University of Southern California Dissertations and Theses

Cognitive attributions for smoking and their roles on subsequent smoking progression and regression

(USC Thesis Other)

Cognitive attributions for smoking and their roles on subsequent smoking progression and regression

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
COGNITIVE ATTRIBUTIONS FOR SMOKING AND THEIR ROLES ON
SUBSEQUENT SMOKING PROGRESSION AND REGRESSION

by
Qian Guo

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PREVENTIVE MEDICINE)

August 2008

Copyright 2008 Qian Guo
ii

Dedication
The author dedicates this work to her husband, her son, her family, and her
friends who have convinced her that she could do this and were patient when she
retreated into dark caves to complete her work. All that the author does and all that
the author will ever do is dedicated to her family, without whose encouragement and
support all would not have been possible.

iii
Acknowledgements
The author wishes to acknowledge the chair of her dissertation committee,
Dr. Carl Anderson Johnson, who served as mentor, role model, shoulder, and driving
force. The author wishes to acknowledge the immense contributions made by all
members in her dissertation committee, including Drs. Stanley Azen, Jennifer Unger,
Iris Chi, and David MacKinnon, who provided invaluable assistance and insight. The
author wishes to acknowledge the University of Southern California
Transdisciplinary Tobacco Use Research Center that supported this research and was
funded by the National Institutes of Health and the Sidney R. Garfield Endowment.
The authors wishes to acknowledge the China Seven Cities Study directors and
project staff at the Centers for Disease Control and Prevention in the cities of
Chengdu, Hangzhou, Harbin, Qingdao, Shenyang, and Wuhan and the Institute for
Health Education in Kunming, China, for assistance with project coordination and
data collection. The author also wishes to acknowledge the principals, physicians,
and teachers in the participating schools for their cooperation. Finally, the author
wishes to express gratitude to the national Chinese Center for Disease Control and
Prevention, Peking University School of Public Health, and the municipal
government, Health Bureau, and Education Committee in each participating city for
their support.

iv
Table of Contents

Signature page

Dedication ii

Acknowledgments iii

List of Tables v

Abstract vi

Chapter One: Introduction 1

Chapter Two: Cognitive Attributions for Smoking: Findings from 25
An Attribution Study Conducted in China Seven Cities

Chapter Three: Roles of Cognitive Attributions for Smoking on 53
Subsequent Smoking Development

Chapter Four: Roles of Cognitive Attribution for smoking on 81
Subsequent Smoking Progression and Regression
Figure 1. Changes of Smoking Status in One Year 85

Chapter Five: Conclusions 107

Bibliography 115

v
List of Tables

Table 1: Demographic Characteristics and Smoking Status of the
Sample

36

Table 2: Cognitive Attributions for Smoking - Factor Pattern Matrix

38

Table 3: Relative Importance of Cognitive Attributions for Smoking,
Ranked by Self-report

39

Table 4: Relative Importance of Cognitive Attributions for smoking,
Ranked by Statistical Results: All Subjects

41

Table 5: Relative Importance of Cognitive Attributions for Smoking,
Ranked by Statistical Results: Males

42

Table 6: Relative Importance of Cognitive Attributions for Smoking,
Ranked by Statistical Results: Females

43
Table 7: Demographic Characteristics of the Sample

66

Table 8: Smoking Status and Smoking Progression over Two Years

68

Table 9: Associations between Cognitive Attributions and
Subsequent Smoking Progression

70

Table 10: Associations between Cognitive Attributions and
Subsequent Smoking Progression, Stratified by Gender and
Year-1 Smoking Status

72

Table 11: Mediation Tests

73
Table 12: Demographic Characteristics and Smoking Status of the
Sample

94

Table 13: Dynamic Changes of Smoking Status in One Year,
Stratified by Gender and Year-1 Smoking Status

95

Table 14: Cognitive Attributions for Smoking Associated with
Dynamic Changes of Smoking Status in One Year
97

vi
Abstract
Cigarette smoking poses a major public health problem worldwide.
Researchers have discovered numerous smoking determinants by linking people’s
personal and environmental factors with actual smoking behaviors. However, few
studies have explored how smokers themselves explain the causes of their smoking
behaviors. According to attribution theory, causal attributions are critical, because
they provide the basis for a person’s future actions. Studies have been undertaken to
ask people directly why they smoke. However, most of them focused on identifying
cognitive attributions for smoking. Few of them have investigated whether cognitive
attributions were associated with actual smoking and influential to subsequent
smoking. To fill in the gaps, the present research was conducted.
Subjects were 14,434 students randomly drawn from middle and high schools
in China seven cities. Self-administration surveys were conducted twice with one
year apart. Demographic characteristics, smoking status, and cognitive attributions
for smoking were assessed. Exploratory factor analysis was employed to categorize
cognitive attributions into themes. The relative importance of each cognitive
attribution was ranked by smoking status and gender. The roles of cognitive
attributions on subsequent smoking development were examined by multilevel
analyses. Susceptibility to smoking’s mediation effects were examined by Baron
and Kenny criteria, Sobel test, and program Prodclin. Polychotomous logistic

vii
regression was applied to examine how each cognitive attribution played its roles in
smoking trajectory.
Eight factors were generated representing eight themes of cognitive
attributions for smoking. They were curiosity, coping, social image, social
belonging, engagement, autonomy, mental enhancement, and weight control. Seven
of them were associated with current smoking. Six of them were associated with
subsequent smoking progression. Most associations were partially mediated by
susceptibility to smoking. Eventually, this research found how each cognitive
attribution influenced smoking at certain points in smoking trajectory and in
particular ways.
An essential recommendation from findings of this research is that anti-
smoking programs should be unique and evidence-based. To prevent adolescents
from smoking in long run, it is necessary to develop and implement comprehensive
anti-smoking programs incorporating primary and secondary smoking prevention
components and smoking cessation components.

1
Chapter One: Introduction
Attribution Theory
“Attribution theory” is a generic term for a cluster of theories designed to
describe how people explain the causes of their own and other people’s behaviors,
particularly with reference to social interactions. This theory assumes a clear
motivation on the part of people to make sense and predictability out of the world by
making rational assessments for the causes of their actions.
Heider is considered the first attributional theorist, first recognized as such by
Jones (Jones & Davis, 1965), although Heider never used the term “attribution
theory”. According to Heider’s analysis, people are motivated to observe, analyze,
and explain behaviors in order to manage their social environment well (Heider,
1958). The theory to describe the explanation process is referred to as “attribution
theory”, and the explanations given by people for behavior are called “attributions”.
Attributions are typically classified into two categories, personal and
situational. The first implies volition or intention on the part of the actor, and the
second has implies contextual or environmental factors that constrain or compel the
behavior. The dynamic of primary interest for the attribution theorists is how much
of causation is attributed to volition and underlying disposition of the actor, and how
much is attributed to contextual constraints that compel the actor to behave in a
particular way. Attribution theorists, especially Edward Jones and Harold Kelly,
view the attributor as a very rational being who follows orderly rules of perception

2
and inference, with the resultant attributions predictable by the information
specifically at hand. Causal attributions are seen as critical because they provide the
basis for one’s future actions, both immediate reaction to behaviors observed, and
later behaviors relevant to the person or context in question.
Heider’s ideas were further extended by Edward Jones and Keith Davis,
Harold Kelley, and Bernard Weiner. According to Jones and Davis’ Correspondent
Inference Theory, people try to infer whether an action corresponds to personal
characteristics of the actor. Three factors are important during the inference process,
which are the degree of choice, the expectedness of behavior, and the intended
effects or consequences of behavior (Jones & Davis, 1965). According to Kelley’
Covariation Principle, three kinds of covariation information can be used to infer
personal or situational attributions. They are consensus, distinctiveness, and
consistency. For behaviors with high in consistency, people usually make personal
attributions if the behaviors are in low consensus and low distinctiveness, and
situation attributions if the behaviors are in high consensus and high distinctiveness.
Behaviors that are low consistency are attributed to passing circumstances (Kelley,
1967). This theory tells what kinds of information elicit personal or situational
attributions in general. According to Weiner’ Three-dimensional Taxonomy,
attributions can be not only internal (personal) or external (situational), but also
stable or unstable, and controllable or uncontrollable. The three features of
attributions imply differently to actors for their self-esteem, expectations, and social

3
emotions (Weiner, 1986). Since it was developed, attribution theory has been used
in the fields of education (Ryan, Szechtman, & Bodkin, 1992), sports psychology
(Greenlees, Lane, Thelwell, Holder, & Hobson, 2005; Rees, 2007), clinical and
counseling psychology (Bardwell, 1986; Daltroy, 1993), interpersonal relations
(Opie & Miller, 1989), environmental psychology (Corrigan, Markowitz, Watson,
Rowan, & Kubiak, 2003), and health psychology (Anderson & Anderson, 1990).
The Attribution theorists acknowledge that causes people infer to explain
their own and other people’s behaviors can be incorrect, and some biases may occur
during the inference process. The most prominent biases that have been recognized
include the Fundamental Attribution Error, which indicates that people tend to
underestimate the situational attributions and overestimate the personal attributions
of other people’s behavior (Ross, 1977), the Actor-Observer Effect, which indicates
that people tend to attribute self behaviors to situational attributions and other
people’s behaviors to personal attributions (Jones & Nisbett, 1971; Monson &
Snyder, 1977), and the False-consensus Effect, which indicates that people tend to
overestimate the consensus for self attributions and behavior (Ross et al., 1977).
However, for the attribution theorists, the task is not to determine the true causes of
these events, but to discern people’s perceptions of the causes. These perceptions of
the causes influence people’s subsequent behaviors.
Attribution Studies Focusing on Cigarette Smoking

4
The adverse physical and social consequences of cigarette smoking have
stimulated much interest in the study of factors influencing people to smoke. Major
attention has been paid to multifaceted factors in people’s micro- to macro-
environments that may influence people to smoke, and people’s personal situations,
such as demographic characteristics, mood, dispositions (including depression,
anxiety, hostility, and sensation seeking), as well as knowledge, attitudes, beliefs in
relation to cigarette smoking. By linking people’s personal and environmental
factors with their cigarette smoking, researchers have identified a number of
determinants of smoking.
In the meantime, no matter whether guided by attribution theory or not,
studies have also been undertaken to directly ask people why themselves and other
people initiate, maintain, or remain addicted to cigarette smoking. Both qualitative
and quantitative strategies have been applied to collect such information. The lists of
reasons for smoking collected have been categorized into themes, either manually
based on the theoretical and conceptual knowledge and group discussions, or by
statistical approaches, such as Exploratory Factor Analysis. This research is of great
importance given the theoretical and conceptual ideas of attribution theory that
individuals are motivated to infer the causes of their own and other people’s
behaviors, and the perceptions of causes of previous and current behaviors influence
future actions, regardless of the accuracy of the perceptions.

5
Therefore, in the following paragraphs, we describe some of the major
attributed reasons that have been identified for adult smoking and adolescent
smoking, the variations of the reasons for smoking attributed by different people for
their own and other people’s smoking, the utility of attribution theory in the
attribution related research, some plausible limitations, and future directions.
Attributed Reasons for Adult Smoking
Addiction and relaxation were the reasons most commonly given for adult
smoking. Other reasons have been found as well by researchers trying to investigate
reasons or motives of people’s own and others’ cigarette smoking. While there were
some similarities among the reasons for adult smoking identified by different
researchers, some differences exist as well. For example, a sample of British adult
and adolescent smokers and ex-smokers were asked to complete a checklist about
their main occasions for smoking. Seven reasons for smoking have been yielded
from EFA, which were nervous irritation smoking, relaxation smoking, smoking
alone, activity accompaniment, food substitution, social smoking, and social
confidence smoking (McKennell, 1970). In an American sample, nine motives of
smoking have been generated by EFA, which were addiction, affective smoking,
forgetful smoking, health-nuisance, relaxation, stimulation, sophistication,
sensorimotor pleasure, and unpleasant habit (Kleinke, Staneski, & Meeker, 1983).
When asking a sample of American smokers about five motives of smoking, which
were psychological addiction, relaxation, physical addiction, pleasant activity, and

6
weight control, subjects perceived psychological addiction as the primary reason for
their own smoking, followed by relaxation (Jenks, 1994a).
A six-factor model for reasons for smoking, including negative-affect
reduction, psychological addictive smoking, habit smoking, pleasure-relaxation,
stimulation, and sensorimotor manipulation, was first developed from Tomkins’
affect management model of smoking (Tomkins, 1966). It has been reported in
many studies (Berlin et al., 2003; Bosse, Garvery, & Glynn, 1980; Coan, 1973;
Costa, McCrae, & Bosse, 1980; Ikard & Tomkins, 1973; Leventhal & Avis, 1976;
O'Connell & Shiffman, 1988; Tate, Pomerleau, & Pomerleau, 1994) and
operationalized as the Reasons for Smoking Scale (RSS). The internal consistency
and test-retest reliability of the RSS were found to be both adequate (Tate, Schmitz,
& Stanton, 1991) and the factor loadings in a Confirmatory Factor Analysis were
found to be good (Currie, 2004).
Attributed Reasons for Adolescent Smoking
Similar to the findings about adult smoking, there are both similarities and
differences on reasons for adolescent smoking across studies. For example, when
examining reasons for smoking and not smoking in a large sample of 13-year-olds
from the general population of New Zealand, three factors were generalized from 18
reasons for smokers to smoke, which were relaxation/pleasure, friends, and image,
and four factors were identified from 17 reasons for nonsmokers’ reasons for not
smoking, which were social context, effects of smoking, access, and health reasons

7
(Stanton, Mahalski, McGee, & Silva, 1993). When asking a sample of 10
th
graders
in rural and urban districts of Washington State why they smoked, curiosity, social
norms, social pressure, pleasure and addiction were mentioned quite often (Sarason,
Mankowski, Peterson, & Dinh, 1992).
Generally speaking, curiosity, image, and social situations were mentioned as
reasons of smoking by a number of studies. Among various stages of smoking,
curiosity has been found to be the most relevant to initiation of smoking (Sarason et
al., 1992). Even at late adolescence, curiosity was still one of the most important
reasons for starting to smoke, reported by more than half of the adolescents in one
study who have started to smoke (Cronan, Conway, & Kaszas, 1991).
The attempt to present a positive image to other people, especially peers, is
another reason commonly given by adolescents for their cigarette smoking. To
achieve the status of cool, tough, and mature has been reported widely as a reason of
smoking (Allbutt, Amos, & Cunningham-Burley, 1995; Rugkasa et al., 2001; Treacy
et al., 2007). A study that followed a group of male recruits for one year after they
entered the Navy Training Center demonstrated that three out of seven reasons,
which were given by more than one third of the recruits about why they started
smoking during their first year in the service, were related to social image: to show I
wasn’t afraid, to want to be cool, and to look like an adult (Cronan et al., 1991).
Image-related attributions remained important for smoking across ages (Stanton &
Silva, 1993), were more important among females than among males (Barton,

8
Chassin, Presson, & Sherman, 1982), and predicted future cigarette smoking (Aloise-
Young, Hennigan, & Graham, 1996).
Social occasions and social relations were reported by several researchers as
a primary reason for adolescent smoking (Allbutt et al., 1995; Cronan et al., 1991;
Rugkasa et al., 2001; Sarason et al., 1992; Stanton & Silva, 1993; Treacy et al.,
2007). Adolescents were motivated to smoke to make friends, maintain friendship,
and achieve group membership and identity. Some studies reported this motivation
as “social pressure”(Sarason et al., 1992). However, other studies found that
adolescents were active actors rather than passive recipients of cigarette smoking
(Rugkasa et al., 2001). According to the later studies, adolescents decided to smoke
not under the physical pressure coming from peers or any other surrounding
environments. Instead, they experienced smoking because they actively wanted to
“join in” peer groups, to create or increase prestigious status, and to obtain group
membership. "Friends" as a reason for smoking has been found to have a small
degree of consistency across age (Stanton et al., 1993). Social norms and social
pressure were found to be the most frequently given reasons for beginning smoking,
reported by more females than males(Sarason et al., 1992).
On one hand, the relative importance for some reasons for adolescent
smoking might not be quite identical to that for adult smoking. For example, coping
with life and addiction were the common reasons given for adult smoking. However,
they were not similarly common reasons given for adolescent smoking. In some

9
studies, adolescents did express that smoking helped when they were stressed
(Allbutt et al., 1995), and helped them calming down (Cronan et al., 1991).
However, in other studies, adolescents felt that coping with life was a reason for
adult smoking, and had nothing to do with children themselves. In their views,
depression, anxiety, and stress are all adult business, which they might experience
only when they grow up, so was addition to smoking that might be resulted from
some mental distress. Adolescents in these studies disagreed that they were
motivated to smoke by personal reasons; instead, they explained that they were
motivated to smoke for social reasons, such as what have been mentioned above, to
gain good social image and group membership (Rugkasa et al., 2001).
On the other hand, some reasons for adolescent smoking are similar to those
for adult smoking in terms of the relative importance. For example, although in a
qualitative focus group interview, several individual adolescents expressed weight
concern as attributions for both their own smoking and parent smoking, such as, “I
was really fat and I lost weight with smoking”, “If my mum stopped smoking, she
will put weight on” (Allbutt et al., 1995), another quantitative study found that the
percentage of adolescents who smoked for the purpose of losing weight was not
high, compared with percentages of adolescents who smoked due to other reasons,
such as curiosity, friend smoking, image, and boredom (Cronan et al., 1991). This is
consistent with findings for adult smoking. For example, in one study, weight loss

10
has been reported to be the least motivation for adult smoking, among 6 motivations
investigated (Jenks, 1994a).
Variation of Reasons for Smoking Given by Different Kinds of People
The inference of reasons for smoking is a cognitive process, depending
heavily upon individuals’ personal dispositions, knowledge, attitudes, beliefs, mood,
as well as micro- to macro- surrounding environments. Therefore, different people
might perceive reasons for smoking quite differently, regardless of their actual
reasons for smoking. It has been reported that reasons for smoking given by
smokers, non-smokers, and ex-smokers were different (Eiser, Sutton, & Wober,
1977; Hines, Fretz, & Nollen, 1998; Kleinke et al., 1983; Sadava & Weithe, 1985;
Stanton et al., 1993), given by smokers for their own and other smokers’ smoking
were different (Jenks, 1994a; Palmqvist & Martikainen, 2005), and given by children
for child smoking and adult smoking were also different (Treacy et al., 2007).
Reasons for smoking given by males and females were found to be identical by some
studies (Grube, Rokeach, & Getzlaf, 1990; Jenks, 1994b; Palmqvist & Martikainen,
2005; Stanton et al., 1993), but not by a few other studies (Anderson & Anderson,
1990; Sarason et al., 1992).
Utility of Attribution Theory as a Tool for Explaining the Variations in Reasons for
Smoking Given by Different Kinds of People
In order to disclose the underlying patterns for the perceptions that different
people have for the reasons of smoking behaviors, attribution theory has been

11
applied. As has been indicated, according to Jones & Nisbett (1971), actors tend to
explain their own behavior more in terms of external factors, and by comparison,
observers are more likely to explain the actors’ behavior by reference to internal
factors (Jones & Nisbett, 1971). Therefore, studies have been conducted to
investigate whether and how “actors” and “observers” gave reasons for own and
other people’s smoking differently. In these studies, when people were asked to give
reasons for their own smoking, they were treated as “actors”, for example, smokers’
giving reasons for their own smoking. In the meantime, when people were asked to
give reasons for other people’s smoking, they were treated as “observers”, for
example, non-smokers’ giving reasons for average smokers’ smoking, or smokers’
giving reasons for other smokers’ smoking. The common hypothesis for these
studies was that “actors” would be more likely to attribute their smoking to external
factors, and in comparison, “observers” would be more likely to attribute other
people’s smoking to internal characteristics.
Most of the attribution studies tried to explain the research findings in terms
of the measures’ internal and external features. Some studies differentiated measures
into internal and external prior to statistical analysis, and then assessed how the
internal and external factors differ among “actors” and “observers” (Sadava &
Weithe, 1985). A few other studies even directly used the terms of attribution
theory, for example, internal and external attributions, and task difficulty, luck,
ability, efforts (Wright, 1980). As a result, the Actor-Observer Error was supported

12
by some studies (Eiser et al., 1977; Eiser, Sutton, & Wober, 1978; Kleinke et al.,
1983; Wright, 1980), but not by others (Jenks, 1994a; Sadava & Weithe, 1985).
Limitations and Future Directions
In review of the attribution studies conducted to date, there are a number of
inconsistencies. The inconsistencies might result from the diversity of the targeted
population in terms of age groups, ethnicities, and countries with diverse political,
economic, cultural and other environmental backgrounds. The inconsistencies might
also come from the study design, measures, coding methods for information
collected, and outcomes of interest. In addition, the smoking status (for example,
smokers, non-smokers, and ex-smokers) was defined quite differently in studies, and
the stages across smoking development were generally not differentiated.
In addition to the inconsistencies on the findings from different attribution
researchers, studies have shown that the reasons expressed by people can be
inaccurate. For example, Kleinke et al (1983) found that, while smokers gave
significantly more agreement to one smoking motive: relaxation, the most important
motives predicting their cigarette consumption were actually addiction and affective
Smoking (Kleinke et al., 1983).
Despite of the inconsistencies and inaccuracy of the reasons that have been
given by people for their own and other people’s cigarette smoking, the attribution
research conducted so far may be of critical importance for understanding smoking
onset and progression. For the attribution theorists, the task is not to determine the

13
true causes of these events, but to discern people’s perceptions of the causes,
because these perceptions of the causes influence people’s subsequent behaviors.
Therefore, it is premature to say that the inconsistency and inaccuracy of the reasons
given to smoking are the “limitations" of attribution studies.
However, based upon the theoretical and conceptual ideas expressed by
attribution theory, there are indeed some limitations among previous attribution
studies.
Insufficient Utilization of Attribution Theory
Regardless of the inference process for the causes of people’s own and
others’ behaviors, the attribution theory contains two basic underlying concepts.
One is that people are motivated to infer the causes of their own and other people’s
behaviors. Another is that the perception of the causes will influence the
individual’s subsequent behaviors. Therefore, the former idea can be helpful for
understanding about smoking, and the latter idea can be helpful for predicting
smoking. In other words, if we can identify the attributed reasons for smoking and
their relative importance to actual smoking behaviors, we will understand why
people smoke; if we know what attributed reasons for smoking influence subsequent
smoking, we will understand what can predict smoking. However, most of the
previous attribution research focused on identifying attributed reasons for smoking
and the self-reported relative importance of the attributed reasons. Only few studies
have linked the attributed reasons for smoking with the actual cigarette smoking

14
behaviors. Even fewer studies have tried to investigate whether cognitive
attributions influence future cigarette smoking behaviors. In one study that we could
identify, it was reported that most of the reasons given by adolescents at 13 years old
(relaxation, friends' smoking and image) did not significantly predict smoking two
years later, while the immediate effects of smoking (taste, smell, feeling sick and
feeling silly) as a reason for not smoking were significantly predictive of non-
smoking at age 15 (McGee & Stanton, 1993). It is unknown whether the failure of
detecting the predictability of given reasons for smoking is due to the specific
sample, the time interval between the collection of reasons for smoking and the
subsequent smoking behavior, or the measures chosen to use. Further investigations
are needed to obtain a clearer understanding of this issue. Longitudinal studies can
provide more clues about causality and prediction.
Unknown Cognitive Attributions for Chinese Adolescent Smoking
Most of the previous smoking related attribution studies were conducted in
European countries, such as England (Eiser et al., 1977, 1978), Scotland (Allbutt et
al., 1995), Ireland (Rugkasa et al., 2001; Treacy et al., 2007), France (Berlin et al.,
2003), and Finland (Palmqvist & Martikainen, 2005), American countries, such as
the United States (Cronan et al., 1991; Jenks, 1994a, 1994b; Kleinke et al., 1983;
Sarason et al., 1992), and a few other countries (McGee & Stanton, 1993; Zoller &
Maymon, 1983). They have rarely been conducted in China, where about one third
of smokers worldwide reside.

15
The prevalence of smoking is very high in China. A population survey
conducted in 30 provinces of China in 1996 among urban and rural residents aged
15-69 years old revealed that 37.6% of Chinese had ever smoked for at least 6
months, 35.3% smoked currently, and 34.1% smoked habitually (Yang et al., 1999).
The prevalence rate of smoking increased 3.4 percentage points from 1984 to 1996
(Yang et al., 1999). The number of smokers increased 30 million from 320 million
in 1996 to 350 million in 2002 (Yang et al., 1999; Yang, Ma, Liu, & Zhou, 2005).
The average number of cigarettes consumed per day increased from only one
cigarette in 1952 to four in 1972, ten in 1992, and fifteen in 1996, remaining about
15 until 2002 (Liu et al., 1998; Yang et al., 2005; Zhang & Cai, 2003). The high
smoking rates have caused more than half of Chinese non-smokers to be exposed to
secondhand smoke at least one day per week and about 15 minutes per day. More
than 60% of female non-smokers who are at the ages of childbearing were passively
exposed to tobacco smoke. If we count smokers and passive smokers together, more
than 600 million Chinese people are either directly or indirectly exposed to cigarette
smoke, accounting for about 72% of the total population.
More serious is the increase of smoking among Chinese adolescents. From
1984 to 2002, smoking among adolescents increased steadily (Yang et al., 1999;
Yang et al., 2005). In 1996, more than 18 million Chinese adolescents aged 15-19
years old had tried smoking, among which 9 million ever smoked for at least 6
months, accounting for 18% of all males and 0.28% of all females (Zhang & Cai,

16
2003). Another population survey conducted among adolescents aged 11-20 years in
12 urban and 12 rural areas located in 16 provinces of China revealed that 47.8% of
males and 12.8% of females ever smoked, and 9.4% of males and 0.6% of females
ever smoked weekly for at least 3 weeks (Yang et al., 2004). While the prevalence
rates of adolescent smoking increased over decades, the age of cigarette smoking
onset decreased. In 1984, 70% of males and 50% of females initiated smoking
before 24 years old, with an average age of 23 years old. However, the average age
of smoking onset decreased 3 years by 1996 among both males and females
respectively (Yang et al., 1999). This is a serious public health problem, because
most smokers started smoking in adolescence. The earlier people initiate smoking,
the more likely they will progress to higher levels of smoking, the more likely to
continue with smoking until adulthood and get addicted to nicotine (Paavola,
Vartiainen, & Puska, 1996), the less likely to quit, and the more likely to suffer from
tobacco related diseases.
Therefore, while more effective interventional campaigns are needed in
China, early intervention is of particular importance for preventing adolescents from
initiating smoking and progressing to higher levels of smoking in order to curtail the
overall smoking all over the country across age groups. To achieve the goals, it is
necessary to understand the causes of Chinese adolescent smoking. Otherwise, if the
current smoking pattern in China remains unchanged, it is estimated that 100 million

17
men currently less than 30 years will die from smoking-related diseases with half
dying in middle age and half in old age (Zhang & Cai, 2003).
Similar to what has been done in the United States and other western
countries, by linking individuals’ personal and environmental factors with smoking
behaviors, empirical studies conducted in China have found that multifaceted factors
influence Chinese adolescent smoking. However, the overwhelming majority of
these studies did not ask adolescents directly what were the reasons for their own and
other people’s smoking behaviors. Previous attribution studies conducted in other
countries have provided pieces of useful information about why people smoke.
However, these studies cannot simply be generalized to China where people live in
different physical and social environments.
For example, some aspects of Chinese culture might place people at higher
risk of smoking. Unlike in most of the developed countries, male smoking is well
accepted in Chinese society. It is quite common that people offer cigarettes to one
another while meeting and talking. It is seen by a lot of people as courteous to
accept cigarette offers and as impolite to refuse them. Cigarettes can be circulated as
gifts during holidays, festivals and other special occasions, and to smoke good
quality and brands of cigarettes even represents a symbol of prestigious and high-
ranked status. For these reasons, it is not too surprising that more than two thirds of
Chinese men smoke, and more than half of male health care professionals and male
teachers, who are taking responsibilities of (health) education for patients and

18
students and acting as role models, smoke (Yang et al., 1999). In the old days,
certain protective effects from the traditional Chinese culture strongly discouraged
women to smoke. However, such protective effects may have reduced and an
obvious increase in smoking among female adolescents and young adults has been
observed during recent years. As a matter of fact, a proportion of smokers might not
really want to smoke at all or certain circumstances. However, living in a society in
which collectivism is more dominant than individualism, they might still light up
cigarettes in order not to feel too distinct from other people who are smoking.
Holding such kinds of cultural and social values, smokers engage in cigarette
smoking, regardless of their knowledge and awareness of a number of health hazards
that might result from cigarette smoking.
It is evident that, in some countries where smoking is socially acceptable,
children are less disapproving of smoking by adults (Zoller & Maymon, 1983), and
children who perceive smoking as a “normal” part of the adult world might be more
likely to expect that they will smoke as adults and might not have the capacity to
resist smoking (Treacy et al., 2007). This might be true in China. In addition, more
than this might put Chinese adolescents at higher risk of smoking. On one hand, the
Chinese academy field is very competent. In order to obtain higher education that is
only available for limited percentage of students who are excellent, adolescents
themselves have to work very hard to achieve and maintain good academic
performance and be superior to others. On the other hand, most of the families in

19
China, urban especially, comply with the one-child family planning policy and have
only-children. Parents who have high expectations about their children’s future
might impose extra pressure such that the only-children have to spend even more
time and effort on study, and other activities such as sports and musical instruments.
Consequently,, the heavy daily burden may make adolescents feel depressed,
stressful, and even angry. Cigarette smoking might be used as a tool to cope with
these mental problems.
Therefore, while some of the attributions for adolescent smoking might be
similar across countries, some attributions might be specific to a certain country.
One study has shown that although smoking behavior of high school students in
Israel is comparable to that of high school students in the United States, the reasons
for smoking and not smoking as perceived by the students themselves were different
in the two countries both qualitatively and quantitatively (Zoller & Maymon, 1983).
In China, we assume that, other than some universal attributions for adolescent
smoking, there are some unique attributions as well, either internal, such as mental
health problems among adolescents which is not common in a lot of other countries,
or external, such as the stressors in the surrounding environments that might result in
adolescents’ mental health problems and the Chinese culture that accepts cigarette
smoking. Therefore, both qualitative and quantitative attribution research will be
helpful for better understanding about the overall and unique reasons for adolescent
smoking there. The knowledge obtained will be important for smoking prevention

20
and control not only for China, but also for the entire world now that one third of
smokers worldwide resides in China.
Stages of Smoking
Smoking has a developmental trajectory from initiation to addiction.
Mayhew et al differentiated 5 stages in the trajectory of smoking development: 1)
never smoking which was characterized by having never smoked no matter whether
or not having thought about smoking, 2) tried smoking which was characterized by
having tried first few cigarettes, 3) experimental smoking which was characterized
by smoking at a gradual increase in the frequency and an increase in the variety of
situations in which cigarettes are used, 4) regular smoking which was characterized
by progression beyond sporadic smoking to smoking on a regular though still
infrequent basis, and 5) established smoking which was characterized by smoking
daily, or almost every day (Mayhew, Flay, & Mott, 2000). This classification of
smoking stages has been applied in numerous studies (Baade & Stanton, 2006).
While some of the attributed reasons for smoking might be common across
stages of smoking, some others might be specific to certain stage(s) of smoking only.
Non- attribution related studies have indicated that the influences of smoking at one
stage might not be identical to those at other stages (Leventhal & Cleary, 1980;
Mayhew et al., 2000). One attribution related study tried to classify smoking into
two stages, beginning smoking and current smoking, and investigated how subjects
in the two groups gave reasons for smoking. As a result, curiosity, social norms, and

21
social pressure were the most frequently given reasons for beginning smoking, and
pleasure and addiction were mentioned most often for current smoking (Sarason et
al., 1992). Therefore, future investigations are needed to explore the attributions of
smoking at each stage in the trajectory of smoking development, rather than at one or
two aggregated or mixed stage(s) only.
Dynamic Changes of Smoking
As we can imagine, each absolute smoking status might result from initiation
of smoking, progression to higher stages, maintenance at the same stages, or
regression to lower stages over time. Of the above conditions, those people who
initiate smoking or progress to higher stages are at highest risk of established or
addicted smoking and smoking related disasters, and should draw the most attention.
Therefore, for future attribution studies, efforts are worthy to be undertaken to
explore attributed reasons for initiation and progression of smoking, other than for
the absolute smoking status only. In addition, since the progression from different
stages of smoking might have different mechanisms or pathways, combining with
the classifications of initial smoking stages, the reasons for smoking identified by
future attribution studies will have profound implications for the development of
primary intervention of smoking, which is to prevent non-smokers from initiating
smoking, and secondary intervention of smoking, which is to prevent smokers from
progressing to higher stages of smoking.

22
Possible Mediation Effects Caused by Susceptibility to Smoking
“Susceptibility” has been defined as “a lack of firm commitment against
cigarette smoking” (Jackson, 1998; Pierce, Choi, Gilpin, Farkas, & Merritt, 1996).
Susceptibility of adolescent smoking has been reported to be predicted by peer
smoking (Gritz et al., 2003; Presson et al., 1984; Straub, Hills, Thompson, &
Moscicki, 2003), family smoking (Presson et al., 1984), exposure to tobacco-related
advertisement (Gritz et al., 2003; Straub et al., 2003), mental distress including stress
(Booker, Gallaher, Unger, Ritt-Olson, & Johnson, 2004; Straub et al., 2003) and
hostility (Hampson, Andrews, & Barckley, 2007), and stressful life events including
negative peer-related and school-related events, and positive personal-related events
(Booker et al., 2004). Susceptibility has also been reported as a cognitive predictor
of cigarette smoking among adolescents. Its prediction to the onset of smoking is
very strong among adolescents who don’t smoke, even stronger than exposure to
family members and friends who smoke cigarettes (Jackson, 1998). However, for
adolescents who have already smoked, susceptibility is less important to predict the
future smoking, compared with variables representing the stages of smoking
(Stanton, Barnett, & Silva, 2005). Studies showed evidence that susceptibility to
smoking was not an independent risk factor for adolescent smoking, but rather a
potential mediating variable (Gritz et al., 2003). As a matter of fact, susceptibility’s
mediating effects are supported by the Theory of Reasoned Action (TRA) (Ajzen &
Fishbein, 1980) and the Theory of Planned Behavior (TPB) (Ajzen, 1985), where

23
attitudes, perceived norms, and perceived behavioral control (in TPB only) influence
behaviors through behavioral intentions. The TRA and TPB have been applied and
approved to be useful to predict smoking behaviors among adolescents (Hanson,
1999; Harakeh, Scholte, Vermulst, de Vries, & Engels, 2004; Maassen, Kremers,
Mudde, & Joof, 2004; O'Callaghan, Callan, & Baglioni, 1999), including Chinese
adolescents (Guo et al., 2007). Therefore, when examining the influences of
cognitive attributed reasons for smoking on subsequent smoking behaviors, it will be
interesting to see whether the influences occur through the mediating effect caused
by susceptibility of smoking. This endeavor will be helpful to understand the
potential pathway that attributed reasons for smoking might influence the subsequent
smoking behaviors.
The Present Research
To fill in the above-mentioned gaps, the overall goal for this research was to
use attribution theory as a tool to investigate cognitive attributions for Chinese
adolescent smoking and their roles on subsequent smoking behaviors. To achieve
the goal, three studies have been conducted. Study One investigated cognitive
attributions for adolescent smoking, their relative importance, and their associations
with actual smoking behaviors. Study Two investigated whether and how cognitive
attributions were influential to subsequent smoking development. Study Three
investigated how cognitive attributions are associated with each of the dynamic
changes of smoking over years, including smoking progression, maintenance, and

24
regression. Through the three studies, we hoped to not only understand why some
Chinese adolescents smoked while others didn’t, but also understand why some
Chinese adolescents tried/experimented smoking, smoked regularly, and smoked
daily, while others didn’t. In the meantime, we hoped to not only investigate the
overall roles of cognitive attributions in subsequent smoking development, but also
the specific role of each of the cognitive attributions in smoking progression from
each of the initial stages to each of the higher stages and smoking regression from
each of the initial stages to each of the lower stages. Findings from this research will
be instructive to development and implementation of effective anti-smoking
programs, especially targeting Chinese adolescents.

25
Chapter Two: Cognitive Attributions for Smoking: Findings from An Attribution
Study Conducted in China Seven Cities

The adverse physical and social consequences of cigarette smoking have
stimulated much interest in the study of factors influencing people to smoke. Major
attention has been paid to multifaceted factors in people’s micro- to macro-
environments that may influence people to smoke, and people’s personal situations,
such as demographic characteristics, social economic status, dispositions (including
depression, anxiety, hostility, and sensation seeking), as well as knowledge,
attitudes, and beliefs in relation to smoking. By linking people’s personal and
environmental factors with their smoking behaviors, researchers have discovered a
number of determinants of smoking (Moolchan, Ernst, & Henningfield, 2000;
Schepis & Rao, 2005; Turner, Mermelstein, & Flay, 2004; Tyas & Pederson, 1998).
However, it remains unclear why people are cognitively motivated to smoke,
regardless of their knowledge and awareness of numerous negative consequences of
smoking, and how smokers themselves perceive the causes of their cigarette
smoking.
According to attribution theory (Heider, 1958; Jones & Davis, 1965; Kelley,
1967), people are motivated to make rational assessments for the causes of personal
behaviors in order to manage social environment well. The causes that people infer
to explain their own behaviors and the behaviors of others can be incorrect, and
biases may occur, such as the fundamental attribution error which indicates that

26
people tend to underestimate the situational attributions and overestimate the
personal attributions (Ross, 1977), the actor-observer effect which indicates that
people tend to attribute self behaviors to situational attributions and other people’s
behaviors to personal attributions (Jones & Nisbett, 1971; Monson & Snyder, 1977),
and the false-consensus effect which indicates that people tend to overestimate the
consensus for self attributions for behavior (Ross et al., 1977). However, the
perceptions of causes are influential to subsequent actions, regardless of their
accuracy. No matter whether guided by attribution theory and other theories or not,
studies have been undertaken to ask people directly why they and other people
initiate, maintain, or become addicted to smoking. Both qualitative and quantitative
strategies have been applied to collect such information and numerous cognitive
attributions for smoking have been identified (Allbutt et al., 1995; Aloise-Young et
al., 1996; Barton et al., 1982; Cronan et al., 1991; Rugkasa et al., 2001; Sarason et
al., 1992; Stanton et al., 1993; Treacy et al., 2007).
While previous attribution studies have provided some useful information
about why people smoke, the majority of them focused on identifying cognitive
attributions for smoking. Some of them have assessed relative importance of
attributions. Only few of them have linked cognitive attributions with actual
smoking behaviors (Kleinke et al., 1983). A comprehensive investigation about
cognitive attributions, their relative importance, and their associations with actual
smoking behaviors will help us obtain better understanding about rationale of

27
smoking, which can further help us with development of effective anti-smoking
strategies. Therefore, the present study was conducted for this purpose.
There are five key features in this study. First, the target population was
Chinese adolescents. Previous attribution studies have been conducted primarily in
European countries (Allbutt et al., 1995; Berlin et al., 2003; Eiser et al., 1977, 1978;
Palmqvist & Martikainen, 2005; Rugkasa et al., 2001; Treacy et al., 2007), the
United States (Cronan et al., 1991; Jenks, 1994a, 1994b; Kleinke et al., 1983;
Sarason et al., 1992), and a few other countries (McGee & Stanton, 1993; Zoller &
Maymon, 1983). They have rarely been conducted in China, where one-third
smokers in the world reside and smoking is extremely prevalent (Yang et al., 1999;
Yang et al., 2004). We can’t simply generalize current research findings to China
where people live in different physical, social, and cultural environments. We
acknowledge that some attributions for smoking can be similar in all countries.
However, some others can be unique in one or some countries only. For example,
male smoking is well accepted but female smoking is strongly discouraged in the
society of China, which is uncommon in most western countries. Zoller et al found
that, although smoking behavior of high school students in Israel was comparable to
that of high school students in the United States, the reasons for smoking and not
smoking as perceived by the students themselves were different in the two countries
both qualitatively and quantitatively (Zoller & Maymon, 1983). Moreover, this
study targeted Chinese adolescents, because adolescence is a critical period when

28
most smokers initiate uptake of cigarettes and understanding about causes of
smoking at this period is essentially important. Second, rather than identifying
sporadic cognitive attributions for smoking, we tried to explore the themes that a
number of individual attributions might represent. Third, based on stages of
smoking, we classified the total sample into four groups: never smokers, lifetime
smokers, past 30-day smokers, and daily smokers, which are equivalent to never
smokers, tried/experimental smokers, regular smokers, and established smokers
classified by Mayhew et al (Mayhew et al., 2000). By exploring cognitive
attributions associated with each stage of smoking, we hoped to understand not only
why some Chinese adolescents smoked while others didn’t, but also why some
Chinese adolescents tried/experimented smoking, smoked regularly, or smoked
daily, while others didn’t. Fourth, while investigating the relative importance of
each cognitive attribution for each stage of smoking, we compared two types of
rankings: those derived from self-reports and those derived from strength of
associations from statistical analyses. By comparing the two types of rankings, we
hoped to evaluate how accurate cognitive attributions were. Lastly, we conducted all
of the above investigations among males and females, respectively. In China,
smoking is primarily a male behavior (Yang et al., 1999; Yang et al., 2004), the
reasons of which are worthy to be explored.
We hypothesized that 1) a number of individual cognitive attributions could
be classified into categories representing particular themes such as curiosity, social

29
images, social relations, coping, engagement, and mental enhancement; 2) cognitive
attributions would vary across stages in the trajectory of smoking uptake; 3) the
relative importance of each attribution ranked by self-report would not be exactly the
same as that ranked by strength of association, since the attributions were the
perceptions of the reasons for smoking, rather than the true reasons; and 4) cognitive
attributions for male smoking and female smoking were different. Given much
higher smoking rates among Chinese males, we assumed that more cognitive
attributions were significantly associated with male smoking, and, for cognitive
attributions that were significantly associated with both male smoking and female
smoking, the associations were stronger among males.

METHODS
The data are part of the China Seven Cities Study (CSCS), a larger project in
China to assess the effects of changing economic and social factors on health
behaviors including tobacco use. The information will be used to develop
community based smoking and other drug abuse prevention programs. The CSCS
includes seven cities in four regions of China: Northeastern (Harbin, Shenyang),
central (Wuhan), southwestern (Chengdu, Kunming), and coastal (Hangzhou,
Qingdao).

30
Participants
Participants were recruited from schools in each of the seven cities. The
schools were selected using a stratification process for (a) administrative district
median income and (b) school academic performance. The administrative districts
with highest, middle, and lowest median income in each city were first identified.
Then the local Education Committees were asked to group the middle and high
schools in each identified district into three levels of academic performance. This
process resulted in a total of nine school clusters from three districts representing
three levels of district income crossed with three levels of academic performance.
One middle school and one traditional high school were randomly selected
from each of the nine clusters in the matrix to participate in the study. One
classroom in the 7
th
and 8
th
grades in the selected middle schools and one classroom
from the 10
th
and 11
th
grades in the selected high schools were recruited to
participate. In addition, one professional high school was selected from each district,
and the three professional high schools selected from three districts in each city
matched on enrollment, type of vocational training, and ratio of male to female
enrollment. Major courses of study within each professional school were randomly
selected, and students in these majors were recruited from the 10
th
and 11
th
grades to
participate in the study.
In summary, 9 middle schools, 9 high schools, and 3 professional schools
were selected from each city, and a total of 147 schools were selected across all

31
seven cities. A total of 15,516 students in the middle and high schools were invited
to participate in the baseline survey. Of this total, 802 students (5.2%) were
excluded because the parent or student declined to participate. Another 253 students
with completed consents from parents (1.6% of those invited to participate) were
absent on the data collection day. Therefore, the participation rate was 93.2%
(14,461 of students invited to participate). Another 27 students who provided
incomplete surveys at baseline were excluded from the study. The total number of
participants included 93.0% (14,434) of the middle and high school students invited
to participate.
Procedures
Survey data were collected from middle school and high school students.
Investigators at the University of Southern California (USC) provided guidance and
training in the research program. The local Center for Disease Control and
Prevention (CDC) in six of the cities and the Institute for Health Education (IHE) in
the city of Kunming helped develop the surveys, gain access to students in the
schools, obtain informed consent, and collect the data. The informed consent and
data collection procedures were reviewed and approved by both the USC and
Chinese Institutional Review Boards. The students completed the surveys in school.
Quality control of the data collection process was monitored by a team that
included faculty members from a university selected in each of the seven cities,
public health officers at the China National CDC, and faculty and staff at USC.

32
Local team members observed data collection in selected classrooms in each of the
seven cities and reported to team members at USC. If any deviation from the
standard protocol was detected, USC team members provided rapid responses and
corrections within 24 hours directly to the associated data collection personnel in the
field.
Measures
The questionnaire was developed through cooperation among USC
researchers and the directors and staff at the CDC’s and IHE in China. Three USC
graduate students fluent in both English and Mandarin individually translated the
questionnaire into Mandarin Chinese, and then, reached a group consensus for the
translation. The questionnaire was also checked and approved by public health
workers and educators in each of the seven cities to ensure comprehensibility. More
details about the methodology of the CSCS were reported elsewhere (Anderson
Johnson et al., 2006; Grenard et al., 2006; Guo et al., 2007).
Demographic characteristics included age, gender, ethnicity, and geographic
regions. Age, gender, and ethnicity were reported on the self-administered surveys
and the geographic regions were the cities where the samples were drawn.
Cognitive Attributions for smoking were measured by one item: “I smoke, (or
might smoke), because: (Circle all that apply)”. Seventeen possible reasons for
smoking were listed as response options. Each of the seventeen response options
was dichotomously recoded as “1”, if the student had circled it, and “0”, if the

33
student hadn’t circled it. A response with a score of “1” means that student has
attributed it as one of the reasons for his or her cigarette smoking. Previous studies
ever used similar item to assess meanings of smoking either qualitatively among
Chinese American college students (Hsia & Spruijt-Metz, 2003) or quantitatively
among multiethnic adolescents in the United States (Spruijt-Metz, Gallaher, Unger,
& Johnson, 2005; Spruijt-Metz, Gallaher, Unger, & Anderson-Johnson, 2004).
Lifetime smoking was assessed by one question: “Have you ever tried
cigarette smoking, even a few puffs?” Two response options were provided as 0
(No) and 1 (Yes).
Past 30-day smoking was assessed by two questions. One was “During the
past 30 days, on how many days did you smoke cigarettes?” Seven response options
were provided, ranging from 1 (0 days) to 7 (All 30 days). Another was “During the
past 30 days, on the days you smoked, how many cigarettes did you smoke per day?”
Seven response options were provided, ranging from 1 (I did not smoke cigarettes
during the past 30 days) to 7 (More than 20 cigarettes per day). This variable was
dichotomously recoded as 1 (Yes) if a student chose response options of 2~7 for
either of the two items, and 0 (No) if a student chose response option of 0 for either
or both of the two items.
Daily smoking was assessed by one item: “Have you ever smoked cigarettes
daily, that is, at least one cigarette every day for 30 days?” Two response options
were provided as 1 (Yes) and 0 (No).

34
Statistical Analyses
Frequencies were calculated to describe demographic characteristics and
smoking status of the samples. Chi-square tests were employed to examine
differences between males and females on these variables.
Exploratory factor analysis (EFA) was applied to generate factors from a list
of variables representing cognitive attributions for smoking. Because these variables
were dichotomous, a tetrachoric correlation matrix of these variables was generated
for EFA. Principal Components Factor methods were used to extract factors. An
oblique rotation was used to allow correlations among the factors. The number of
factors was determined based on the theoretical knowledge and the statistical rule
of "eigenvalue greater than 1". The models with different number of factors were
compared to choose a model that better addressing the fundamental questions of this
research. Once the EFA was done, variables that load on the same factors were
summed to create new variables representing the themes of a list of attributions for
smoking. Each of the new variables was dichotomously recoded as “0” if the value
of the sum was “0”, and as “1” if the value of the sum was equal to or greater than
“1”.
The relative importance of each attribution for smoking was ranked in two
ways. First, the rankings were determined by the magnitudes of frequencies of
attributions, by stage of smoking, and among all subjects, males, and females
respectively. Second, the rankings were determined by strength of associations

35
obtained from polychotomous logistic regression that examined associations between
each attribution and smoking behaviors. Polychotomous logistic regression was
employed among all subjects, males, and females respectively, adjusting for gender,
age, geographic region, district economy rank, and school academic rank. The
outcome was a new variable created to represent four stages of smoking, which was
coded as “0” if a student had ever smoked, “1” if he/she had ever smoked but didn’t
smoke during the past 30 days, “2” if he/she had smoked during the past 30 days but
didn’t smoke daily, and “3” if he/she had smoked daily during the past 30 days.

RESULTS
Demographic Characteristics and Smoking Status of the Sample
Table 1 shows that the sample of this study contained slightly more females
(51.4%) than males (48.6%). The distribution of ethnicity was not significantly
different among males and females (p=0.27), but the distributions of age groups
(p<0.001) and geographic regions (p=0.03) were significantly different. The
prevalence of smoking was higher among males than among females at all stages of
smoking (p<0.001).
Themes of Cognitive Attributions for Smoking
Eight factors were generalized by EFA from a list of 17 cognitive attributions
for smoking. The factor pattern matrix was presented by Table 2. We proposed that
the factors represented eight themes of cognitive attributions for smoking, consisting

36
Table 1: Demographic Characteristics and Smoking Status of the Sample
All Male Female Gender
n (%) n (%) n (%) Difference
6987 (48.6) 7381 (51.4)
Age
12 years or younger 1232 (8.6) 543 (7.8) 689 (9.3) χ
2
(5)=46.8
13 2692 (18.7) 1360 (19.5) 1332 (18.0) p<0.0001
14 1899 (13.2) 1009 (14.4) 890 (12.1)
15 1502 (10.4) 660 (9.4) 842 (11.4)
16 3714 (25.9) 1759 (25.2) 1955 (26.5)
17 years or older 3329 (23.2) 1656 (23.7) 1673 (22.7)
Ethnicity
Han 13736 (95.7) 6665 (95.5) 7071 (95.9) χ
2
(1)=1.2
Others 612 (4.3) 311 (4.5) 301 (4.1) p=0.27
City
Chengdu 2198 (15.2) 1047 (15.0) 1141 (15.4) χ
2
(6)=14.2
Hangzhou 1902 (13.2) 929 (13.2) 970 (13.1) p=0.03
Shenyang 2327 (16.1) 1117 (16.0) 1188 (16.1)
Wuhan 2106 (14.6) 995 (14.2) 1111 (15.0)
Harbin 1852 (12.8) 893 (12.8) 950 (12.9)
Kunming 1992 (13.8) 1040 (14.9) 949 (12.9)
Qingdao 2057 (14.3) 972 (13.9) 1079 (14.6)
Smoking status
Never smoker 8760 (61.2) 3517 (50.8) 5215 (71.1) χ
2
(3)=811.4
Lifetime smoker 3524 (24.6) 1947 (28.0) 1562 (21.3) p<0.0001
Past 30-day smoker 1544 (10.8) 1053 (15.2) 489 (6.6)
Daily smoker 484 (3.4) 413 (6.0) 70 (1.0)

37
of curiosity, coping, social image, social belonging, engagement, autonomy, mental
enhancement, and weight control. These themes were similar to but not the same as
what have been reported by a previous study that used similar measures to assess
meanings of smoking among a multiethnic sample of eight graders in great Los
Angeles (Spruijt-Metz et al., 2005). In that study, four meanings of smoking were
identified from 19 individual items, which were personal (including mental
enhancement, engagement, and social belonging in this study), functional (equivalent
to coping in this study), social image, and weight control. Two items that were left
out of factors were excluded, which were curiosity and autonomy in this study.
Relative Importance of Cognitive Attributions for Smoking, Ranked by self-reports
Table 3 shows that, by self-report, curiosity was ranked the 1
st
and 2
nd
for
lifetime smoking and past 30-day smoking respectively, but only 4
th
for daily
smoking. Other than curiosity, coping was ranked the 1
st
across all stages, including
lifetime smoking, past 30-day smoking, and daily smoking. Social image and social
belonging were ranked as more important at earlier stages than later stages of
smoking. Engagement and mental enhancement were ranked as more important at
later stages than earlier stages of smoking. Autonomy and weight control were
ranked as least important across all stages of smoking.

38
Table 2: Cognitive Attributions for Smoking - Factor Pattern Matrix
Factor
I smoke (or might smoke), because: Coping
Weight
Control
Mental
Enhancement
Social
Image
Engagement
Social
Belonging
Autonomy Curiosity
It helps me forget my problems 0.93 0.07 -0.10 -0.03 -0.06 0.12 0.05 -0.04
It helps me deal with anger 0.88 0.01 -0.06 0.03 0.07 0.04 0.00 0.07
It helps me deal with stress 0.84 0.04 0.23 -0.05 0.00 -0.05 -0.02 -0.03
It helps me to relax 0.66 -0.05 0.25 0.10 0.13 -0.08 0.00 -0.05
It helps me keep my weight down 0.08 0.92 0.08 0.07 -0.09 -0.05 -0.10 0.08
It keeps me from eating too much -0.01 0.85 0.04 -0.04 0.12 -0.03 0.07 0.05
It helps me concentrate 0.12 0.10 0.79 -0.07 0.04 0.03 0.02 0.02
It gives me more energy 0.16 -0.08 0.77 0.07 -0.01 0.03 0.12 0.00
It helps me study -0.02 0.39 0.64 0.01 -0.01 0.08 -0.03 -0.07
It makes me look good 0.05 0.02 -0.02 0.92 0.00 -0.08 0.00 0.11
I would have more friends -0.03 0.05 0.06 0.66 0.04 0.30 0.00 -0.07
It keeps me from being bored 0.16 -0.02 -0.05 0.02 0.89 0.01 -0.06 0.05
It gives me something to do 0.02 0.06 0.15 0.00 0.75 0.04 0.04 -0.01
I don't like to refuse when someone
gives me a cigarette
0.06 -0.10 0.14 -0.01 -0.01 0.91 -0.12 0.18
I don't want to make another person
smoke alone
-0.01 0.26 -0.13 0.05 0.11 0.51 0.32 -0.17
I feel like I'm making my own
decisions
0.04 -0.05 0.10 0.00 -0.03 -0.06 0.95 0.10
I'm curious what it's like -0.03 0.08 -0.04 0.05 0.03 0.08 0.08 0.90
Note: Factor loadings greater than 0.50 are shown in bold

39

Table 3: Relative Importance of Cognitive Attributions for Smoking, Ranked by
Self-report
Lifetime Smoker Past 30-day Smoker Daily Smoker
n (%) Rank n (%) Rank n (%) Rank
All Subjects
Curiosity 1212 (34.6) 1 510 (33.3) 2 134 (27.9) 4
Coping 695 (19.9) 2 639 (41.8) 1 331 (68.8) 1
Social Image 288 (8.2) 3 246 (16.1) 5 127 (26.4) 6
Social Belonging 264 (7.5) 4 272 (17.8) 4 133 (27.7) 5
Engagement 250 (7.1) 5 332 (21.7) 3 192 (39.9) 2
Autonomy 202 (5.8) 6 187 (12.2) 6 81 (16.8) 7
Mental
Enhancement 131 (3.7) 7 173 (11.3) 7 160 (33.3) 3
Weight Control 62 (1.8) 8 64 (4.2) 8 34 (7.1) 8

Males
Curiosity 626 (32.4) 1 370 (35.5) 2 120 (29.3) 5
Coping 344 (17.8) 2 462 (44.3) 1 298 (72.7) 1
Social Image 201 (10.4) 3 205 (19.7) 5 121 (29.5) 4
Social Belonging 174 (9.0) 4 215 (20.6) 4 118 (28.8) 6
Engagement 160 (8.3) 5 259 (24.9) 3 178 (43.4) 2
Autonomy 128 (6.6) 6 143 (13.7) 7 76 (18.5) 7
Mental
Enhancement 85 (4.4) 7 146 (14.0) 6 148 (36.1) 3
Weight Control 29 (1.5) 8 48 (4.6) 8 29 (7.1) 8

Females
Curiosity 583 (37.5) 1 140 (28.8) 2 14 (20.0) 3
Coping 349 (22.5) 2 176 (36.2) 1 32 (45.7) 1
Social Image 86 (5.5) 5 41 (8.4) 6 5 (7.1) 6
Social Belonging 89 (5.7) 4 57 (11.7) 4 15 (21.4) 2
Engagement 90 (5.8) 3 72 (14.8) 3 13 (18.6) 4
Autonomy 73 (4.7) 6 44 (9.1) 5 5 (7.1) 6
Mental
Enhancement 45 (2.9) 7 27 (5.6) 7 12 (17.1) 5
Weight Control 33 (2.1) 8 16 (3.3) 8 5 (7.1) 6

40

Relative Importance of Cognitive Attributions for Smoking, Ranked by Strength of
Associations
Table 4 shows that, by its strength of association with smoking behaviors,
curiosity remained ranked the 1
st
and 2
nd
for lifetime smoking and past 30-day
smoking respectively, but only 4
th
for daily smoking. Other than curiosity, coping
was consistently ranked the 1
st
across all stages of smoking, including lifetime
smoking, past 30-day smoking, and daily smoking, followed by engagement, social
belonging, and social image which were about equally important across all stages of
smoking. Autonomy and mental enhancement were significantly associated with
smoking, but less important compared with the others. Weight control was not
significantly associated with adolescent smoking.
Cognitive Attributions for Male Smoking and Female Smoking
By comparing Table 5 and Table 6, we can observe that the relative
importance of most of the attributions for smoking was not identical among males
and females. However, there are a few exceptions. For example, both males and
females ranked curiosity as the 1st for lifetime smoking, and coping as the 1
st
for
past 30-day smoking and daily smoking. More attributions were significantly
associated with male smoking than female smoking. The numbers of significant
attributions among males and females were 7 versus 5 for lifetime smoking, 6 versus
5 for past 30-day smoking, and 7 versus 3 for daily smoking. Among attributions

41

Table 4: Relative Importance of Cognitive Attributions for Smoking, Ranked by
Statistical Results: All Subjects
Ranking
β(se) OR (95% CI) p LTS PTS DS
Curiosity
Lifetime Smoking 2.05 (0.06) 7.76 (6.85, 8.80) <.0001 1
Past 30-day Smoking 1.63 (0.08) 5.11 (4.34, 6.02) <.0001 2
Daily Smoking 1.23 (0.14) 3.44 (2.64, 4.48) <.0001 4
Coping
Lifetime Smoking 0.86 (0.08) 2.36 (2.04, 2.74) <.0001 2
Past 30-day Smoking 1.72 (0.09) 5.59 (4.71, 6.63) <.0001 1
Daily Smoking 2.30 (0.13) 9.96 (7.66, 12.94) <.0001 1
Social Image
Lifetime Smoking 0.49 (0.13) 1.63 (1.27, 2.11) 0.0002 5
Past 30-day Smoking 0.64 (0.14) 1.89 (1.43, 2.50) <.0001 6
Daily Smoking 0.84 (0.18) 2.31 (1.62, 3.29) <.0001 5
Social Belonging
Lifetime Smoking 0.67 (0.13) 1.96 (1.53, 2.52) <.0001 4
Past 30-day Smoking 1.21 (0.14) 3.36 (2.57, 4.39) <.0001 4
Daily Smoking 1.30 (0.17) 3.68 (2.62, 5.17) <.0001 3
Engagement
Lifetime Smoking 0.77 (0.13) 2.17 (1.67, 2.82) <.0001 3
Past 30-day Smoking 1.46 (0.14) 4.32 (3.31, 5.65) <.0001 3
Daily Smoking 1.69 (0.17) 5.42 (3.92, 7.49) <.0001 2
Autonomy
Lifetime Smoking 0.40 (0.12) 1.50 (1.18, 1.90) 0.0010 6
Past 30-day Smoking 0.73 (0.14) 2.07 (1.59, 2.70) <.0001 5
Daily Smoking 0.61 (0.19) 1.84 (1.27, 2.65) 0.0011 7
Mental Enhancement
Lifetime Smoking -0.54 (0.15) 0.58 (0.44, 0.77) 0.0002 7
Past 30-day Smoking -0.12 (0.15) 0.89 (0.66, 1.19) 0.4256 --
Daily Smoking 0.75 (0.17) 2.11 (1.50, 2.96) <.0001 6
Weight Control
Lifetime Smoking -0.30 (0.21) 0.74 (0.49, 1.12) 0.1591 --
Past 30-day Smoking -0.35 (0.24) 0.71 (0.44, 1.12) 0.1423 --
Daily Smoking -0.47 (0.31) 0.62 (0.34, 1.13) 0.1202 --

42

Table 5: Relative Importance of Cognitive Attributions for Smoking, Ranked by
Statistical Results: Males
Ranking
β(se) OR (95% CI) p LTS PTS DS
Curiosity
Lifetime Smoking 2.23 (0.10) 9.29 (7.57, 11.41) <.0001 1
Past 30-day Smoking 2.00 (0.12) 7.41 (5.84, 9.40) <.0001 2
Daily Smoking 1.65 (0.17) 5.20 (3.75, 7.22) <.0001 3
Coping
Lifetime Smoking 0.98 (0.12) 2.65 (2.10, 3.35) <.0001 2
Past 30-day Smoking 2.00 (0.13) 7.42 (5.80, 9.48) <.0001 1
Daily Smoking 2.64 (0.16) 13.95 (10.10, 19.27) <.0001 1
Social Image
Lifetime Smoking 0.80 (0.18) 2.23 (1.57, 3.17) <.0001 5
Past 30-day Smoking 0.91 (0.19) 2.49 (1.71, 3.61) <.0001 5
Daily Smoking 1.19 (0.22) 3.28 (2.12, 5.05) <.0001 5
Social Belonging
Lifetime Smoking 0.88 (0.18) 2.41 (1.70, 3.42) <.0001 3
Past 30-day Smoking 1.38 (0.18) 3.96 (2.76, 5.68) <.0001 4
Daily Smoking 1.41 (0.22) 4.10 (2.67, 6.28) <.0001 4
Engagement
Lifetime Smoking 0.81 (0.18) 2.24 (1.57, 3.22) <.0001 4
Past 30-day Smoking 1.53 (0.19) 4.60 (3.20, 6.61) <.0001 3
Daily Smoking 1.82 (0.21) 6.14 (4.08, 9.24) <.0001 2
Autonomy
Lifetime Smoking 0.55 (0.17) 1.73 (1.25, 2.39) 0.001 6
Past 30-day Smoking 0.87 (0.18) 2.39 (1.68, 3.39) <.0001 6
Daily Smoking 0.85 (0.22) 2.35 (1.51, 3.64) 0.0001 6
Mental Enhancement
Lifetime Smoking -0.51 (0.20) 0.60 (0.41, 0.89) 0.0111 7
Past 30-day Smoking -0.06 (0.20) 0.94 (0.64, 1.40) 0.7681 --
Daily Smoking 0.79 (0.22) 2.20 (1.44, 3.36) 0.0003 7
Weight Control
Lifetime Smoking 0.33 (0.39) 1.40 (0.65, 3.01) 0.3973 --
Past 30-day Smoking 0.32 (0.41) 1.38 (0.62, 3.06) 0.4283 --
Daily Smoking 0.08 (0.46) 1.08 (0.44, 2.65) 0.8642 --

43

Table 6: Relative Importance of Cognitive Attributions for Smoking, Ranked by
Statistical Results: Females
Ranking
β(se) OR (95% CI) p LTS PTS DS
Curiosity
Lifetime Smoking 1.99 (0.08) 7.34 (6.22, 8.65) <.0001 1
Past 30-day Smoking 1.29 (0.13) 3.64 (2.82, 4.70) <.0001 3
Daily Smoking 0.50 (0.34) 1.65 (0.85, 3.21) 0.1369 --
Coping
Lifetime Smoking 0.86 (0.10) 2.36 (1.93, 2.88) <.0001 2
Past 30-day Smoking 1.50 (0.14) 4.48 (3.44, 5.85) <.0001 1
Daily Smoking 1.71 (0.32) 5.51 (2.96, 10.26) <.0001 2
Social Image
Lifetime Smoking 0.06 (0.20) 1.06 (0.72, 1.57) 0.7684 --
Past 30-day Smoking 0.15 (0.25) 1.16 (0.71, 1.90) 0.5478 --
Daily Smoking -0.51 (0.57) 0.60 (0.20, 1.84) 0.3712 --
Social Belonging
Lifetime Smoking 0.39 (0.19) 1.48 (1.02, 2.15) 0.0393 4
Past 30-day Smoking 0.95 (0.22) 2.59 (1.68, 4.02) <.0001 4
Daily Smoking 1.76 ()0.38 5.79 (2.75, 12.21) <.0001 1
Engagement
Lifetime Smoking 0.79 (0.20) 2.21 (1.49, 3.28) <.0001 3
Past 30-day Smoking 1.42 (0.22) 4.14 (2.69, 6.36) <.0001 2
Daily Smoking 1.28 (0.41) 3.60 (1.62, 8.00) 0.0017 3
Autonomy
Lifetime Smoking 0.26 (0.19) 1.30 (0.90, 1.87) 0.1646 --
Past 30-day Smoking 0.65 (0.22) 1.91 (1.24, 2.96) 0.0034 5
Daily Smoking 0.08 (0.53) 1.08 (0.38, 3.05) 0.8865 --
Mental Enhancement
Lifetime Smoking -0.72 (0.22) 0.49 (0.32, 0.76) 0.0014 5
Past 30-day Smoking -0.50 (0.28) 0.60 (0.35, 1.04) 0.067 --
Daily Smoking 0.73 (0.41) 2.08 (0.93, 4.68) 0.0763 --
Weight Control
Lifetime Smoking -0.52 (0.26) 0.59 (0.35, 1.00) 0.0481 --
Past 30-day Smoking -0.60 (0.35) 0.55 (0.28, 1.08) 0.0835 --
Daily Smoking 0.02 (0.57) 1.02 (0.33, 3.14) 0.9723 --

44

that were associated with both male and female smoking, the strengths of
associations were all stronger among males than among females.

DISCUSSION
Cognitive Attributions for Smoking and Their Relative Importance
This study investigated cognitive attributions for Chinese adolescent
smoking, as well as the relative importance of each attribution and its association
with actual smoking behaviors. Results from exploratory factor analysis indicates
that Chinese adolescents attributed reasons for their smoking to curiosity about
smoking, coping, social image, social belonging, engagement, autonomy, mental
enhancement, and weight control.
Curiosity was found to be the most important reason for lifetime smoking,
which is consistent with previous findings (Sarason et al., 1992). Cronan et al found
that, even at late adolescence, curiosity was still an important reason for starting to
smoke, reported by more than half of the adolescents who have smoked (Cronan et
al., 1991). The present study further found that curiosity became less important at
more advanced stages of smoking.
To achieve the status of being cool, tough, and mature has been widely
reported by adolescents as another reason for smoking (Allbutt et al., 1995; Rugkasa
et al., 2001; Treacy et al., 2007). One study demonstrated that three out of seven
reasons given by more than one third of recruits in a Navy Training Center about

45

why they started smoking during their first year in the service were related to social
image: to show I wasn’t afraid, to want to be cool, and to look like an adult (Cronan
et al., 1991). Images seemed to keep important for smoking across ages (Stanton &
Silva, 1993), more important among females than males (Barton et al., 1982), and act
as an indicator of future cigarette smoking (Aloise-Young et al., 1996). Consistent
with previous findings, this study indicates that Chinese adolescents perceived social
image as an important reason for their smoking as well. It was ranked as more
important by males than by females and more important for earlier stage of smoking.
The positive images that adolescents perceived for smoking might be derived from
the mass media, such as TV, movies, and magazines, where cigarette smoking was
usually presented as fashionable, elegant, cool, and sophisticated. However,
rankings by strength of associations showed equal importance across all stages of
smoking. An image of autonomy was pulled out from other images in this study and
turned out to be a relatively unimportant reason for smoking. Maybe in China where
collectivism is more dominant than individualism, autonomy is not perceived as
appealing so smoking is no longer needed as a tool to present it.
At a focus group interview, one student said, “If you’ve got a drink or a
cigarette, it’s easier to talk to somebody.”(Allbutt et al., 1995) Indeed, social
occasions and relations were reported by a lot of adolescents as primary reasons for
their smoking (Allbutt et al., 1995; Cronan et al., 1991; Rugkasa et al., 2001; Sarason
et al., 1992; Stanton & Silva, 1993; Treacy et al., 2007). Adolescents were

46

motivated to smoke to make friends, maintain friendship, and achieve group
membership and identity. Some studies reported this motivation as “social pressure”
(Sarason et al., 1992). However, other studies found that adolescents were active
actors rather than passive recipients of cigarette smoking (Rugkasa et al., 2001).
According to the later studies, adolescents decided to smoke not under the physical
pressure coming from peers or any other surrounding environments. Rather, they
experienced smoking because they actively wanted to “join in” peer groups, to create
or increase prestigious status, and to obtain group membership. "Friends" as a
reason for smoking has been found to have a small degree of consistency across age
(Stanton et al., 1993). This study indicates that Chinese adolescents also perceived
social belonging as an important reason for smoking, especially for male adolescents
and at earlier stages of smoking.
The role of coping on adolescent smoking was inconsistently reported in
previous attribution studies. In some studies, adolescents did express that smoking
helped when they were stressed (Allbutt et al., 1995), and helped them calming down
(Cronan et al., 1991). However, in other studies, adolescents felt that coping with
life was a reason for adult smoking and had nothing to do with children themselves
(Rugkasa et al., 2001). In their views, depression, anxiety, and stress were all adult
business, which they might experience only when they grew up, so was addition to
smoking that might result from some mental distress. Adolescents disagreed that
they were motivated to smoke by personal reasons; instead, they explained that they

47

were motivated to smoke for social reasons to gain good social image and group
membership. In the present study, coping was almost the most important reason for
all stages of smoking. This indicates that, unlike adolescents in most other countries,
Chinese adolescents might experience more psychological problems and cigarette
smoking has been widely used as a tool to cope with these problems.
Engagement was found to be a reason for smoking in this study, which is
consistent with previous findings (Cronan et al., 1991). We further found that
engagement was ranked to be more important for later stages of smoking among
males by self-report, however, equally important across all stages of smoking and
among each gender by statistical analysis. Mental enhancement was found to be
unimportant by statistical analyses, which is good sign indicating that Chinese
adolescents might have not been addictive to nicotine so it was unnecessary to
smoking to refresh their mind. However, male daily smokers considered mental
enhancement as an important reason for smoking, which is not good. According to
attribution theory, if people think mental enhancement can be a reason for smoking,
they might smoke when they need to refresh mind or increase concentration.
Weight control was reported as the least reason for smoking in this study,
which is consistent with previous findings. Although at a qualitative focus group
interview, several individual adolescents expressed weight concern as a motive for
both their own smoking and parent smoking (Allbutt et al., 1995), another
quantitative study found that the percentage of adolescents who smoked for the

48

purpose of losing weight was not high, compared with percentages of adolescents
who smoked due to other reasons, such as the curiosity, friend smoking, images, and
boredom (Cronan et al., 1991).
Comparisons of Relative Importance of Cognitive Attributions between Rankings by
Self-report and Rankings by Strength of Associations
One of the few studies that examined associations between cognitive
attributions and smoking behaviors found that, while smokers gave significantly
more agreement to relaxation as a smoking motive, the most important motives
associated with cigarette consumption were actually addiction and affective Smoking
(Kleinke et al., 1983). In our study, curiosity and coping were consistently ranked
by both self-reports and strength of associations as the most important attribution,
and autonomy and weight control were consistently ranked by the two means as the
least important attributions of smoking. This implies that the most obvious and
unobvious reasons that Chinese adolescents given for their smoking are reliable and
should not be overlooked. However, the relative importance of other reasons is not
so accurately consistent. For example, compared with social image and social
belonging, engagement was detected as more important by strength of associations
but ranked less important by self-report.
Cognitive Attributions for Various Stages of Smoking
Numerous non-attribution studies have demonstrated that reasons for various
stages of smoking were different (Baade & Stanton, 2006; Mayhew et al., 2000).

49

One attribution research also found that the adolescents frequently attributed
curiosity, social norms, and social pressure as reasons for beginning smoking, and
pleasure and addiction as reasons for current smoking (Sarason et al., 1992). This
study got consistent results that the relative importance of each attribution was
different at various stages of smoking. For example, curiosity was the most
important reason at earlier stage, mental enhancement was the most important reason
at later stage, and coping was the most important reason across all stages of smoking.
This brings to us an important implication that intervention components should be
different when the participants’ stages of smoking are different. No single
intervention program can fit all audience.
Cognitive Attributions for Male Smoking and Female Smoking
Smoking is much more prevalent among Chinese male adolescents.
Attributions for male smoking and female adolescent smoking were found to be
identical by some previous studies (Grube et al., 1990; Jenks, 1994b; Palmqvist &
Martikainen, 2005; Stanton et al., 1993), but not by a few others (Anderson &
Anderson, 1990; Sarason et al., 1992). In this study, attributions for male smoking
and female smoking were not exactly the same. In addition, more reasons were
significantly associated with male smoking and in stronger strength. These findings
can probably explain why Chinese boys are more likely to smoke than Chinese girls.
Moreover, they remind us that, to prevent Chinese adolescents from smoking, the
most commonly used approaches at classroom settings are not sufficient. Tailoring

50

should be considered, because more intensive endeavor should be invested among
male adolescents.
Summary
This study investigated cognitive attributions for smoking, their relative
importance, and their associations with actual smoking behaviors. Eight themes of
cognitive attributions for Chinese adolescent smoking were identified. Compared
with those reported by previous attribution studies, some of them were similar but
others were unique to this sample of Chinese adolescents. Seven out of eight
cognitive attributions were truly associated with actual smoking behaviors. The
relative importance of each cognitive attribution differs among various stages in the
trajectory of smoking development and between males and females. This study
extends the results of most of other previous attribution studies in that it not only
identified cognitive attributions for smoking, but also examined the relative
importance of each cognitive attribution for various stages in gender-specific
trajectories of smoking development. Therefore, the findings provide more complete
information about the causes of Chinese adolescent smoking. This study will act as
the first initiative for a series of endeavor to be further implemented for exploring the
roles of cognitive attributions on subsequent smoking progression and regression
among Chinese adolescents.
Limitations and Future Directions

51

We should acknowledge that there exist a couple of limitations in this study.
First, data used were from an ongoing study and not specifically designed for the
present investigation. Therefore, the measures for defining cognitive attributions for
smoking were not ideal. Peer smoking (Chen, Stanton et al., 2006; Grenard et al.,
2006; Hesketh, Ding, & Tomkins, 2001; Zhang, Wang, Zhao, & Vartiainen, 2000;
Zhang, Wang, & Zhou, 2005; Zhu, Liu, Shelton, Liu, & Giovino, 1996) and familial
smoking (Chen, Stanton et al., 2006; Hesketh et al., 2001; Hu et al., 1990; Xiang et
al., 1999; Yang et al., 2004; Ye & Lin, 1984; Zhang et al., 2000; Zhang et al., 2005;
Zhu et al., 1996) have been found to be strong factors influencing Chinese
adolescents to smoking in non-attribution studies. Unfortunately, these important
reasons were not included in the current investigation. Further studies need to
investigate broader attributions of smoking covering intrapersonal, interpersonal,
institutional, community, and policy levels. Guided by attribution theory, future
studies also need to categorize attributions of smoking in terms of their locus
(personal and situational), stability (stable and unstable), and controllability
(controllable and uncontrollable). According to Weiner, the three features of
attributions imply differently to actors for their self-esteem, expectations, and social
emotions (Weiner, 1986). Second, although this retrospective study has successfully
identified cognitive attributions for Chinese adolescent smoking, their relative
importance, and their associations with actual smoking behaviors, longitudinal

52

studies are needed to reveal the influences of these cognitive attributions on
subsequent smoking behaviors.

53

Chapter Three: Roles of Cognitive Attributions for Smoking on
Subsequent Smoking Development

Cigarette smoking is a major public health problem worldwide. Its adverse
physical and social consequences have been widely documented. By linking
people’s intrapersonal, interpersonal, social, cultural, and environmental factors with
their smoking behaviors, researchers have discovered numerous determinants of
smoking (Moolchan et al., 2000; Schepis & Rao, 2005; Turner et al., 2004; Tyas &
Pederson, 1998). However, few studies have explored how smokers themselves
attribute the causes of their smoking behaviors.
According to attribution theory, which describes how people explain the
causes of their own behaviors and the behaviors of others, causal attributions are
critical because they provide the basis for a person’s future actions, both immediate
reaction to behaviors observed, and later behaviors relevant to the person or context
in question (Heider, 1958; Jones & Davis, 1965; Kelley, 1967). This theory assumes
a clear motivation on the part of people to make sense and predictability out of the
world by making rational assessments of the causes of personal action. Attribution
theorists acknowledge that causes people infer to explain their own and other
people’s behaviors can be incorrect and biases may occur during the inference
process (Jones & Nisbett, 1971; Monson & Snyder, 1977; Ross, 1977; Ross et al.,
1977). However, for attribution theorists, the task is not to determine the true causes

54

of these events, but to discern people’s perceptions of the causes, because these
perceptions influence people’s subsequent behaviors.
Studies inspired by attribution theory and other theories have been conducted
to ask people directly why they and other people smoke. While findings from these
attribution studies have provided some useful information about smoking among
adults (Jenks, 1994a; Kleinke et al., 1983; McKennell, 1970; Tomkins, 1966) and
adolescents (Allbutt et al., 1995; Aloise-Young et al., 1996; Barton et al., 1982;
Cronan et al., 1991; Rugkasa et al., 2001; Sarason et al., 1992; Stanton et al., 1993;
Treacy et al., 2007), most of them focused on identifying cognitive attributions for
smoking and their relative importance. Few studies have linked the cognitive
attributions with the actual smoking behaviors (Kleinke et al., 1983). Even fewer
studies have explored whether cognitive attributions influence subsequent smoking
behaviors (McGee & Stanton, 1993). To fill in the gaps, Guo et al took the first
initiative to investigate cognitive attributions for smoking, their relative importance
across various stages in smoking development, and their associations with actual
smoking behaviors (cite Paper One). In this study, a total of eight themes of
cognitive attributions for adolescent smoking, which consisted of curiosity, coping,
social image, social belonging, autonomy, engagement, mental enhancement, and
weight control, were identified from a list of seventeen individual cognitive
attributions. As a continued endeavor, the present study was conducted to

55

investigate the roles of cognitive attributions on subsequent smoking behaviors.
There are three key features in this study.
First, most previous attribution studies have been conducted in European
countries (Allbutt et al., 1995; Berlin et al., 2003; Eiser et al., 1977, 1978; Palmqvist
& Martikainen, 2005; Rugkasa et al., 2001; Treacy et al., 2007), the United States
(Cronan et al., 1991; Jenks, 1994a, 1994b; Kleinke et al., 1983; Sarason et al., 1992),
and a few other countries (McGee & Stanton, 1993; Zoller & Maymon, 1983). They
have rarely been done in China, where about one third of smokers in the world
reside. The high smoking rates in China have caused more than 600 million Chinese
people are either directly or indirectly exposed to cigarette smoke, accounting for
about 72% of the total population and including 60% female non-smokers of
childbearing age (Yang et al., 1999). Most smokers start smoking in adolescence.
The earlier they initiate smoking, the more likely they progress to higher stages of
smoking and get addicted to nicotine (Paavola et al., 1996). However, in China
adolescent smoking has increased steadily since 1984, whereas the age of smoking
onset has decreased steadily (Yang et al., 1999; Yang et al., 2005). An
understanding about why so many Chinese adolescents are cognitively motivated to
smoke is essentially important for prevention and control of smoking all over the
country or even the world. Unfortunately, while still being quite limited, findings
from previous attribution studies cannot simply be generalized to China where
people live in a different physical, social, and cultural environment. Therefore, this

56

study selected Chinese adolescents as the target population. In addition, smoking
rates are dramatically higher among Chinese males than among Chinese females
(Yang et al., 1999; Yang et al., 2004). The causes of smoking might differ between
genders. Hence, this study investigated males and females separately.
Second, any subsequent smoking status may result from initiating smoking,
progressing to higher stages, maintaining at the same stage, or regressing to lower
stages of smoking. Of the above conditions, people who initiate smoking or progress
to higher stages of smoking are at highest risk of established smoking and suffering
from smoking related disasters. They should draw the most attention from public
health officials and researchers. Therefore, the present study used smoking initiation
and progression rather than an absolute smoking status as the outcomes of interest.
By investigating cognitive attributions associated with smoking progression from
each of the initial stages to higher stages, we hoped to provide evidence for
development of primary prevention programs to prevent people from initiating
smoking and secondary prevention programs to prevent people from progressing to
higher stages of smoking.
Third, other than investigating the influences of cognitive attribution on
subsequent smoking progression, this study further explored whether susceptibility
to smoking produced any mediation effects on the associations between cognitive
attributions and subsequent smoking progression. Factors, such as peer smoking
(Gritz et al., 2003; Presson et al., 1984; Straub et al., 2003), family smoking (Presson

57

et al., 1984), and psychological problems (Booker et al., 2004; Hampson et al., 2007;
Straub et al., 2003), have been reported to predict susceptibility to adolescent
smoking, which was also reported as a predictor of the initiation of adolescent
smoking (Jackson, 1998; Stanton et al., 2005). Gritz et al showed evidence that
susceptibility to smoking was not an independent risk factor but rather a potential
mediating variable for adolescent smoking (Gritz et al., 2003). Susceptibility’s
mediating effects have been incorporated into the Theory of Reasoned Action (TRA)
(Ajzen & Fishbein, 1980) and the Theory of Plan Behavior (TPB) (Ajzen, 1985),
where attitudes, perceived norms, and perceived behavioral control (in TPB only)
influence behaviors through behavioral intention. TRA and TPB have been applied
to explain and predict smoking behaviors among adolescents (Hanson, 1999;
Harakeh et al., 2004; Maassen et al., 2004; O'Callaghan et al., 1999), including
Chinese adolescents (Guo et al., 2007). Therefore, when trying to test the mediation
effects that might occur on the pathway from cognitive attributions to subsequent
smoking progression, susceptibility to smoking was considered as a potential
mediator.
We hypothesized that 1) the roles of cognitive attributions on smoking
progression differed according to initial smoking status. When adolescents had
never smoked, curiosity and autonomy were significantly associated with initiation
of smoking. When adolescents had tried smoking, engagement, social image, social
belonging, and coping were significantly associated with smoking progression.

58

When adolescents had already smoked during the past 30 days, mental enhancement
was significantly associated with smoking progression. Weight control was not
significantly associated with smoking progression. 2) more cognitive attributions
were associated with smoking progression among males than among females; and 3)
associations between cognitive attributions and subsequent smoking progression
were mediated by susceptibility to smoking, either partially or entirely.

METHODS
The data are part of the China Seven Cities Study (CSCS), a larger project in
China to assess the effects of changing economic and social factors on health
behaviors including tobacco use. The information will be used to develop
community based smoking and other drug abuse prevention programs. The CSCS
includes seven cities in four regions of China: Northeastern (Harbin, Shenyang),
central (Wuhan), southwestern (Chengdu, Kunming), and coastal (Hangzhou,
Qingdao).
Participants
Participants were recruited from schools in each of the seven cities. The
schools were selected using a stratification process for (a) administrative district
median income and (b) school academic performance. The administrative districts
with highest, middle, and lowest median income in each city were first identified.
Then the local Education Committees were asked to group the middle and high

59

schools in each identified district into three levels of academic performance. This
process resulted in a total of nine school clusters from three districts representing
three levels of district income crossed with three levels of academic performance.
One middle school and one traditional high school were randomly selected
from each of the nine clusters in the matrix to participate in the study. One
classroom in the 7
th
and 8
th
grades in the selected middle schools and one classroom
from the 10
th
and 11
th
grades in the selected high schools were recruited to
participate. In addition, one professional high school was selected from each district,
and the three professional high schools selected from three districts in each city
matched on enrollment, type of vocational training, and ratio of male to female
enrollment. Major courses of study within each professional school were randomly
selected, and students in these majors were recruited from the 10
th
and 11
th
grades to
participate in the study.
In summary, 9 middle schools, 9 high schools, and 3 professional schools
were selected from each city, and a total of 147 schools were selected across all
seven cities. A total of 15,516 students in the middle and high schools were invited
to participate in the baseline survey. Of this total, 802 students (5.2%) were
excluded because the parent or student declined to participate. Another 253 students
with completed consents from parents (1.6% of those invited to participate) were
absent on the data collection day. Therefore, the participation rate was 93.2%
(14,461 of students invited to participate). Another 27 students who provided

60

incomplete surveys at baseline were excluded from the study. The total number of
participants included 93.0% (14,434) of the middle and high school students invited
to participate. At the one-year follow-up, 85.8% (12,382) of the year-one students
completed surveys.
Procedures
Two waves of surveys were administered with one year apart. Survey data
were collected from middle school and high school students. Investigators at the
University of Southern California (USC) provided guidance and training in the
research program. The local Center for Disease Control and Prevention (CDC) in six
of the cities and the Institute for Health Education (IHE) in the city of Kunming
helped develop the surveys, gain access to students in the schools, obtain informed
consent, and collect the data. The informed consent and data collection procedures
were reviewed and approved by both the USC and Chinese Institutional Review
Boards. The students completed the surveys in school.
Quality control of the data collection process was monitored by a team that
included faculty members from a university selected in each of the seven cities,
public health officers at the China National CDC, and faculty and staff at USC.
Local team members observed data collection in selected classrooms in each of the
seven cities and reported to team members at USC. If any deviation from the
standard protocol was detected, USC team members provided rapid responses and

61

corrections within 24 hours directly to the associated data collection personnel in the
field.
Measures
The questionnaire was developed through cooperation among USC
researchers and the directors and staff at the CDC’s and IHE in China. Three USC
graduate students fluent in both English and Mandarin individually translated the
questionnaire into Mandarin Chinese, and then, reached a group consensus for the
translation. The questionnaire was also checked and approved by public health
workers and educators in each of the seven cities to ensure comprehensibility. More
details about the methodology of the CSCS were reported elsewhere (Anderson
Johnson et al., 2006; Grenard et al., 2006; Guo et al., 2007).
Demographic characteristics included age, gender, ethnicity, and geographic
regions. Age, gender, and ethnicity were reported on self-administered surveys.
Geographic regions were the cities where the samples were drawn.
Cognitive attributions for smoking included curiosity about smoking (for
example, “I'm curious what it's like”), coping (for example, “It helps me deal with
stress”), social image (for example, “It makes me look good”), social belonging (for
example, “I don't like to refuse when someone gives me a cigarette”), engagement
(for example, “It keeps me from being bored”), autonomy (for example, “I feel like
I'm making my own decisions”), mental enhancement (for example, “It helps me
concentrate”), and weight control (for example, “It helps me keep my weight

62

down”). They were generated from seventeen individual items through exploratory
factor analyses, representing eight themes of cognitive attributions rather than any
individual items. Each of these cognitive attributions was dichotomously coded as 0
(No) and 1 (Yes). More detailed methodology about how the eight themes of
cognitive attributions were generalized and coded can be seen elsewhere (cite Paper
One).
Lifetime smoking was assessed by one question: “Have you ever tried
cigarette smoking, even a few puffs?” Two response options were provided as 0
(No) and 1 (Yes).
Past 30-day smoking was assessed by two questions. One was “During the
past 30 days, on how many days did you smoke cigarettes?” Seven response options
were provided, ranging from 1 (0 days) to 7 (All 30 days). Another was “During the
past 30 days, on the days you smoked, how many cigarettes did you smoke per day?”
Seven response options were provided, ranging from 1 (I did not smoke cigarettes
during the past 30 days) to 7 (More than 20 cigarettes per day). The past 30-day
smoking variable was dichotomously recoded as 1 (Yes) if a student chose response
options of 2~7 for either of the two items, and 0 (No) if a student chose response
option of 1 for either or both of the two items.
Daily smoking was assessed by one item: “Have you ever smoked cigarettes
daily, that is, at least one cigarette every day for 30 days?” Two response options
were provided as 1 (Yes) and 0 (No).

63

Susceptibility to smoking was assessed by one question: “At any time in the
next 12 months, do you think you will smoke a cigarette?” Four response options
were provided, including 1 (Yes, definitely), 2 (Maybe yes), 3 (Maybe no), and 4
(No, definitely not) (Pierce et al., 1996). At the analytical stage, this variable was
dichotomously re-coded as “0” if a student chose the response option of 4 (No,
definitely no) and “1” if a student chose any of other response options. This
recoding method was based on definition of “susceptibility to smoking” made by
Jackson et al as “a lack of firm commitment against cigarette smoking” (Jackson,
1998).
Statistical Analyses
Frequencies were calculated to describe demographic characteristics,
smoking status, smoking susceptibility, and smoking progression of the sample.
Chi-square tests were employed to examine differences on these variables between
males and females, and between students who provided data at both years and others
who provided data at year one but not at year two.
Cross-sectional datasets obtained from the year-one and year-two surveys
respectively were used to create two new variables representing four stages of
smoking at each of the two years. The new variables were coded as “0” if a student
had never smoked, as “1” if he/she had ever smoked but didn’t smoke during the past
30 days, as “2” if he/she had smoked during the past 30 days but didn’t smoke daily,
and as “3” if he/she had smoked daily during the past 30 days. Longitudinal datasets

64

were used to create a new variable representing the status of smoking progression in
one year. The new variable was coded as “1” if a student progressed smoking to a
higher stage from year one to year two, and “0” if he/she did not.
Associations between each cognitive attribution and subsequent smoking
progression were tested by multilevel analyses, taking into account the clustering of
individuals within same unit(s). Intraclass correlations (ICC’s) were calculated at
the city, district, school, and class levels, and used for determination of which
level(s) of unit should be counted in multilevel analyses. Year-one smoking status,
gender, age, geographic region, district economy rank, and school academic rank
were adjusted for. In order to test the potential moderation effects caused by gender
and year-one smoking status, interaction terms for each cognitive attribution X
gender and year-one smoking status were added into the model. If the interaction
terms were significant at p<0.05, the sample was stratified by gender and/or year-one
smoking status, and the models were retested. Otherwise, if the interaction terms
were insignificant at p>0.05, no further stratification analyses were performed.
The plausible mediation effects that susceptibility to smoking may produce
on the pathway from cognitive attributions to subsequent smoking progression were
examined based on Baron and Kenny’s criteria (Baron & Kenny, 1986). Multilevel
analyses were performed for the steps, stratified by year-one smoking status and
adjusting for gender, age, geographic region, district economy rank, and school
academic rank. The significance of the mediation effects was further tested with

65

Sobel tests (Krull & MacKinnon, 1999; MacKinnon & Dwyer, 1993). Although the
difference between the total effect and direct effect is not algebraically equivalent to
β
a
*β
b
in multilevel models, Krull and MacKinnon (1999) found that the absolute
value of their discrepancy became vanishingly small as the number of groups and
group size increased, implying that they were equivalent for very large
samples(Krull & MacKinnon, 1999). The significance of mediation effects was also
verified by program Prodclin (distribution of the PRODust Confidence Limits for
INdirect effects). In recent years, this program is considered as more powerful than
other commonly used mediation tests and having more accurate Type 1 error rates,
because it computes asymmetric confidence limits based on the distribution of the
product rather than based on the normal distribution (MacKinnon, Fritz, Williams, &
Lockwood, 2007). The mediated effects were the differences of the total effects and
direct effects dividing by the total effect.

RESULTS
Demographic Characteristics of the Sample
The sample contained slightly more females than males (51.5% versus
48.5%) (Table 7). The gender distribution did not vary significantly across cities
(p=0.22). There were about 96% Han ethnicity among both males and females
(p=0.48). Age distribution was significantly different between genders (p<.0001).

66

Table 7: Demographic Characteristics of the Sample
All Male Female Gender
n (%) n (%) n (%) Difference
5988 (48.5) 6354 (51.5)
Ethnicity
Han 11850 (96.1) 5731 (96.0) 6097 (96.2) χ
2
(1)=0.5
Others 481 (3.9) 239 (4.0) 238 (3.8) p=0.48
Age

12 Years or
Younger 1174 (9.5) 509 (8.5) 650 (10.2) χ
2
(5)=38.3
13 years 2515 (20.3) 1239 (20.7) 1273 (20.0) p<0.0001
14 years 1759 (14.2) 913 (15.3) 843 (13.3)
15 years 1342 (10.8) 589 (9.8) 752 (11.8)
16 years 3132 (25.3) 1487 (24.8) 1641 (25.8)
17 Years or Older 2460 (19.9) 1251 (20.9) 1195 (18.8)
City
Chengdu 1866 (15.1) 918 (15.3) 939 (14.8) χ
2
(6)=8.2
Hangzhou 1720 (13.9) 826 (13.8) 892 (14.0) p=0.22
Shenyang 1756 (14.2) 851 (14.2) 892 (14.0)
Wuhan 1961 (15.8) 908 (15.2) 1053 (16.6)
Harbin 1486 (12.0) 720 (12.0) 758 (11.9)
Kunming 1748 (14.1) 884 (14.8) 861 (13.6)
Qingdao 1845 (14.9) 881 (14.7) 959 (15.1)

67

Attrition Analyses
From year one to year two, about 2,052 students (14.2%) were lost to
attrition. There was no significant difference on gender between students followed
and lost (p=0.52). However, the students lost were about one year older on the
average than those followed (mean age of 15.7 versus 14.8 years, p<0.001). The
percentages of students lost to follow-up were significantly different among the
seven cities, ranging from 6.9% to 24.5% (p<0.001). Students of non-Han ethnicities
were more likely to be lost, compared with those of Han ethnicity (19.8% versus
13.9%, p<0.001).
Smoking Progression from Year One to Year Two
At year one, the prevalences of lifetime smoking, past 30-day smoking, and
daily smoking were 24.2%, 9.8%, and 2.5% respectively (Table 8). However, the
rates increased to 33.9%, 10.6%, and 3.7% respectively at year two. The overall
percentage of students whose smoking progressed from year one to year two was
17.0%. Compared with females, significantly more males smoked at all stages in
both years and progressed their smoking to higher stages in one year (p<.0001 for
all).
Cognitive Attributions and Subsequent Smoking Initiation and Progression
Intraclass Correlations (ICC’s) at the city, school and classroom levels for
smoking behavior were 0.02, 0.07 and 0.07 respectively, indicating a low level of
clustering of smoking behavior at the city level but a large effect of clustering at the

68

Table 8: Smoking Status and Smoking Progression over Two Years
All Male Female Gender
n (%) n (%) n (%) Difference

Smoking at Year One
Non-smoker 7801 (63.5) 3182 (53.6) 4596 (72.9) χ
2
(3)=594.2
Lifetime Smoker 2976 (24.2) 1680 (28.3) 1283 (20.3) p<0.0001
Past 30-day Smoker 1207 (9.8) 829 (14.0) 377 (6.0)
Daily Smoker 294 (2.5) 243 (4.1) 51 (0.8)

Susceptibility to Smoking at Year One
Yes 1890 (15.4) 1286 (21.6) 600 (9.5) χ
2
(1)=346.6
No 10409 (84.6) 4658 (78.4) 5717 (90.5) p<0.0001

Smoking at Year Two
Non-smoker 6379 (51.8) 2425 (40.8) 3934 (62.1) χ
2
(3)=734.7
Lifetime Smoker 4181 (33.9) 2281 (38.4) 1885 (29.8) p<0.0001
Past 30-day Smoker 1310 (10.6) 857 (14.4) 450 (7.1)
Daily Smoker 451 (3.7) 383 (6.4) 66 (1.0)

Smoking Progression from Year One to Year Two
Among All Subjects 2076 (17.0) 1222 (20.7) 847 (13.5) χ
2
(1)=113.8
p<0.0001
1439 (18.6) 756 (24.0) 679 (14.8) χ
2
(1)=103.5 Among Year-one Non-
smokers
p<0.0001
498 (16.8) 339 (20.2) 157 (12.3) χ
2
(1)=33.2 Among Year-one Lifetime
Smokers
p<0.0001
139 (11.5) 127 (15.4) 11 (2.9) χ
2
(1)=39.6 Among Year-one Past 30-
day Smokers
p<0.0001

69

school and classroom levels. Since the ICC’s were identical at the school and
classroom levels, school was used as the level 2 variable in the multilevel analyses.
After controlling for demographic characteristics and year-one smoking
status, curiosity, coping, social image, and engagement were positively associated
with smoking initiation and progression from year one to year two (p<0.05 for all)
(Table 9). After adding interaction teams between each of the cognitive attributions
and gender and year-one smoking status, moderation effects caused by gender and
year-one smoking status were detected. For example, curiosity was more associated
with smoking progression from earlier stages of smoking ( β=-0.04, p=0.006), and
engagement was more associated with smoking progression among males ( β=0.10,
p=0.010). Therefore, associations between cognitive attributions and smoking
progression were re-tested among males and females, and among students at
different stages of smoking.
Cognitive Attributions and Subsequent Smoking Initiation and Progression,
Stratified by Year-one Smoking Status and Gender
The effects of attributions on smoking progression differed according to
initial smoking status (Table 10). Among adolescents who had never smoked,
curiosity ( β=0.11, p<.001) and autonomy ( β=0.08, p=0.019) were positively
associated with initiation of smoking. Among adolescents who had tried smoking,
coping ( β=0.07, p<0.001) and social image ( β=0.10, p=0.000) were positively

70

Table 9: Associations between Cognitive Attributions and Subsequent Smoking
Progression
1,2

Main Effect
3
Moderation Effect
4

β (se) p β (se) p
Curiosity 0.03 (0.01) 0.002 0.09 (0.02) <.0001
Coping 0.04 (0.01) 0.000 0.03 (0.02) 0.089
Social Image 0.05 (0.02) 0.004 0.06 (0.04) 0.134
Social Belonging 0.02 (0.02) 0.360 0.03 (0.03) 0.439
Engagement 0.04 (0.02) 0.020 0.02 (0.04) 0.675
Autonomy 0.02 (0.02) 0.197 0.04 (0.03) 0.197
Mental Enhancement 0.03 (0.02) 0.193 0.03 (0.04) 0.473
Weight Control -0.03 (0.03) 0.358 0.01 (0.05) 0.786
Smoking Status -0.09 (0.01) <.0001 -0.07 (0.01) <.0001
Gender 0.09 (0.01) <.0001 0.09 (0.01) <.0001
Curiosity*Smoking Status -0.04 (0.01) 0.006
Coping*Smoking Status 0.00 (0.01) 0.747
Social Image*Smoking Status -0.03 (0.02) 0.121
Social Belonging*Smoking Status -0.01 (0.02) 0.487
Engagement*Smoking Status -0.03 (0.02) 0.139
Autonomy*Smoking Status -0.02 (0.02) 0.238
Mental Enhancement*Smoking Status 0.02 (0.02) 0.432
Weight Control*Smoking Status -0.03 (0.03) 0.389
Curiosity*Gender -0.04 (0.02) 0.058
Coping*Gender 0.00 (0.03) 0.979
Social Image*Gender 0.06 (0.04) 0.167
Social Belonging*Gender 0.02 (0.04) 0.626
Engagement*Gender 0.10 (0.04) 0.010
Autonomy*Gender 0.01 (0.04) 0.822
Mental Enhancement*Gender -0.03 (0.05) 0.483
Weight Control*Gender 0.00 (0.07) 0.990
Notes:
1. Longitudinal study: attributions for smoking were measured at year one; smoking
progression was assessed by the increase of smoking stages from year one to year two.
2. Multilevel analyses were applied, taking into accounts clustering of individuals within
same schools.
3. Attributions for smoking adjusted for the following covariates: year-one smoking status,
geographic region, district economy rank, school academy rank, gender, and age.
4. Attributions for smoking adjusted for geographic region, district economy rank, school
academy rank, and age.

71

associated with smoking progression. Among adolescents who had already smoked
quite often, social image (β=0.05, p=0.043), engagement ( β=0.07, p=0.003), and
mental enhancement ( β=0.15, p<.001) were positively associated with smoking
progression. The significant cognitive attributions identified were not the same
among males and females. More cognitive attributions were associated with
smoking initiation and progression among males than among females.
Mediation Effects Caused by Susceptibility to Smoking
Most of the significant cognitive attributions identified, such as curiosity,
autonomy, coping, engagement, and mental enhancement, influenced smoking
initiation and progression partially through the mediated effects caused by
susceptibility to smoking. The magnitudes of the mediation effects differed, ranging
from 4.3% for curiosity to 30.8% for coping. However, social image did not
influence subsequent smoking progression through susceptibility to smoking (Table
11).

DISCUSSION
The percentages of students in this sample who progressed smoking to higher
stages were 18.6%, 16.8%, and 11.5% respectively among year-one non-smokers,
lifetime smokers, and daily smokers. Apparently, adolescence is an important period
for smoking initiation and progression in China. One of the few longitudinal
attribution studies that we could identify reported that most of attributions given by

72

Table 10: Associations between Cognitive Attributions and Subsequent Smoking
Progression, Stratified by Gender and Year-1 Smoking Status
All Male Female
β (se) p β (se) p β (se) p
Year-one Non-smokers
Curiosity 0.11 (0.02) <.0001 0.01 (0.04) 0.905 0.15 (0.02) <.0001
Coping 0.00 (0.02)0.854 -0.06 (0.05) 0.156 0.02 (0.03) 0.480
Social Image 0.02 (0.04) 0.578 0.09 (0.07) 0.208 0.02 (0.05) 0.678
Social Belonging 0.03 (0.04) 0.469 0.06 (0.07) 0.374 -0.01 (0.05) 0.785
Engagement 0.04 (0.04)0.368 0.19 (0.07) 0.010 -0.08 (0.06) 0.169
Autonomy 0.08 (0.03) 0.019 0.13 (0.05) 0.017 0.04 (0.04) 0.278
Mental Enhancement 0.01 (0.04) 0.742 -0.06 (0.06) 0.393 0.06 (0.04) 0.181
Weight Control 0.08 (0.06) 0.160 0.06 (0.14) 0.687 0.08 (0.06) 0.150

Year-one Lifetime Smokers
Curiosity -0.02 (0.01)0.299 -0.01 (0.02) 0.604 -0.02 (0.02) 0.407
Coping 0.07 (0.02) <.0001 0.10 (0.03) 0.000 0.05 (0.02) 0.060
Social Image 0.10 (0.03) 0.000 0.12 (0.04) 0.001 0.06 (0.04) 0.220
Social Belonging 0.03 (0.03) 0.350 0.04 (0.04) 0.311 0.00 (0.04) 0.979
Engagement 0.04 (0.03)0.166 0.04 (0.04) 0.350 0.03 (0.04) 0.423
Autonomy -0.05 (0.03)0.130 -0.05 (0.04) 0.212 -0.03 (0.05) 0.581
Mental Enhancement -0.01 (0.04) 0.832 -0.01 (0.05) 0.889 -0.05 (0.06) 0.388
Weight Control -0.10 (0.06) 0.074 -0.09 (0.09) 0.306 -0.09 (0.07) 0.161

Year-one Past 30-day Smokers
Curiosity 0.01 (0.02)0.745 0.01 (0.03) 0.690 -0.01 (0.02) 0.728
Coping 0.04 (0.02)0.053 0.05 (0.03) 0.082 0.05 (0.02) 0.043
Social Image 0.05 (0.03) 0.043 0.07 (0.03) 0.028 -0.03 (0.03) 0.340
Social Belonging 0.02 (0.03) 0.490 0.00 (0.03) 0.937 0.07 (0.03) 0.015
Engagement 0.07 (0.02) 0.003 0.10 (0.03) 0.002 -0.03 (0.03) 0.303
Autonomy 0.03 (0.03)0.307 0.02 (0.04) 0.530 0.01 (0.03) 0.803
Mental Enhancement 0.15 (0.03) <.0001 0.16 (0.04) <.0001 0.07 (0.05) 0.104
Weight Control -0.07 (0.04) 0.096 -0.05 (0.06) 0.337 -0.10 (0.05) 0.041

73
Table 11: Mediation Tests
Path a Path b Path c Sobel Test Prodclin
β (se) p β (se) p β (se) p z p 95% CL
Mediated
Effect

Year-one Non-smokers
Curiosity 0.05 (0.01) <.0001 0.12 (0.02) <.0001 0.12 (0.02) <.0001 3.61 0.000 (0.003, 0.010) 4.3%
Autonomy 0.09 (0.02) <.0001 0.08 (0.03) 0.015 0.09 (0.03) 0.006 2.21 0.022 (0.001, 0.014) 11.5%
Susceptibility to Smoking 0.11 (0.02) <.0001

Year-one Lifetime Smokers
Coping 0.23 (0.02) <.0001 0.05 (0.02) 0.004 0.07 (0.02) <.0001 2.28 0.015 (0.004, 0.020) 30.8%
Social Image 0.02 (0.03) 0.449 0.10 (0.03) 0.000 0.10 (0.03) 0.000 2.59 0.513 (-0.003, 0.008) --
Susceptibility to Smoking 0.10 (0.02) <.0001

Year-one Past 30-day Smokers
Social Image 0.19 (0.04) <.0001 0.04 (0.03) 0.108 0.06 (0.03) 0.021 1.54 0.199 (-0.001, 0.018) --
Engagement 0.25 (0.03) <.0001 0.06 (0.02) 0.017 0.08 (0.02) 0.001 2.28 0.005 (0.003, 0.026) 24.3%
Mental Enhancement 0.23 (0.04) <.0001 0.14 (0.03) <.0001 0.16 (0.03) <.0001 3.54 0.000 (0.016, 0.051) 12.0%
Susceptibility to Smoking 0.09 (0.02) <.0001
Notes:
Path a: cognitive attributions for smoking predict susceptibility to smoking.
Path b: susceptibility to smoking predicts smoking progression, after controlling for cognitive attributions for smoking.
Path c: cognitive attributions for smoking predict smoking progression.

74
adolescents as causes of their own smoking (for example, relaxation, friends'
smoking, and image) did not significantly predict their smoking two years later
(McGee & Stanton, 1993). However, the present study demonstrated that six out of
eight cognitive attributions given by Chinese adolescents as causes of their own
smoking, which include curiosity, autonomy, social image, coping, engagement, and
mental enhancement, were influential to their subsequent smoking progression one
year later. These findings help to explain why Chinese adolescents initiate and
progress smoking.
Possible Explanations about Why Chinese Adolescents Initiate and Progress
Smoking
Adolescents are in a maturation period – the transition from childhood to
adulthood. They are curious about adult behaviors, including smoking behavior, and
tend to imitate. They also want to show others that they are becoming mature and
independent. Cigarette smoking, which has been widely used by adults, especially
Chinese adults, becomes a tool for them to achieve the goal.
In western countries, some adolescents reported that smoking helped relieve
stress (Allbutt et al., 1995). However, other adolescents felt that coping with life
could not be a reason for their smoking because they would not experience mental
problems until they grew up (Rugkasa et al., 2001). Among Chinese adolescents,
coping with anger, stress, and other problems was an important reason for smoking,
almost the top reason for smoking across all stages (i.e., lifetime smoking, past 30-

75
day smoking, and daily smoking) (cite Paper One). The present study demonstrates
that coping was also associated with subsequent smoking progression. This
phenomenon can probably be explained by some unique aspects of the Chinese
society. On one hand, the academy field in China is very competent. In order to
obtain higher education that is only available for limited percentage of students who
are extremely excellent, adolescents themselves are motivated to work very hard to
achieve and maintain good academic performance and be superior to others. On the
other hand, most families in China, especially in urban areas, comply with the one-
child family planning policy enacted by the Chinese government. Parents who have
high expectations about the children’s future might impose extra pressure such that
the only-children have to spend even more time and effort on study and other skill-
building activities. Consequently, heavy daily burdens may make adolescents feel
depressed, stressful, and even angry. They may resort to cigarette smoking due to
lack of other better ways to cope with these mental problems. Under high pressure
to study hard, adolescents might also try smoking to refresh their mind and increase
their concentration.
It is not surprising that social image and engagement were associated with
smoking progression. Adolescents like to present cool and other favorable images to
other people (especially their peers) and to engage while feeling bored, and cigarette
smoking might become a tool for both situations (Allbutt et al., 1995; Cronan et al.,
1991; Rugkasa et al., 2001; Stanton et al., 1993; Treacy et al., 2007). However,

76
social belonging, which has been widely reported as one of the most important
reasons for adolescent smoking (Allbutt et al., 1995; Cronan et al., 1991; Rugkasa et
al., 2001; Sarason et al., 1992; Stanton et al., 1993; Stanton & Silva, 1993; Treacy et
al., 2007), including Chinese adolescent smoking (cite Paper One), was not
associated with smoking progression. The mechanisms about how social belonging
may have influenced smoking need to be further explored.
Variations of Cognitive Attributions for Different Stages in Smoking Development
Although six cognitive attributions were found to be influential to subsequent
smoking progression, this study found that each of them influenced only one or two
periods in the trajectory of smoking development. For example, curiosity and
autonomy were positively associated with initiation of smoking, coping was
positively associated with smoking progression from lifetime smoking to higher
stages, engagement and mental enhancement were positively associated with
smoking progression from past 30-day smoking to higher stages, and social image
was positively associated with smoking progression from lifetime and past 30-day
smoking to higher stages. These findings have profound implications for anti-
smoking efforts. For primary smoking prevention endeavor aiming to prevent
adolescents from lighting up the first cigarette, we may consider how to prevent
adolescents from being curious about smoking and trying to use cigarettes to show
mature and independence. For secondary smoking prevention endeavors aiming to
prevent adolescents from progressing their smoking to higher stages, strategies can

77
be varied depending upon initial smoking status. When adolescents are at earlier
stages, we may consider how to prevent them from using cigarettes to present
positive social images, how to reduce and minimize psychological problems, and
how to manage psychological problems when they really occur. When adolescents
have been at higher stages of smoking, strategies about how to deal with boredom
and how to concentrate should be incorporated and addressed.
Gender Differences
Compared with female adolescents, many more male adolescents initiated
smoking and progressed to higher stages of smoking. Some previous studies found
that attributions for male smoking and female smoking were identical (Grube et al.,
1990; Jenks, 1994b; Palmqvist & Martikainen, 2005; Stanton et al., 1993); however,
a few other studies got inconsistent results (Anderson & Anderson, 1990; Sarason et
al., 1992). Among Chinese adolescents, Guo et al found that more cognitive
attributions were associated with male smoking; and for attributions that were
associated with both male smoking and female smoking, the strength of associations
was all stronger among males (cite Paper One). The present study indicated that
cognitive attributions that were associated with smoking progression were also
different between gender and more cognitive attributions were associated with
smoking progression among males. For example, while engagement and autonomy
were positively associated with initiation of smoking among males, only curiosity
about smoking was positively associated with initiation of smoking among females;

78
while coping and social image were positively associated with smoking progression
among year-one lifetime male smokers, no cognitive attributions was identified to be
significantly associated with smoking progression among year-one lifetime female
smokers; while mental enhancement, engagement, and social image were positively
associated with smoking progression among year-one past 30-day male smokers,
coping, social belonging, and weight control were associated with smoking
progression among year-one past 30-day female smokers. These findings imply that,
in a context like Chinese society where the gender differences in smoking behaviors
and expectations about smoking are larger, even if adolescents are at the same
smoking status, anti-smoking intervention components should not be the same
among males and females.
Mediated Effects
This study found that susceptibility to smoking acted as one of the
mechanisms, through which most of cognitive attributions may lead to subsequent
smoking progression. The magnitudes of medication effect varied, being relatively
large for some cognitive attributions (such as, coping and engagement), but very
small for others (such as, curiosity). By utilizing longitudinal datasets, this study
provided evidence for causal relationships between cognitive attributions and
subsequent smoking progression. By identifying this mediation pathway, this study
demonstrated more clues about causal mechanisms.

79
Summary
Findings from this study support the general idea underlying attribution
theory that people’s perceptions of causes of behaviors are influential to subsequent
actions. By employing a longitudinal design, conducting investigations by initial
smoking status and by gender, and using smoking progression as the outcome of
interest, this study disclosed some roles that cognitive attributions may have played
in subsequent smoking initiation and progression among males and females who are
at earlier or later stages of smoking. The findings are instructive to development of
primary and secondary smoking prevention, especially targeting Chinese
adolescents. An essential recommendation coming from this study is that smoking
intervention programs should be unique and evidence-based, because no single
program can fit all audiences.
Limitations and Future Directions
We should acknowledge that the less than ideal measure for cognitive
attributions is the biggest limitation for this study. This measure was in narrower
dimension and didn’t include a lot of other important reasons for adolescent
smoking. Further studies need to investigate broader cognitive attributions of
smoking. In addition, these attributions should be able to be categorized in terms of
their locus (personal and situational), stability (stable and unstable), and
controllability (controllable and uncontrollable), because each of which implies
differently to actors for their self-esteem, expectations, and social emotions (Weiner,

80
1986). Another limitation came from the attrition during follow-ups. While students
lost were not significantly different from those followed on some demographic
characteristics such as gender, they were about one year older than students followed
and distributed differently across the seven cities. However, given the large sample
size and relatively low attrition rate, this difference might not affect the study
findings in a significant level. Lastly, although susceptibility to smoking has been
successfully tested and proven to be one of the mechanisms through which some
cognitive attributions led to subsequent smoking development. More mediation
pathways are worthy to be explored in the future.

81
Chapter Four: Roles of Cognitive Attributions for Smoking on
Subsequent Smoking Progression and Regression

Cigarette smoking poses a major public health problem worldwide.
Adolescent smoking has drown special attention from researchers and public health
officials, given the fact that most smokers initiate smoking at adolescence and that
the earlier they start smoking, the more likely they continue smoking until adulthood,
and get addicted to nicotine (Chassin, Presson, Sherman, & Edwards, 1990). By
examining associations between people’s personal and environmental factors and
their smoking behaviors, researchers have discovered multifaceted determinants of
adolescent smoking (Moolchan et al., 2000; Schepis & Rao, 2005; Turner et al.,
2004; Tyas & Pederson, 1998).
In the meantime, other studies explored how adolescents themselves
attributed the reasons for their smoking behaviors (Allbutt et al., 1995; Aloise-Young
et al., 1996; Barton et al., 1982; Cronan et al., 1991; Rugkasa et al., 2001; Sarason et
al., 1992; Stanton et al., 1993; Treacy et al., 2007). Although these studies are few in
number, they are important. According to attribution theory, people are motivated to
observe, analyze, and explain their own behaviors and the behaviors of others in
order to manage their social environment well (Heider, 1958). The explanations
given by people for behaviors are called “attributions”. Attributions are typically
classified into two categories, personal and situational. The first implies volition or
intention on the part of the actor, and the second implies contextual or environmental

82
factors that constrain or compel the behavior. The dynamic of primary interest for
the attribution theorists is how much of causation is attributed to volition and
underlying disposition of the actor, and how much is attributed to contextual
constraints that compel the actor to behave in a particular way. Later approaches
indicated that attributions could be not only internal (personal) or external
(situational), but also stable or unstable, and controllable or uncontrollable. The
three features convey different information about the actors’ self-esteem,
expectations, and social emotions (Weiner, 1986). Attribution theorists, especially
Edward Jones and Harold Kelly, view the attributor as a very rational being who
follows orderly rules of perception and inference, with the resultant attributions
predictable by the information specifically at hand (Jones & Davis, 1965; Kelley,
1967). Causal attributions are seen as critical because they provide the basis for
one’s future actions, both immediate reaction to behaviors observed, and later
behaviors relevant to the person or context in question.
No matter whether guided by attribution theory and other theories or not,
studies have been conducted to ask adolescents directly why they smoked. The
reasons commonly given by adolescents as causes of their smoking include curiosity
about smoking (Cronan et al., 1991; Sarason et al., 1992), social image (Allbutt et
al., 1995; Aloise-Young et al., 1996; Barton et al., 1982; Cronan et al., 1991;
Rugkasa et al., 2001; Treacy et al., 2007), social relations (Allbutt et al., 1995;
Cronan et al., 1991; Rugkasa et al., 2001; Sarason et al., 1992; Stanton et al., 1993;

83
Treacy et al., 2007), and boredom (Cronan et al., 1991). While most previous
attribution studies mainly investigated cognitive attributions for adolescent smoking
and their relative importance, few others have examined whether cognitive
attributions were associated with actual smoking behaviors (Kleinke et al., 1983) and
could influence subsequent smoking behaviors (McGee & Stanton, 1993). Among
these are a set of studies conducted among Chinese adolescents who live in a country
with one third of smokers in the world residing (cite Paper One and Paper Two). In
these studies, it was found that the seventeen cognitive attributions that Chinese
adolescents gave as reasons for their smoking behaviors could be categorized into
eight groups, representing eight themes of cognitive attributions, such as curiosity
about smoking, coping with life, social image, social belonging, engagement,
autonomy, mental enhancement, and weight control (cite Paper One). Seven of them
were associated with current smoking behaviors (cite Paper One), and six of them
were associated with subsequent smoking development (cite Paper Two). Findings
from these studies and other previous attribution studies are instructive for
development of appropriate anti-smoking strategies by providing information about
why adolescents smoked and what influenced them to progress their smoking to
higher stages. However, there remain some uncertainties.
First, although cognitive attributions associated with lifetime smoking, past
30-day smoking, and daily smoking respectively have been identified (cite Paper
One), cognitive attributions for each of the stages can be aggregated. For instance,

84
daily smokers experience a trajectory in smoking development from initiation to
more advanced stages. Attributions given by Chinese adolescents for their daily
smoking could be a combination of reasons for all stages across their smoking
development. Therefore, it remains unknown what attributions made by Chinese
daily smokers were for their current smoking status and what others were for their
previous smoking status. Second, cognitive attributions that were associated with
smoking progression among year-one non-smokers, lifetime smokers, and past 30-
day smokers respectively have been identified (cite Paper Two). However, smoking
can progress gradually to one higher stage, or even more rapidly to more than one
higher stage. What factors can cause smoking to progress in different speeds?
Third, as time goes by, smoking can be not only initiated and progressed to higher
stages, but also maintained at the same stages and regressed to lower stages. What
factors can influence the dynamic changes of smoking in distinct directions? To
answer these questions, the present study was conducted.
The goal of this study was to illustrate the comprehensive influences that
cognitive attributions have on each of the dynamic changes in smoking. Unlike most
previous studies, which usually used absolute smoking status as the outcome of
interest, this study paid attention to dynamic changes in smoking. It investigated
what cognitive attributions could influence smoking to progress from each of the
initial stages to each of the higher stages, and what cognitive attributions could

85
prevent smoking from regressing from each of the initial stages to each of the lower
stages. The detailed pathways can be seen in Figure 1.

Figure 1. Changes of Smoking Status in One Year
Year-one Smoking Year-two Smoking

We hypothesized that 1) some cognitive attributions were associated with
earlier stages of smoking, and some others were associated with later stages of
smoking; and 2) some cognitive attributions influenced smoking to progress, some
Daily
Smoking
Never
Smoking
Past 30-day
Smoking
Ever
Smoking
Daily
Smoking
Past 30-day
Smoking
Ever
Smoking
Never
Smoking
t
1

e
2
e
1
n
3
n
2
n
1
t
-1
d
-2
d
-1
n
0
e
0
t
0
d
0

86
prevented smoking from regressing, and some others had dual effects. We hoped
that findings from this study could provide informative implications for development
and implementation of effective anti-smoking programs.

METHODS
The data are part of the China Seven Cities Study (CSCS), a larger project in
China to assess the effects of changing economic and social factors on health
behaviors, including tobacco use. The information will be used to develop
community based smoking and other drug abuse prevention programs. The CSCS
includes seven cities in four regions of China: Northeastern (Harbin, Shenyang),
central (Wuhan), southwestern (Chengdu, Kunming), and coastal (Hangzhou,
Qingdao).
Participants
Participants were recruited from schools in each of the seven cities. The
schools were selected using a stratification process for (a) administrative district
median income and (b) school academic performance. The administrative districts
with highest, middle, and lowest median income in each city were first identified.
Then the local Education Committees were asked to group the middle and high
schools in each identified district into three levels of academic performance. This
process resulted in a total of nine school clusters from three districts representing
three levels of district income crossed with three levels of academic performance.

87
One middle school and one traditional high school were randomly selected
from each of the nine clusters in the matrix to participate in the study. One
classroom in the 7
th
and 8
th
grades in the selected middle schools and one classroom
from the 10
th
and 11
th
grades in the selected high schools were recruited to
participate. In addition, one professional high school was selected from each district,
and the three professional high schools selected from three districts in each city
matched on enrollment, type of vocational training, and ratio of male to female
enrollment. Major courses of study within each professional school were randomly
selected, and students in these majors were recruited from the 10
th
and 11
th
grades to
participate in the study.
In summary, 9 middle schools, 9 high schools, and 3 professional schools
were selected from each city, and a total of 147 schools were selected across all
seven cities. A total of 15,516 students in the middle and high schools were invited
to participate in the baseline survey. Of this total, 802 students (5.2%) were
excluded because the parent or student declined to participate. Another 253 students
with completed consents from parents (1.6% of those invited to participate) were
absent on the data collection day. Therefore, the participation rate was 93.2%
(14,461 of students invited to participate). Another 27 students who provided
incomplete surveys at baseline were excluded from the study. The total number of
participants included 93.0% (14,434) of the middle and high school students invited

88
to participate. At the one-year follow-up, 85.8% (12,382) of the year-one students
completed surveys.
Procedures
Two waves of surveys were administered with one year apart. Survey data
were collected from middle school and high school students. Investigators at the
University of Southern California (USC) provided guidance and training in the
research program. The local Center for Disease Control and Prevention (CDC) in six
of the cities and the Institute for Health Education (IHE) in the city of Kunming
helped develop the surveys, gain access to students in the schools, obtain informed
consent, and collect the data. The informed consent and data collection procedures
were reviewed and approved by both the USC and Chinese Institutional Review
Boards. The students completed the surveys in school.
Quality control of the data collection process was monitored by a team that
included faculty members from a university selected in each of the seven cities,
public health officers at the China National CDC, and faculty and staff at USC.
Local team members observed data collection in selected classrooms in each of the
seven cities and reported to team members at USC. If any deviation from the
standard protocol was detected, USC team members provided rapid responses and
corrections within 24 hours directly to the associated data collection personnel in the
field.
Measures

89
The questionnaire was developed through cooperation among USC
researchers and the directors and staff at the CDC’s and IHE in China. Three USC
graduate students fluent in both English and Mandarin individually translated the
questionnaire into Mandarin Chinese, and then, reached a group consensus for the
translation. The questionnaire was also checked and approved by public health
workers and educators in each of the seven cities to ensure comprehensibility. More
details about the methodology of the CSCS were reported elsewhere (Anderson
Johnson et al., 2006; Grenard et al., 2006; Guo et al., 2007).
Demographic characteristics included age, gender, ethnicity, and geographic
regions. Age, gender, and ethnicity were reported on the self-administered surveys
and the geographic regions were the cities where the samples were drawn.
Cognitive attributions for smoking included curiosity about smoking (for
example, “I'm curious what it's like”), coping (for example, “It helps me deal with
stress”), social image (for example, “It makes me look good”), social belonging (for
example, “I don't like to refuse when someone gives me a cigarette”), engagement
(for example, “It keeps me from being bored”), autonomy (for example, “I feel like
I'm making my own decisions”), mental enhancement (for example, “It helps me
concentrate”), and weight control (for example, “It helps me keep my weight
down”). They were generated from seventeen individual items through exploratory
factor analyses, representing eight themes of cognitive attributions rather than any
individual items. Each of these cognitive attributions was dichotomously coded as 0

90
(No) and 1 (Yes). More detailed methodology about how the eight themes of
cognitive attributions were generalized and coded can be seen elsewhere (cite Paper
One).
Lifetime smoking was assessed by one question: “Have you ever tried
cigarette smoking, even a few puffs?” Two response options were provided as 0
(No) and 1 (Yes).
Past 30-day smoking was assessed by two questions. One was “During the
past 30 days, on how many days did you smoke cigarettes?” Seven response options
were provided, ranging from 1 (0 days) to 7 (All 30 days). Another was “During the
past 30 days, on the days you smoked, how many cigarettes did you smoke per day?”
Seven response options were provided, ranging from 1 (I did not smoke cigarettes
during the past 30 days) to 7 (More than 20 cigarettes per day). The past 30-day
smoking variable was dichotomously recoded as 1 (Yes) if a student chose response
options of 2~7 for either of the two items, and 0 (No) if a student chose response
option of 1 for either or both of the two items.
Daily smoking was assessed by one item: “Have you ever smoked cigarettes
daily, that is, at least one cigarette every day for 30 days?” Two response options
were provided as 1 (Yes) and 0 (No).
Statistical Analyses

91
Frequencies were calculated to describe demographic characteristics,
smoking status, and changes of smoking in one year. Chi-square tests were
employed to examine differences on these variables between males and females.
Cross-sectional datasets obtained from the year-one and year-two surveys
respectively were used to create two new variables representing four stages of
smoking at each of the two years. The new variables were coded as “0” if a student
had never smoked, as “1” if he/she had ever smoked but didn’t smoke during the past
30 days, as “2” if he/she had smoked during the past 30 days but didn’t smoke daily,
and as “3” if he/she had smoked daily during the past 30 days.
Longitudinal datasets were used to create new variables representing dynamic
changes of smoking status over one year, including smoking progression,
maintenance, and regression. The new variables were named and coded based on
three factors, which were year-one smoking status, changes of smoking status during
one year, and number of stages that smoking had changed. Based on the smoking
status at year one, the new variables were named as “N” if subjects had never
smoked (even for a few puffs), as “E” if they had ever smoked but didn’t smoke
during the past 30 days, as “T” if they had smoked during the past 30 days but didn’t
smoke daily, and as “D” if they had smoked daily during the past 30 days. Based on
changes of smoking status during one year, the new variables were coded as “+” if
subjects had initiated smoking or progressed to higher stages of smoking, as “0” if
they maintained smoking status at the same stages, and as “-” if they regressed their

92
smoking to lower stages. Additionally, the new variables were coded as “1” if
subjects had progressed or regressed smoking to the next higher or lower stage, and
as “2” if they had progressed or regressed smoking to adjacent two stages. For
example, if a subject was a past 30-day smoker at year one and became a daily
smoker at year two, his or her change of smoking status in one year was named as
“T
1
” and coded as “1”. Otherwise, if a subject was a daily smoker at year one but
became a lifetime smoker at year two, his or her change of smoking in one year was
named as “D
-2
” and coded as “-2”.
Once the outcome measures were created and coded, polychotomous logistic
regression was applied to examine associations between cognitive attributions for
smoking and dynamic changes of smoking status in one year. Examinations were
conducted stratified by year-one smoking status (such as non-smokers, lifetime
smokers, past 30-day smokers, and daily smokers) and adjusting for gender, age,
geographic region, district economy rank, and school academic rank.

RESULTS
Demographic Characteristics and Smoking Status of the Sample
The sample contained slightly more females than males (51.5% versus
48.5%). The gender distribution did not vary significantly across cities (p=0.22).
There were about 96% Han ethnicity among both males and females (p=0.48). Age
distribution was significantly different between genders (p<.0001). The prevalence

93
rates of lifetime smoking, past 30-day smoking, and daily smoking all increased
from year one to year two among both males and females (Table 12).
Dynamic Changes of Smoking Status in One Year
Overall, the percentages of students whose smoking status progressed,
maintained, and regressed from year one to year two were 17.0%, 76.3%, and 6.7%
respectively. However, among year-one past 30-day smokers and daily smokers, the
percentages of students whose smoking regressed to lower stages were 55.4% and
53.8% respectively, higher than those of students whose smoking progressed to
higher stages (i.e., 11.5%) or maintained at the same stages (i.e., 33.1%). Compared
with females, males more likely changed their smoking status and less likely
maintained their smoking status at the same stages (p<.0001). More details about the
dynamic changes of smoking status among year-one non-smokers, lifetime smoking,
past 30-day smokers, and daily smokers can be observed in Table 13.
Associations between Cognitive Attributions and Dynamic Changes of Smoking
Status in One Year
Seven out of eight cognitive attributions, consisting of curiosity, autonomy,
social image, belonging, coping, engagement, and mental enhancement, were
significantly associated with changes of smoking status in one year (Table 14), but
attributions differed depending on level of progression and regression. Among
adolescents who had never smoked, curiosity about smoking was positively

94
Table 12: Demographic Characteristics and Smoking Status of the Sample
All Male Female Gender
n (%) n (%) n (%) Difference
5988 (48.5) 6354 (51.5)
Ethnicity
Han 11850 (96.1) 5731 (96.0) 6097 (96.2) χ
2
(1)=0.5
Others 481 (3.9) 239 (4.0) 238 (3.8) p=0.48
Age (Years)
12 or Younger 1174 (9.5) 509 (8.5) 650 (10.2) χ
2
(5)=38.3
13 2515 (20.3) 1239 (20.7) 1273 (20.0) p<0.0001
14 1759 (14.2) 913 (15.3) 843 (13.3)
15 1342 (10.8) 589 (9.8) 752 (11.8)
16 3132 (25.3) 1487 (24.8) 1641 (25.8)
17 or Older 2460 (19.9) 1251 (20.9) 1195 (18.8)
City
Chengdu 1866 (15.1) 918 (15.3) 939 (14.8) χ
2
(6)=8.2
Hangzhou 1720 (13.9) 826 (13.8) 892 (14.0) p=0.22
Shenyang 1756 (14.2) 851 (14.2) 892 (14.0)
Wuhan 1961 (15.8) 908 (15.2) 1053 (16.6)
Harbin 1486 (12.0) 720 (12.0) 758 (11.9)
Kunming 1748 (14.1) 884 (14.8) 861 (13.6)
Qingdao 1845 (14.9) 881 (14.7) 959 (15.1)
Smoking at Year One
Non-smoker 7801 (63.5) 3182 (53.6) 4596 (72.9) χ
2
(3)=594.2
Lifetime Smoker 2976 (24.2) 1680 (28.3) 1283 (20.3) p<0.0001
Past 30-day Smoker 1207 (9.8) 829 (14.0) 377 (6.0)
Daily Smoker 294 (2.5) 243 (4.1) 51 (0.8)
Smoking at Year Two
Non-smoker 6379 (51.8) 2425 (40.8) 3934 (62.1) χ
2
(3)=734.7
Lifetime Smoker 4181 (33.9) 2281 (38.4) 1885 (29.8) p<0.0001
Past 30-day Smoker 1310 (10.6) 857 (14.4) 450 (7.1)
Daily Smoker 451 (3.7) 383 (6.4) 66 (1.0)

95
Table 13: Dynamic Changes of Smoking Status in One Year, Stratified by Gender
and Year-1 Smoking Status
All Male Female Gender
n (%) n (%) n (%) Difference
All Subjects
Smoking Regressed 824 (6.7) 504 (8.6) 320 (5.1) χ
2
(2)=194.3
Smoking Maintained 9319 (76.3) 4168 (70.7) 5121 (81.4) p<0.0001
Smoking Progressed 2076 (17.0) 1222 (20.7) 847 (13.5)
Non-smokers
Smoking Maintained
Path n
0
6314 (81.4) 2396 (76.0) 3899 (85.2) χ
2
(3)=122.0
Smoking Progressed p<0.0001
Path n
1
907 (11.7) 469 (14.9) 435 (9.5)
Path n
2
451 (5.8) 226 (7.2) 224 (4.9)
Path n
3
81 (1.1) 61 (1.9) 20 (0.4)
Lifetime Smokers
Smoking Maintained
Path e
0
2472 (83.2) 1336 (79.8) 1125 (87.8) χ
2
(2)=33.6
Smoking Progressed p<0.0001
Path e
1
403 (13.6) 272 (16.2) 130 (10.1)
Path e
2
95 (3.2) 67 (4.0) 27 (2.1)
Past 30-day Smokers
Smoking Regressed
Path t
-1
667 (55.4) 391 (47.3) 276 (73.2) χ
2
(2)=80.4
Smoking Maintained p<0.0001
Path t
0
398 (33.1) 308 (37.3) 90 (23.9)
Smoking Progressed
Path t
1
139 (11.5) 127 (15.4) 11 (2.9)
Daily Smokers
Smoking Regressed
Path d
-2
106 (36.3) 67 (27.8) 39 (76.5) χ
2
(2)=43.7
Path d
-1
51 (17.5) 46 (19.1) 5 (9.8) p<0.0001
Smoking Maintained
Path d
0
135 (46.2) 128 (53.1) 7 (13.7)

96
associated with initiation of smoking (OR=2.23, p<0.0001). Attempts to show
autonomy along with curiosity about smoking made adolescents progress their
smoking to two higher stages (OR=1.80, p=0.03). But nothing was found to
associate with smoking progression to more than two higher stages (p>0.05 for all).
Among adolescents who had tried smoking, social image (OR=1.85, p=0.00) and
coping (OR=1.81, p<0.0001) were positively associated with smoking progression to
a higher stage. Among adolescents who had already smoked during the past 30 days,
coping (OR=0.55, p=0.00) and social belonging (OR=0.66, p=0.03) were negatively
associated with smoking regression to a lower stage. In the meantime, engagement
(OR=1.62, p=0.04) and mental enhancement (OR=2.19, p=0.01) were positively
associated with smoking progression to a higher stage. Among adolescents who had
smoked daily, engagement was negatively associated with smoking regression back
to the previous stage (OR=0.39, p=0.02) or even lower stages (OR=0.41, p=0.05,
marginally significant).

DISCUSSION
Findings from his study sketched a comprehensive profile about how
cognitive attributions have played roles across stages in the smoking trajectory and
in particular ways among Chinese adolescents. This study provides new information
about what can influence smoking to progress to various higher stages or prevent

97
Table 14: Cognitive Attributions for Smoking Associated with Dynamic Changes of
Smoking Status in One Year
Path β (se) OR (95% CI) p
Year-one Non-smokers
n
1
Curiosity 0.80 (0.15) 2.23 (1.65, 3.02) <.0001
Coping -0.13 (0.18) 0.88 (0.62, 1.25) 0.47
Social Image 0.16 (0.29) 1.18 (0.66, 2.10) 0.58
Social Belonging 0.12 (0.30) 1.13 (0.63, 2.03) 0.68
Engagement 0.04 (0.34) 1.05 (0.53, 2.05) 0.90
Autonomy 0.31 (0.23) 1.36 (0.86, 2.16) 0.19
Mental Enhancement -0.10 (0.30) 0.90 (0.50, 1.62) 0.73
Weight Control 0.31 (0.43) 1.36 (0.59, 3.13) 0.47

n
2
Curiosity 0.46 (0.21) 1.58 (1.04, 2.41) 0.03
Coping 0.14 (0.23) 1.16 (0.74, 1.80) 0.52
Social Image -0.24 (0.42) 0.79 (0.35, 1.80) 0.57
Social Belonging 0.30 (0.37) 1.35 (0.66, 2.78) 0.41
Engagement 0.24 (0.40) 1.27 (0.58, 2.76) 0.55
Autonomy 0.59 (0.28) 1.80 (1.05, 3.09) 0.03
Mental Enhancement 0.29 (0.34) 1.34 (0.69, 2.59) 0.38
Weight Control 0.81 (0.44) 2.25 (0.94, 5.37) 0.07

n
3
Curiosity -0.02 (0.68) 0.98 (0.26, 3.75) 0.98
Coping -0.42 (0.70) 0.66 (0.17, 2.57) 0.55
Social Image 0.81 (0.90) 2.24 (0.38, 13.18) 0.37
Social Belonging -0.42 (1.17) 0.66 (0.07, 6.56) 0.72
Engagement 1.02 (0.94) 2.76 (0.44, 17.39) 0.28
Autonomy 0.09 (0.79) 1.09 (0.23, 5.17) 0.91
Mental Enhancement 0.65 (0.84) 1.91 (0.37, 9.94) 0.44
Weight Control -- -- --

Year-one Lifetime Smokers
e
1
Curiosity -0.03 (0.12) 0.97 (0.77, 1.23) 0.81
Coping 0.59 (0.14) 1.81 (1.37, 2.38) <.0001
Social Image 0.62 (0.19) 1.85 (1.28, 2.69) 0.00
Social Belonging 0.20 (0.21) 1.22 (0.81, 1.82) 0.34

98
(Table 14, Continued)

Engagement 0.22 (0.21) 1.25 (0.83, 1.87) 0.29
Autonomy -0.49 (0.27) 0.61 (0.36, 1.04) 0.07
Mental Enhancement -0.05 (0.30) 0.95 (0.53, 1.69) 0.86
Weight Control -0.69 (0.49) 0.50 (0.19, 1.30) 0.16

e
2
Curiosity -0.46 (0.25) 0.63 (0.39, 1.02) 0.06
Coping 0.03 (0.29) 1.03 (0.59, 1.83) 0.91
Social Image 0.49 (0.37) 1.64 (0.78, 3.41) 0.19
Social Belonging 0.22 (0.40) 1.25 (0.57, 2.76) 0.58
Engagement 0.37 (0.39) 1.45 (0.67, 3.13) 0.34
Autonomy 0.38 (0.42) 1.46 (0.64, 3.35) 0.37
Mental Enhancement 0.20 (0.57) 1.22 (0.40, 3.68) 0.73
Weight Control -- -- --

Year-one Past 30-day Smokers
t
-1
Curiosity 0.08 (0.15) 1.08 (0.81, 1.45) 0.60
Coping -0.60 (0.16) 0.55 (0.40, 0.75) 0.00
Social Image -0.05 (0.20) 0.96 (0.64, 1.43) 0.82
Social Belonging -0.42 (0.19) 0.66 (0.45, 0.96) 0.03
Engagement -0.13 (0.19) 0.87 (0.61, 1.26) 0.47
Autonomy -0.09 (0.22) 0.91 (0.59, 1.41) 0.68
Mental Enhancement -0.40 (0.25) 0.67 (0.41, 1.09) 0.11
Weight Control -0.03 (0.33) 0.98 (0.51, 1.88) 0.94

t
1
Curiosity 0.09 (0.22) 1.09 (0.71, 1.69) 0.68
Coping 0.16 (0.23) 1.17 (0.74, 1.85) 0.50
Social Image 0.46 (0.27) 1.59 (0.94, 2.69) 0.08
Social Belonging 0.06 (0.25) 1.06 (0.65, 1.75) 0.81
Engagement 0.48 (0.24) 1.62 (1.02, 2.57) 0.04
Autonomy 0.18 (0.28) 1.20 (0.68, 2.09) 0.53
Mental Enhancement 0.78 (0.28) 2.19 (1.26, 3.78) 0.01
Weight Control -0.66 (0.47) 0.52 (0.21, 1.29) 0.16

99
(Table 14, Continued)

Year-one Daily Smokers
d
-2
Curiosity 0.20 (0.47) 1.22 (0.49, 3.06) 0.67
Coping -0.57 (0.42) 0.57 (0.25, 1.30) 0.18
Social Image 0.38 (0.50) 1.46 (0.55, 3.84) 0.45
Social Belonging -0.32 (0.50) 0.72 (0.27, 1.92) 0.51
Engagement -0.89 (0.45) 0.41 (0.17, 0.98) 0.05
Autonomy -0.32 (0.61) 0.73 (0.22, 2.38) 0.60
Mental Enhancement -0.51 (0.45) 0.60 (0.25, 1.45) 0.26
Weight Control -1.46 (1.06) 0.23 (0.03, 1.84) 0.17

d
-1
Curiosity 0.06 (0.44) 1.06 (0.45, 2.49) 0.90
Coping -0.37 (0.42) 0.69 (0.31, 1.57) 0.38
Social Image 0.28 (0.45) 1.33 (0.56, 3.18) 0.52
Social Belonging 0.03 (0.45) 1.03 (0.43, 2.50) 0.95
Engagement -0.94 (0.41) 0.39 (0.17, 0.88) 0.02
Autonomy -1.13 (0.55) 0.32 (0.11, 0.95) 0.04
Mental Enhancement 0.25 (0.40) 1.28 (0.58, 2.82) 0.54
Weight Control 0.56 (0.72) 1.75 (0.43, 7.11) 0.44

100
smoking from regressing to various lower stages, and how each of the subsequent
smoking status may result from previous higher or lower stages.
Social influences have been widely reported as one of the most important
reasons for adolescent smoking (Allbutt et al., 1995; Cronan et al., 1991; Rugkasa et
al., 2001; Sarason et al., 1992; Stanton et al., 1993; Stanton & Silva, 1993; Treacy et
al., 2007). Consistent with previous findings, Guo et al found that social image,
through which adolescents may want to make friends and obtain favorable respect
from friends, and social belonging, through which adolescents may want to maintain
friendship and group identity, were both significantly associated with Chinese
adolescent smoking (cite Paper One). However, it is surprising that, while social
image was significantly associated with subsequent smoking development, social
belonging was not (cite Paper Two). How the social image and social belonging,
which are both in relation to social situations, would have played different roles in
subsequent smoking behaviors then? The present study obtains an answer by finding
that social image attribution was positively associated with smoking progression to a
higher stage among adolescents who had tried smoking; and social belonging
attribution was negatively associated with smoking regression to a lower stage
among adolescents who had smoked during the past 30 days. In other words,
adolescents may smoke in order to present good images to others and to make
friends. Once they have smoked to a certain stage, it will be difficult for them quit.
When their friends smoke, they might want to smoke too in order to participate in an

101
activity with them and maintain good relationship. When their friends offer
cigarettes, they feel reluctant to refuse, being afraid of losing friendship by providing
an unfavorable response. This can probably explain why friend smoking and being
offered cigarettes by friends have been widely reported as reasons for Chinese
adolescent smoking (Chen, Fang, Li, Stanton, & Lin, 2006; Grenard et al., 2006;
Hesketh et al., 2001; Zhang et al., 2000; Zhu et al., 1996).
This study indicates that each of the cognitive attributions actually influenced
smoking at a critical point across the smoking trajectory. Some cognitive
attributions had influences on earlier stages of smoking. For example, curiosity and
autonomy were positively associated with initiation of smoking; social image was
positively associated with smoking progression from lifetime to past 30-day
smoking; social belonging was negatively associated with smoking regression from
past 30-day to lifetime smoking; and coping was positively associated with smoking
progression from lifetime to past 30-day smoking, as well as negatively associated
with smoking regression backward to previous lower stage. Some other cognitive
attributions had influences on later stages of smoking. For example, mental
enhancement was positively associated with smoking progression from past 30-day
to daily smoking, and engagement was positively associated with smoking
progression from past 30-day to daily smoking, and negatively associated with
smoking regression backward to previous lower stages.

102
This study also indicates that each of the cognitive attributions influenced
smoking in a particular way. Some cognitive attributions influenced smoking to
progress. For example, curiosity and autonomy were positively associated with
initiation of smoking; social image was positively associated with smoking
progression from lifetime to past 30-day smoking; and mental enhancement was
positively associated with smoking progression from past 30-day to daily smoking.
Some cognitive attributions prevented smoking from regressing. For example, social
belonging was negatively associated with smoking regression from past 30-day to
lifetime smoking. Some other cognitive attributions influenced smoking to progress
and prevent smoking from regressing. These cognitive attributions produced dual
effects and should draw the most attention from researchers and public health
officials. For example, coping was positively associated with smoking progression
from lifetime to past 30-day smoking, and negatively associated with smoking
regression backward; engagement was positively associated with smoking
progression from past 30-day to daily smoking and negative associated with smoking
regression backward.
Therefore, the variation of the roles that various cognitive attributions have
played in subsequent smoking progression and regression across smoking trajectory
leads to an important recommendation for the future that the components of anti-
smoking programs should be targeted and stage-matched. When designing an anti-
smoking program, one should first understand the smoking status of target

103
population. What stages are they in the trajectory of smoking development?
Findings from this study suggest that, if they are at earlier stages of smoking,
curiosity, autonomy, social image, social belonging, and coping should be taken into
account; and if they are at later stages of smoking, mental enhancement and
engagement should be taken into account. One should also consider the purpose of
the intervention efforts. Will it be a smoking prevention program or a cessation
program? Findings from this study suggest that, if it is a primary prevention
program aiming to prevent people from initiating smoking, curiosity and autonomy
should be taken into account; if it is a secondary prevention program aiming to
prevent people from progressing to higher stages of smoking, coping, social image,
engagement, and mental enhancement should be taken into account; and if it is a
smoking cessation program, coping, social belonging, and engagement should be
taken into account. Only with better understanding about the target population and
purpose of intervention efforts, can one expect to design evidence-based, unique, and
effective smoking intervention programs. Through these programs, we want to get
people change their cognitive attributions for smoking behaviors. To achieve the
goal, we should reduce or rule out the above-mentioned risk factors, so that people
will no longer or less likely attribute these factors as reasons for their smoking
behaviors.
Findings from this study contain more profound implications. Adolescents
are at an unstable period. According to findings from this study, while many

104
adolescents initiated smoking and progressed to higher stages of smoking, others
regressed their smoking to lower stages. Because the smoking habit tends to be
unstable during adolescence, adolescence is a critical period for anti-smoking
intervention. It is necessary to avoid lighting up the first cigarette and smoking more
cigarettes. However, it is equally important to pull back those adolescents who have
smoked to smoke fewer cigarettes and in less frequency. Findings from this study
indicate that it should not be too difficult now that smoking in this period is still at a
dynamic stage and addiction to nicotine hasn’t yet been established. If nothing is
done to foster their smoking to regress, adolescents might continue with engagement
of smoking until adulthood. Once their smoking habits and addition have been
established, it would be very hard to intervene even if many more human and
financial resources are invested. Therefore, traditional programs that primarily aim
at preventing adolescents from initiating smoking or progressing to higher stages of
smoking seem to be insufficient. Components that foster smoking to regress to
lower stages should be incorporated. To effectively prevent and control smoking
progression among adolescents in long run, comprehensive programs with both
prevention and cessation components are needed and will be more efficient than
other programs with any single component only.
In summary, this study took the advantages of longitudinal data to investigate
the roles that cognitive attributions played in the dynamic changes of smoking
behaviors. It discovers that each of the cognitive attributions can actually tell a story

105
about how it influences smoking at a special point in the smoking trajectory and in a
particular way. By identifying cognitive attributions associated with smoking
initiation and progression, this study provides evidences for primary and secondary
smoking prevention programs. By identifying cognitive attributions associated with
smoking progression from each initial stage to each of higher stages, this study helps
us understand why some adolescents progressed their smoking gradually to an
advanced stage, but some others progressed more rapidly to even more advanced
stages within same time period. By identifying cognitive attributions associated with
smoking regression from each initial stage to each of the lower stages, this study tells
us how the cognitive attributions played different roles in smoking regression and
provides us with evidences for smoking cessation programs. This study proposes
that anti-smoking programs should be unique and evidence-based. Moreover, to
prevent adolescents from smoking in long run, this study recommends developing
and implementing comprehensive anti-smoking programs that incorporates primary
and secondary smoking prevention components and smoking cessation components.
We should acknowledge that there are several limitations in this study. First,
as has been acknowledged previously elsewhere, the measures for cognitive
attributions were not specifically designed for the current investigations. Therefore,
they were less than ideal, not covering a lot of other important reasons for adolescent
smoking and not allowing for testing more aspects of cognitive attributions in terms
of their locus, stability, and controllability (cite Paper One and Paper Two). Second,

106
although the large sample size of this longitudinal study provided a good opportunity
to investigate the roles that cognitive attributions play on dynamic changes of
smoking status. The power was not sufficient enough for investigating males and
females respectively. In China, smoking rates are dramatically different between
genders (Yang et al., 1999; Yang et al., 2004), the reasons of which are worthy being
explored. In addition, in this sample, about 1.1% of year-one non-smokers and 3.2%
of year-one lifetime smokers progressed their smoking rapidly to daily smoking in
one year. However, no significant cognitive attributions were identified for such
rapid progression. While it is possible that none of the tested variables were
contributed to the rapid progression, it is also possible that the number of students in
this progression was too small and there was no sufficient power to detect the
significance of the progression. Third, samples were drawn from middle and high
schools in urban areas of seven large China cities. By employing purposive
sampling and non-purposive sampling strategies (such as stratified random sampling
and cluster sampling), this study has tried to cover big across-city variations and
maximized within-city variations and made the study sample more representative for
Chinese adolescents. However, we can’t generalize study findings to non-school
adolescents and middle and high school students in rural areas of China, which this
study didn’t cover, neither to adolescents in other countries who live in totally
different physical, social, and cultural environments.

107
Chapter Five: Conclusions
Guided by attribution theory, this research demonstrated the importance of
cognitive attributions on subsequent actions, and proceeded with a series steps to
investigate the roles of cognitive attributions for smoking in subsequent smoking
development. A total of three studies have been conducted.
Study One took the first initiative to investigate cognitive attributions for
smoking, their relative importance, and their associations with actual smoking
behaviors. In this study, eight themes of cognitive attributions for Chinese
adolescent smoking were identified. Compared with those reported by previous
attribution studies, some of them were similar but others were unique to this sample
of Chinese adolescents. Seven out of eight cognitive attributions were truly
associated with actual smoking behaviors. The relative importance of each cognitive
attribution differs among various stages in the trajectory of smoking development
and between males and females. This study extends the results of most of other
previous attribution studies in that it not only identified cognitive attributions for
smoking, but also examined the relative importance of each cognitive attribution for
various stages in gender-specific trajectories of smoking development. Therefore,
the findings provide more complete information about the causes of Chinese
adolescent smoking.
Study Two used longitudinal datasets to investigate the roles of the cognitive
attributions in subsequent smoking behaviors. As time goes by, smoking can be

108
initiated, progressed to higher stages, maintained, or regressed to lower stages. Of
the above conditions, people who initiate smoking or progress to higher stages of
smoking are at highest risk of established smoking and suffering from smoking
related disasters. They should draw the most attention from public officials and
researchers. Therefore, this study used smoking progression rather than an absolute
subsequent smoking status as the outcome of interest. It was found that six out of
eight cognitive attributions were associated with subsequent smoking progression.
However, cognitive attributions that were associated with smoking progression were
different among year-one non-smokers, lifetime smokers, and past 30-day smokers,
and between males and females. This study is superior to most previous studies in
that it not only investigated associations between cognitive attributions and
subsequent smoking behaviors, but also examined the association between each
cognitive attribution and smoking progression from each of the initial stages to
higher stages. Therefore, the findings suggest implications for future anti-smoking
programs. For example, cognitive attributions identified to be associated with
smoking progression among year-one non-smokers are instructive for development
of primary smoking prevention that is to prevent people from initiating smoking.
Cognitive attributions identified to be associated with smoking progression among
lifetime smokers and past 30-day smokers are instructive to secondary smoking
prevention that is to prevent people from progressing to higher stages of smoking.
The intervention endeavor is not to simply get people to change their cognitive

109
attributions for smoking behaviors. Rather, all what adolescents attributed as reasons
for their smoking can be seen as risk factors of smoking, which should be reduced or
ruled out so that adolescents will no longer or less likely attribute them as reasons for
their smoking. This study further demonstrated that susceptibility to smoking acted
as one of the mediation pathways through which some cognitive attributions led to
smoking progression.
Study Three went beyond Study Two and conducted a more advanced
investigation about the roles of cognitive attributions in subsequent smoking
behaviors. It took advantage of longitudinal data to investigate the roles of cognitive
attributions in all dynamic changes of smoking behaviors, including both smoking
progression and smoking regression. Similar to Study Two, the total sample was
classified into four subgroups: non-smokers, lifetime smokers, past 30-day smokers,
and daily smokers, based on the smoking status at baseline. However, unlike Study
Two that investigated the roles of cognitive attributions on overall smoking
progression from each initial stage to higher stages, this study investigated the roles
of cognitive attributions on smoking progression from each initial stage to each of
the higher stages. In addition, other than smoking progression, this study paid
attention to smoking regression as well. It investigated the roles of cognitive
attributions on smoking regression from each initial stage to each of the lower
stages. This study revealed that each of the cognitive attributions can actually tell a
story about how it influences smoking at a specific point across the smoking

110
trajectory and in a particular way. Some of them influenced earlier stages of
smoking, and some other influenced later stages of smoking. Some of them
influenced smoking to progress, and some others prevented smoking to regress. The
findings produce many more implications for the future. By identifying cognitive
attributions associated with smoking progression from each initial stage to each of
higher stages, this study provides more detailed evidence for the development of
primary prevention programs and secondary prevention programs, and additionally,
helps us understand why some adolescents progressed their smoking gradually to an
advanced stage, but some others progressed more rapidly to even more advanced
stages within the same time period. By identifying cognitive attributions associated
with smoking regression from each initial stage to each of the lower stages, this
study tells us how the cognitive attributions played different roles in smoking
regression and provides us with evidence for development of smoking cessation
programs.
In summary, three studies in this research collectively indicate that cognitive
attributions were not only associated with current smoking behaviors, but also
influential to subsequent smoking development. The associations and influences
differed among students at various stages of smoking and between males and
females. The findings of three studies further demonstrate not only some
mechanisms through which cognitive attributions played their roles in subsequent
smoking development, but also more comprehensive profile about how each of the

111
cognitive attributions played a special role in smoking trajectory in a particular way.
The findings are instructive for development of primary smoking prevention
programs, secondary smoking prevention programs, and smoking cessation
programs, especially targeting Chinese adolescents. Overall, an essential
recommendation coming from the three studies is that anti-smoking programs should
be unique and evidence-based, because no one single program can fit all audience.
Saying specifically, future anti-smoking programs should be gender-specific, stage-
matched, and purposive. To achieve the goal, the following steps can be taken for
design and implementation of an effective anti-smoking program.
Step 1. Differentiate the audience into males and females.
Step 2. Among each gender, classify the audience into sub-groups (for
example, non-smokers, lifetime smokers, past 30-day smokers, and daily smokers),
based on the smoking status.
Step 3. Define the goals and objectives. Determine whether to prevent the
audience from initiating and progressing smoking or to foster the audience to reduce
and quit smoking.
Step 4. Determine issues to address in the anti-smoking program, based on
findings from previous research (for example, taking into accounts factors that are
associated with smoking progression from each of the initial smoking stages to each
of the higher stages, and that are associated with smoking regression from each of
the initial stages to each of the lower stages).

112
Step 5. Consider all feasibilities to implement the anti-smoking program
either among all audience (for example, lectures) or individually (for example,
tailoring using personal counseling, mails, emails, internet, flyers, phone calls, and
other mechanisms).
To prevent adolescents from smoking in long-run, this research also
recommends developing and implementing comprehensive anti-smoking programs
that incorporate primary and secondary smoking prevention components and
smoking cessation components.
Limitations and Future Directions
However, we should acknowledge that there are some limitations in this
research.
First, data used were from an ongoing study and not specifically designed for
the present investigation. Therefore, the measures for defining cognitive attributions
for smoking are not ideal and didn’t cover a lot of other important reasons for
Chinese adolescent smoking, such as peer smoking and familial smoking. Further
studies need to investigate broader attributions of smoking covering intrapersonal,
interpersonal, institutional, community, and policy levels. Guided by attribution
theory, future studies also need to categorize attributions of smoking in terms of their
locus (personal and situational), stability (stable and unstable), and controllability
(controllable and uncontrollable), which may imply differently to actors for their
self-esteem, expectations, and social emotions (Weiner, 1986).

113
The second limitation came from the attrition of this longitudinal study.
About 14% students were lost to follow-up. While students lost were not
significantly different from those followed on some demographic characteristics such
as gender, they were about one year older than students followed and distributed
differently across the seven cities. However, given the large sample size and
relatively low attrition rate, this difference might not affect the study findings in a
significant level.
Third, although the large sample size of this longitudinal study provided a
good opportunity to investigate the roles that cognitive attributions play on dynamic
changes of smoking status. The power was not sufficient enough for conducting
investigations among males and females respectively. In China, smoking rates are
dramatically different between the two genders (Yang et al., 1999; Yang et al.,
2004), the reasons of which are worthy being explored. In addition, in this sample,
about 1.1% of year-one non-smokers and 3.2% of year-one lifetime smokers
progressed their smoking rapidly to daily smoking at year two. However, no
significant predictors were identified for such rapid progressions. While it is
possible that none of the tested variables were contributed to these progressions, it is
also possible that there was no sufficient power to detect the significance of the
progressions.
Fourth, samples were drawn from middle and high schools in urban areas of
seven large China cities. By employing purposive sampling and non-purposive (for

114
example, stratified random sampling and cluster sampling) sampling strategies, this
study has tried to cover big across-city variations and maximized within-city
variations and made the sample more representative for Chinese adolescents. The
generalizability of research findings can be enhanced accordingly. However, we
can’t generalize study findings to non-school adolescents and middle and high school
students in rural areas of China, which this study didn’t cover, neither to adolescents
in other countries who live in more different physical, social, and cultural
environments.

115
Bibliography
Ajzen, I. (1985). From decisions to actions: A theory of planned behavior. In
J. Kuhl & J. Beckmann (Eds.), Action-control: From conngnition to behavior (pp.
11-39). New York: Springer.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting
social behavior. Inglewood Cliffs, NJ: Prentice Hall.

Allbutt, H., Amos, A., & Cunningham-Burley, S. (1995). The social image of
smoking among young people in Scotland. Health Education Research, 10, 443-454.

Aloise-Young, P. A., Hennigan, K. M., & Graham, J. W. (1996). Role of the
self-image and smoker stereotype in smoking onset during early adolescence: a
longitudinal study. Health Psychology, 15(6), 494-497.

Anderson, R. C., & Anderson, K. E. (1990). Success and failure attributions
in smoking cessation among men and women. AAOHN Journal, 38(4), 180-185.

Baade, P. D., & Stanton, W. R. (2006). Determinants of stages of smoking
uptake among secondary school students. Addictive Behaviors, 31(1), 143-148.

Bardwell, R. (1986). Attribution theory and behavior change: ideas for
nursing settings. J Nurs Educ, 25(3), 122-124.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: conceptual, strategic, and statistical
considerations. Journal of Personality and Social Psychology, 51(6), 1173-1182.

Barton, J., Chassin, L., Presson, C. C., & Sherman, S. J. (1982). Social image
factors as motivators of smoking initiation in early and middle adolescence. Child
Development, 53(6), 1499-1511.

Berlin, I., Singleton, E. G., Pedarriosse, A. M., Lancrenon, S., Rames, A.,
Aubin, H. J., & Niaura, R. (2003). The Modified Reasons for Smoking Scale:
factorial structure, gender effects and relationship with nicotine dependence and
smoking cessation in French smokers. Addiction, 98(11), 1575-1583.

Booker, C. L., Gallaher, P., Unger, J. B., Ritt-Olson, A., & Johnson, C. A.
(2004). Stressful life events, smoking behavior, and intentions to smoke among and
multiethnic sample of sixth graders. Ethn Health, 9(4), 369-397.

116
Bosse, R., Garvery, A. J., & Glynn, R. J. (1980). Age and addiction to
smoking. Addict Behav, 5(4), 341-351.
Chassin, L., Presson, C. C., Sherman, S. J., & Edwards, D. A. (1990). The
natural history of cigarette smoking: predicting young-adult smoking outcomes from
adolescent smoking patterns. Health Psychol, 9(6), 701-716.

Chen, X., Fang, X., Li, X., Stanton, B., & Lin, D. (2006). Stay away from
tobacco: a pilot trial of a school-based adolescent smoking prevention program in
Beijing, China. Nicotine Tob Res, 8(2), 227-237.

Chen, X., Stanton, B., Fang, X., Li, X., Lin, D., Zhang, J., Liu, H., & Yang,
H. (2006). Perceived smoking norms, socioenvironmental factors, personal attitudes
and adolescent smoking in China: a mediation analysis with longitudinal data. J
Adolesc Health, 38(4), 359-368.

Coan, R. W. (1973). Personality variables associated with cigarette smoking.
J Pers Soc Psychol, 26(1), 86-104.

Corrigan, P., Markowitz, F. E., Watson, A., Rowan, D., & Kubiak, M. A.
(2003). An attribution model of public discrimination towards persons with mental
illness. J Health Soc Behav, 44(2), 162-179.

Costa, P. T., Jr., McCrae, R. R., & Bosse, R. (1980). Smoking motive factors:
a review and replication. Int J Addict, 15(4), 537-549.

Cronan, T. A., Conway, T. L., & Kaszas, S. L. (1991). Starting to smoke in
the Navy: when, where and why. Soc Sci Med, 33(12), 1349-1353.

Currie, S. R. (2004). Confirmatory factor analysis of the Reasons for
Smoking Scale in alcoholics. Nicotine Tob Res, 6(3), 465-470.

Daltroy, L. H. (1993). Doctor-patient communication in rheumatological
disorders. Baillieres Clin Rheumatol, 7(2), 221-239.

Eiser, J. R., Sutton, S. R., & Wober, M. (1977). Smokers, non-smokers and
the attribution of addiction. British Journal of Social & Clinical Psychology, 16(4),
329-336.

Eiser, J. R., Sutton, S. R., & Wober, M. (1978). Smokers' and non-smokers'
attributions about smoking: a case of actor-observer differences? British Journal of
Social & Clinical Psychology, 17(2), 189-190.

117
Green, D. E. (1977). Psychological factors in smoking. NIDA Res
Monogr(17), 149-156.
Greenlees, I., Lane, A., Thelwell, R., Holder, T., & Hobson, G. (2005).
Team-referent attributions among sport performers. Res Q Exerc Sport, 76(4), 477-
487.

Grenard, J. L., Guo, Q., Jasuja, G. K., Unger, J. B., Chou, C. P., Gallaher, P.
E., Sun, P., Palmer, P., & Anderson Johnson, C. (2006). Influences affecting
adolescent smoking behavior in China. Nicotine Tob Res, 8(2), 245-255.

Gritz, E. R., Prokhorov, A. V., Hudmon, K. S., Mullin Jones, M., Rosenblum,
C., Chang, C. C., Chamberlain, R. M., Taylor, W. C., Johnston, D., & de Moor, C.
(2003). Predictors of susceptibility to smoking and ever smoking: a longitudinal
study in a triethnic sample of adolescents. Nicotine Tob Res, 5(4), 493-506.

Grube, J. W., Rokeach, M., & Getzlaf, S. B. (1990). Adolescents' value
images of smokers, ex-smokers, and nonsmokers. Addictive Behaviors, 15(1), 81-88.

Guo, Q., Johnson, C. A., Unger, J. B., Lee, L., Xie, B., Chou, C. P., Palmer,
P. H., Sun, P., Gallaher, P., & Pentz, M. (2007). Utility of the theory of reasoned
action and theory of planned behavior for predicting Chinese adolescent smoking.
Addict Behav, 32(5), 1066-1081.

Hampson, S. E., Andrews, J. A., & Barckley, M. (2007). Predictors of the
development of elementary-school children's intentions to smoke cigarettes: hostility,
prototypes, and subjective norms. Nicotine Tob Res, 9(7), 751-760.

Hanson, M. J. (1999). Cross-cultural study of beliefs about smoking among
teenaged females. West J Nurs Res, 21(5), 635-647; discussion 647-651.

Harakeh, Z., Scholte, R. H., Vermulst, A. A., de Vries, H., & Engels, R. C.
(2004). Parental factors and adolescents' smoking behavior: an extension of The
theory of planned behavior. Prev Med, 39(5), 951-961.

Heider, F. (1958). The psychology of interpersonal relations: New York:
Wiley.

Hesketh, T., Ding, Q. J., & Tomkins, A. (2001). Smoking among youths in
China. Am J Public Health, 91(10), 1653-1655.

118
Hines, D., Fretz, A. C., & Nollen, N. L. (1998). Regular and occasional
smoking by college students: personality attributions of smokers and nonsmokers.
Psychological Reports, 83(3 Pt 2), 1299-1306.

Hsia, F. N., & Spruijt-Metz, D. (2003). The meanings of smoking among
Chinese American and Taiwanese American college students. Nicotine Tob Res,
5(6), 837-850.

Hu, J. F., Liu, R. Z., Zhang, H. L., Xu, X. F., Li, K., Yang, R. Z., Li, S. X., &
Zhang, Z. T. (1990). A survey of cigarette smoking among middle school students in
1988. Public Health, 104(5), 345-351.

Ikard, F. F., & Tomkins, S. (1973). The experience of affect as a determinant
of smoking behavior: a series of validity studies. J Abnorm Psychol, 81(2), 172-181.

Jackson, C. (1998). Cognitive susceptibility to smoking and initiation of
smoking during childhood: a longitudinal study. Prev Med, 27(1), 129-134.

Jenks, R. J. (1994a). Attitudes and perceptions toward smoking: smokers'
views of themselves and other smokers. Journal of Social Psychology, 134(3), 355-
361.

Jenks, R. J. (1994b). Smoking and satisfaction and motivations: a comparison
of men and women. Journal of Social Psychology, 134(6), 847-849.

Johnson, C. A., Palmer, P. H., Chou, C. P., Pang, Z., Zhou, D., Dong, L.,
Xiang, H., Yang, P., Xu, H., Wang, J., Fu, X., Guo, Q., Sun, P., Ma, H., Gallaher, P.
E., Xie, B., Lee, L., Fang, T., & Unger, J. B. (2006). Tobacco use among youth and
adults in Mainland China: The China Seven Cities Study. Public Health, 120(12),
1156-1169

Jones, E. E., & Davis, K. E. (1965). From acts to dispositions: the attribution
process in person perception. Adv. Exp. Soc. Psychol., 2, 219-266.

Jones, E. E., & Nisbett, R. E. (1971). The actor and the observer: divergent
perceptions of the causes of behavior. Morristown: New Jersey: General Learning
Press.

Kelley, H. H. (1967). Attribution theory in social psychology. Nebr. Symp.
Motiv., 15, 192-238.

119
Kleinke, C. L., Staneski, R. A., & Meeker, F. B. (1983). Attributions for
smoking behavior: comparing smokers with nonsmokers and predicting smokers'
cigarette consumption. Journal of Research in Personality, 17, 242-255.

Krull, J. L., & MacKinnon, D. P. (1999). Multilevel mediation modeling in
group-based intervention studies. Eval Rev, 23(4), 418-444.

Leventhal, H., & Avis, N. (1976). Pleasure, addiction, and habit: factors in
verbal report of factors in smoking behavior? J Abnorm Psychol, 85(5), 478-488.

Leventhal, H., & Cleary, P. D. (1980). The smoking problem: a review of the
research and theory in behavioral risk modification. Psychol Bull, 88(2), 370-405.

Liu, B. Q., Peto, R., Chen, Z. M., Boreham, J., Wu, Y. P., Li, J. Y.,
Campbell, T. C., & Chen, J. S. (1998). Emerging tobacco hazards in China: 1.
Retrospective proportional mortality study of one million deaths. Bmj, 317(7170),
1411-1422.

Maassen, I. T., Kremers, S. P., Mudde, A. N., & Joof, B. M. (2004). Smoking
initiation among Gambian adolescents: social cognitive influences and the effect of
cigarette sampling. Health Educ Res, 19(5), 551-560.

MacKinnon, D. P., Fritz, M. S., Williams, J., & Lockwood, C. M. (2007).
Distribution of the product confidence limits for the indirect effect: program
PRODCLIN. Behav Res Methods, 39(3), 384-389.

MacKinnon, J. P., & Dwyer, J. H. (1993). Estimating mediated effects in
prevention studies. Evaluation Review, 17, 144-158.

Mayhew, K. P., Flay, B. R., & Mott, J. A. (2000). Stages in the development
of adolescent smoking. Drug Alcohol Depend, 59 Suppl 1, S61-81.

McGee, R., & Stanton, W. R. (1993). A longitudinal study of reasons for
smoking in adolescence. Addiction, 88(2), 265-271.

McKennell, A. C. (1970). Smoking motivation factos. Br J Soc Clin Psychol,
9(1), 8-22.

McNeill, A. D., Jarvis, M. J., Stapleton, J. A., West, R. J., & Bryant, A.
(1989). Nicotine intake in young smokers: longitudinal study of saliva cotinine
concentrations. Am J Public Health, 79(2), 172-175.

120
Monson, T. C., & Snyder, M. (1977). Actors, observers, and the attribution
process. J. Exp. Soc. Psychol., 13, 89-111.

Moolchan, E. T., Ernst, M., & Henningfield, J. E. (2000). A review of
tobacco smoking in adolescents: treatment implications. J Am Acad Child Adolesc
Psychiatry, 39(6), 682-693.

O'Callaghan, F. V., Callan, V. J., & Baglioni, A. (1999). Cigarette use by
adolescents: attitude-behavior relationships. Subst Use Misuse, 34(3), 455-468.
O'Connell, K. A., & Shiffman, S. (1988). Negative affect smoking and
smoking relapse. J Subst Abuse, 1(1), 25-33.

Opie, N. D., & Miller, E. T. (1989). Attribution for successful relationships
between severely disabled adults and personal care attendants. Rehabil Nurs, 14(4),
196-199.

Paavola, M., Vartiainen, E., & Puska, P. (1996). Predicting adult smoking:
the influence of smoking during adolescence and smoking among friend and family.
Health Education Research, 11, 309-315.

Palmqvist, R. A., & Martikainen, L. K. (2005). Changes in reasons given for
adolescent smoking, 1984-1999. Subst Use Misuse, 40(5), 645-656.

Pederson, L. L., & Lefcoe, N. M. (1987). Short- and long-term prediction of
self-reported cigarette smoking in a cohort of late adolescents: report of an 8-year
follow-up of public school students. Prev Med, 16(3), 432-447.

Pierce, J. P., Choi, W. S., Gilpin, E. A., Farkas, A. J., & Merritt, R. K.
(1996). Validation of Susceptibility as a Predictor of Which Adolescents Take up
Smoking in the United States. Health Psychology, 15(5), 355-361.

Presson, C. C., Chassin, L., Sherman, S. J., Olshavsky, R., Bensenberg, M.,
& Corty, E. (1984). Predictors of adolescents' intentions to smoke: age, sex, race, and
regional differences. Int J Addict, 19(5), 503-519.

Rees, T. (2007). Main and interactive effects of attribution dimensions on
efficacy expectations in sport. J Sports Sci, 25(4), 473-480.

Ross, L. (1977). The intuitive psychologist and his shortcomings: distortions
in the attribution process. Adv. Exp. Soc. Psychol., 10, 174-220.

121
Ross, L., Greene, D., & House, P. (1977). The false consensus effect: an
egocentric bias in social perception. J. Educ. Psychol., 13, 279-301.

Rugkasa, J., Knox, B., Sittlington, J., Kennedy, O., Treacy, M. P., &
Abaunza, P. S. (2001). Anxious adults vs. cool children: children's views on smoking
and addiction. Social Science & Medicine, 53(5), 593-602.

Ryan, E. B., Szechtman, B., & Bodkin, J. (1992). Attitudes toward younger
and older adults learning to use computers. J Gerontol, 47(2), P96-101.

Sadava, S. W., & Weithe, H. (1985). Maintenance and attributions about
smoking among smokers, nonsmokers, and ex-smokers. International Journal of the
Addictions, 20(10), 1533-1544.

Sarason, I. G., Mankowski, E. S., Peterson, A. V., Jr., & Dinh, K. T. (1992).
Adolescents' reasons for smoking. Journal of School Health, 62(5), 185-190.

Schepis, T. S., & Rao, U. (2005). Epidemiology and etiology of adolescent
smoking. Curr Opin Pediatr, 17(5), 607-612.

Spruijt-Metz, D., Gallaher, P., Unger, J. B., & Johnson, C. A. (2005). Unique
contributions of meanings of smoking and outcome expectancies to understanding
smoking initiation in middle school. Ann Behav Med, 30(2), 104-111.

Spruijt-Metz, D., Gallaher, P. E., Unger, J. B., & Anderson-Johnson, C.
(2004). Meanings of smoking and adolescent smoking across ethnicities. J Adolesc
Health, 35(3), 197-205.

Stanton, W. R., Barnett, A. G., & Silva, P. A. (2005). Adolescents' intentions
to smoke as a predictor of smoking. Prev Med, 40(2), 221-226.

Stanton, W. R., Mahalski, P. A., McGee, R., & Silva, P. A. (1993). Reasons
for smoking or not smoking in early adolescence. Addictive Behaviors, 18(3), 321-
329.

Stanton, W. R., & Silva, P. A. (1993). Consistency in children's recall of age
of initiating smoking. International Journal of Epidemiology, 22(6), 1064-1069.

Straub, D. M., Hills, N. K., Thompson, P. J., & Moscicki, A. B. (2003).
Effects of pro- and anti-tobacco advertising on nonsmoking adolescents' intentions to
smoke. J Adolesc Health, 32(1), 36-43.

122
Tate, J. C., Pomerleau, C. S., & Pomerleau, O. F. (1994). Pharmacological
and non-pharmacological smoking motives: a replication and extension. Addiction,
89(3), 321-330.

Tate, J. C., Schmitz, J. M., & Stanton, A. L. (1991). A critical review of the
Reasons for Smoking Scale. J Subst Abuse, 3(4), 441-455.

Tomkins, S. S. (1966). Psychological model for smoking behavior. Am J
Public Health Nations Health, 56(12), Suppl 56:17-20.

Treacy, M. P., Hyde, A., Boland, J., Whitaker, T., Abaunza, P. S., & Stewart-
Knox, B. J. (2007). Children talking: emerging perspectives and experiences of
cigarette smoking. Qualitative Health Research, 17(2), 238-249.

Turner, L., Mermelstein, R., & Flay, B. (2004). Individual and contextual
influences on adolescent smoking. Ann N Y Acad Sci, 1021, 175-197.

Tyas, S. L., & Pederson, L. L. (1998). Psychosocial factors related to
adolescent smoking: a critical review of the literature. Tobacco Control., 7(4), 409-
420.

Weiner, B. (1986). An attributional theory of motivation and emotion: New
York: Springer-Verlag.

Wright, S. J. (1980). Actor-observer differences in attributions for smoking:
introducing ex-actors and attributions for failure to give up. Br J Soc Clin Psychol,
19(1), 49-50.

Xiang, H., Wang, Z., Stallones, L., Yu, S., Gimbel, H. W., & Yang, P.
(1999). Cigarette smoking among medical college students in Wuhan, People's
Republic of China. Prev Med, 29(3), 210-215.

Yang, G., Fan, L., Tan, J., Qi, G., Zhang, Y., Samet, J. M., Taylor, C. E.,
Becker, K., & Xu, J. (1999). Smoking in China: findings of the 1996 National
Prevalence Survey. Jama, 282(13), 1247-1253.

Yang, G., Ma, J., Chen, A. P., Brown, S., Taylor, C. E., & Samet, J. M.
(2004). Smoking among adolescents in China: 1998 survey findings. Int J Epidemiol,
33(5), 1103-1110.

123
Yang, G., Ma, J., Liu, N., & Zhou, L. (2005). An investigation on smoking
and second-hand smoking investigation among Chinese population in 2002. Chinese
Journal of Epidemiology, 26(2), 77-83.

Ye, G. S., & Lin, W. S. (1984). Cigarette smoking among Beijing (Peking)
high schoolers. World Smoking Health, 9(1), 15-18.

Zhang, H., & Cai, B. (2003). The impact of tobacco on lung health in China.
Respirology, 8(1), 17-21.

Zhang, L., Wang, W., Zhao, Q., & Vartiainen, E. (2000). Psychosocial
predictors of smoking among secondary school students in Henan, China. Health
Educ Res, 15(4), 415-422.
Zhang, L., Wang, W. F., & Zhou, G. (2005). A cross-sectional study of
smoking risk factors in junior high school students in Henan, China. Southeast Asian
J Trop Med Public Health, 36(6), 1580-1584.

Zhu, B. P., Liu, M., Shelton, D., Liu, S., & Giovino, G. A. (1996). Cigarette
smoking and its risk factors among elementary school students in Beijing. Am J
Public Health, 86(3), 368-375.

Zoller, U., & Maymon, T. (1983). Smoking behavior of high school students
in Israel. Journal of School Health, 53(10), 613-617.

Abstract (if available)

Abstract Cigarette smoking poses a major public health problem worldwide. Researchers have discovered numerous smoking determinants by linking people's personal and environmental factors with actual smoking behaviors. However, few studies have explored how smokers themselves explain the causes of their smoking behaviors. According to attribution theory, causal attributions are critical, because they provide the basis for a person's future actions. Studies have been undertaken to ask people directly why they smoke. However, most of them focused on identifying cognitive attributions for smoking. Few of them have investigated whether cognitive attributions were associated with actual smoking and influential to subsequent smoking. To fill in the gaps, the present research was conducted.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The role of depression symptoms on social information processing and tobacco use among adolescents

PDF

Genetic variants and smoking progression in Chinese adolescents

PDF

The effects of mindfulness on adolescent cigarette smoking: Measurement, mechanisms, and theory

PDF

Smoke-free housing policies and secondhand smoke exposure in low income multiunit housing in Los Angeles County

PDF

Cultural risk and protective factors for tobacco use behaviors and depressive symptoms among American Indian adolescents in California

PDF

The effect of an educational fotonovela on the prevention of secondhand smoke

PDF

Intrapersonal and environmental factors associated with Chinese youth alcohol use experimentation and binge drinking behaviors

PDF

Adolescent social networks, smoking, and loneliness

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

Effects of flavorings in electronic cigarettes on the use and appeal of e-cigarettes among youth and adults

PDF

Using a structural model of psychopathology to distinguish relations between shared and specific features of psychopathology, smoking, and underlying mechanisms

PDF

Role transitions, past life events, and their associations with multiple categories of substance use among emerging adults

PDF

Anxiety symptoms and nicotine use among adolescents and young adults

PDF

Mixed methods investigation of user engagement with a smoking cessation app

PDF

Non-steroidal anti-inflammatory drugs, anti-acid drugs and passive smoking as impact factors for esophageal and gastric adenocarcinomas

PDF

DNA methylation groups determined by GATA5 gene methylation level are correlated with tumor subtype, sex, smoking status, and body mass index in esophageal and gastric adenocarcinoma

PDF

Effects of sugar and fiber consumption in minority adolescents and self-tracking as a potential dietary intervention tool

PDF

Tobacco and marijuana surveillance using Twitter data

$A phase I study of vincristine, escalating doses of irinotecan, temozolomide and bevacizumab (VIT-B) in pediatric and adolescent patients with recurrent or refractory solid tumors of non-hematopo...$

PDF

A phase I study of vincristine, escalating doses of irinotecan, temozolomide and bevacizumab (VIT-B) in pediatric and adolescent patients with recurrent or refractory solid tumors of non-hematopo...

PDF

Examining tobacco regulation opinions and policy acceptance among key opinion leaders and tobacco retailers in low socioeconomic status African American, Hispanic, and non-Hispanic White communities

Asset Metadata

Creator Guo, Qian (author)

Core Title Cognitive attributions for smoking and their roles on subsequent smoking progression and regression

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior)

Publication Date 07/28/2008

Defense Date 06/06/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Adolescent,attribution,OAI-PMH Harvest,Smoking

Place Name China (countries)

Language English

Advisor Johnson, Carl Anderson (committee chair), Azen, Stanley Paul (committee member), Chi, Iris (committee member), MacKinnon, David (committee member), Unger, Jennifer B. (committee member)

Creator Email qguo8@yahoo.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1430

Unique identifier UC1448111

Identifier etd-Guo-20080728 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-194864 (legacy record id),usctheses-m1430 (legacy record id)

Legacy Identifier etd-Guo-20080728.pdf

Dmrecord 194864

Document Type Dissertation

Rights Guo, Qian

Type issue, texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu