111-1119.
Carey, W. B., & McDevitt, S. C. (1978). Revision of the Infant Temperament
Questionnaire. Pediatrics, 61, 735-739.
Chen, X.
(2000). Growing up in a collectivist culture: Socialization and
socioemotional development in Chinese children. In A. L. Comunian, & U.
P. Gielen (Eds.), International perspectives on human development (pp.
331-353). Lengerich, Germany: Pabst Science Publishers.
Chess, S., & Thomas, A. (1984).
Origins and evolution of behavior disorders:
From infancy to adult life. New York: Brunner/Maze!.
Chess, S., & Thomas, A. (1991). Temperament and the concept of goodness of
fit.
In J. Strelau & A. Angleitner (Eds.), Explorations in temperament:
International perspectives on theory and measurement (pp. 15-28). New
York: Plenum Press.
126
Derryberry, D., & Rothbart, M. K. (1988).
components of temperament.
Arousal, affect, and attention as
Journal of Personality and Social
Psychology, 55,958-966.
DeVrise, M. W. (1994).
Kids in contexts: Temperament in cross-cultural
perspective. In W. B. Carey & S. C. McDevitt (Eds.), Prevention and early
intervention: Individual differences as risk factors for the mental health of
children (pp. 126 -139). New York: Brunner/Maze!.
Ellis,
L.
K.
(2002).
Individual
differences
and
adolescent
psychosocial
development. Dissertation Abstracts International, 63 (08), 3956B. (UMI
No. 3061943)
Ellis, L. K. & Rothbart, M. K. (2001). Revision of the early adolescent
temperament questionnaire. Poster session presented at biennial meeting
of the Society for Research
in
Child
Development,
Minneapolis,
Minnesota.
Falbo, T., Postern, D. L., & Feng, X. T. (1996). The academic, personality, and
physical outcomes of Chinese only children: A review.
In S. Lau (Ed.),
Growing up the Chinese way: The role of culture in socialization (pp. 265287). Hong Kong: Chinese University Press.
Freedman, F. G., & Freedman, N. A. (1969). Differences in behavior between
Chinese-American and European-American newborns.
Nature, 224,
1227.
Garstein, M., Putnam, S., Bechen-Jones, L., & Rothbart, M. K. (2002, April).
Infant Behavior Questionnaire-Revised: New evidence in support of
127
reliability and validity.
In Proceedings of the Biennial Meeting of the
International Society for Infant Studies, Toronto, Canada.
Goldsmith, H. H. (1996). Studying temperament via construction of the Toddler
Behavior Assessment Questionnaire. Child Development, 67,218-225
Goldsmith, H. H., Buss, A H., & Lemery, K. S. (1997). Toddler and childhood
temperament:
Expanded
content,
stronger
genetic
evidence,
new
evidence for the importance of environment. Developmental Psychology,
33 (6),891-905.
Goldsmith, H. H., Buss, A H., Plomin, R, Rothbart, M. K., Thomas, A, Chess,
S., Hinde, R
A, & McCall, R
B. (1987).
Roundtable: What is
temperament? Four approaches. Child Development, 58, 505-529.
Goldsmith, H. H., & Rothbart, M.K. (1991).
Contemporary instruments for
assessing early temperament by questionnaire and in the laboratory. In J,
Strelau & A Angleitner (Eds.), Exploration in temperament: International
perspectives on theory and measurement (pp. 249-272).
New York:
Plenum.
Ho, D. Y. F. (1986). Chinese patterns of socialization: A critical review. In M. H.
Bond (Ed.), The psychology of the Chinese people (pp. 1-37). New York:
Oxford University Press.
Hsu, C., Soong, W., Stigler, J. W., Hong, C., & Liang, C. (1981). The
temperamental characteristics of Chinese babies. Child Development, 52,
1337-1340.
128
Hubert, N. C., Wachs, 1. D., Perters-Martin, P., & Gandour, M. J. (1982). The
study of early temperament: Measurement and conceptual issues. Child
Development, 53, 571-600.
Kagan, J. (1994). Galen's prophecy: Temperament in human nature. New York:
Basic Books.
Kagan, J. (1998). Biology and child. In W. Damon (Series Ed.) & N. Eisenberg
(Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional and
personality development (5 th ed., pp. 177-235). New York: Wiley.
Kagan, J., Arcus, D., Snidman, N., Wang, Y., Hendler, J. & Greene, S. (1994).
Reactivity in infants: A cross-national comparison.
Developmental
Psychology, 30 (3), 342-345.
Kagan, J., Kearsley. R. B., Zelaro, P.R. (1978).
Infancy: Its place in human
development. Cambridge, MA: Harvard University Press.
Kerr, M. (2001). Culture as a context for temperament: Suggestions from the life
course of shy Swedes and Americans.
In 1. D. Wachs, & G. A.
Kohnstamm (Eds.), Temperament in context (pp. 139-152). Mahwah, NJ:
Erlbaum.
Kohnstamm, G. A. (1989). Temperament in childhood: Cross-cultural and sex
differences. In G. A. Kohnstamm, J. E. Bates, & M. K. Rothbart (Eds.),
Temperament in childhood (pp. 483-508.). New York: Wiley.
Lee, Y., & Chen, Y. (1998). The temperamental characteristics of the Chinese
adolescents in Taiwan. Journal of Adolescence, 21 (3), 337-340.
129
Liu, Xiaodong (2004, July). Community effects on child adjustment in rural China.
Poster session
presented
at the
annual
American
Psychological
Association convention. Honolulu, Hawaii.
Maccoby, E E, & Jacklin, C. N. (1974).
The development of sex differences.
Stanford, CA: Stanford University Press.
Marsella, A. J., Dubanoski, J., Hamada, W., & Morse, H. (2000).
The
measurement of personality across cultures: Historical, conceptual, and
methodological issues and considerations. American Behavioral Scientist,
44 (1), 41-62.
Marsella, A. J., Purcell, I. P., & Carr, S. C. (2004). Reflections on qualitative and
quantitative research methods in international and cross-cultural research.
Manuscript submitted for publication.
Marsella, A. J., & Leong, F. T. (1995). Cross-cultural issues in personality and
career assessment. Journal of Career Assessment, 3 (2), 202-218.
Martin, R. P., & Halverson, C. F. Jr. (1991).
Mother-father agreement in
temperament ratings: A preliminary investigation.
In J. Strelau & A.
Angleitner (Eds.), Exploration in temperament: International perspectives
on theory and measurement (pp. 235-248). New York: Plenum Press.
Nunnally, J. (1978). Psychometric theory. New York: McGraw-Hili
Posner, M. I., & Rothbart, M. K. (2000).
Developing mechanisms of self-
regulation. Development and Psychopathology, 12, 427-441.
130
Prior, M. R., Sanson, A
Y., and Oberklaid, F. (1989). The Australian
temperament project. In G. A Kohnstamm, J. E. Bates, & M. K. Rothbart
(Eds.), Temperament in childhood (pp. 537-554). New York: Wiley.
Rothbart, M. K. (1981).
Measurement of temperament in infancy.
Child
Development, 52, 569-578.
Rothbart, M. K. (1986).
Longitudinal observation of infant temperament.
Developmental Psychology, 22, 356-365
Rothbart, M. K. (1989). Temperament in childhood. In G. A. Kohnstamm, J. E.
Bates, & M. K. Rothbart (Eds.), Temperament in childhood (pp. 59-73).
New York: Wiley.
Rothbart, M. K., & Ahadi, S. A (1994). Temperament and the development of
personality. Journal of Abnormal Psychology, 103, 55-66.
Rothbart, M. K., Ahadi, S. A, & Evans, D. E. (2000).
personality: Origins and outcomes.
Temperament and
Journal of Personality and Social
Psychology, 78, 122-135.
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, 1394-1408.
Rothbart, M.K., & Bates, J. E. (1998). Temperament. In W. Damon (Series Ed.)
& N. Eisenberg (Volume Ed.), Handbook of child psychology: Vol. 3.
Social, emotional and personality development (5 th ed., pp. 105-176).
New York: Wiley.
131
Rothbart, M. K., & Derryberry, D. (1981). Development of individual differences
in temperament.
In M. E. Lamb & A. L. Brown (Eds.), Advances in
developmental psychology (Vol. 1, pp. 37-86). Hillsdale, NJ: Erlbaum.
Siabach, E. H., Morrow, J., & Wachs, T. D. (1991). Questionnaire measurement
of infant and child temperament.
In J. Strelau & A. Angleitner (Eds.),
Exploration in temperament: International perspectives on theory and
measurement (pp. 205-234). New York: Plenum Press.
Strelau, J. (1998).
Temperament: A psychological perspective.
New York:
Plenum Press.
Stevens, J. (2002). Applied multivariate statistics for the social sciences (4 th ed.).
Mahwah,NJ: Erlbaum.
Stevenson, H. W., Lee, S., Chen, C., Stigler, J. W., Hsu, C., & Kitamura, S.
(1990).
Contexts of achievement.
Monographs of the Society for
Research in Child Development, 55 (Serial No. 221).
Stevenson, H. W., Lee, S. Y., & Stigler, J. W. (1987). Mathematics achievement
of Chinese, Japanese, and American children. In S. Chess & A. Thomas
(Eds.), Annual progress in child psychiatry and child development (pp.
220-237). New York: Brunner/Mazel.
Super, C. M., & Harkness, S. (1986). Temperament, development, and culture.
In R. Plomin & J. Dunn (Eds.), The study of temperament: Continuities,
changes, and challenges (pp. 131-149). Hillsdale, NJ: Erlbaum.
Talwar, R., Nitz, K., Lerner, J. V., & Lerner, R. M. (1991).
The functional
significance of organismic individuality: The sample case of temperament.
132
In J. Strelau & A
Angleitner (Eds.), Exploration in temperament:
International perspectives on theory and measurement (pp. 29-42). New
York: Plenum Press.
Thomas, A, Chess, S., Birch, H. G., Hertzig, M. E., & Korn, S. (1963). Behavioral
individuality in early childhood. New York: New York University Press.
Thomas, A, Chess, S., & Birch, H. G. (1968).
Temperament and behavior
disorders in children. New York: New York University Press.
Thomas, A, & Chess, S. (1977).
Temperament and development.
New York:
Bruner/Maze!.
Wachs, T. D., & Kohnstamm, G. A
(2001). The bi-directional nature of
temperament-context links. In T. D. Wachs, & G. A Kohnstamm (Eds.),
Temperamentin context (pp. 201-222). Mahwah, NJ: Erlbaum.
Windle, M., Iwawaki, S., & Lerner, R. M. (1987). Cross-cultural comparability of
temperament among Japanese and American early- and late-adolescents.
Journal of Adolescence Research, 2,423-446.
Windle, M., Iwawaki, S., & Lerner, R. M. (1988). Cross-cultural comparability of
temperament
among
Japanese
and
American
preschool
children.
International Journal of Psychology, 23, 547-567.
Windle,
M., & Lerner,
R.
M. (1986).
Reassessing the dimensions of
temperamental individuality across the life span: The Revised Dimensions
of Temperament Survey (DOTS-R). Journal of Adolescent Research, 1,
213-230.
133
Xu, Y., Farver, J. A. M., Chang, L., Zhang, Z., Yu, L., & Cai, B. (2004). What you
see is what you get? Differentiating shyness in Chinese children.
Manuscript submitted for publication.
Yang, K. (1986). Chinese personality and its change. In Bond, M. H. (Ed.), The
psychology of the Chinese people (pp. 106-170).
New York: Oxford
University Press.
Zeng, Q. (1999). An exploratory study of child temperament and parenting: A
Chinese perspective. Dissertation Abstracts International, 60 (02), 0860B
(UMI No. 9921366).
134