«.3

WWW.” . .
.35 . .

.3. mm . . ..7
‘ . . A. . gm,

‘ . . in . ‘ :u .r.:.

. . , 3kg... .wm. .53.

a
..
5

3?;

< ’ . , _ .
. ."mﬁumr
m;

. , .. m ..
.Ivs — . ka’ hwpsuﬁ‘v- x .wAL-i.
7 . :1. V . .5.
.. 3:34;“ L2. .3 . ‘ r... S? , ..
3‘ 7. 9m us. Emmy? . _ aw}? i
, , r..¢......m§.. . . , :1... m... tr?
v. . 2.....5. 1., . .. a. . 3...}. .
:Qu. .5» . “m V £9...
Evita.
. anal?
swim. ,
«53m.

‘3;

a...”

.

.2.‘
. . v .
.t»:~v

.
’51‘
.44? “Just.” .th ... .
. . . . . .
. s... L?
. ,
_ﬂr””w<
, .1.‘ .3 . ,
1 .113 3....
:5ﬂﬂcriw... .93..an
. . .ﬁ»lﬁ4 1
x
.03. . ‘
.1.
if. . . 32:73
- . y
1.732..“ 2. V
av .p .3 ‘
X .ID‘Q‘R 0 an
a? P Q
:1 . .
5.01::
1:, 915%
k. uurn...
. \
5:13.
i .1}.
awn
. .
.«N

 

5...}
.i.‘ .

..
3“...
.3 .3}: I
4.1!..Vb1. an.

.
...~..:
.23.?
—v5

.éxmgs

.iweﬁgggs ..

 

This is to certify that the
dissertation entitled

The Effect of Physician Empathy on Patient
Outcomes (Patient Satisfaction and Compliance)
In Korea

presented by

Sung Soo Kim

has been accepted towards fulﬁllment
of the requirements for

Ph.D. Sociology

 

 

degree in

M ajor professd/

Date October 13, 2000

 

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove

this checkout from your record.

TO AVOID FINES return on or before date due.

MAY BE RECALLED with ear

lier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 cJCIRODatoDUOpGS-p.“

THE EFFECT OF PHYSICIAN EMPATHY ON PATIENT OUTCOMES (PATIENT
SATISFACTION AND COMPLIANCE) IN KOREA

By

SUNG SOO KIM

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Department of Sociology

2000

ABSTRACT

The Effect of Physician Empathy on Patient Outcomes (patient
satisfaction and compliance) in Korea

By
SUNG soo KIM

Physician empathy has been reported to be one of the most important
determinants of patients’ outcomes. Like American patients, empathic physicians may be
desired by Korean patients as well, but rarely found in medical settings in Korea.
Furthermore, the effects of Korean physician empathy on patients’ outcomes are little
known up to now. The purpose of this study was to evaluate a new empathy model with
dualistic view (cognitive informational and affective empathy). The empathy model
focuses on the physician-patient communication processes by determining how each
empathy component is speciﬁcally related to patient outcomes (satisfaction and
compliance). Participants in this study were recruited ﬁom a large, multifaceted
university hospital in Pusan, Korea in 1999. Five hundred ﬁﬁy (550) outpatients were
participated in this survey. Structural equation analysis using EQS software program
(Windows 5.7b version) was utilized to test the empathy model.

The results showed that the physician’s empathic communication skills
signiﬁcantly inﬂuenced patient satisfaction and patient compliance via the mediating
factors such as partnership and perception of physician expertise. Above all, for Korean
patients, emotional aspects of physicians’ communicative behaviors played the most
important roles in their compliance and satisfaction (e.g., partnership and affective

empathy). Interestingly, the fact that partnership was the most signiﬁcant factor

inﬂuencing both patient compliance and satisfaction may reﬂect Korean patients’ strong
yearning for an egalitarian relationship with their physicians in the current hierarchical
authority structure of the medical consultation processes in Korea. Finally, important

implications for medical providers in Korea were suggested.

Copyright by
SUNG soo KIM
2000

To my parents (Yong Jung Kim and Kyung Ja Lee), my dedicated wife (So Young Seo),
my daughter (J i Ye Kim), and my newborn son (Sang Young Kim).

ACKNOWLEDGEMENTS

I would like to thank my dissertation committee: Dr. Kaplowitz, Chairman, Dr.
Conner, Dr. Gold, Dr. Calantone, and Dr. Perlstadt for their outstanding direction and
assistance in my dissertation.

I am most grateful to Dr. Kaplowitz for his constant encouragement and help
throughout my entire study.

My special thanks go to Dr. Calantone for teaching me the structural equation
modelings with great passion and enthusiasm as well as for his critical and helpful
comments on my data analysis.

I am greatly indebted to Dr. Thio, my lifetime mentor, for his unconditional
support. He has been the one whom I turned to whenever I was in difﬁcult situations. He
showed me the real meaning of trust and honesty.

I am also grateful to Dr. Marger for his kindness to my family. He also showed
me excellent teaching skills.

I would also like to thank my friends, Dr. Sung Hoon Lee, Kang Yuan Fu, and
Chien-Juh On, for their ﬁ‘iendship and support.

Finally, I would like to sincerely thank my parents for their patience and support,
my lovely wife for her dedication, and my children for their big smiles. Without them, I
could not have achieved my goal.

vi

TABLE OF CONTENTS

List of Tables .............................................................................................................. x
List of Figures .............................................................................................................. xi
CHAPTER I .................................................................................................................. 1
INTRODUCTION ........................................................................................................ 1
Purpose of the Study ........................................................................................ 1
CHAPTER II ................................................................................................................ 4
LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT .......................... 4
Models of Physician-Patient Relationships, Patient Compliance and
Satisfaction, and Cultural Variations ............................................................. . 4
Physicians in Korea: Cultural Context ............................................................. 9
Structure of Empathy ........................................................................................ 11
Deﬁnition .......................................................................................................... l3
Physician Empathy in Medical Encounter ........................................................ l7
Physician’s Cognitive and Affective Empathy .................................... I7
Cognitive Information Exchange .......................................................... 17
Partnership ........................................................................................... 18
Patients’ Trust: Perception of Physician Expertise and Interpersonal
Trust ...................................................................................................... 20
Cognitive Information Exchange, Perception of Physician Expertise,
and Compliance ................................................................................... 22
Partnership, Interpersonal Trust, and Satisfaction ................................ 24
Satisfaction and Compliance ................................................................ 26
Hypotheses ........................................................................................................ 26
CHAPTER III .............................................................................................................. 29
METHODOLOGY ....................................................................................................... 29
Research Design and Data Collection 29
Translation ............................................................................................ 29
Sampling ............................................................................................... 30
Obtaining approval for survey .............................................................. 30
Printing Questionnaire .......................................................................... 31
Recruiting and Training ........................................................................ 32
Modiﬁcations in the survey procedures ................................................ 32
Measures ........................................................................................................... 34
Physician empathy ................................................................................ 34

vii

Satisfaction Measure ............................................................................. 35

Compliance Measure ........................................................................... 36
Patients’ Questionnaire ......................................................................... 36
Cognitive Information Exchange and Partnership ................................ 36

Patient’s Trust: perception of physician expertise and interpersonal
trust ..................................................................................................... 36
Analysis ............................................................................................................ 37
Use of Structural Equation Modelings (SEM) .................................... 37
Structural Equation Modeling Software Program (EQS) ..................... 40
Assessment of Model Fit ..................................................................... 40
Preliminary Analyses ........................................................................... 41
Two-Step Modeling Approach ....................................................................... 42
Conﬁrmatory Measurement Model Test (CFA) .................................. 42
Measure Puriﬁcation ................................................................ 44
Reliability and Validity ........................................................... 44
Full Structural Model Test (SEM) ....................................................... 45
CHAPTER IV .............................................................................................................. 46
RESULTS ..................................................................................................................... 46
Data. ................................................................................................................ 46
Assessment of Model Fit ................................................................................ 49
Normality Test ................................................................................................ 49
Univariate normality checked for skewness, kurtosis .......................... 49
Multivariate normality ......................................................................... 50

Results of Conﬁrmatory Factor Analysis Measurement Model Test

(CFA) ............................................................................................................. 51
Measure Puriﬁcation ........................................................................... 5]
Reliability and Validity ...................................................................... 56
Results of Full Structural Model Test .............................................................. 60
Hypotheses Test .................................................................................. 61
Total Effects ....................................................................................... 65
CHAPTER V ................................................................................................................ 70
SUMMARY, DISCUSSION, AND IMPLICATIONS ................................................ 70
Summary ......................................................................................................... 70
Discussion of Major Findings .......................................................................... 71
Implications for Practitioners and Educators .................................................... 78
Limitation and Future Research ...................................................................... 80
REFERENCES ............................................................................................................. 84
APPENDIX I : Initial Empathy Questionnaire ........................................................... 92

viii

APPENDD( H: Final Set of Variables Used For the Data

APPENDIX 111: References for Questionnaire
Innns.. ."uuuuu. .uu..uuuuuuuu.n.u.u.n.n.n.u.u.n.u.n.u.H.H.H.H.H.nn 100

ix

LIST OF TABLES

Table 1 - Age ............................................................................................................. .47
Table 2 - Sex of Patient .............................................................................................. 47
Table 3 - Patient Education ........................................................................................ 48
Table 4 - Patient Income ............................................................................................ 48
Table 5 - Sociodemographic Indicators in Korea: age, sex, education, and

household ................................................................................................... 49
Table 6 - Skewness and Kurtosis ............................................................................... 50
Table 7 - Initial Number of Indicators for Constructs ................................................ 52
Table 8 - Final Number of Indicators for Constructs ................................................. 54
Table 9 - Goodness of Fit Results for CFA ............................................................... 55
Table 10 - Test Results of Measurement Model (CFA) ............................................. 58
Table 11 - Test of Discriminant Validity: Xd (1)2 test between the constrained and

unconstrained models ................................................................................. 59
Table 12 - Correlation / Discriminant Validity Measurement: correlation among

constructs .................................................................................................. 60
Table 13 - Goodness of Fit Results for Full Structural Model ................................... 61
Table 14 - Results of Hypotheses Test ...................................................................... 63

Table 15 - The Order of the Importance of Predictors on Compliance: total effects" 68
Table I6 - The Order of the Importance of Predictors on Satisfaction: total effects" 68

LIST OF FIGURES

Figure l - Hypothesized Empathy Model (causd phases in the relationship between

empathy and patient outcomes) .................................................................... ‘ 27

Figure 2 - An Exemplary Path diagram of a Single Latent Variable with Four
Indicators 38
Figure 3 - Hypothesized Full Empathy Model ............................................................ 40
Figure 4 - Full Model SEM Results ............................................................................. 62
Figure 5 - Total Effects on Patient Compliance .......................................................... 66
Figure 6 - Total Effects on Patient Satisfaction ........................................................... 67

xi

Chapter 1

INTRODUCTION

Accumulating research shows the important effect that physician’s
communication behaviors have on patient outcome. Particularly, the physician empathy
has been proposed as one of the most important determinants of patient satisfaction and
illness improvement (Olson, 1995). A physician’s use of empathy not only helps him or
her to obtain a reliable history of the patient’s illness, but also to more effectively
understand the illness from the patient’s point of view. This, in turn, enhances the
patient’s faith in the physician and creates an environment in which patients are
encouraged to interact more with their physicians. Thus, the physician empathy provides
an important conceptual framework for a better systematic theory of physician-patient
communication which enables us to explain how and why physician empathy is one of the

most important determinants of patient satisfaction and compliance.

Purpose of the study

Although much of the research agrees that the presence of empathy is an essential
variable in successful physician-patient communication with positive outcomes, there has
been a very few studies with the dualistic view (cognitive and affective) of empathy.
Among those few studies, Squier (1990) proposed one of the most comprehensive
empathy models by making theoretical links between dual components (cognitive and
affective) of physician empathy and patient adherence. However, Squier’s model suffers

several shortcomings. First, it remains only a theoretical model since it has not been

empirically tested (consequently, no proper measurements have been developed). Second,
Squier’s model doesn’t consider another important outcome variable such as patient
satisfaction. Failing to incorporate other outcome variables into his model may have
missed the signiﬁcant link of the patient satisfaction to patient compliance. Third,
Squier’s model can’t determine the degree to which each component of physician
empathy is contributing to patient outcomes (satisfaction and compliance). Therefore, it
needs an empirical test as well as further conceptual improvement.

Furthermore, while over the last couple of decades there has been impressive
increase of empathy studies in the United States, the medical ﬁeld in Korea has rarely
seen physician-patient relationship studies, let alone the effect of the physician empathy
on the patient outcomes. Most of the studies in Korea are related to Korean health care
systems, showing a tendency to focus on such structural issues as ﬁnancial, cost
containment measures for the reformation of the Korean National Health Insurance
(KN I-H) with little concerns with the physician-patient communication issues.

Therefore, this study tested a new empathy model, derived loosely from the
Squier’s model, but built on the recognition of the shortcomings of the Squier’s model.
The empathy model for this study ﬁrst attempted to develop and test a new empathy scale
to measure two aspects (cognitive and affective) of physician empathy. Next, it attempted
to give a ﬁrst-kind of insight into the mechanism or process of the physician-patient
communication in Korea. It did so by examining the effect of Korean physician empathy
on patient outcomes from dualistic view of empathy. The dualistic view of the physician
empathy provides us with a powerful explanatory mechanism of how each component of

the empathy construct plays a part in patient behavior such as patients’ satisfaction and

compliance. In other words, it determines how and what component of the empathy
construct accounts for different levels of medical outcomes.

The Korean government is currently attempting to reform the health insurance
system due to rapid growth of health care expenditure, proliferation and duplication of
medical technology, and lack of access for low-income groups (Bong-Min Yan, 1996).
Further impetus for this reform comes from the fact that the fees for health services
healthcare institutions in Korea tend to provide low quality care (Youngsoo Shin, 1995).
The knowledge provided by this study provided both physicians and medical educators in
Korea with a scientiﬁc basis by which to foster a better physician-patient relationship. It
suggested ways to eventually improve medical outcomes for patients without resorting to
additional increase of medical cost. Additionally, it also provided Western readers with a
rare opportunity to grasp the physician-patient relationship in Korea.

The premise adopted here was that the patient interpretation or impression of
physician empathy during a medical encounter is a key variable in determining patient
outcomes. Empathy underlies this impression made upon the patient. That impression
results from the physician’s connectedness to the patient by the conscious and
unconscious conﬂuence of thoughts and emotions (Brock, & Salinsky, 1993). The
patient’s feeling of being empathically understood will eventually inﬂuence both the

patient’s satisfaction and his or her compliance.

Chapter II

LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT

Models of Physician-Patient Relationships, Patient Compliance and Satisfaction,
and Cultural Variations

The quality of medical care depends largely on the interaction and communication
between the patient and the doctor. Physician-patient relationships can be categorized
according to their communication styles. Two important works which discussed the
models of the physician-patient relationship were written by Szasz and Hollender (1956)
and Veatch (1991). Szasz and Hollender (1956) described physician communication
styles according to the nature of the patient illness. Their models improved and reﬁned
the Parsonian functionalistic and paternalistic views (Parsons, 1951) on the physician-
patient relationships by encompassing different types of illness, other than the most acute
(as in the Parsons’ sick role model). Szasz and Hollender identiﬁed three different types
of physician-patient interaction: activity-passivity (appropriate for acute illness);
guidance-participation (for less acute illness); mutual cooperation (for chronic illness).

Concerned with the issue of moral partnership between physicians and patients,
Veatch (1991) proposed four broader models for the physician-patient relationship
models. The ﬁrst type, the Engineering model, argues that the physician as ‘a plumber’
without moral integrity provides simply the facts to the patient, who then decides whether

to use medical science to solve his or her problem. The second type, the Priestly model,

which is similar to the paternalistic model by Parsons (1951), and the activity-passivity
model by Szasz and Hollender (1956), views the physician as a ‘priest’ who takes care of
all decision-making and medical processes. The third type, the Collegial model,
emphasizes a more balanced power association. In this case, both parties pursue the
common goal of the well being of the patient. The last type, the Contractual model, «
which is developed based on the concept of contract, argues that physicians and patients
can truly share ethical authority and responsibility.

Although both Szasz and Hollender (1956), and Veatch (1951) appear to agree on
the much greater role of physician authority (or paternalism) in the traditional physician-
patient relationship, which is still seen as the most common, patients have some means to
control the relationship. Patient noncompliance can be viewed as an expression of patient
power. Indeed, one study by Hayes-Bautista (1976) examined patient noncompliance in
relation to the physician-patient relationship. In the study of Chicano patients, Hayes-
Bautista (1976) attempted to analyze physician-patient relationships with focus on power
and authority. Hayes-Bautista argued that the modiﬁcation of medical regiments becomes
an important sources of conﬂict between physician-patient communication. Hayes-
Bautista observed two major tactics (‘convincing’ and ‘countering’) utilized by patients
and physicians respectively. Hayes-Bautista found that patients used non-compliance as a
means of taking control of the physician-patient relationship to achieve satisfaction with
their treatment. In another patient-noncompliance study, Svarstad (1986) proposed the
‘health communication model’, which strongly emphasized the quality of physician-
patient communication. The core of the model asserts that patient compliance is

inﬂuenced mainly by two critical factors (patient comprehension and recall of regimen,

and patient motivation to follow regimen). Svarstad implied that physician
socioemotional communication skills may increase patient compliance while the
physician’s dominant communication skills do otherwise. Although the Svarstad model
remains only a conceptual model, it provides physicians with a variety of useful strategies
for enhancing patient compliance (e. g., explicit direction and friendliness or
approachability).

Another health outcome, patient satisfaction, was also related to physician-patient
communication, particularly physician communication styles. For example, in their study
of physician interviewing style and patient satisfaction, Rowland-Morin and Carroll
(1990) found that 27% of the variance in patient satisfaction scores were explained by
three aspects of a physician communication style (use of silence, language reciprocity and
interruptions). Rowland-Morin and Carroll concluded that these verbal communication
behaviors mostly reﬂect patient-centered interviewing style, involvement expressiveness,
lack of dominance, and reciprocity.

Indeed, others (Allman, et al., 1993; Wyatt 1991) also found similar results in
their literature review on physician-patient communication. Alhnan and his colleagues
(1993) argue that the differences between physicians’ and patients’ agendas for medical
encounters are the most serious obstacles to forming a satisfactory relationship between
physicians and patients. For example, patients wish more information while physicians
underestimate the patients’ information desire. Moreover, they continued to ﬁnd evidence
to assert that the physician’s communication style is directly related to patient satisfaction
and patient compliance. In a similar study, Wyatt (1991) conducted a content analysis of

the 168 articles published between 1983 and 1989 regarding physician-patient

relationships. Wyatt found that less than 1% of the medical literature was related to
physician-patient relationships. Wyatt attributed this to the continuing power of the
biomedical model in medicine. Wyatt implied that effective communication with more
humanistic concerns between physicians and patients is the key to patient satisfaction.
Different cultural (or ethnic) groups communicate their illness to physicians in
different ways by using their own cultural and contextual language of the disease
(Helman 1994). Thus, cultural variations in patients can be an important factor that might
affect physician-patient communication. One of the classic studies on cultural variation in
physician-patient communication was done by Zola (1966). In the study of the
presentation of the illness by the Italian and Irish Americans, Zola found that that the
Italian Americans tended to dramatize their illness with many more symptom complaints
while the Irish tended to downplay their symptoms. The physicians also showed a
tendency to diagnose the patients differently, based on the patients’ origins. Even without
any physical disease present, the Italian was more often diagnosed neurotic than the Irish.
Zborowski’s (1952) study’also reported similar ﬁndings in the responses to pain by
different ethnic groups (Irish-, Italian-, Jewish-American patients, ‘Old Americans’).
Zborowski found that different ethnic groups had different attitudes toward pain. For
example, the Italians tended to be more concerned with the actual pain experience, while
Jewish patients were worried more about the symptomatic meaning of pain in relation to
their health. Zborowski pointed out some factors which might cause these differences
among different ethnic groups such as the degree of Americanization and education.

Zborowski concluded that these attitude differences toward pain were the result of the

patients’ early socialization. The researcher emphasized the importance of the cultural
context of the patients.

In conclusion, with the help of the models provided by Szasz and Hollender
(1956), and by Veatch (1991), we are able to better understand the nature of physician-
patient communication. Szasz and Hollender’s mutual participation model and the '
Veatch’s contractual model provide an important implication for this study. Both models
have one important element in common: egalitarianism. Like the physicians in these
models, an empathic physician might be more likely to perform egalitarian behaviors
(e.g., sharing power and responsibility). The relationship between the empathic physician
and the patient might fall under these models (mutual participation and contractual
model). I believe that the physician’s egalitarian behaviors are the consequences of the
physician’s affective empathy.

Next, the physicians' socioemotional and communicative behaviors ( or empathic
behaviors) were shown to be directly related to such outcomes of health care as patient
satisfaction and compliance (Rowland-Morin and Carroll,l 990; Svarstad, 1986). More
importantly, the studies by Rowland-Morin and Carroll (1990) and Svarstad (1986)
provide a signiﬁcant implication for a possible link between the physician’s empathic
behaviors and the medical outcomes (patient satisfaction and compliance). In the
Svarstad’ health communication model for patient compliance, there is a link between
patient compliance and the two factors (patient comprehension and recall of regimen, and
patient motivation to follow regimen). This implies that there are two distinct physician
behaviors that affect patient compliance: cognitive behaviors for the patient

comprehension and recall, and affective behaviors for patient motivation. Lastly, Zola

(1966) and Zborowski (1952) showed that there were clear cultural variations in
physician-patient communication, depending on the patient’s ethnic origin. With the
important role of culture kept in mind, in the next section, I will examine how the Korean

culture inﬂuences the physician-patient relationship.

Physicians in Korea: Cultural Context

Although little has been known about the relationship between physicians and
patients in Korea, physicians in Korea are still widely considered a prestigious profession.
Physicians enjoy quite a lot of both autonomy in their expertise and authority over the
patients and the patients usually accept that. However, this seemingly high degree of the
physicians' autonomy and authority in Korea can be explained not only by the technical
knowledge gap between the physicians and the patients, but also partially by Koreans’
traditional cultural norms toward authority ﬁgures such as doctors, teachers, scholars,
government officials. Indeed, having reviewed many studies (e.g., Choi, 1979;
Henderson, 1968; Kim, 1983; Lirn, 1983) on Korean value system, Lee (1993) concluded
that the traditional Korean value pattern is an authoritarian one, emphasizing deference to
the traditional symbols and holders of authority.

The public perception that doctors can be self-interested and greedy rather than
just interested in the public’s well-being, and the level of the public’s consciousness
about their rights have signiﬁcantly increased in the last a few decades. However, the
ideas that the holders of authority-doctors, teachers, scholars, and government officials--
should deserve authority and that people should not question about their authority, are

still widely embedded in the general public's mind. There are still many Korean patients

who see themselves as passive objects in the medical encounter. This high public
tolerance of physicians’ authority may contribute to the lack of Korean physicians’
empathic behaviors. Therefore, experiencing authoritarian Korean physicians in medical
settings is not uncommon. To Korean patients who have been relatively less exposed to
empathic physicians (compared to patients in the United States), experiencing empathic
behaviors from their physicians might play a greater role in patients’ outcomes
(satisfaction and compliance) than in the US.

Furthermore, the Korean health care system rapidly expanded national health
insurance in the last three decades (0k, 1991). As a result, new challenges emerged. One
of them is patient dissatisfaction. Due to the rapid social and economic growth in Korea,
many Koreans began to see themselves as active health care customers rather than passive
recipients. This change in the patients’ perspective brought forth a new meaning in the
doctor-patient relationship. The patients no longer want to be in an unequal hierarchical
structure. The patients are trying to see it as a supplier-customer relationship between
equals. Furthermore, since most patients prefer to be treated in general hospitals,
overcrowded general hospitals became a serious concern for patient satisfaction (Yang,
1996). The general hospitals’ inefficient bureaucratic management has left many patients
with unpleasant experiences such as long-waiting time, unfriendly services, and above all
impersonal treatment of patients by physicians.

Therefore, given the above situation, the increase of patient dissatisfaction with
the quality of the services has become a critical factor for the survival of medical

providers in the highly competitive Korean health care market.

10

Structure of Empathy/Deﬁnition

Despite the widespread recognition of empathy as an important quality of human
characteristics, little consensus has been reached among the theorists on its deﬁnition.
Mead (1934) contended that empathy, as a role-taking ability, is the essence of social and
moral development as well as of social intelligence. Freud (1949) deﬁned empathy as the
mechanism that enables us to take up any attitude toward another’s mental life.

For Jung, empathy was viewed as one’s ability to project as a conﬂuence of the viewer’s
and the viewed’s thoughts. Others contended that the accurate understanding of another
person’s feelings is the key factor in empathic interaction (Bellet & Maloney, 1991).
Although this diversity of deﬁnitions of empathy seems confusing, the underlying
denominator concerns two important elements: cognitive ability to understand the other
person’s perspective and affective ability to respond and improve the others’ emotional
state.

Indeed, historically, a controversy exists over whether empathy is an affective or
cognitive construct, or both. Psychologists studying empathy use either one or the other
of these general deﬁnitions. Some researchers took empathy to be a cognitive process
with the emphasis on cognitive role taking or perspective taking. (e. g., Hogan, 1969).
Others investigators took it to mean a primarily affective process (e.g., Mehrabian &
Epstein, 1972). Still others, primarily in the medical area, focused on a communicatory
and information-giving or information-gathering function of empathy (e.g., Roter et al.
1987; Dean, 1993; Roter, et. al., 1987; Glanz, 1997; Comstock, et. al., 1982; Waitzkin

I984).

11

As a result of the debate surrounding these perspectives, some researchers are now
integrating what is most valuable in each view into a multidimensional consideration of
the empathy construct (Eisenberg and Strayer, 1990). For example, some researchers like
Davis (1983) approached empathy from a multidimensional perspective — a combination
of the cognitive role-taking and vicarious arousal. In fact, the most prevailing current
view among empathy theorists and researchers holds that empathy entails both affective
and cognitive elements and that only the clear recognition of the existence of both
affective and cognitive aspects can improve our understanding of empathy (Davis, 1983;
Squier, 1 990).

Likewise, in medical research on empathy in relation to patient outcomes, the
most predominant trend in the ﬁeld of the physician-patient relationship appears to lean
toward the multifaceted aspects of empathy, emphasizing empathy as a communicatory
process. For example, Squier (1990) deﬁned empathy as consisting of two components: a
cognitive-informational aspect and an affective-motivational component and saw both as
enhancing patients’ compliance. Brock and Salinsky (1993) also suggested a similar
biphasic structure of empathy: empathic understanding and empathic response. In
somewhat similar vein, Carmel and Glick (1996) viewed physician empathy as having
two-dimensional constructs: what they call the scientiﬁc-technical and socio-emotional.

Therefore, conceptualizing the dual faceted empathy (cognitive and affective) has the
most relevant implications for the present study. First, physicians’ affective empathic
behaviors alone can rarely produce such desired patient behaviors as compliance and
satisfaction. They may make patients feel liked and friendly, but this is not sufﬁcient to

achieve the patients’ compliance and satisfaction. Feeling friendly cannot alone motivate

12

the patients to attend to information and comply with a treatment regimen (Glanz, K. et.
al. [eds], 1997). Another beneﬁt from the dual faceted conceptualization is that it shows
exactly how each component of the empathy construct is related to patient outcomes --
patients’ satisfaction and compliance (Is each component equally related to the outcomes?
or does one component play a greater role in the outcomes?) Thus, this study will attempt
to develop and evaluate a new empathy construct which enables us to measure two

separate components (cognitive and affective) of physician empathy.

Deﬁnition

For the present study, some of the operational deﬁnitions used in the study should
be clariﬁed. The operational deﬁnition of physician empathy is a patient’s feelings of
being understood and accepted by a physician. It consists of two components: cognitive
and affective. The cognitive aspect of physician empathy is deﬁned as the physicians’
ability to apprehend accurately the mental state of patients (ability to take another
person’s point of view), and to communicate effectively this perspective back to the
patients. The affective aspect of physician empathy is deﬁned as physicians’ ability to
respond and improve patients’ emotional state. From now on, I will use physician

empathy for patient-perceived physician-empathy for a stylistic reason.

13

Critique of Ben-Sira’s Social Interaction Model and Squier’s Model of Empathic
Understanding and Adherence to Treatment Regimens

Among numerous studies of empathy, the two important theoretical models
deserve attention for the present study. Both Ben-Sira’s (1976) Social Interaction Model
and Squier’s (1993) model have attempted to examine the link between physician
behaviors (affective or empathic) and patient outcome. In his attempt to explain the effect
of physician affective behavior on patient satisfaction, Ben-Sira (1976, 1980) suggested a
social interaction model. In his social interaction model, Ben-Sira argued that due to a
patient’s anxiety about the condition of the illness, lack of medical technical knowledge,
and inability to judge better treatment, the patient’s immediate satisfaction will be the
results of the physician’s affective communication behavior. Ben-Sira has contributed
signiﬁcantly to our understanding of the roles of physician empathy (or “caring” in
patient outcome with his insightful social interaction model. However his model contains
some conceptual shortcomings in explaining why physician empathy (or affective
behavior for Ben-Sira) is one of the most important determinants of patient satisfaction
and compliance.

First, Ben-Sira’s social interaction model (1976) implicitly indicated that those
patients who are educated about medical procedures, and not too emotionally involved in
the illness, would depend only on a physician’s technical performance of medical
procedures when evaluating their medical care. This ignores an important role that
physician empathy plays in these technical procedures. For more effective care, these
technical procedures should depend on the use of physician empathy to more effectively

ascertain the patient’s medical condition and a reliable history, provide a more accurate

l4

diagnosis, give more clear-cut treatment directives, and understand more clearly the
meaning of the illness to the patient (Buller and Buller, 1987).

Second, Ben-Sira overlooked the importance of the nonverbal aspects of
physicians’ affective behavior in his model. A physician’s empathic verbal statements
are not the only factor inﬂuencing a patient’s deﬁnition of the medical situation or
interpretation of a physician’s communication behavior. Nonverbal aspects of a
physician’s empathic behavior are actually the key factors with which patients are able to
judge the genuineness of a physician’s communication (Goffrnan, 1959). In support of
this argument, Bensing (1991) argued that the nonverbal aspects of affective behavior
(eye contact and showing interest) were strongly related to the quality rating of
psychosocial care.

In contrast to Ben-Sira, Squier (1990) proposed the most comprehensive model of
physician empathy and patient outcome. His model hypothesizes that physician’s
empathic understanding is a prerequisite to patient compliance with medical regimens.
The strength of his theoretical model is that it takes a dualistic view of empathy
(cognitive-informational and affective-motivational). As discussed previously, unlike
many other studies, which take one-dimensional view of empathy (either affective or
cognitive), Squier’s model emphasizes that affective aspect of physician empathy is not
alone sufficient enough to produce desired patient outcomes. Both dimensions (cognitive-
informational and affective) must be present to achieve complete empathy.

Despite this model’s theoretical strengths, as previously discussed, Squier’s model
suffers several shortcomings. First, it still remains a conceptual model since it has not

been empirically tested (consequently, no proper measurements have been developed).

15

Second, Squier’s model doesn’t consider another important outcome variable, patient
satisfaction. Failing to incorporate that variable into his model may have missed the
signiﬁcant link between the patient satisfaction and patient compliance. Third, Squier’s
model didn’t conceptualize patients’ trust in physicians and its effect on patient
outcomes. Patients’ trust in their physicians have been recently acknowledged as a
signiﬁcant factor in patient-physician relationships (Anderson and Dedrick, 1990) as well
as in successﬁrl treatment (Mechanic, 1998). However, Squier’s model seems to be
ignorant of the importance of this variable in regard to the relationship processes between
physician empathy and patient outcomes. Fourth, without an empirical test, the model can
not determine the degree to which each component of physician empathy is contributing
to patient outcomes (satisfaction and compliance). This is an important drawback since
knowing this would help an individual physician to realize what aspect of his or her
empathic communication skills he/she needs to improve to maximize patients’
satisfaction and compliance. Thus, the Squier’s model needs an empirical test as well as
further conceptual improvement.

In sum, despite the important contribution of Ben-Sira’s social interaction model
and Squier’s model to understanding of the relationship between empathic physician
communication and patient outcome, the shortcomings of their work necessitate a new

and improved conceptualization.

16

Physician Empathy in Medical Encounter
Physician ’3 Cognitive and Aﬂ'ective Empathy

The physician should attempt to understand the patients’ illness experience, both
cognitively and affectively (Squier, 1990). The cognitive and affective components of the
physician’s empathy work in concert to form a more informed understanding of the
patient. First, cognitive empathy is the physician’s ability to accurately discern the
mental state of the patient (the patient’s point of view), and communicate effectively this
perspective back to the patient, which explains most of cognitive consequences of
physician’s empathy. Second, affective empathy is the physician’s ability to respond and
improve the patient’s emotional status, which accounts mainly for the affective beneﬁt to
patients.
Cognitive Information Exchange

The cognitive component of the physician’s empathy facilitates the cognitive
information exchange between physician and patient. Most of the traditional roles of the
physician (e.g. Parsons’ [1951] universalism, functional speciﬁcity, and affective
neutrality) take place in the domain of cognitive information exchange. These roles are
mainly technically based skills (“expertness”) for which patients are consulted (Roter and
Hall, 1997). These roles include information-giving, thoroughness, data-gathering,
accurate diagnosis and appropriate treatment. Accuracy of the physician’s cognitive
understanding of the patient’s illness experience as well as the physician’s other cognitive
informational behaviors (e.g., sufﬁcient information giving, physicians’ thoroughness)
provides the patient with a feeling of being understood. This stimulates the patient to

more accurately report past and current symptoms as well as medically relevant

l7

experiences (Roter, 1997), thereby increasing insight into the nature of the illness (Squier,
1990;More and Milligan, 1994). A study by Young (1980) strongly supports the above
argument. Young’s study found that when patients perceived their physicians as highly,
technically competent, their willingness to disclose all types of symptoms increased.

When the patient provides detailed information, it helps the physician obtain a
reliable history, understand more accurately the patient’s illness from the patient’s point
of view, and provide more information to the patient (Bellet and Maloney, 1991). The
cognitive information exchange, facilitated by physician cognitive empathy, is mutually
beneﬁcial to both the physician and the patient. The functions of the cognitive
information exchange are signiﬁcant not only “in furthering the physicians’ clinical
agenda to diagnose and treat, but also in furthering the patient’s agenda to understand and
make sense of the frightening vulnerability of illness as well as to feel that his or her
experience and perspective are understood.”1

Thus, it is hypothesized that the cognitive component of physician empathy leads

to better exchange of cognitive information (Hypothesis l-a).

Partnership

While the cognitive component of physician empathy expedites the cognitive
information exchange, the affective aspect of physician empathy stimulates a physician to
deﬁne his or her relationship with the patient as a partnership. The partnership is deﬁned

as a physician’s willingness to labor jointly with a patient to promote the same goal- the

 

' Glanz, K. et al. (Eds) 1997. Health Behavior and Health Education: Theory, Research, and Practice (2mI
edition) Jossey—Bass Publishers, San Francisco. Pp. 215.

18

well-being of the patient. This type of relationship enables the physician to share power
and responsibility with the patient for the patient’s health and medical decisions as a
partner. In particular, partnership helps the physician encourage the patient to express or
indicate his or her feelings, concerns, opinions, and questions. The partnership also
prevents passive or noncommunicative behaviors from the patients, since it engages them
more fully in the medical conversation (Roter and Hall, 1993).

Without an empathetic partnership, passive, inactive, or dominant physicians may
elicit negative reactions from patients. As in any informal social interaction (e.g. social
conversation), patients may expect more egalitarian interaction or more partnership-like
relationship with physicians. In such a relationship, both parties make reciprocal
contributions to the medical communication rather than having the physician control and
dominate the content and direction of the medical discourse (Street et al., 1987). For
example, Waitzkin (1984) pointed out that a physician—dominated style of communication
often overlooks patient concerns and correlates to subsequent patient dissatisfaction with
physicians. Other ﬁndings are also consistent with Waitzkin’s (1984) research (e.g.,
Buller and Bullet, 1986; Hall, et. al., 1981).

In sum, physician’s affective empathy is a crucial factor in building partnership
between physicians and patients. Partnership enables both physicians and patients to
make reciprocal contributions to the medical communication, which become an important
basis for the patient’s satisfaction and compliance. Thus, it is hypothesized that the

aﬁ’ective aspect of physician empathy leads to partnership (Hypothesis l-b).

19

Patients ’ Trust: Perception of Physician Expertise and Interpersonal Trust

The patient’s trust in his or her physician has recently been recognized as a critical
factor in the patient-physician relationship (Anderson and Dedrick, 1990) as well as in
successful treatment (Mechanic, 1998). However, few studies have attempted to actually
measure and examine the functions of the patient’s trust in their physician in relation to
patient outcomes. Here, I deﬁne patients’ trust in physicians as another important factor
in producing better patient outcomes. It consists of two components: the ﬁrst is what I
call ‘perception of physician expertise’ and the second, ‘ interpersonal trust.’ The
conceptualization of the dualistic structure of trust for this study is supported mainly by
work of Thom and Campbell (1997). In their study of patients’ trust in their physicians,
Thom and Campbell classiﬁed physician behavior affecting trust into nine categories.
They found that two of the nine categories were associated with technical competence
(e.g., thoroughness in evaluation) and that ﬁve were related primarily to interpersonal
behaviors (e.g., building partnership/sharing power). The other two categories were
predisposing factors (e.g., age, sex) and structural/stafﬁng factors (e.g., courtesy of ofﬁce
staff). I excluded the last two categories (predisposing factors and structural/stafﬁng
factors) since they are not related to the actual patient-physician communication. A
careful examination of the rest of the categories revealed that trust might consist of two
major components: perception of physician expertise and of trustworthiness.

Thus, based on the implications of the above study, the perception of physician
expertise is deﬁned as the patient’s general assessment of whether he or she can trust the
physician as a medical expert who will provide a proper treatment of their illness. The

patient’s perception of physician expertise is derived primarily from the patient’s

20

impression of physician’s technical behaviors in the cognitive information exchange
domain. During a medical encounter patients are constantly assessing whether they can
trust their physicians as well qualiﬁed to treat their medical problems, though they lack
medical technical knowledge to judge accurately. Perception of physician expertise is an
important basis for inspiring the patient’s conﬁdence in the physician, thereby giving
patients hope for a fast recovery. Thus, I hypothesize that better exchange of cognitive
information leads to increased perception of physician expertise. (Hypothesis 2-a).
Another aspect of trust is what I call ‘ interpersonal trust.’ The interpersonal trust
is deﬁned as a patient’s general perception of whether he or she can trust a physician
emotionally. It is derived primarily from the partnership domain. Because of the gap
between physicians and patients in regard to knowledge and power, trust becomes a very
critical concern of the patient (DiMatteo et. al, 1985). Thus, the partnership practiced by
an affectively empathic physician creates a feeling of trust in the patient. Stile et al.
(1979) showed that physicians who acknowledged and understood the importance of the
patient’s own knowledge and abilities were found to elicit more trust and gain more
involvement in the consultation process. This is particularly important when we consider
a patient’s “emotional orientation” toward his or her relationship with the physician
(Segall et. al., 1980). In this view, patient expectations go beyond the mere desire for
physicians' technical competencies, because as patients visit doctors they expect not only
physical relief, but also social and psychological relief. The patient’s expression of
socio-emotional or, in Waitzkin’s (1991) words, ‘contextual problems,’ is not typically
facilitated in the medical discourse because of modern medicine’s emphasis on technical

and biomedical orientation. Thus, the partnership leads to increased interpersonal trust by

21

remain

enhancing physicians’ ‘interpersonal competence’ which, in turn, allows patients to
comfortably express ‘contextual problems’ or socio-emotional concerns (W aitzkin, 1991;
Bellet and Maloney, 1991). Patients need perception of physician expertise or conﬁdence
in their physicians’ technical performances, but they also need interpersonal trust in their
physicians. The interpersonal trust strengthens the physician-patient relationship and
substantiates the patient’s belief that the physician is working on their side and in their
best interest (Roter and Hall, 1993).

Therefore, I hypothesize that partnership leads to increased interpersonal trust

(Hypothesis 2-b).

Cognitive Information Exchange, Perception of Physician Expertise, and Compliance
Cognitive information exchange (e.g., information-giving, accurate report of
symptoms), facilitated by the physician’s cognitive empathy, coupled with perception of
physician expertise is more likely to lead patients to compliance. Although the literature
on compliance shows that both the cognitive and the affective components of physicians’
empathetic behaviors are related to patients’ compliance (e.g., Hall, et. a1, 1988),
physicians’ partnership behaviors is not enough to cause patients to attend to information,
change lifestyles, and comply with medical regimens. In addition, much of the patient’s
non-compliance has been traced to the patient’s difficulties of understanding the
treatment regimen as well as miscommunication between the physician and the patient
(Stanton, 1987; Svarstad, 1976; Freund, E. S. and McGuire, M. B. 1995). For example,
one study found that patients' noncompliance increased from 11 percent to 24 percent

when physicians failed to fulﬁll the patients' desire for expected information disclosures

22

about their illnesses (Francis, Korsch, and Morris 1969). In their meta-analysis study of
communication dynamics, Roter and Hall (1997) found the weak relationship between
physicians’ socioemotional behaviors (e.g., partnership building, positive talk, and etc.)
and patients’ task behavior (e.g., compliance and recall). Thus, it is hypothesized that
better exchange of cognitive information has a greater eﬁ’ect on the patient ’s compliance
than does partnership (Hypothesis 3-a)

Furthermore, I argue that perception of physician expertise leads to more
compliance than satisfaction. For the patient to comply with the medical advice of the
physician, the patient should not only have a clear understanding of medical regimens,
but perception of physician expertise (or conﬁdence) in his or her physician. Without
perception of physician expertise, the physician’s instructions are not perceived as
credible, which increases non-compliance. Although no studies have examined the direct
link between perception of physician expertise and compliance as deﬁned here, some
research provided valuable implications on the two variables. For example, DiMatteo
(1994) in his literature review on compliance argued:

“Patients typically follow only the recommendations they really believe in
when patients doubt the usefulness of a medication, for example, they purposely
ignore it or conveniently forget to take it. It is up the physician to elicit all of
the patient’s concerns about the recommended treatment and to provide enough
information for the patient to believe the treatment is right for him or her.”2

Thus, from the implications of the above empirical statement, we could argue that if

patients trust their physicians as reliable medical experts, they are more likely to believe

in their physicians’ recommendations and comply with them.

 

2 DiMatteo, M. R. 1994. “Enhancing patient adherence to medical recommendations” JAMA, Vol. 271(1):pp.
79, 79-83.

23

Furthermore, since patients build their perception of physician expertise mainly
from their impression of physicians’ technical behaviors (e.g., physicians’ information-
giving, accurate diagnosis) which was shown to be more associated with compliance than
with patient satisfaction, it could be inferred that perception of physician expertise might
also be more associated with patient compliance than with patient satisfaction. Therefore,
it is hypothesized that perception of physician expertise has a greater eﬂect on the

patient ’s compliance than on satisfaction (Hypothesis 4-a).

Partnership, Interpersonal Trust, and Satisfaction

An empathic (affective) physician establishes and maintains partnership with a
patient and a patient’s interpersonal trust. Partnership and the interpersonal trust become
strong bases particularly for the patient’s satisfaction. Many studies revealed that
physicians’ partner-like behaviors (statements of agreement, social conversation, and
more eye contact) were the most signiﬁcant factor in determining patients’ satisfaction.
(Hall, Roter, and Katz, 1988;Bensing 1991; Buller and Buller, 1987). Quine and Rutter’s
(1994) study of satisfaction in relation to medical communication at the time of diagnosis
of a child’s severe mental or physical disability is especially relevant here. They tested
two models of doctor-patient communication: Korsch’s (1968) affective model and Ley’s
(1977) cognitive model. Ley’s model concentrated on cognition (e.g., understanding of
treatment and recall), while Korsch’s model focused on affect and social interaction (e.g.,

“whether the doctor had a sympathetic manner; whether he or she understood the

24

mother’s concern; and whether he was direct, approachable, and a good communicator“).
Quine and Rutter (1994) found that Korsch’s affective model was a much stronger
predictor of parental satisfaction than Ley’s cognitive model. Indeed, in their meta-
analysis study of communication dynamics, Roter and Hall (1997) also found that the
physician socioemotional behaviors (positive, social, and negative talk, interpersonal
competence and partnership building) were more correlated with patient satisfaction than
with patient compliance

In sum, the implications of the above studies indicate that partnership appears to
be a strong predictor of patient satisfaction. Thus, it is hypothesized that partnership has
a greater eﬂect on the patient ’s satisfaction than does cognitive information exchange
(Hypothesis 3-b).

No empirical study has used the same deﬁnition of ‘ interpersonal trust’ as
employed in this study to examine its link to patient satisfaction. However, since
empirical evidence showed a signiﬁcant link between partnership and patient satisfaction
(Roter and Hall, 1997) and since ‘ interpersonal trust’ was conceptualized to be derived
mainly from the patient’s impression of the physician’s partnership-like behaviors (e.g.,
the statement of agreement and approval), it is inferred that the interpersonal trust along
with partnership can contribute signiﬁcantly to patient satisfaction.

Thus, it is hypothesized that interpersonal trust has a greater eﬂect on the

patient ’s satisfaction than on compliance (Hypothesis 4-b).

 

3 Quine, L. and Rutter, D. R. 1994 “First Diagnosis of Severe Mental and Physical Disability: A Study of
Doctor-Parent Communication. ” Journal of Child Psychology and Psychiatry and Allied Disciplines Vol. 35(7):
1276.

25

Satisfaction and Compliance

Although all of the relevant variables (cognitive information exchange,
partnership, perception of physician expertise and interpersonal trust) in the medical
consultation should also be present and combined to produce patients’ satisfaction and
compliance, patients’ satisfaction is more likely to be inﬂuenced by the ‘partnership’ and
‘ interpersonal trust’ variables. And more satisﬁed patients are more likely to comply
with the medical regimens. Indeed, there is evidence that patient satisfaction with a
physician’s aﬁ‘ective behavior leads to a greater patient compliance with the medical
regimens (Haynes, 1976; Segall et. a1, 1980; Ley, 1986; Becker et. al., 1975). Many other
studies also revealed that the more satisﬁed patients are, the more compliant they are with
their doctors’ drug regimens (Weisman, C. S. and Nathanson C. A., 1985; Linn M. W. et.
a1. 1982; Barlett, E. E. et. al., 1984; Korsch, B. M. et. a1, 1968; Hulka, B. S. et. al.
1976). The implications of all of the above studies would suggest a hypothesis that

patient ’s satisfaction has a positive influence on their compliance (Hypothesis 5).

Hypotheses

Based on the literature review, it is hypothesized:

I a) The cognitive component of physician empathy leads to better exchange of
cognitive information.
b) The aﬂective aspect of physician empathy leads to partnership.

2 a) Better exchange of cognitive information leads to increased perception of physician
expertise.

b) Partnership leads to increased interpersonal trust.

3 a) Better exchange of cognitive information has a greater eﬁ'ect on patient 's
compliance than does partnership.

26

 

"A

b) Partnership has a greater eﬂect on patient ’3 satisfaction than does cognitive
information exchange.

4 a) Perception of physician expertise has a greater eﬁizct on patient ’s compliance than
on satisfaction.
b) Interpersonal trust has a greater eﬂect on patient ’5 satisfaction than on
compliance.

5 Patient ’s satisfaction will have a positive inﬂuence on their compliance.

Figure l
Hypothesized Empathy Model (causal phases in the relationship between empathy and

patient outcomes)

Cognitive Better Exchange of
Empathy . :4 itive lnformation _

 
  

 
    
 

Perception of

Physician expertise ....... 1w; ..»- ,3”...-
.......................................... “A
Interpersonal """""""""
Trust
Affective P hip

 

Empathy

‘ Line (-—-) indicates weak causal effect while line (— ) indicates strong causal effect between the two variables.

The causal model relating physician empathy to patient satisfaction and
compliance is shown in Figure 1. It provides a powerful explanatory tool in understanding
the processes in which each component of physician empathy is related to patient
outcomes. The model shows that physician empathy is conceptualized as consisting of
dual structure (cognitive and affective). And although all of the variables (cognitive

information, partnership, and perception of physician expertise and trustworthiness) must

27

be present and combined to produce stronger patient satisfaction and compliance,
cognitive information exchange and perception of physician expertise will have stronger
effect on compliance than on satisfaction, while partnership and interpersonal trust will
have better effect on satisfaction than compliance. Furthermore, the relationship between
physician empathy, and patient satisfaction and compliance is expected to be mediated by
other factors (better exchange of cognitive information, partnership, perception of
physician expertise, and interpersonal trust). Thus, highly empathic physicians (both
highly cognitive and affective) will result in more satisﬁed patients as well as more

compliant patients.

28

Chapter III

Methodology

Research Design and Data Collection
Translation

1 developed the questionnaire based solely on the US studies written in English.
For this reason, I had to control for the cultural factor (Calantone etc., 1996) in
measurement, that is, 1 accurately translated the original English version of the
questionnaire into Korean.

The two-parallel double-translation method was used to translate the
questionnaire. Four Korean translators were chosen because of their ﬂuency in both
English and Korean. Two were Ph.D. students in sociology at a US university and the
other two majored in English at a Korean University and now work as professional
translators. The process consisted of three steps. First, two of the four people were asked
to translate the original English version questionnaire into Korean and then the other two
were asked to translate the Korean version into English. Finally after comparing the
back-translated English version with the original English version of the questionnaire
showed considerable consistency, I met with all four of them to correct some minor
inconsistencies. Based on intensive discussions with them, we developed the ﬁnal Korean

version of the questionnaire.

29

Sampling

Participants in this study were recruited from a large, multifaceted university
hospital in Pusan, Korea. I chose this site because the patients represent a wide variety of
illnesses, physician-patient relationships, and types and lengths of medical examination.

The six interviewers started to recruit the subjects in the pharmacy lobby, where
people waited to pick up their prescriptions after seeing the doctors. The interviewers
approached and asked patients whether they had seen the doctor within two weeks of their
present visit. If the patient said yes, he or she was asked to participate in the survey. The
interviewers promised that the patient would remain anonymous and that physicians
would not see individual responses. Aﬁer two weeks of the survey, I collected 550

questionnaires.

Obtaining approval for survey

To learn more about the proper way of doing the survey in Korea, I talked with a
couple of experienced professors in sociology and in social work at a national university
in Korea. Their general suggestions were: 1) make the questionnaire look very
professional; 2) recruit the best female interviewers; 3) prepare giﬁs for patient-subjects
to increase response rates; 4) obtain approval from the hospital using personal
connections.

Based on their advice, the ﬁrst major task was to obtain the approval from the
hospital for the survey. As the professor in social work suggested, a personal connection
appears to be the key factor in getting approval from the hospital in Korea. First, I

targeted the head of the hospital in which I wanted to conduct the survey. 1 was told by

30

many that if I could get his cooperation, I would have no difﬁculties in conducting the
survey. Fortunately, with the help of my family’s social connections, a meeting was
arranged to see the head of the hospital. I met him at his ofﬁce to explain the nature of
my survey. He agreed to cooperate and convened all the staff members to explain my
survey. The four most important staff members came in and I explained the purpose of
this survey and asked for their help. All of them agreed to help, although one of them
raised an interesting question about the real intention of my survey. She asked me if this
was a preliminary stage for a foreign hospital chain to open its hospital in Korea. I
explained that this survey has nothing to do with anything like that. I explained that this
was a purely scholarly investigation to understand the relationship between doctors and
patients in Korea and the data collected will be used only for my dissertation purpose. In
the end, she seemed convinced. Coincidentally, the head of the hospital said that his
hospital was about to do a general annual patient-survey and that they would replace their
survey with my survey. As a result of the meeting, I obtained their approval as well as
their endorsement, whichlenabled me to indicate that this survey was being ofﬁcially
conducted by the hospital on the cover page of the questionnaire. I think that this

legitimization of the survey increased the patients’ response rate.

Printing Questionnaire

The next task was to contact a printer that had many experience in printing survey
questionnaires. They showed me many choices of printing formats for the questionnaire.
After a couple of meetings with them, I chose the one that looked the most professional. I

asked them to print 600 copies.

31

Recruiting and Training

Meanwhile, I started to recruit interviewers. From the previous consultation with
a professor in social work, I learned that many senior students in social work have
valuable experiences in surveying patients. I decided to recruit these students as
interviewers because of their previous patient-involved survey experience. I contacted the
social work department of a well-known national university and got eight students who
showed interest in the survey. I decided to have an orientation to screen out the
unqualiﬁed interviewers. The orientation was held in a classroom at the university. The
orientation consisted of several parts to help them understand the importance of this
survey and its procedures. It was composed of the following: 1) rules for interviewers
such as how to talk and dress; 2) the importance of this survey; 3) how to proceed with
the survey; and 4) an exercise where the candidates interviewed each other. In the end, I
decided to hire six of them.

To ensure the reliability of the data collection process, I hold two debrieﬁng
sessions with the interviewers, one before and one after the survey. The interviewers
were repeatedly asked to stop and discard the questionnaire whenever the patients seemed

uneasy or insincere about answering the questions.

Modiﬁcations in the survey procedures

The original survey procedures had to be modiﬁed due to the unexpected
circumstances in the university hospital. Initially, I had planned to conduct the survey on
people who had previously seen the doctor (within two weeks) and were now in the

waiting room for their second visit. However, the university started an appointment

32

system a few years ago. The head of the hospital pointed out some difﬁculties I might
face with the original survey procedures since the appointment-based visits reduced the
previously experienced waiting time substantially for the patient and that there might not
be enough time for the patients to complete the questionnaire. Furthermore, he added that
not enough patients might be available for the survey due to the appointment system in
the doctors’ waiting rooms.

However, the head of the hospital suggested that it might be a good strategy to get
the subjects in the pharmacy lobby since the average waiting time to pick up a
prescription was around 40 minutes. There was a huge pool of subjects waiting there.
A careful consideration based on the previous consultation with several people who have
done the survey in Korea convinced me of changing some of the survey procedures as
follows:
1. I should utilize a small giﬁ to enhance the response rate and reliability. As a result, I
ordered six hundred gift items ﬁom a wholesale store at a cost of one dollar per gift.
2. I also utilized two interviewing methods: one is that patients ﬁlled out the
questionnaire by themselves and the other is that the interviewers read the questionnaire
for them. The professors who I consulted with pointed out that Koreans are not familiar
with the former. They said that this might cause some uneasiness as well as difﬁculties of
understating the questionnaire among the subjects. Not many outpatients at a big hospital
in Korea are well educated enough to read and understand the questionnaire. Therefore, I
decided to use both methods to conduct the survey. The decision to choose between the

two methods depended on the subjects’ preferences.

33

Measures

In order to measure each one of the important variables (physician empathy,
cognitive information exchange, partnership, perception of physician expertise and
interpersonal trust, patient satisfaction, and patient compliance) in the model, 7 measures
will be utilized as described below. They were designed to be ﬁlled out less than 20 '
minutes by the participating patients. While a number of previously used scales were

used, all were modiﬁed because of the different deﬁnitions of the concepts for this study.

Physician empathy

Since there has been no instruments to exactly measure the two constructs
(cognitive and affective) of physician empathy, I decided to construct a new empathy
scale. The new empathy scale was designed to speciﬁcally measure each of the two
components (cognitive and affective) of physician empathy. In the process of constructing
the new empathy scale, ﬁrst, several items were selected and modiﬁed from the existing
well-known empathy instrument (Barrett-Lennard, 1981 ,The Barrett-Lennard
Relationship Inventory) which ﬁt the theoretical deﬁnition of each component employed
for this study. Second, based on the literature review and discussion with a number of
colleagues, additional items were created to assess cognitive and affective empathy. This
new scale asks patients to rate physician empathy. In other words, it measures the
patient’s perception of his or her physician’s empathy.

Using this scale to measure physician empathy has a signiﬁcant advantage over
other empathy measures. A majority of empathy measures (e.g. Hogan’s [1969] Empathy

Scale [EM], Mehrabian and Epstein’s [1972] Emotional Empathy [QMEE] ) employ

34

subject (physician) ratings of their own empathic concerns. This hardly reﬂects actual
patient feelings and these feelings have important implications for patient outcomes.
Furthermore, Squier (1990) found clear evidence in his review of psychotherapy
literature that empathy must be perceived and felt by patients in order to be effective. In
their comparative study of empathy rated by three different groups-~patients, therapists,
and clinical supervisors, Free, et al. (1985) found that only the patient perceived empathy
was signiﬁcantly related with patient outcomes. In a similar study, Kurtz (1972) also
found that therapist self-reported empathy is not a useful measure since it is unrelated to
outcome; only client-perceived empathy is highly associated with therapy outcomes.
Therefore, it is more reasonable to use a patient-perceived empathy scale to measure

physician empathy.

Satisfaction Measure

The scale for patient satisfaction is adopted ﬁ-om the existing measures (the
Satisfaction Questionnaire suggested by Comstock et al., 1982 and Doctor Satisfaction
Scale by Kaplowitz [unpublished]). The four-item questionnaire is scored on a S-interval
scale with a score of 1 indicating lowest satisfaction and a score of 5 indicating highest
satisfaction. The four questions are: 1) Overall, I am satisﬁed with this doctor; 2) I have
conﬁdence in this doctor? 3) Overall, I am discontented with this doctor; 4) compared to

the other doctors you have seen, this doctor is very good.

35

Compliance measure

A modiﬁed form of the General Adherence Scale suggested by DiMatteo et a1
(1993) is used to measure patient tendencies overall to adhere to medical
recommendations, allowing comparison of respondents across different treatment
regimens. The authors found Internal consistency reliabilities of .89 and higher for the
General Adherence Scale. Two items compose this modiﬁed scale; 1) I followed my

doctor’s suggestions exactly; 2) I exactly complied with this doctor’s drug regimens.

Patienm ’ Questionnaire
The questionnaire given to patients to collect information about the patient

includes items about the patient’s age, sex, education, and income.

Cognitive Information Exchange and Partnership

In order to construct the measure of the cognitive information exchange variable
and partnership, ﬁrst, a number of instruments were reviewed (Roter, et. al., 1987;
Kaplowitz [unpublished]). Then, items were written, based upon the deﬁnitions of the

two variables employed here. The actual items are included in the appendix at the end of

this paper.

Patient ’s Trust: perception of physician expertise and interpersonal trust
Anderson and Dedrick (1990) developed a Trust in Physician Scale (TPS) to
measure patients’ interpersonal trust in their primary-care physicians. First, a number of

items from the TPS, which were considered to best tap perception of physician expertise

36

and interpersonal trust, were selected. Additionally, a few more items were written to

measure interpersonal trust.

Analysis

Use of Structural Equation Modelings (SEM)

The works of Karl Jbreskog and his associates (e.g., Jbreskog, 1977; Jbreskog and
Van Thillo, 1973) helped the social and behavioral researchers more easily access general
SEM techniques since the 19708. Considerable use of SEM has been noticed in social
science research settings such as in the areas of psychology, sociology, and education
(Anderson and Gerbing, 1988; Mueller, 1996). One important reason for this is that SEM,
as a powerful multivariate data analysis tool, can equip researchers with a comprehensive
means to assess and modify theoretical models (Anderson and Gerbing, 1988).
Furthermore, with the help of SEM computer programs readily available such as LISREL
(Ibreskog and Sorbom, 1993) and EQS (Bentler, 1993; Bentler and Wu, 1993), today
SEM is a well-established and respected data analysis method for understanding aspects
of social and behavioral phenomena.

Unlike simple and multiple linear regression techniques (often used in the social
science literature as data analytical tools for prediction and variance explanation), SEM
allows us to model the effects of unobserved or latent factors that are imperfectly
measured by observed indicator variables. In other words, measurement error in observed
variables can be taken into account when assessing the effect of a set of independent

variables on one or more dependent variables (Mueller, 1996).

37

In multiple regression, or AN OVA (analysis of variance), for instance, we learn
that the regression coefﬁcients and the error variance estimates are computed ﬁom the
minimization of the sum of squared differences between the predicted and observed
dependent variable for each case. Residual analyses display discrepancies between
predicted and observed values for every case of the sample (Bollen, 1989). However,
SEM uses covariances rather than individual cases. Instead of minimizing functions of
observed and predicted individual values, SEM minimizes the difference between the
sample covariances and the covariances predicted by the model. The observed
covariances minus the predicted covariances yield the residuals. The fundamental axiom
for these structural equation procedures is that the covariance matrix of the observed
variables is a function of a set of parameters. If the model were correct and if we know
the parameters, the population covariance matrix would be exactly reproduced.

2 = 2(6) ( Bollen, 1989)“.

Figure 2 Path Diagram of a Single
Latent Variable

 

 

 

25
x
N
x
w
2‘

 

 

 

 

 

 

 

 

 

 

 

 

‘ 2 (sigma) is the population covariance matrix of observed variables, 9 (theta) is a vector that contains the
model parameters, and 2 (9) is the covariance matrix written as a function of 0.

38

The covariances implied by the model are computed with covariance algebra.
Figure 2 shows an example of a path diagram of a single latent variable with four
indicators. Thus, COV (x.,x4) is

COV(x1,x4) = cov (M. a. + 51, mg, + 5.05

= 7&1 11414511

This indicates that the COV(x1,x4) is a product of the effects of £1 on x1 and x4
(e.g., it” and 1.41) and of the variance of the latent variable £1.

The fundamental axiom of SEM offers a uniﬁed way of encompassing many of
the most widely used statistical techniques in the social sciences (e.g., regression analysis,
path analysis, conﬁrmatory analysis, AN OVA, and analysis of covariance).

Like most other model-development processes, SEM analysis starts with the
identiﬁcation of relevant variables and hypothesizes its relations to exogenous and
endogenous latent variables. An exogenous (independent) latent variable is one whose
causes lie outside of the model. On the other hand, an endogenous variable is determined
by variables within the model (Bollen, 1989).

For instance, in Figure 3, cognitive empathy and affective empathy are exogenous

variables and the rest of the variables are endogenous.

 

5 i = latent exogenous variables; 1 = coefficient relating to x to 5,; 5 = measurement errors for x (Bentler,
1995).

39

Figure 3

Hypothesized Model

Cognitive Better Exchange of

Empathy o; itivelnformation

Interpersonal """""""" ..
Trust

Affective Partnership
Empathy

 

 

.‘

’ Line (-—-) indicates weak causal effect while line (— ) indicates strong causal effect between the two variables.

Structural Equation Modeling Software Program (EQS)

EQS (Bentler, 1989) version 5.7b for Windows was used for statistical analysis.
EQS is gaining popularity for its straightforward programming language. It is also easy to
learn and apply. It has a unique capability of dealing with corrected tests and statistics
(e.g., a robust version of the Bentler Compared Fit Index [Bentler, 1990]). In EQS,
observed variables are termed as Vs; latent variables, Fs (or factors). Residuals in

observed (or measured) variables are called errors, or Es.

Assessment of Model Fit
The major purpose of assessing a structural equation model is determining
whether the hypothesized model ﬁts the sample data. In general, the overall model ﬁt has

been based on the x2 statistic. However, since the x2 statistic has been known to be

40

sensitive to variations of sample size and since it reﬂects the discrepancy between the
observed and reproduced sample covariances, many different alternative ﬁt indexes have
been developed to help researchers reach a better judgment about the model ﬁt. EQS
provides several goodness-of-ﬁt indexes. The )8 statistic is ﬁrst given for the
hypothesized model. EQS also gives a robust test statistic called the scaled x2 with a
similar interpretation of the x2 statistic (Satorra & Bentler, 1988) as an option. The ML
(Maximum likelihood) estimator which is used for this study has multivariate normal
distributional assumption. The robust test statistic is designed to modify the usual test
statistic so that it can more closely approximate x2 distribution when this normal
distributional assumption is violated (Bentler, 1993). Hu et al. (1992) argued that the
scaled x2 was the most reliable in their study. EQS yields a list of ﬁt indexes (e.g., NFI,
NNFI [Bentlter & Bonett, 1980], CFI [Bentler, 1990]).

For this study, CPI and Root Mean Square Error of Approximation (RMSEA;
Steinger 1990; 1989) were used. I selected CFI which uses the sample noncentrality
statistic because Bentler (1992) recommends it as the best index. RMSEA was chosen
because it is increasingly gaining popularity in recent years (Wang etc., 1999). In
addition, RMSEA and CFI were reported to be the least sensitive to sample size (Wang
etc. 1999).

In general, for CFI, a value greater than .90 indicates an acceptable ﬁt to the data
(Bagozzi and Yi, 1988; Bentler and Bonett[1980]). For RMSEA, values less than .05

indicates a very good ﬁt and those below .10 indicates a reasonable ﬁt (Steinger 1989).

41

Preliminary Analyses

An essential prerequisite to SEM is an assessment of the statistical assumptions
on which an estimation method (ML was used here in this study) is based. ML is based
upon the assumption that the data are multivariately normal. EQS provides a number of
univariate and multivariate statistics in this regard. For the univariate distribution,
skewness and kurtosis for each variable in a model is automatically printed in EQS. For
multivariate distribution, EQS provides multivariate measures of skewness and kurtosis

developed by Mardia (1970).

Two-Step Modeling Approach

Following the suggestions of Anderson and Gerbing (1988), I used a
comprehensive two-step modeling approach to make meaningﬁrl inferences about
theoretical constructs and their interrelations. The two-step modeling approach consists of
tests of conﬁrmatory measurement (CF A) and latent variable model. Prior to the full
structural model test, the test of measurement model provides a comprehensive,
conﬁrmatory assessment of both convergent validity and discriminant validity. Once
convergent and discriminant validity are achieved in the measurement model, the test of
the latent variable model (full structural equation model) enables a conﬁrmatory

assessment of nomological validity (Anderson and Gerbing, 1988).

1. Conﬁrmatory Measurement Model Test (CFA)
EQS soﬁware version 5.7b (Bentler, 1998) was used with raw data as input to test

the measurement model using CFA prior to assessing the full structural relationships.

42

Maximum likelihood (ML) was used as the estimation method because it has been the
predominant estimation method (under the assumption of a multivariate normal
distribution). ML has been shown to be unbiased, consistent, and efﬁcient (Anderson and
Gerbing, 1988). F urtherrnore, ML parameter estimates were shown to have some
robustness for the nonnormal data (Browne, 1984; Harlow, 1985; Tanaka, 1984).

One of the major advantages of using CF A is that it allows researchers to develop a
prespeciﬁed, theory-driven model for evaluating to what degree a certain data set
‘conﬁrms’ what is theoretically believed to be its underlying structure (Mueller, 1996).

Thus, unlike other variable reduction methods such EFA (exploratory factory
analysis) or PCA (principal component analysis), the CFA approach to multivariate data
analysis enables the researchers to theorize an underlying structure and examine whether
the data ﬁts this prespeciﬁed model. In doing so, “CFA provides a framework for
addressing some of the problems associated with traditional ways of assessing a
measure’s validity and reliability“.

According to Anderson and Gerbing (1982), we must ensure that all constructs are
unidimensional and reliable prior to assessing the full structural relationships among
them. The purpose of the measurement model test using CFA is to show that the
measurement model has a satisfactory level of validity and reliability (F omell and Larker,

1981).

43

Measure Puriﬁcation
Since no prior testing of the operational constructs used in the model has been done, a
series of steps was taken to ensure the unidirnensionality of the multiple-item constructs
and improve reliability and validity as follows:
I Removing unreliable items that had cross-loadings on multiple constructs or had low
coefﬁcient loadings on its factor.
I Eliminating outliers (ﬁve cases [multivariate outliers] signiﬁcantly contributing to

multivariate kurtosis are identiﬁed in EQS program [EQS is unique in this regard]).

After removing questionnaire items that loaded on multiple constructs or had low
factor loadings, CPI, and RMSEA were used to assess the degree of ﬁts of the CFA to the

data.

Reliability and Validity

Reliability (internal consistency) refers to how consistently the instrument measures
whatever it was designed to measure (Mueller, 1996). Reliabilities were tested using
Cronbach’s alpha.

The validity of a measure refers to the measure’s overall property of indeed measuring
what it was designed to measure. For this study, two validity measures, convergent and

discriminant, were used:

 

6 Mueller, Ralph O. 1996. Basic principles of structural equation modeling: an introduction to LISREL and
EQS: p. 62.

Convergent validity: convergent validity was assessed from the measurement
model by determining whether each indicator’s loading on its prespeciﬁed
construct factor is signiﬁcant (Anderson, and Gerbing, 1988).

Discriminant validity: Discriminant validity tests whether measures of distinct
constructs perfectly correlate or not. This was done in two ways. First,
discriminant validity test was conducted by a chi-square difference test on the
values obtained for the constrained and unconstrained models using two estimated
constructs (Joreskog, 1971). “A signiﬁcantly lower )8 value for the model in
which the trait correlations are not constrained to unity would indicate that the
traits are not perfectly correlated and that discriminant validity is achieved.”7
Second, a complementary assessment of discriminant validity was done by
examining whether the conﬁdence interval around the correlation estimate

between the two factors includes 1.0 (Anderson and Gerbing, 1988).

2. Full Structural Model Test

Once convergent and discriminant validity were achieved in the measurement

model, the full structural model was tested. First, the model’s overall goodness of ﬁt to

the sample data was assessed by using the x2 test statistic, CPI, and RMSEA measures.

Then, the standardized parameter estimates (structural path) and t-values were examined

to test the individual hypotheses.

 

7 Baggozi, R. R. and Phillips, L.W., 1982. “Representing and testing organizational theories: a holistic
construal.” Administrative Science Quarterly, vol.27: p.476.

45

Chapter IV

RESULTS

Data
Sampling was conducted from outpatients at a university hospital in Korea. Six

trained interviewers collected 550 survey questionnaires with a 5 point-Lickert scale.
Interviewers reported that the response rates were about 65% of those who were
approached. After removing unusable questionnaires, we obtained 522 suitable
questionnaires. The data were entered with SPSS windows programs. The data with
missing values were replaced with variable mean.

Table 1, 2, 3, and 4 show the descriptive results of the sociodemographic
characteristics (age, sex, education, and income) of the patients. Table 5 shows
sociodemographic statistics about Korea. From Table l we see that there is a high
concentration of patients in their 405 and 505 (about 66%), compared to the average age
(32.2 years) in Korea. There is a relative lack of younger and older patients. About 68%
of the patients were female; 32% were male (see Table 2). This high female concentration
was due to the fact that most of the patients who were present (in the pharmacy’s lobby)
were female. The majority (69%) of the patients had at least a high school education (see
Table 3). Interestingly, there is a very high concentration of patients with a high school
education (about 45%), which is about the average education for Koreans (see table 5).
Regarding patient income (see Table 4), about 50% of the patients reported that they
earned above the average household income per month (about $1,500). And about 50%

reported below average earnings. However, over 32% of the patients did not want to

46

reveal their income. The reasons for their refusal were unclear and we did not ask them

any further questions on the issue.

 

 

 

 

 

 

 

 

 

Table 1 Age
A90
Valid
Freguency Percent
Valid 10 - 19 6 1.2
20 - 29 61 11.7
30 - 39 72 13.8
40 - 49 120 23.1
50 - 59 133 25.6
60 - 69 92 17.7
70 - 79 33 6.3
80 - 89 3 .6
Total 520 100.0
Missing System 2
Missing
Total 2
Total 522
Table 2 Sex of Patient
Sex of Patient
Valid
Frequency PereenL
Valid female 356 68.2
male 166 31.8
Total 522 100.0
Total 522

 

 

 

 

 

47

Table 3 Patient Education

 

 

 

 

 

 

 

 

 

 

Patient Education
Valid
Frequency Percent
Valid elementary 69 13.9
middle
school 84 16.9
high school 223 44.8
2yr college 31 6.2
4 yr college 83 16.7
graduate
school 8 1.6
Total 498 100.0
Missing System 24
Missing
Total 24
Total 522
Table 4 Patient Income
Patient Income
Valid
Frequency Percent
Valid ' below 55% 38 10.7
$501-$1.000 59 16.7
$1.001-51.500 86 24.3
$1,501-52,000 89 25.1
52001-52500 38 10.7
above$2.5001 44 12,4
Total 354 100.0
Missing 9 1
System
Missing 167
Total 168
Total 522

 

 

 

 

48

 

Table 5

Sociodemographic Indicators in Korea: age, sex, education, and household income.

 

 

 

Age (years) Sex (rate) Education (years) Household income
Male 31.1 101.4 11.8
Female 33.4 100.0 9.37
Average 32.2 10.25 About $1,500

 

"' Data was drawn from the 1998 Census study by the Korean National Statistical Office.

Assessment of Model Fit
Normality Test

1. Univariate normality checked for skewness, kurtosis

Since the method of estimation used (ML) assumes multivariate normality, test
results of univariate normality and multivariate normality for the observed indicators
were examined. The univariate statistics in EQS represent the mean, standard deviation,
skewness, and kurtosis. As Table 6 shows, skewness for all the indicators was close to
1.0, which is reasonable. All of the indicators had a kurtosis of less than 2.0 except for
four that were a little over 2.0. In general, skewness and kurtosis do not drastically

deviate from univariate normality.

49

 

dew—a a

@5158... 2:. 5:48?

 

 

 

 

 

 

 

 

 

 

 

 

 

Valoza Om. 0mm nmu >m_ >m~ >Mu >mA >mu >ma >ma 9m. 9mm 9mm 9mg 9mm
wroisomm 9.24 94mm .9m _ u .930 .9an .9meu -9ch .93: -995 .9.“ 3 .930 .chcc -9qu .663 .9me
~950me 9~No 9%: 98m -993 9a.: 73¢ 98.» -986 955 98m 9 _ we .9 _ ma -9 _ MN 9 Eu -935
<33? 1mm. 11m“ Ecmw 3.— SM Eu :5 35.2 _ ENHZN 35.2w 25.2.. ENHZM
mwosaom .9auq -_ .93 .93a .90 _ w .9m8 but .9uoc -_ bum .9uac .90 _ c .9muq .9a _ w
m
558% rm .o 953 79 5 98m .3» 3 958 93w Nbun Noun 9cm.» Fume 9&3
$5929 v0— v8 mm. an 1mm vma
£82.53 .92: .90”: .93m .925 .909» .9u3
558$ _._ B 9mg 92: 9qu 93° 9 3m

 

 

 

50

2. Multivariate normality

Next, for the multivariate normal distribution test, variants of Mardia’s (1970)
coefﬁcients of multivariate kurtosis (the multivariate statistics reported by EQS) were
examined. Mardia’s coefﬁcient (62, P) was 282.21 and the normalized estimate was
66.62 since the observed sample was distributed in large samples from a multivariate
normal population as a normal variate so that large positive values indicated signiﬁcance
(pp. 145 Hoyle; Bentler [1995, manual]). Romeu and Ozturk (1993) published empirical
critical values (values above which the population would fail to meet the assumption of
multivariate normality) for Mardia’s coefficients. According to them, for the sample size
with over 200, 2.58 was considered a critical value. Therefore, our obtained normalized
estimate for Mardia’s coefﬁcient (66.62) would indicate substantial non-normality of the
data. In other words, the normal distribution assumption seems to be violated according
to Romeu and Ozturk’s( 1993) criteria.8 Thus, the scaled 12 (Satorra and Bentler, 1988),

a robust test statistic in EQS, was used to treat this problem.

Results of Conﬁrmatory Factor Analysis Measurement Model Test (CFA)

EQS software version 5.7b (Bentler, 1998) was used with raw data as input to test
the measurement model using CFA prior to assessing the full structural relationships.
Measure Puriﬁcation

Since no prior testing of the operational constructs used in the model has been done, a

series of steps was taken to ensure unidimensionality of the multiple-item constructs and

51

improve reliability and validity. Table 7 shows the number of initial items (or indicators)

for each construct. The original questionnaire had 57-items and 8 construct measures.

Table 7

Initial Number of Indicators for Constructs

 

 

 

Constructs Number of Indicators
Cognitive Empathy 7

Affective Empathy 9

Cognitive Information Exchange 13

Perception of Physician Expertise 6

Interpersonal Trust 6

Partnership 10

Patient Compliance 2

Patient Satisfaction 4

 

 

Next, based upon the assessments of the results of largest standardized residuals and a
multivariate LaGrange Multiplier test (see Bentler’s EQS [1995] manual for more
details), I puriﬁed unreliable items that had cross-loadings on multiple constructs or had
low coefficient loadings (s. 50) on its factor.

The residual covariance matrix is the difference between sample covariance and
covariance implied by the model. Large values in the residual covariance matrix indicate
probable inadequacy in the structural model. However, due to its difﬁculty in
interpretation, a standardization is performed on the residual covariance matrix so that
problems with a model would be more easily identiﬁed than in the residual covariance

matrix (Bentler, 1995). The elements from the Standardized Residual Matrix are ordered

 

' According to Romeu and Ozturk (1993), Marida’s coefficient values over 2.58 might indicate the
violation of the normal distribution assumption with the sample size of larger than 200.

52

 

from large to small in absolute value, and EQS provides the largest twenty of these along
with a designation of which pairs of variables are involved. Large standardized residuals
tend to be more inﬂuential in lack of model ﬁt (Bentler, 1995).

Following Calantone’s suggestion, for puriﬁcation, we considered standardized
residuals greater than 2.0 to indicate poor model ﬁt. Out of over 1,500 residuals, only
three were greater than 2.0 in the initial run. Those variables associated with these larger
residuals were eliminated for better model ﬁt.

Second, the LaGrange Multiplier (LM) test was used in the process of removing
unreliable items. The LM test indicates the values of including more free parameters to
improve the model ﬁt. It does so by relying on the differences in chi-square test;
comparing the nested hypothesized model M, to one with more ﬁee parameters, say M.
It indicates the expected decrease in the overall chi-square value when a presently ﬁxed
parameter (or set of parameters) is freed in a subsequent analysis. When there is a large
decrease in chi-square by freeing the relationship between two variables (latent or
observed), this shows a signiﬁcant relationship between those variables. However,
if there is an empirically strong relationship between a factor and an indicator (for
example, F3 [a latent factor] and V13 [an indicator]) but, theoretically there is no
relationship between them, we consider removing the indicator, V13. Thus, the LM is
often used for removing unreliable indicators.

By examining both largest residuals and LM test results, 1 reduced the indicators
from 57 to 33 items with 8 constructs as shown in Table 8, which reveals the number of

the ﬁnal indicators for each construct.

53

Table 8

Final Number of Indicators for Constructs

 

 

 

Constructs Items eliminated Final Number of
(# refers to initial instrument) Indicators

Cognitive Empathy l, 2, 4, 5 3

Affective Empathy 5, 8 7

Cognitive Information 1, 2, 3, 6, 7, 8, 12, 13 5
Exchange

Perception of Physician 4, 5, 6 3
Expertise

Interpersonal Trust 1, 2 4

Partnership 2, 3, 4, 6, 7 5

Patient Compliance 2

Patient Satisfaction 4

 

 

In addition, EQS identiﬁes ﬁve cases (multivariate outliers) that signiﬁcantly
contributed to multivariate kurtosis. The program automatically prints out the ﬁve cases
with the largest contribution most to Mardia's multivariate kurtosis coefficient. An outlier
is identiﬁed based on the estimate for one case presented for one case relative to those for
the other four cases. However, no absolute value on which to make this judgement has
been established. Thus, none of the ﬁve cases could be actually an outlier. (Hoyle 1995).
In order to determine whether a case with largest contribution to kurtosis created a
problem for the analysis, the job is resubmitted with the case with the largest contribution
removed. I ran the same procedure until case contributions are similar in size which
means no indication of another outlier (p. 122, EQS manual). In the process, seven items
were removed.

After purifying questionnaire items that loaded on multiple constructs or had low

54

factor loadings, the xz statistic CFI9 and RMSEAlo were used to assess the degree of ﬁts
of the CFA to the data. First, the results easily met the preliminary ﬁt criteria suggested
by Bagozzi and Yi (1988), showing the absence of negative error variances as well as
correlations greater than one. The model converged very rapidly (6 iterations) and I had
no problem estimating the model parameters. As shown in Table 9, the chi-square test
was signiﬁcant, therefore unsatisfactory. However, since the x2 statistic has been known
to be sensitive to variations of sample size, other ﬁt indexes should be examined other
than the chi-square statistics to ensure the best assessment of model ﬁt (Fan, etc., 1999;
Bentler 1990; Bagozzi and Yi, 1988). As Table 9 shows, both CPI and CPI-Robust were
well above .90 which indicates an acceptable model ﬁt to the data. RMSEA was .043,
which indicates a very good ﬁt since values less than .05 shows a very good ﬁt according
to Steinger (1989). As a result, the model appears to ﬁt the data well.

Table 9

Goodness of Fit Results for CFA

Chi-square . 910.986
d.f. 467
p < 0.001
Scaled Chi-square 733.293
p < 0.000
CH .948
CF l-Robust .957
RMSEA 0.043
90% conﬁdence interval (0.039, 0.047)

 

 

9 Incremental ﬁt index CF I (comparative ﬁt index) is algebraically deﬁned as
CFI = 1 - max{( TT- dfr). Ol/maXK Tr - dfr). (Ta - dfa). 0]
Where T3 .. T statistic for the baseline model; TT 3 T statistic for the target model;
a = degrees of freedom for the baseline model; dfT = degrees of freedom for the target model
(Bentler, 1989, 1990).
‘° Absolute ﬁt index RMSEA (root-mean-square error of approximation) is algebraically deﬁned as
RMSEA = \JFO/df, W, F, = max[ I, - dnyuv — 1), 0] (Hub, Bentler, and Steiger, 1993).

55

Reliability and Validity

According to Anderson and Gerbing (1982), the researcher must ensure that all
constructs are unidimensional and reliable prior to assessing the full structural
relationships among them. In other words, the measurement model should show a
satisfactory level of reliability and validity.

Reliability was estimated using Cronbach’s alpha. As shown in Table 10, the
coefficients for all the scales were well above .60, indicating the usefulness of the scale
(Nannally, 1978). Some might argue that having only two items for patient compliance
causes a low reliability problem. The formula for the standardized a,

a = kr_/1 + (k — 1)r_ : k = number of items, r_ = average correlation
Therefore, while holding r— constant, more items lead to higher (1. However, it is not so
simple. It is also true that a) two items can have high a if their r is high and that b)
adding additional items which result in a lower r- can reduce reliability.

To determine convergent validity (whether each indicator’s loading on its
prespeciﬁed construct factor is signiﬁcant [Anderson, and Gerbing, 1988]), each
indicator’s coefﬁcient on its prespeciﬁed construct factors was examined and they all
were signiﬁcantly different from zero. All standardized factor loadings were above .60,
as shown in Table 10. Thus, convergent validity was satisfactory.

To achieve discriminant validity (whether measures of distinct constructs
perfectly correlate or not), ﬁrst, a chi-square difference test was conducted on the values
obtained for the constrained models (with correlations between constructs or latent

variables ﬁxed at 1.0) and the unconstrained models using two constructs whose

56

correlation is estimated (Joreskog, 1971). Each chi-square test examines a null
hypothesis that the two constructs are perfectly correlated. If the null hypothesis is
rejected, it means that the constructs are not perfectly correlated. As Table 11 shows, all
of each pair of the chi-square tests yielded signiﬁcant lid (1)2 values which are signiﬁcant
at p < 0.05. This shows that all construct measures are distinctly developed, measuring
different concepts. Thus, even though some of constructs are highly correlated, the

structural model is capable of separating out their effects on other variables.

57

Table 10

Test Results of Measurement Model (CFA)

 

 

Constructs/ Standardized Parameter Reliabilities
Items ((1)
Cognitive Empathy (F l) .68
CE] .600
CE2 .662
CE3 .693
Affective Empathy (F2) .87
AB] .707
AE2 .700
AE3 .675
AE4 .719
ABS .694
AE6 .746
AE7 .748
Cognitive Information Exchange (F 2) .80
ClEl .608
C1E2 .770
C1133 .713
CIE4 .647
CIES .604
Perception of Physician Expertise (F4) .70
PPEl .614
PPE2 .690
PPE3 .762
Interpersonal Trust (F5) .79
IT] .663
1T2 .727
1T3 .765
1T4 .709
Partnership (F6) .79
PRTNl .646
PRTN2 .715
PRTN3 .644
PRTN4 .706
PRTN5 .626
Patient Compliance (F7) .78
PC] .910
PC2 .705
Patient Satisfaction (F8) .87
PSI .852
PS2 .805
PS3 .806
PS4 .752

 

‘All standardized coefficients were signiﬁcant at a =0.05
’A item number refers to Appendix II.

58

Table 11

Test of Discriminant Validity: Xd (1, 2 test between the constrained and

 

 

unconstrained models

Constructs CE AE CIE PPE IT PRNT PC

CE

AE 35.85

CIE 27.64 59.87

PPE 49.57 78.79 66.26

IT 61.91 128.77 98.18 49.90

PRNT 38.48 42.13 20.28 47.05 129.59

PC 193.72 206.79 192.90 147.27 172.42 167.08

PS 58.90 157.36 99.34 21.39 106.73 51.01 176.75

 

" All mm 2 were signiﬁcant at or level of .05. Each test had one degree of freedom.

Next according to Anderson and Gerbing’s (1988) recommendations, a
complementary assessment of discriminant validity was also done by examining whether
the conﬁdent interval around the correlation estimate between the two factors includes
1.0. As Table 12 shows, none of them included 1.0. That is, they were not perfectly
correlated indicating the distinctiveness of each measure. As results, the discriminant
validity was achieved. For example, the largest correlation was .911 (see Table12). If we
transform it (Fisher’s r) to T(r), T(.911) is 1.533. The upper conﬁdence limit for T(p) is
1.574 which is computed by adding 1.96 standard errors (standard error = .021) to 1.533.
Then, if we take the reverse transformation of 1.574, we ﬁnd the upper conﬁdence limit
of p (the population correlation) to be .918. Therefore, the conﬁdence internal does not

include 1.0.

59

 

 

Table 12
Correlation / Discriminant Validity Measurement: correlation among constructs
Constmct (l) (2) (3) (4) (5) (6) (7)
(1) Cognitive
Empathy 1.0
(2) Affective .891 1.0
Empathy (.026)
(3) Cognitive Inf. .820 .866 1.0
Exchange (.033) (.021)
(4) Perception of .710 .756 .742 1.0
Physicians Expertise (.043) (.031) (.034)
(5) Interpersonal .698 .750 .744 .774 1.0
Trust (.039) (.028) (.031) (.033)
(6) Partnership .770 .895 .911 .779 .679 1.0
(.036) (.019) (.021) (.033) (035)
(7) Patient Compliance .266 .300 .246 .402 .377 .362 1.0
(.055) (.048) (.051) (.051) (.049) (.049)
(8) Patient Satisfaction .740 .809 .813 .882 .786 .867 .381
(.034) (.022) (.024) (.023) (.026) (.021) (.047)

 

" Values in parenthesis are standard errors.

Results of Full Structural Model Test

Since convergent and discriminant validity were achieved in the measurement

model, the test of full structural model proceeded. The model’s overall goodness of ﬁt to

the sample data was assessed by using the 12 test statistic, CFI, CPI-Robust, and RMSEA

measures. As Table 13 shows, examination of overall ﬁt statistics showed acceptable ﬁt

of the model to the data. The x2 test statistic (chi-squared of 1066.992, 481 df, p< 0.001)

60

Table 13

Goodness of Fit Results for Full Structural Model

 

Chi-square 1066.992
d.f. 481
p < 0.001
Scaled Chi-square 857.0709
p < 0.000
CF] .931
CFl-Robust .940
RMSEA 0.049
90% conﬁdence interval (0.045, 0.053)

 

was not satisfactory, but as previously pointed out, the x2 test statistic was known to be
sensitive to the sample size. Therefore, the best assessment should be based on other ﬁt
indexes rather than the 12 test statistic alone. CPI and CFI-Robust were well above .90
and RMSEA was 0.049. Thus, overall, ﬁt indexes indicated the model had adequacy and
satisfactory ﬁt. Next, each structural path and t-values were examined to test the

hypotheses.

Hypotheses Test

The empathy model was developed on theoretical grounds and previous empirical
ﬁndings. It contains several hypotheses. They were tested by estimating the full structural
model as shown in Figure 4. Figure 4 reveals the key parameter values for testing
hypotheses in the full structural equation model.

Table 14 shows the summary of the results of hypothesis test. The

following discussion was based on these results.

61

Figure 4

Full Model SEM Results

 

Cognitive 95 l ‘ ‘ ‘
Empathy

.320

 

A

Better Exchange of
Cognitive Information _

 

Affective 95 1“"
Empathy

-.783”

  
 

8L 0‘.

.503‘ Compliance

Perception of .............. ”,1
Physrc1an expertise .......... -'-‘..._ f -. 150
............. ,..-".... "-5-... 1...,-
....".....- 'I. .:501::". '--..‘e.t' ..""vQ-: 62

interpersonal Satisfaction
Trust . "
..,,..-.639‘
.76 m . 519°“
Partnership

a =p<‘05, u =p<,00],and "* =p<.001.

Hypothesis 1a: The cognitive component of physician empathy leads to better
exchange of cognitive information.

Hypothesis 1 b: The aﬂective aspect of physician empathy leads to partnership.

Hypothesis la predicted that the cognitive component (CE) of physician empathy

leads to better exchange of cognitive information (CIE). The effect of CE on CIE was

positive and signiﬁcant (path coefficient CE—>CIE =.951, t=10.262, p<0.001). The strong

relationship between CE and CIE showed that CE is a very good predictor of CIE. Thus,

the result supported the hypothesis la. The effect of the physician’s affective empathy

(AB) had a signiﬁcant positive impact on partnership (PRTN) (path coefﬁcient

AE->PRTN= .951, t=12.192. p<0.001). Physicians who are more affectively empathic

are more partnership-oriented. Thus, hypothesis lb was also supported.

62

Table 14

Results of Hypotheses Test

 

Hypothesis

Standardized Parameter
Estimate

t-values

Conclusion

 

Hla: The cognitive
component of physician
empathy leads to better
exchange of cognitive
information.

.951 (CE—>CIE)

10.26?”

Hla supported

 

Hlb: The affective
aspect of physician
empathy leads to
partnership.

.951 (AB—9PM)

12.19?"

Hlb supported

 

H2a: Cognitive
exchange of information
leads to increased
perception of physician
expertise.

.804 (ClE—tPPE)

8525""

H2a supported

 

H2b: Partnership leads
to increased
interpersonal trust.

.767 (Prtn—HT)

9930“"

H2b supported

 

H3a: Cognitive
information exchange of
has a greater effect on
patient’s compliance than
does partnership.

-.783 (CIE—)PC)

.639 (Pm-)PC)

-2.712”

2.034“

H3a not supported

 

H3b: Partnership has a
greater effect on
patient’s satisfaction
than does cognitive
information exchange.

.519 (Pm—W8)

-.162 (CIE—)PS)

4.136“”'”'l

-l.186

H3b supported

 

H4a: Perception of
physician expertise has a
greater effect on
patient’s compliance
than on satisfaction.

.503 (PPE—WC)

.507 (PPE—)PS)

2.011“

5351""

I-I4a not supported

 

H4b: Interpersonal trust
has a greater effect on
patient’s satisfaction
than on compliance.

.172 (IT-)PS)

.246 (IT—>PC)

2.854"

2.124‘

H4b not supported

 

 

H5: Patient’s
satisfaction will have a
positive inﬂuence on
their compliance.

 

-.150 (PS-)PC)

-.565

 

 

H5 not supported

 

 

r =p<.05, " =p<.001,and m =p<.001.
Hypothesis 2a: Better exchange of cognitive information leads to increased

perception of physician expertise.

63

Hypothesis 2b: Partnership leads to increased interpersonal trust.

The results indicated that cognitive information exchange (CIE) strongly
inﬂuenced the perception of physician expertise (PPE) (path coefﬁcient CIE—>PPE =
.804, F8525, p<0.001). Partnership (Prtn) also had a strong effect on interpersonal trust
(IT) (Prtn—)IT=. 767, t=9.930, p< 0.001). CIE and Prtn are strong predictors of PPE and
IT respectively. Thus, both hypotheses were supported.

Hypothesis 30: Better exchange of cognitive information has a greater eﬂect on

patient ’s compliance than does partnership.

Hypothesis 3b: Partnership has a greater eﬂect on patient ’s satisfaction than
does cognitive information exchange.

The results revealed that cognitive information exchange (CIE) had a greater
effect on the patient’s compliance (PC) than does partnership (Prtn) (CIE->PC=-. 783, t=-
2.712, p<0.05; Prtn—)PC=.639, t=2.034, p<0.05). However, contrary to the predicted
positive effect of CIE on PC, CIE had a strong negative impact on PC. Thus, the
hypothesis 3a was not supported. On the other hand, Prtn showed a greater positive effect
on the patient’s satisfaction (PS) than does CIE (Prtn—)PS=. 519, t=4.135, p< 0.001;
CIE—>PS=-. 162, p> 0.05).

Conﬁdence intervals for the two coefficients (Prtn->PS and CIE—)PS ) were
computed to see if they are signiﬁcantly different from each other. The conﬁdence
intervals of these coefficients did not overlap each other ( .196 = lower point of
conﬁdence interval for Prtn—>PS, .148 = upper point of conﬁdence interval for CIE->PS).
The two coefﬁcients were signiﬁcantly different from each other. In conclusion, Prtn

strongly inﬂuenced PS. H3b was supported.

64

ire-fl

Hypothesis 4a: Perception of physician expertise has a greater effect on patient ’3
compliance than on satisfaction.

Hypothesis 4b: Interpersonal trust has a greater eﬂect on patient ’s satisfaction
than on compliance.

Contrary to the hypothesis (4a), the results showed that the patient’s perception of
physician expertise (PPE) had almost the same effects on both PC and PS (the ’
coefﬁcients from PPE—>PC = .503; PPE—>PS = .507). The results (4b) also revealed that
although the inﬂuence of IT on both PS and PC was relatively meager and not
substantially different, interpersonal trust (IT) had a little more inﬂuence on PC than on
PS (IT-)PS = .172; lT—>PC = .246). Thus, the hypothesis H4b was not supported.

Hypothesis 5: Patient’s satisfaction will have a positive inﬂuence on their
compliance.

Contrary to the hypothesis, the results indicated that the patient’s satisfaction (PS)
did not have any signiﬁcant inﬂuence on the patient’s compliance (PC)(PS-)PC = -. 150,
t =-. 565, p > 0.05). PS was not a good predictor of PC. Thus, the hypothesis was not

supported.

Total Effects

However, we should be cautious about using only the direct effect of a variable
for the above interpretations. It may be misleading because it is calculated while

controlling for all the variables that affect a given endogenous variable. “That is, the

65

variables that mediate the effect of a causal variable on an endogenous variable are also

controlled when the direct effect of the former on the latter is calculated.”ll

Figure 5

Total Effects on Patient Compliance

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

,, n394 .221
Cognitive Interpersonal
Empathy Trust
Affective .695
Empathy \
Patient ‘_7_30__ Partnership

Cognitive Inf. _ 415 Compliance
Exchange '
P t' f Patient

mi". '0" O .427 -.150 Satisfaction
Physrcran 2 ,
Expertise

 

 

 

 

” Pedhazur, Elazar I. 1982. Multiple regression in behavioral research (2"‘1 edition): p.604.

66

Figure 6 Total Effects on Patient Satisfaction

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cognitive .232 .507 Perception of
Empathy Physician

Expertise
Affective 620 Patient 172 Interpersonal
Empathy ' w Satisfaction ‘ ' Trust

Partn h'
Cognitive Inf. 244 f \ 651 ers 1p
Exchange ' '

 

 

 

 

 

 

Thus, considering the total effect (= indirect effect + direct effect) of each latent
variable on PC and PS is more appropriate when we try to discern the order of the
importance of each predictor on PC and PS. Figure 5 and 6 above show the total effect of
each latent variable on PC and PS. Table 15 and 16 below show the order of the
importance of these predictors on patient compliance and satisfaction.

First, for patient compliance (PC), partnership (Prtn) has the strongest effect on
PC, followed by affective empathy (AE), physician expertise (PPE), cognitive
information exchange (CIE), cognitive empathy (CE) and interpersonal trust (IT). Both
perception of physician expertise (PPE) and cognitive information exchange (CIE)
indicated a similarly moderate effect on PC while cognitive empathy (CE) and
interpersonal trust (IT) showed a small effect on PC. Contrary to my expectations, CE

and CIE had moderate negative total effects on PC. Although CIE had a moderate

67

-_
.3

1

negative total effect, it also had a positive indirect effect (.368) on PC via perception of

physician expertise.

Table 15

The Order of the Importance of Predictors on Compliance: total effects

 

 

 

 

 

Order Predictor Total Effect

1 Partnership .730

2 Affective Empathy .695

3 Perception of Physician .427
Expertise

4 Cognitive Information -.415
Exchange

5 Cognitive Empathy -.3 94

6 Interpersonal Trust .221

Table 16

The Order of the Importance of Predictors on Satisfaction: total effects

 

 

 

Order Predictors Total Effects

1 Partnership .651
Affective Empathy .620

3 Perception of Physician .507
Expertise

4 Cognitive Information .244
Exchange

5 Cognitive Empathy .232

6 Integlersonal Trust .172

 

 

Second, for patient satisfaction (PS), like patient compliance (PC), partnership

(Prtn) was again the most important predictor of PS, followed by affective empathy (AE),

perception of physician expertise (PPE), cognitive information exchange, cognitive

empathy (CE), and interpersonal trust (IT). While Prtn, AE, and PPE showed a strong

68

positive total effect on PS, other factors (CIE, CE, and IT) had a small total inﬂuence on

PS.

69

 

Chapter V

SUMMARY, DISCUSSION AND IMPLICATIONS
Summary

The effects of Korean physician empathy on patient outcomes have been
understudied until now. This study attempted to shed light on the communication
processes between physicians and patients in Korea. The main purpose of this study was
to propose and evaluate a new empathy model with a dualistic view (cognitive and
affective empathy). The empathy model showed how each empathy component was
speciﬁcally related to patient outcomes (satisfaction and compliance). The model was
developed loosely based on Squier’s empathic understanding model. This study also
developed and tested a new empathy measure, which taps two aspects of physician
empathy (cognitive and affective).

The premise tested here was that the patient’s interpretation or impression of the
physician’s level of empathy during a medical encounter is a key factor in determining
patient outcomes. Empathy underlies this impression made upon the patient.

The ﬁndings of this study were based on the data collected in a university hospital
in Korea, 1999. Five hundred ﬁfty (550) outpatients were participated in this survey. I
chose this site because the patients represent a wide variety of illnesses, doctor-patient
relationships, and types and lengths of medical examination.

Hypotheses were careﬁllly developed based on theoretical and previous empirical
studies. Due to the lack of existing measures for the constructs for this study, prior to the
testing of the full structural model, new scales were developed and examined for their

reliability and validity. Coefﬁcient alpha was used to examine the reliability of each

70

‘P‘ .15.;
I

measure. All of the measures had an acceptable level of coefﬁcient alpha (> .60).
Convergent validity was assessed from the measurement model by determining whether
each indicator’s factor loadings on its prespeciﬁed construct factor was signiﬁcant and of
satisfactory size via conﬁrmatory factor analysis (CF A). All standardized factor loadings
were above .60. Discriminant validity was examined by a chi-square difference test on
the values obtained for the constrained and unconstrained models. All possible pairs of
the constructs were tested for discriminant validity. All the chi-square difference tests
showed signiﬁcantly lower x2 value for the model in which the trait correlations are not
constrained to unity. A complementary assessment of discriminant validity was also done
by examining whether the conﬁdence interval around the correlation estimate between the
two factors includes 1.0. The results showed that none of them included 1.0. As results,
convergent and discriminant validities were achieved.

After convergent and discriminant validity were achieved, the overall goodness of
ﬁt to the observed data was tested by using ﬁt indexes (e.g., the x2 test statistic, CFI, CF1-
Robust, and RMSEA measures). Examination of overall ﬁt statistics showed acceptable
ﬁt of the model to the data. Hypothesis tests proceeded. They were tested by estimating

the key parameter values and their signiﬁcance in the full structural equation model.

Discussion of Major Findings
The major purpose of this study was to examine how each empathy component
was speciﬁcally related to patient outcomes (satisfaction and compliance) through critical

mediating variables.

71

rat-
I
.‘

Cognitive empathy was found to have a strong positive effect on cognitive
information exchange. Physicians who were able to accurately discern the mental state of
the patients and communicate effectively this perspective back to the patients were the
ones who more effectively exchanged cognitive information between themselves and
patients. Physicians with high cognitive empathy tended to not only provide more
information to the patients but also stimulate the patients to more accurately report their
symptoms as well as medically relevant experiences.

The affective component of physicians’ empathy was a strong predictor of
partnership. Physicians who had greater affective empathy were more likely to display
willingness to work jointly with a patient to promote his or her well being. Thus, for
Korean patients, the physician’s affective empathy was a key factor in building
partnership between physicians and patients.

Cognitive information exchange was strongly associated with the patient’s
perception of the physician’s expertise. Patients appeared to have formed their perception
of physician expertise based on the cognitive information exchange domain. The
physicians’ technical behaviors (e.g., sufficient information giving, physicians’
thoroughness) as well as their strong efforts to discern patients’ medical problems were
determinants of patients’ perception of physician expertise.

On the other hand, the patients’ interpersonal trust (emotional) was strongly
positively inﬂuenced by partnership. This conﬁrms the previous ﬁndings of Stile et al.
(1979). Because there is a gap between physicians and patients regarding knowledge and
power, patients’ emotional trust in physicians may be a more critical concern of the

patient. The partnership practiced by physicians may have provided an important tool to

72

the patients to assess whether their physicians could be trusted emotionally. Patients may
need perception of physician expertise or conﬁdence in their physicians’ technical
performance, but they may also need interpersonal trust in their physicians. Therefore,
for Korean patients, trust may consist of two distinct components (perception of physician
expertise [PPE] and interpersonal trust [IT]). The results showed that the distinctly -
different predictors (cognitive information exchange and partnership) strongly inﬂuenced
PPE and IT respectively.

Partnership had a strong direct effect on satisfaction while cognitive information
exchange had insigniﬁcant inﬂuence on patient satisfaction. This is consistent with many
previous studies (Hall, Roter, and Katz, 1988; Bensing 1991; Buller and Bullet, 1987;
Quine and Rutter, 1994). The results showed that the highly partnership-oriented
physicians tended to have more satisﬁed patients.

Perception of physician expertise (PPE) was strongly positively associated with
both patient compliance (PC) and with patient satisfaction (PS). The effect size (direct) of
PPE on both PC and PS were almost the same magnitude. This partially supported the
hypothesis that PPC would have effects on both PC and PS, but it was inconsistent with
the initial prediction that PPE would have more effect on PC than on PS. Although no
studies examined the direct linkage between PPC and PC, the prediction was based on the
implication of the previous studies (DiMatteo, 1994). It was hypothesized that if patients
trust their physicians as reliable medical experts, they would be more likely to believe in
their physicians’ recommendations and comply with them. Furthermore, as previously
pointed out, compared to patient compliance, patient satisfaction has been shown to be

more associated with affective and partner-like behaviors of physicians than task-oriented

73

behaviors (physician expertise). However, the results showed perception of physician
expertise (PPE) as equally good predictors of both patient compliance and patient
satisfaction. Physicians with a high rating of expertise by Korean patients tended to have
more compliant and satisﬁed patients.

The ﬁnding that interpersonal trust had a weaker inﬂuence on patient satisfaction
(PS) than on patient compliance (PC) ran counter to the prediction that it would have
more inﬂuence on patient satisfaction. The direct effect size of interpersonal trust on
both PS and PC was found to be relatively small, compared to that of PPE on PC and PS.
One possible explanation for this might be related to the measure of interpersonal trust
(IT). Some of the measurement items for IT may have been outside the reality of the
Korean patients. For instance, one of the items was ‘I feel I can call him or her if
something goes wrong and I need him or her’. Since a family physician system has been
rarely instituted in Korea, most of the patients may never have internally developed the
idea of ‘my doctor or my physician’. Thus, it is not very common for the patients to call a
speciﬁc doctor even in the case of an emergency. They would simply go to the hospital in
that situation. Thus, for Korean patients, some items for the interpersonal trust as deﬁned
for this study may not be very meaningful.

A signiﬁcant inﬂuence of patient satisfaction (PS) on patient compliance (PC)
was not found. Although this ﬁnding was inconsistent with many of the previous studies
in US. (Haynes, 1976; Segall et. al, 1980; Ley, 1986; Becker et. al., 1975; Weisman, C.
S. and Nathanson C. A., 1985; Linn M. W. et. al., 1982; Barlett, E. E. et. al., 1984;
Korsch, B. M. et. al., 1968; Hulka, B. S. et. al., 1976), it is consistent with one study

arguing that patient satisfaction with their medical visit was not necessarily a good

74

$-

predictor of patient compliance (Speedling & Rose, 1985). Lastly, the large negative
direct effect of cognitive information exchange (CIE) on compliance (PC) was a very
unexpected ﬁnding. CIE was indeed found to have greater direct inﬂuence on PC than
did partnership (Prtn) in absolute value. But I expected CIE to have a positive effect on
PC, not a negative effect. The result was inconsistent with the previous study that when
the doctor offers more information, the patient is more likely to be compliant (Hall,
Roter, & Katz, 1988).

Some possible explanations might be found in the Korean cultural context. First,
it may have to do with Korean patients’ expectation about the physician-patient
relationship. As previously discussed, the traditional Korean value pattern is a
collectivistic and authoritarian one, emphasizing deference to the traditional symbols and
holders of authority (Lee, 1993). The ideas that the holders of authority-doctors, teachers,
scholars, and government ofﬁcials-should deserve authority and that people should not
question their authority, are still widely embedded in the public mind. Physicians as
authority ﬁgures in Korea enjoy quite a lot of both autonomy in their expertise and
authority over the patients. Therefore, for Korean patients, it may be very typical to
expect a paternalistic physician-patient relationship (passive patients and dominant
physicians) in their medical encounters.

Furthermore, as Freidson (1970) pointed out, the physician-patient relationship is
inevitably asymmetric because physicians possess what patients do not have, the
professional medical expertise, and patients need that expertise for their health problems.
This gap between physicians and patients along with the Korean patients’ expectation of

paternalistic physician-patient relationship may heighten physician-dominant behaviors.

75

Thus, for Korean patients, it may be part of their social norms to accept the power and
authority of physicians.

Like American patients, Korean patients may have a great desire for information
and explanation about their medical conditions. However, because of the culture and the
asymmetric power in medical encounters, Korean patients may not often expect to receive
a lot of information and explanations from physicians. At the same the patient does not
expect the physicians to create conditions which stimulate to fully report their current
medical problems. In other words, the institutionalized passive patient roles may have
prevented Korean patients from actively getting information from their doctors and
criticizing physicians for not providing that information.

On the other hand, the Korean physicians, because of the social norms assigned to
them (dominant roles), might have been trying to maintain their expected authority and
dominance in the physician-patient relationship by limiting the ﬂow of information to
patients.

Consequently, as the Korean patients encounter physicians who behave otherwise,
they may begin to question the physicians’ authority. As Street (1987) pointed out, highly
expressive physician behaviors might connote insincerity. To Korean patients, this might
violate their expectations about the physicians. The patients may think that the physicians
are uncertain about the nature of the illnesses and therefore trying to cover up their
uncertainty by giving too much medical information, most of which is not understood by
the patients anyway. Subsequently the patients might be less likely to believe in and
comply with their medical recommendations because the physician’s behavior did not ﬁt

the typical image of physicians in the hierarchical authority structure of the Korean

76

medical settings.

However, as previously pointed out, we should be cautious about using only the
direct effect of a variable for the above interpretations. It may be misleading because it is
calculated while controlling for all the variables that affect a given endogenous variable.

The total effects of each variable on PC and PS were examined to discern the
order of the importance of each predictor on PC and PS in the previous section. For both
patient compliance (PC) and patient satisfaction (PS), partnership has the strongest effect,
followed by affective empathy (AE), physician expertise (PPE), cognitive information
exchange (CIE), cognitive empathy (CE) and interpersonal trust (IT). Contrary to my
expectations, CE and CIE had moderate negative total effects on PC. While Prtn, AE, and
PPE showed a strong positive total effect on PS, other factors (CIE, CE, and IT) had a
small total inﬂuence on PS.

In sum, the results of this study showed that the physician’s empathic
communication skills signiﬁcantly inﬂuenced patient satisfaction and patient cempliance
via the mediating factors such as partnership and perception of physician expertise.
However, the ﬁndings that cognitive empathy and cognitive information exchange had
some negative effects on patient compliance were contrary to the hypotheses. As
previously discussed, I attempted to ﬁnd some explanation for this in the Korean cultural
context (social norms or roles in an authoritarian culture). Above all, for Korean patients,
emotional aspects of the physicians’ communicative behaviors played the most important
roles in their compliance and satisfaction (e.g., partnership and affective empathy).

Unexpectedly, perception of physician expertise was one of the best predictors of both

77

patient satisfaction and compliance, although Korean patients were more concerned with
their physicians’ affective aspects of communicative skills.

Interestingly, the fact that partnership was the most signiﬁcant factor that
inﬂuenced both patient compliance and satisfaction may reﬂect Korean patients’ strong
yearning for an egalitarian relationship with their physicians in the current hierarchical

authority structure of the medical consultation processes in Korea.

Implications for Practitioners and Educators

The ﬁndings of this study have several implications for both physicians and
medical educators in Korea. The ﬁndings revealed that the patients’ perception of the
physician’ empathy was indeed a key to patient outcomes. Each component of physician
empathy (cognitive and affective) was signiﬁcantly associated with patient outcomes via
such mediating factors as partnership and perception of physician expertise, although
cognitive empathy had an unexpected negative effect on patient compliance. In general,
physicians’ emotional aspects of communicative behaviors signiﬁcantly affected Korean
patients’ satisfaction (PS) and compliance (PC). In order to improve patient outcomes,
Korean physicians and medical educators should ﬁrst identify the weak areas of
communication for each individual physician. Special attention should be given to those
areas. For example, physicians with low affective empathy or with low partnership
should reexamine whether their communication style is too dominant. Above all, they
ﬁrst honestly ask themselves if they feel superior to their patients. The long established
ideology of this kind might have been so irnbedded in the Korean physicians’ rrrind that

they might have taken it for granted. If the answer is ‘yes’, they should start to educate

78

and shift themselves toward a more equal and mutual doctor-patient relationship and
more compassionate concerns for their patients’ well-being.

Since it may be very difﬁcult for the established physicians to adjust to the new
kind of egalitarian doctor-patient relationship, it is more reasonable to start to expose
medical students to the egalitarian doctor-patient concept as well as empathic
communication skills in the early medical school training. Speciﬁcally, for medical
educators, the ﬁnding that emotional aspects of physicians’ communicative skills
(affective empathy and partnership) were the most important factors in patient satisfaction
and compliance suggests that those interpersonal communication skills should be
incorporated in their medical curriculum. And special emphasis should be put on teaching
empathic communicative skills to medical students.

Regarding patient compliance, the established physicians may argue that they do
not have enough time for empathy and may not see the immediate beneﬁts from being
empathic. However, as the ﬁndings of this study showed, the physicians’ partnership and
affective empathy were strongly associated with patient compliance. This suggests that
physicians with highly empathic communicative skills are more likely to make patients
obey their medical recommendations, which could result in saving much time and
expense for both sides.

In sum, every physician and medical student in Korea should learn to improve the
effective use of empathy because as Donald Light (1979) points out, too much emphasis
on clinical judgments and techniques might cause physicians to become insensitive to
complexities in diagnosis, treatment, and relations witll patients. This can result in errors

and malpractice suits. Moreover, given the rapid grth of health care expenditures, the

79

proliferation and duplication of medical technology, the lack of access for low-income
groups, and the low quality care (Bong-Min Yan, 1996; Youngsoo Shin, 1995), the
effective use of empathic communicative skills may be one of the best alternatives to
improving patient outcomes without increasing medical costs that are often caused by
frequently switching doctors, patient non-compliance, and malpractice suits. Empathy
will help physicians make patients more satisﬁed and more compliant. Consequently,
patients have better chances for recovery with proper treatment. In addition, as one of
management strategies, hospitals in Korea should adopt a program to enhance their
physicians’ empathic skills. Hospitals that have more empathic physicians have an
advantage over hospitals that have fewer empathic physicians because they are more

satisfying to patients.

Limitation and Future Research

There are several limitations to this study. First, the measure of compliance did
not include an item about the difficulty of following the medical regimen. For example,
one study showed that patients showed low compliance rates with the treatment regimens
that were very complex, time consuming, and demanded major behavior changes (Turk
and Meichenbaum, 1989).

Other item questions addressing these issues should be added. For example,
“How difficult was the medical regimen to follow?” “How many times did you have to
take your medicine each day?” and “How long did you have to follow the treatment

regimen?”

80

Second, question items in the interpersonal trust (IT) measure should be modiﬁed
to be more culturally meaningful to Korean patients. For example, as previously
explained, one of the items of IT, “ I feel I can call him or her if something goes wrong
and I need him or her”, should be changed since it was later found that for Korean
patients, the idea of calling a doctor personally appears to be relatively uncommon or
even foreign to them. An item that is more closely related to the reality of Korean
patients should be developed to tap the core of the interpersonal trust construct.

Third, this study was done solely on outpatients, not on inpatients, in a big
hospital in Korea. Further research is needed to examine possible differences between
inpatients and outpatients as well as between big hospital patients and small clinic
patients.

Fourth, this study focuses only on physicians’ abilities to be empathic and their
effects on patient outcomes. However, each physician may have different levels of
empathy. And these variations in physician empathy might be inﬂuenced not only just by
the physicians’ individual-empathic abilities but also by patient’s communication
behaviors as well. For example, two important elements of patient’s communication
behaviors -- question-asking and negative affect expressing - might be capable of
inﬂuencing a physician’s communication skills. The previous research shows physicians’
dominant role in medical communication. For instance, Waitzkin (1984; 1985) found
that physicians do not prefer patient-initiated questions and discourage them by
interrupting and changing the topic. Thus, patients may appear to have little control over
the content of communication with their physicians. However, like many other

conversations, medical discourse in itself is a mutual interactant process between the

81

participants. According to a symbolic interactionist perspective, for an interaction to be
successful, an actor in a social interaction does not arbitrarily deﬁne the situation and act
upon it all by himself or herself. The actor is required to act in line with the other actor’s
response. Thus, in a medical discourse, when asked, physicians may feel obligated to
give at least some information to patients. In support of this argument, several studies
documented that physicians provided more information to patients who asked more
questions (Street, 1991; 1992; Beisecker and Beisecker, 1990). Furthermore, given the
fact that the idea of patients as health care customers is rapidly spreading among the
general public, Korean patients might be more willing to show aggressive communication
styles more than ever before. Therefore, future study should take this into consideration.
patients’ communication behaviors need to be considered in the future study for further
investigation.

Fifth, although the comprehensive review of the US literature on the relationship
between sociodemographic factors and patient’s satisfaction with medical care shows
conﬂicting results, the characteristics (sex, age, etc.) about both patients and physicians
might be related to patient outcomes in Korea. This study did not include these factors in
the empathy model. Future research is recommended to look into the possible link
between these factors and patient outcomes.

Sixth, this study does not include a patient expectation about physician
communication styles (for instance, expectation of authoritarian versus egalitarian styles).
Some patients might prefer authoritarian physicians while others do not (Street, 1990).
Korean patients might in general, have different expectations about physicians’

communication styles from those in other countries. Even within Korea there may be

82

important variations in patients’ expectations about physicians’ communication styles.
Future researchers should consider patients’ expectation as another mediating factor in
the model.

Lastly, one should be cautious about the interpretation and generalization of the
results until the ﬁndings can be replicated in other countries. Future research might want
to replicate this study in the US or other countries and compare them to see cross—cultural

differences and similarities.

83

REFERENCES

Allman, R., Yoels, W., and Clair, J. 1993. "Reconciling the agendas of physicians and
patients." pp. 29-48 in Sociomedical perspective on Patient Care by Jeffrey Clair
and Richard Allman (eds.) Lexington KY Univ of Kentucky Press.

Anderson, J. C. and Gerbing, D. W. 1982. “Some methods for respecifying
measurement.” Journal of Marketing Research, vol. 19 (Nov): 453-460.

Anderson, J. C. and Gerbing, D. W. 1988. “Structural equation modeling in practice: a
review and recommend two-step approach.” Psychological Bulletin vol. 103(3):
411-423.

Anderson, L. A. and Dedrick, R. F. 1990. “Development of the trust in physician scale: a
measure to assess interpersonal trust in patient-physician relationship.”
Psychological Reports, Vol. 67: 1 091-1 100.

Bagozzi, R.P., and Yi, Y. 1988. “On the evaluation of structural equation models.”
Journal of the Academy of Marketing Science, Spring, vol. 16(1): 74-94.

Barlett, E. E., Grakyson, M., and Barker, R., et al., 1984. “The effects of physician
communications skills on patient satisfaction, recall, and adherence, Journal of
Chronic Diseases vol. 37: pp. 755—764.

Barrett-Lennard, G. T. 1981 “Empathy cycle: reﬁnement of a nuclear concept.” Journal
of Counselling Psychology, Vol. 28(2): 91-99.

Becker, M. H. and Mairnan, L. A. 1975. “Sociobehavioral Determinants of
Compliance with Health and Medical Care Recommendations” Medical Care
Vol. 13: 10-24.

Beisecker A. E. and Beisecker T. D. 1990. “Patient information-seeking behaviors when
communicating with doctors.” Medical Care, Vol. 28: 19-28.

Bellet, P. S. and Maloney, M. J. 1991. “The importance of empathy as an interviewing
skill in medicine.” JAMA Vol. 266(13): 1831-1832.

Bensing, J. 1991. “Doctor-patient communication and the quality of care.” Social
Science and Medicine Vol. 32(11): 1301-1310.

Ben-Sira Z. 1976. "The Function of the Professional's Affective Behavior in Client

Satisfaction: A Revised Approach to Social Interaction Theory." Journal of
Health and Social Behavior 17:3-1 1.

84

Ben-Sira Z. 1980. “Affective and instrumental components in the physician-patient
relationship: an additional dimension of interaction theory.” Journal of Health
and Social Behavior, vol. 21(June): 170-180.

Ben-Sira, Z. 1986. Stress, disease and primary medical care. Gower, England.

Bentler 1989, 1990). Bentler, P. M. (1989). EQS structural equations program manual.
Los Angeles: BMDP Statistical Soﬂware. Bentler, P. M. (1990). Comparative ﬁt
indexes in structural models. Psychological Bulletin, 107, 238-246. ‘

Bentler, P. M. 1995. EQS Structural Equation Program Manual. Multivariate Software,
Inc.: Encino, CA.

Bentler, P. M. (1993). EQS structural equations program manual. Los Angeles: BMDP
Statistical Software.

Bentler, P. M. 1989. EQS structural equations program manual. Los Angeles: BMDP
Statistical Software.

Bentler, P. M. 1990. “Comparative ﬁt indexes in structural models.” Psychological
Bulletin, 107:238-246.

Bentler, P. M. 1992. “On the ﬁt of models to covariances and methodology to the
Bulletin.” Psychological Bulletin, vol. 112 (Nov) : 400-404.

Bentler, P. M., and Bonett, D. G. 1980. “Signiﬁcance tests and goodness of ﬁt in the
analysis of covariance structures.” Psychological Bulletin, vol. 88: 588-606.

Bentler, P. M., and Wu, E.J.C. 1993. EQS/Windows: User ’s Guide. Los Angeles: BMDP
Statistical Soﬁware.

Bollen, K. A. 1989. Structural Equations with Latent Variables. Wiley: New York.

Brock, C. D. and Salinsky, J. V. 1993. “Empathy: an essential skill for understanding
the physician-patient relationship in clinical practice.” Family Medicine Vol.
25: 245-248.

Browne, M. W. 1984 “Asymptotically distribution-ﬂee methods for the analysis of
covariance structures.” British Journal of Mathematical and Statistical
Psychology, Vol. 37:62-83.

Buller, M. K. and Buller, D. B..1987. "Physicians' Communication Style and Patient
Satisfaction." Journal of Health and Social Behavior 28:375-3 88.

85

Calantone, R. J ., Schmidt, J. B., and Song, X. M. 1996. “Controllable factors of new
product success: a cross-national comparison.” Marketing Science vol. 15(4):
341-358.

Carmel, S. and Glick, S. 1996. “Compassionate-empathic physicians: Personality
traits and social-organizational factors that enhance or inhibit this behavior
pattern.” Social Science and Medicine Vol. 43(8): 1253-1261.

Comstock, L. M., Hooper, E. M., Goodwin, J. M., and Goodwin, J. S. 1982. "Physician
Behaviors That Correlates With Patient Satisfaction." Journal of Medical
Education Vol. 57:105-112.

Davis, M. H. 1983. “Measuring Individual Differences in Empathy: Evidence for a
Multidimensional Approach.” Journal of Personality and Social Psychology,
Vol. 44(1): 113-126.

Dean, M. 1993. “Effective communication with patients.” The Lancet, Vol.
342(8885): 1477.

DiMatteo et al, 1985. ‘Affect an neutrality in physician behavior: a study of patients’
values and satisfaction’ Journal of Behavioral Medicine, vol. 8(4): 397-409.

DiMatteo, M. R., et al. 1993. “Patient adherence to cancer control regimes: scale
development and initial validation.” Psychological Assessment Vol. 5: 102-112.

Eisenberg N. and Strayer, J. (Eds) 1990. Empathy and its development Cambridge
University Press, Cambridge.

Fan, X., Thompson, B., and Wang, L. 1999. “Effects of sample size, estimation methods,
and model speciﬁcation on structural equation modeling ﬁt indexes.” Structural
Equation Modeling, vol. 6(1): 56-83.

Fornell, C. and Larker, D. F. 1981. “Evaluating structural equation models with
unobservable variables and measurement errors.” Journal of Marketing
Research, vol. 18 (Feb): 39-50.

Francis, V., Korsch, B. M., and Morris, M. J. 1969. "Gaps in Doctor-Patient
Communication: Patients' Response to Medical Advice." New England Journal
of Medicine 280: 535-540.

Free, N. K. et a1. 1985 “Empathy and Outcome in Brief Focal Dynamic Therapy”
American Journal of Psychology, Vol. 142(8): 917-921.

F reidson, Eliot. 1970. Professional Dominance. New York. Atherton.

Freud, S. 1949. The psycho-analytic theory of neurosis. New York: Norton.

86

F reund, E. S. and McGuire, M. B. 1995. Health, Illness, and The Social Body: 0 critical
sociology (2“d edition) Prentice Hall, Englewood Cliffs, New Jersey.

Glanz, K. et al. (eds.) 1997. Health Behavior and Health Education: Theory,
Research, and Practice (2"d edition) Jossey-Bass Publishers, San Francisco.

Goffrnan, Erving 1959 The Presentation of Self in Everyday Life Doubleday.

Hall, J. A. et al. 1988. “Meta-Analysis of Correlates of Provider Behavior in Medical
Encounters.” Medical Care, Vol. 26: 657-675.

Hall, J. A., Roter, D. and Rand, C. S.. 1981. "Communication of Affect between Patient
and Physician." Journal of Health and Social Behavior 22: 18-30.

Hall, J. A., Roter, D. L., and Katz, N. R. 1988. “Meta-analysis of correlates of provider
behavior in medical encounters.” Medical Care, vol. 26: 657-675.

Harlow, LL. 1985. “Behavior of some elliptical theory estimators with nonnormal data
in a covariance structures framework: a Monte Carlo study.” Dissertation
Abstracts International, vol.46: 2495B.

Hayes-Bautista, David E. 1976. "Modifying the treatment: Patient Compliance, patient
control and medical care." Social Science and Medicine 10:233-238.

Haynes, R. B. 1976 “A Critical Review of the Determinants of Patient Compliance with
Therapeutic Regimens” In Compliance With Therapeutic Regimens by Sackett D.
L. and Haynes, R. B. (Eds) Johns Hopkins University Press, Baltimore, Md.

Hogan, R. 1969 “Development of An Empathy Scale.” Journal of Consulting and
Clinical Psychology Vol. 33: 307-3 16.

Hu, L. T., Bentler, P. M., and Kano, Y. 1992. “Can test statistics in covariance structure
analysis be trusted?” Psychological Bulletin, vol. 112: 351-362.

Hub Li-tze; Bentler,Peter M. 1998 “Fit indices in covariance structure modeling:

Sensitivity to underparameterized model misspeciﬁcation.” Psychological
Methods, Vol. 3 (4) Dec: 424-453.

Hulka, B. S., Cassel, J. C., and Kupper, L. L., et. al. 1976. “Communication, compliance,
and concordance between physicians and patients with prescribed medications.”
American Journal of Public Health, vol. 66: 847-853.

Jbreskog, K. G. 1971. “Statistical analysis of sets of congeneric tests.” Psychometrika,
vol. 36: 109-133.

Jbreskog, K. G., and Sbrbom, D. 1993. LISREL 8 User ’s Reference Guide. Chicago:
Scientiﬁc Software International.

87

Jbreskog, K. G., and Van Thrillo, M. 1973. LISREL — a general computer program for
estimating a linear structural equation system involving multiple indicators of
unmeasured variables. Research Report 73-5, Department of Statistics, Uppsala
University, Sweden.

Kalfas, N. S. 1974. “Client-perceived therapist empathy as a correlate of outcome.”
Dissertation Abstracts International vol. 34: 5633A.

Kaplowitz, S. et al (unpublished) “Desire for Prognosis Information by Cancer Patients.”

Korsch, B et al., 1968. “Gaps in doctor-patient communication. 1: Doctor-patient
interaction and patient satisfaction.” Pediatrics, Vol. 42:855-871.

Kurtz, R. R. 1972. “Different Approaches to the Measurement of Therapist Empathy
and Their Relationship to Therapy Outcomes.” Journal of Consulting and
Clinical Psychology, Vol. 39(1): 106-1 15.

Lee, Ale-Ric 1993. “Culture and Popular Protest in South Korea.” Comparative
Political Studies vol. 26(1): 63-80.

Ley, P. 1977. “Psychological studies of doctor-patient communication.” In S.
Rachmann (Ed.) Contributions to medical psychology, Vol. 1 Oxford: Pergarnon
Press: 9-42.

Ley, Philip 1986 “Cognitive Variables and Noncompliance” The Journal of
Compliance in Health Care Vol. 1(2) : 171-188.

Light, D. Jr. 1979. “Uncertainty and control in professional training.” Journal of Health
and Social Behavior, vol. 20: 310-322.

Linn M. W., Linn B. S., and Stein S. R. 1982. “Satisfaction with ambulatory care and
compliance in older patients.” Medical Care, vol. 20: 606-614.

Mardia, K.V. 1970. “Measures of multivariate skewness and kurtosis with applications.”
Biometrika, vol. 57: 519-530.

Mead, G. H. 1934. Mind, Self and Society University of Chicago Press.

Mechanic, D. 1998. “Public trust and initiatives for new health care partnerships ” The
Mibank Quarterly Vol. 76(2): 281-302.

Mehrabian, A. and Epstein, N. 1972. “A measure of emotional empathy.” Journal of
Personality, vol. 40: 525-543.

More, E. S. and Milligan, M. A. (Eds) 1994. The Empathic Practitioner: Empathy,
Gender, and Medicine. Ruters Uninversity Press: New Brunswick, NJ.

88

Mueller, R. O. 1996. Basic Principle of Structural Equation Modeling: an introduction
to LISREL and EQS. Spring: New York.

Nunnally, J. c. 1978. Psychometric Theory, 2“d Edition, McGraw-Hill: New York.

Ok, Ryun Moon 1991. “The Korean Health insurance program: analysis of ﬁnanical and
cost containment measures.” Asia-Pacific Journal of Public Health vol.5 (1): 66-
78.

Olson, J. K. 1995. “Relationships between nurse-expressed empathy, patient-perceived
empathy and patient distress” Image: Journal of Nursing Scholarship Vol. 27(4):
3 1 7-3 22.

Parsons, T. 1951. The Social System. Free Press: New York.

Quine, L. and Rutter, D. R. 1994 “First Diagnosis of Severe Mental and Physical
Disability: A Study of Doctor-Parent Communication. ” Journal of Child
Psychology and Psychiatry and Allied Disciplines Vol. 35(7): 1273-1287.

Romeu, J. L., and Ozturk, Aydin. 1993. “A comparative study of goodness-of-ﬁt tests
for multivariate normality.” Journal of Multivariate Analysis, vol. 46: 309-334.

Roter, D. and Hall, J. A. 1997. “Patient-provider communication.” In Glanz et al (Eds),
Health Behavior and Health Education: Theory, Research, and Practice. J ossey-
Bass Publishers: San Francisco: 206-226.

Roter, D. L. and Hall J . A. 1993. Doctors Talking with Patients / Patients Talking with
Doctors: improving communication in medical visits. Auburn House, London.

Roter, et. al., 1987. “Relations between physicians’ behaviors and analogue patients’
satisfaction, recall, and impressions.” Medical Care, vol. 25(5): 437-451.

Rowland-Morin, PA and Carroll, J G 1990. “Verbal communication skills and patient
satisfaction: A study of doctors-patient interviews.” Evaluation and the Health
Professions 13 (2 ):168-185.

Satorra, A., and Bentler, P. M. 1988. “Scaling corrections for chi-square statistics in
covariance structure analysis. Proceeding s of the American Statistical

Association: 308-3 13 .

Segall, A. and Burnett, M.. 1980. "Patient Evaluation of Physician Role Performance."
Social Science and Medicine 14A: 269-278.

Shin, Youngsoo 1995. “The health insurance system in Korea and its implications.”
World Hospitals and Health Services, Vol. 31(3): 3-9.

89

Speedling, E. J ., & Rose, D. N. “Building an effective doctor-pateint relationship: From
patient satisfaction to patient participation” Social Science and Medicine, vol.
21: 115-120.

Squier, R. 1990 “A Model of Empathic Understanding and Adherence to Treatment
Regimens in Practitioner-Patient Relationships.” Social Science and Medicine
Vol. 30(3): 325-339.

Stantton, A. L. 1987. “Determinants of adherence to medical regimens by hypertensive
patients.” Journal of behavioral medicine, Vol. 10:377-394.

Steiger, J. H. (1989). EzPA TH: A supplementary module for SYSTAT and SYGRAPH.
Evanston, IL: SYSTAT.

Steiger, J. H. 1990. “Structural model evaluation and modiﬁcation: an interval
estimation approach.” Multivariate Behavioral Research, vol. 25: 173-180.

Stiles, W. et al, 1979. “Interaction Exchange Structure and Patient Satisfaction with
Medical Interviews.” Medical Care Vol. 17: 667-679.

Street, Jr. et. al., 1987. “Patient satisfaction with physician’s interpersonal involvement,
expressiveness, and dominance.” Communication Yearbook, vol. 10: 591-612.

Street, R. L. 1990. “Communication in medical consultation: a review essay.”
Quarterly Journal of Speech, vol. 76: 315-357.

Street, R. L., Jr. 1991. “Inforrnation-giving in medical consultations: the inﬂuence of
patients' communicative styles and personal characteristics.” Social Science and
Medicine Vol. 32(5): 541-548.

Street, KL. 1992. “Communicative styles and adaptations in physician-parent
consultations.” Social Science and Medicine Vol. 34(10): 1155-1163.

Svarstad, B. L. 1976. “Physician-patient communication and patient conformity with
medical advice.” In D. Mechanic (Ed.), The growth of bureaucratic medicine: An
inquiry into the dynamics of patient behavior and the organization of medical
care. New York: Wiley.

Svarstad, B. L. 1986. "Patient practitioner relationships and compliance with prescribed
medical regimens," pp. 43 8-459 in Applications of Social science to clinical
medicine and health policy L. Aiken and D. Mechanic (eds) New Brunswick, NJ
Rutgers Univ. Press.

Szasz, T. and Hollender, M. H. 1956. “The basic models of the doctor-patient
relationship.” Archieves in Internal medicine 97: 585-592.

90

Tanaka, J. C. 1984. “Some results on the estimation of covariance structure models.”
Dissertation Abstracts International, vol. 45: 924B.

Thom, D. H. and Campbell, B. 1997. “Patient-physician trust: an exploratory study.”
The Journal of Family Practice, Vol. 44(2): 169-176.

Turk, D. C., and Meichenbaum, D. 1991. “Adherence to self-care regimens: the patient’s
perspective.” In R. H. Rozensky, J. J. Sweet, and S. M. Tovian (Eds), Handbook
of clinical psychology in medical settings, Plenum: New York: 249-266.

Waitzkin, H. 1984. "Doctor-Patient Communication: Clinical Implications of Social
Scientiﬁc Research." Journal. American Medical Association 252: 2441-2446.

Waitzkin, H.1985. "Information Giving in Medical Care." Journal of Health and Social
Behavior 26:81-101.

Waitzkin, Howard 1991. The Politics of Medical Encounters: How Patients and Doctors
Deal With Social Problems New Haven and London Yale University Press.

Weisman CS, Nathanson CA. Professional satisfaction and client outcomes: a
comparative organizational analysis. Medical Care 1985, vol. 23: 1179-1192.

Wyatt, N. 1991. "Physician-patient relationships--What do doctors say? Health
Communications 3 (3): 157—174.

Veatch, RM 1991. The patient-physician relation: The patient as partner, part 2.
Bloomington IN Indiana Univ. Press.

Yan, Bong-Min 1996. “The Role of health insurance in the growth of the private health
sector in Korea.” International Journal of Health Planning and Management,
vol. 11:231-252.

Young, J. W. 1980. “The effects of perceived physician competence on patients’
symptom disclosure to male and female physicians.” Journal of Behavioral

Medicine, Vol. 3(3): 279-290.

Zborowski, M. 1952. "Cultural components in response to pain" Journal of
Social Issues 8:16-30.

Zola, Irving K. 1966. "Culture and Symptoms-- An Analysis of patients' presenting
complaints.” American Sociological Review 31 :615-630.

91

APPENDIX I

Initial Empathy Survey Questionnaire

You are asked to ﬁll out this survey on completely voluntary basis. This survey is
completely anonymous. Nowhere on this form will an identify mark or number be
placed. If you feel uncomfortable with completing the survey, you may simply stop it.
You indicate your voluntary consent to participate by completing this questionnaire. '

Background Information about Patients:
1. Your sex: ...... Male _ Female
2. How old are you now? ...... Years
3. Have you ﬁnished _ ?
1) Elementary school 2) Middle school 3) High school
4) Some college or associates’ degree 5) Bachelor’s degree 6) Graduate Degree
4. What is your family’s income a month?
1) below 50 manwon 2) 51-100 manwon 3) 101-150 manwon
4)151-200 manwon 5) 201-250 manwon 6) above 251 manwon

Below is a list of thoughts and feelings people sometimes have about their doctors

Each item is a statement with which you may agree or disagree. Beside each statement is
a scale that ranges from strongly agree (1) to strongly disagree (5). For each item please
circle the number that represents the extent to which you agree or disagree with the
statement.

Please make sure that you answer every item and that you circle only one number per
item. It is important that you respond according to what you actually believe and not
according to how you feel you should believe or how you think we may want you to
respond.

1 = Strongly disagree

2 = Disagree
3 = Neutral
4 = Agree

5 = Strongly agree

Cognitive Empathy

1. This doctor tries to see things through my eyes. 1 2 3 4 5
2. This doctor does not understand the way I feel. 1 2 3 4 5
3. This doctor nearly always knows exactly what I mean. 1 2 3 4 5

92

At times he/she jumps to incorrect conclusions about
my feelings.

This doctor understands what my experiences feel like to me.
When I am not clear, this doctor still understands me.

This doctor is interested in knowing what my experiences
mean to me.

Affective Empathy

1.

2.

This doctor responds to me mechanically.

This doctor tries to keep me from worrying.

. This doctor respects my feelings.

This doctor shows caring about my psychological well-being.

This doctor is able to remain calm even though I worry so
much about my problems.

This doctor cares about me.
This doctor is interested in me.
Sometimes this doctor is not at all comfortable with me.

This doctor shows great concern for my well-being.

Cognitive Information Exchange

1.

2.

This doctor clearly explained what my troubles are.
This doctor told me exactly what he planned to do next.
This doctor told me why certain tests would be done.

This doctor clearly explained why I should do the things he

93

l

—I

asked me to do.
5. This doctor does a good job of explaining things to me.

6. I understand what this doctor is telling me about my
condition.

7. I understand what the doctor wants me to do.

8. It is sometimes hard to know what the doctor is telling me
to do.

9. This doctor clearly explains my medical condition to me so
that I understand it perfectly.

10. I have a feeling that I was able to accurately report current
and past symptoms and medically relevant experiences.

11. The way this doctor talks stimulated me to fully report my
current medical problems.

12. This doctor understands my medical problem very well.

13. This doctor tries to get thorough information about my
physical condition.

Perception of Physician Expertise

1. If this doctor tell me something is so, then it must be true.

2. I sometimes distrust my doctor’s opinion and would like
a second one.

3. I trust my doctor’s judgments about my medical care.

4. I feel my doctor does not do everything he/she should for
my medical cares.

5. I trust my doctor to put my medical needs above all other

considerations when treating my medical problems.

6. This doctor is a real expert in taking care of medical
problems like urine.

94

Interpersonal trust

1.

2.

This doctor is usually considerate of my needs.

I doubt that this doctor really cares about me as a person.

. I feel I can call him or her if something goes wrong and I

need him or her.
I feel this doctor trusts me.
I have conﬁdence that he/she is on my side.

I believe that this doctor is working in my best interest.

Partnership

1.

9.

This doctor often makes acknowledgment that he or she has
heard the statement that I have just made.

This doctor treats me as an equal.
This doctor regards me as a disagreeable patient.

This doctor encourages me to ask questions.

. This doctor answers my questions respectfully.

This doctor often agrees with my opinion.

This doctor encourages me to express my own feelings
or concerns.

I feel comfortable asking this doctor questions about
my illness.

This doctor was interested in hearing about my symptoms.

10. This doctor is very ﬂexible, open to my suggestions or ideas.

95

Patient Compliance

1.

2.

I followed my doctor’s suggestions exactly.

I exactly complied with this doctor’s drug regimens

Patient Satisfaction

l.

2.

Overall, I am satisﬁed with this doctor.
I have conﬁdence in this doctor.
Overall, I am discontented with this doctor.

Compared to the other doctors you have seen, this doctor is
very good.

96

APPENDIX II

Final Set of variables Used For the Data Analysis

Cognitive Empathy
1. This doctor nearly always knows exactly what I mean.
2. When I am not clear, this doctor still understands me.

3. This doctor is interested in knowing what my experiences
mean to me.

Affective Empathy

1. This doctor responds to me mechanically.

2. This doctor tries to keep me from worrying.

3. This doctor respects my feelings.

4. This doctor shows caring about my psychological well-being.
5. This doctor cares about me.

6. This doctor is interested in me.

7. This doctor shows great concern for my well-being.

Cognitive Information Exchange

1. This doctor clearly explained why I should do the things he
asked me to do.

2. This doctor does a good job of explaining things to me.

3. This doctor clearly explains my medical condition to me so
that I understand it perfectly.

4. I have a feeling that I was able to accurately report current
and past symptoms and medically relevant experiences.

97

5.

The way this doctor talks stimulated me to ﬁllly report my
current medical problems.

Perception of Physician Expertise

I.

2.

3.

If this doctor tell me something is so, then it must be true.

I sometimes distrust my doctor’s opinion and would like
a second one.

I trust my doctor’s judgments about my medical care.

Interpersonal trust

1.

2.

3.

4.

I feel I can call him or her if something goes wrong and I
need him or her.

I feel this doctor trusts me.
I have conﬁdence that he/she is on my side.

I believe that this doctor is working in my best interest.

Partnership

1.

This doctor oﬁen makes acknowledgment that he or she has
heard the statement that I have just made.

This doctor answers my questions respectﬁrlly.

I feel comfortable asking this doctor questions about
my illness.

This doctor was interested in hearing about my symptoms.

This doctor is very ﬂexible, open to my suggestions or ideas.

Patient Compliance

1.

2.

I followed my doctor’s suggestions exactly.

I exactly complied with this doctor’s drug regimens

98

Patient Satisfaction

1. Overall, I am satisﬁed with this doctor.

2. I have conﬁdence in this doctor.

3. Overall, I am discontented with this doctor.

4. Compared to the other doctors you have seen, this doctor is
very good.

99

Appendix III: References for Questionnaire Items

The following table shows where some of the initial questionnaire items had been
adopted from (all the items had been modiﬁed for the study).

 

 

 

Empathy Scale: BLRI Stan’s Satisfaction Scale DiMatteo’s General
(Barrett-Lennard, 1981) Adherence Scale
CE: 1,2, 3, 5, 6, and 7. PS: 1,3. PC: 1,2.

AB: 1, 5.

 

 

 

100

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

lulllslmlmgnmglljllul