.1
1.', "‘JJ“ JJ‘J‘ 1111I11JJJ-. .111111JIIII """1'J'1"JJ‘ 1‘111I WJJJ'J. 'I'I'JJJ’J "'JJJ 11111|J
”11' 1 JJ 1

I.1111I1' 11111.11 “"1 11. .1
‘JJJ” JI'JJJJ' 'JfiJJJJ'JJJJ“ "'J‘ 1'1 IJJJJJ ‘JJJJJJ"" I 'JJJ" '1'". 11 .

.1111‘J.‘11111-11111""1"11111I.III'I“ 11 '
I'J1'1'11"“""J' 11"1“' "JJIJJJJJJJJ""‘ 11"r II

["1 11'.’J 1 '1'1‘ 1 ‘JJIJJ

J111IIII'J1'I1I
JJJ JJ . , 1
I.

11'1' '
J' II

1I'1I I'I1
11
1 'JJ 1'
‘I1
"1'I""I1"I"'1"'I"II'""
11'1'

         
   
     
   
    

 

'1
'J IJI1I..IJI

        

 

  
       
  
   
 
  
  
    

   
  
  
   
       

 

    
 

  
 
  
 

   
   

      
  
  

    
  

       
       
   
    

    

  

      
   
  
  

  

"J'I'JJ ' 'J'1":1JIJI "'JJ IIII 1'1J'I:
‘ JIJI"J'IJ IIII-I'1I1'I‘I JJJJ‘I‘JJ JJJ'J1‘I' "J‘ II'IJ 1'1.‘ II'I1IJI 1 1JII1'I1' ‘1 IJI III
'1.11"' JJ‘J'HJ.‘ ' ‘
J'IJ‘I'I' "'J1"" 1 IW'JJ“ "'11 '
:' """'“"'J"J'"' ' ""11" 1"‘ 1'" J‘J1J1““1 3:; I13, III” I:
'.' 1"“1'1'I'J .1 1. '.11'11 1'11J1JI'I'J‘1 "'1;'1‘1‘J1'J1",,"" 1:1111f'.JI'11-'.'1"J.1‘ ‘1"; iii-III ‘ .¢~JIJ‘1'1LJJ‘ J‘ *1 VHJ‘
., J '1 '1 ‘1 1 1:11 1" .11:111.1111111111111“11-1.1 '1“ "11" Jr" 1 1;." 11
i' :‘1 1'1'1'J1F'1 ‘1‘1'11 ’111‘1'1' 'r1:1“1i'1111‘,"1"1‘"JE ' ‘ ' .";. "11 . .
1‘1 'J'J I J :I 111. '1'JJJ'J J..""11'1"J1,.'J1"" 1JI1JI1‘1‘JI'I1 J‘1J'1J:£s "13." III .‘I I 1 {va1 VJ
J' ' ' IJJJJJJ‘JJJIJ '1'"' IJJJJ' 'JJJIJ JJJI' 'IJJ'JJ' JJHJ JJJJ Jag \Jfi” II] ‘ \J‘k “I?!”
" "' "" "‘""' ""1'1' ‘JJJ J‘JJ "' """"J“J“" """" 1"“ J'w'flw' $1”ng 1,11‘JJJ" 1.'-11'1JJI'J1'J1“ i'1'1“1
'1 1I1'I1II1I1IIIJJJ1II'1JJ1IIJ JJJ JJJJJJIIJ JJ‘IJIJ‘J‘J JJJJJJ “WM“ J‘JJ,‘ III ‘JIJJJJL'J'JJJJIW I 'JJJ‘JI" JJJJ ;J'(
‘ J1' 1 ,.1. 1‘11
1.1 111'..1 J 1‘1 1111111
" J 1JI'JIJ1'I‘ ‘J“"‘J‘1 1" ‘JJ‘J‘J J11111‘JIJJI‘ """ " '1‘“ “J "J' .11)
‘ J J 1' 1111‘1

  

   
   

  
        
 
   
 

IIJIIJIII

J‘11'1J

'Jl

J"" ' J" 'IJ‘I’JJJ .
1J J J'J'J 1 1'“ J1 :JJ'JIIJJ {1"{1‘ II1‘III1IIII.
1|J‘J‘JJ'JJIJ1J JIIL5J ‘JJJ“..J "a“ 111V '{Ifw :‘II'JIIII t‘ y& I '9 JJJJme'I'I‘I'I'fiJ)"J"—"‘ J 'IJI'I.«1:I‘.IJJ11'I‘I:IIIJ“"'I:{ JJJ“!
JJ1J‘JJ11'J1E'1I L ' 1 1|
"'1J‘JJJ"'1

J .‘JII I ' H‘J" II‘IJ’M JJ‘
JJJJ11111111'111111

'IJJ'J
JJJ

  
  
 

""'}'J111JJJ11‘ "'IJ
"1'1 1."1'11‘ I1III'IJ11J 'JJJ "'J
" 'J'.,'1'111 J

11
111111

 

 

II Luann; lllzllflifllllil Lu! J1”! "MI wing; llljllzl || ~

 

 

OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

 

 

 

@ Capyright by
FRANCES WIDEHA STOTT

1979

 

CORRESPONDENCE BETWEEN PERSONALITY
DIMENSIONS AND EXPRESSIONS OF

FACIAL AFFECT

BY

Frances Wideha Stott

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Counseling, Personnel Services
and Educational Psychology

1979

ABSTRACT

CORRESPONDENCE BETWEEN PERSONALITY
DIMENSIONS AND EXPRESSIONS OF
FACIAL AFFECT

BY

Frances W. Stott

The purpose of the study was to investigate whether
there are significant differences in expressions of facial
affect as recorded on videotape and measured by trained
judges between persons with specific personality character-
istics as measured by pencil/paper personality instruments.
The assumption that personality characteristics related to
emotional style are discernable in the face by specific ex-
pressions of facial affect was the basis for this research.
The potential use of such information for subsequent re-
search efforts and therapeutic intervention provided impe-
tus for the study.

One hundred sixty-eight undergraduate students at
the University of Texas-Austin were administered a battery
of pencil/paper personality instruments: the Plutchik

Emotions Profile Index, the gysenck Personality Inventory,

 

 

and the Birkman Method. Sixty-two subjects were subse-

 

quently selected to participate in the interview phase, on

Frances W. Stott

the basis of their high or low scores across the nine per-
sonality dimensions that were used as the independent
variables for the study. Subjects were divided into two
groups (high and low) for each of the personality dimensions
used, based on their scores on one of the personality dimen-
sions at a time.

The interviewer was the same for all subjects, and
was instructed to engage in a responsive but non-confrontive
manner. Subjects were asked to select an item of personal
interest from the Moonenyroblem Check List's "Personal
Psychological Relations Scale," and then to talk with the
interviewer about that topic for 15 minutes.

The interviews were videotaped, and a systematic
sampling procedure was used to prepare the data for rating.
There were 30 samples of facial affect per subject, or a
total of 1,680 samples for the study. Three females were
trained as raters, to recognize the categories of Enjoy,
Anger, Surprise, Interest, Disgust, and Distress. These
six categories, plus calculations regarding frequency of
change from one affect category to another, and number of
affect categories used, provided the total of 8 dependent
measures for the study.

The fifteen hypotheses used in the study were stated
so as to test the relationship between performance on speci-

fic personality dimensions and the type, range, or frequency

Frances W. Stott

of change of facial affect expressed by the subject. Hypo-
theses were tested by a comparison of groups on the affective
measures, using t-tests, with an alpha level set at .05.
Supplementary analyses from the multivariate, univariate,
and correlational aspects were also used for further explor-_
ation of differences and relationships.

Two of the 15 hypotheses were significant. A sig—
nificant difference was found between high and low Dis-
trustful groups on the affective measure of Disgust (p=.008)
indicating that persons who score high on Distrustful tend
to display more disgust than do persons who score low on
this scale. A significant difference was also found be-
tween high and low groups of Getting Along With Others on
the affective measure Range of Affect (p=.028), indicating
that persons who score high on Getting Along With Others
tend to display more types of affect than do persons scoring
low on this scale.

Supplementary analyses did indicate further differ-
ences between groups: the independent variables Aggressive,
Gregarious, Distrustful, and Getting Along With Others all
showed evidence of (non-hypothesized) significant dfiienames
between their respective high and low groups on at least
one affective measure at p<.05.

The discussion section noted implications for this

study and for future research, as well as methodological

Frances W. Stott

observations and recommendations for studies which use

moving facial affect in the sampling procedures.

To my parents,

Wideha and Russell Stott

who long ago provided me with the
affective basis for my interest in this re-
search area,

who instilled values of intellectual
curiosity and academic achievement that have
served me in undertaking this degree program,

and most importantly, who now celebrate
with me in the pride and satisfaction of its

completion.

ii

ACKNOWLEDGMENTS

As this dissertation is the culmination of a degree
program in Counseling Psychology, so it began not with a
research proposal but at the Listening Ear Crisis Inter-
vention Center of East Lansing, Michigan. Consequently, I
wish first to acknowledge June Jacobson and Dozier Thornton,
founders of the Center and initiators of my enthusiasm and
understanding for applied psychology. I would like to ack-
nowledge the depth and richness of their personal friend-
ships for me, and to thank them for the values of personal
integrity, professional excellence, and caring that they
imparted to us all.

I wish also to express many thanks to Bill Hinds,
who was my first formal instructor in graduate psychology,
who supported my application to this degree program, who
served as academic advisor and dissertation chairperson
throughout, and whose friendship I value highly.

I would like to acknowledge and thank each of my
committee members, each of whom has had personal impact on
my graduate career, and all of whom have offered thoughtful
criticisms as well as interest and support toward this
research project:

Bill Hinds, who introduced me to the area of facial

iii

affect, graciously endured the agonies of long distance
consultation, and responded both expediently and thoroughly
to letters, calls, forms, and drafts:

Judith Taylor, who was singularly indispensable in
her ability to make this research area a manageable one,
to make research in general an intriguing and rewarding pro-
cess, and in her generosity of time, expertise, personal
interest, and an invaluable sense of humor;

Dozier Thornton, who consistently offered studied
and helpful suggestions as well as a personal interest in
my research efforts;

Joanne Hamachek, who conscientiously critiqued and
consistently shared her enthusiasm;

Bill Farquahr, who thought through with me in the
important, initial stages and has maintained a supportive
stance throughout.

In addition, I would like to acknowledge Tim
Sadler of Birkman and Associates, who provided computer re-
sources and helpful suggestions along the way.

As this research was conducted on the University of
Texas-Austin campus, there were numerous persons from that
University involved with its progress. I would especially
like to acknowledge Ira Iscoe of the Psychology Department

who mobilized local resources, acted as a local sponsor,

iv

and cared fervently that I was able to finish.

I would like to acknowledge the Departments of
Communication and Psychology, which allowed me to solicit
subjects from among their students; the College of Educa-
tion which permitted me the use of its videotape equipment,
and particularly Jim Spear, Royal Embree, and Coystal Stone
who facilitated that process.

The University of Texas Counseling-Psychological
Services Center allowed me the use of its facilities for
the interviewing and videotaping of subjects, as well as
for the drafting of the dissertation. I wish particularly
to thank David Drum as its Director, Gus Baron who oversaw
the project, and Rich Klein, who helped in setting up the
equipment and literally abandoned his office space to two
weeks of videotapes and clutter.

Frank Karass patiently interviewed all 62 subjects,
'and offered support and friendship, as always. Dusty Humes.
Karen Hill, and Elizabeth Sackton worked conscientiously as
raters, shared my enthusiasm for the research, and at times
restored it. Patsy Stice waded through the data with me,
helping to tabulate, compile, ponder, and yet maintain a
perspective sense of humor. My parents, Wideha and
Russell Stott, contributed vacation days in Austin to re-

cording data, proofing tables, and generally cheering me on.

Susie Stiegelbauer contributed her artistry by drawing
graphs and illustrations. To each, as well as to the 168
students who completed the instruments and the 62 who came
for interviews, I express thanks and appreciation.

To the many friends in East Lansing who were
pleased to note my progress, I express thanks as well as
delight in our continued friendships; and particularly to
Becky Hollingsworth and June Jacobson, who provided un-
wavering long distance encouragement, loved me throughout
the project (and sometimes despite it) I am most grateful.
Likewise, I express thanks to that small group of special
Austin friends who insisted that I take time out to enjoy
our friendships, and who generally kept my sense of warmth
and playfulness alive.

Finally, I would like to mention Roger Hall, who
had no direct contact with this research project, but who
has gently shaped my attitudes toward professionalism and
psychotherapy, and who has had profound impact on me and
on my life throughout this period of time. I wish to ack-
nowledge him for his clinical skills as well as his
artistry in his work as a psychotherapist, and to thank

him for valuing me as I have come to know and value myself.

vi

Chapter

I.

II.

III.

IV.

TABLE OF CONTENTS

INTRODUCTION . . . . . . . . . . . .
REVIEW OF LITERATURE . . . . . . . .

Emotion . . . . . . . . . . . . . .
Facial Expression of Emotion . . . .
The Scoring of Facial Affect . . . .
Affect Categories to be Used in This
Summary . . . . . . . . . . . . . .

RESEARCH DESIGN AND METHODOLOGY . .
Sample . . . . . . . . . . . . . . .
Instrumentation . . . . . . . . . .
Selection of Interview Sample . . .
Interview and De-briefing Procedures
The Interviewer . . . . . . . . . .

Videotape Apparatus Used During Interviews

Videotape Editing . . . . . . . . .
Rater Sample . . . . . . . . . . . .
Rater Training . . . . . . . . . . .
Rater Reliability . . . . . . . . .
Rater Agreement by Affect Category .
Research Design . . . . . . . . . .
Measurement of Facial Affect . . . .
Research Hypotheses . . . . . . . .
Statistical Analyses and Procedure .
Summary . . . . . . . . . . . . . .

ANALYSIS OF DATA . . . . . . . . . .
Results of Hypothesis Tests . . . .

vii

Page

13

13
24
40
44
55

58
58
60
68
69
72
73
74
75
76
80
82
91
92
94
95
96

100
101

Chapter

Further Analyses of Group Differences . . .
Multivariate Analysis of Variance . .
Further Analyses of Interest . . . . .

Correlational Analyses . . . . . . . . . .
Within Instrument Correlations . . . .
Between Instrument Correlations . . .

Correlations Within the Affective
Measures . . . . . . . . . . . . .

Correlations Between Test Instruments
and Affective Measures . . . . . .

Correlations Between Methodological
Measures and Affective Measures .

Summary . . . . . . . . . . . . . . . . . .

V. SUMMARY, CONCLUSIONS, AND DISCUSSION . . .
Summary . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . .

Findings . . . . . . . . . . . . . . .
Methodology . . . . . . . . . . . . .
Implications . . . . . . . . . . . . . . .
Suggestions for Further Research . . .

Implications for Therapy . . . . . . .

APPENDICES C O O O O O O O O O O O O O O O O O O O

A: Subject Consent Forms for Research
Participation . . . . . . . . . . .
B: Interview Materials . . . . . . . . . .
C: Rater Training Material . . . . . . .
D: Rater and Scoring Forms Used in
Measurement of Facial Affect . . . .
E: Supplementary Univariate Analyses . . .
F: 24 X 24 Inter-Correlational Matrix . .

BIBLIOGRAPHY O O O O O O O O O O O O O O O O O O 0

viii

Page

112
112
113
119
120

128

131

133

135
136

141
141
146
148
148
153
158
158
160

162

162
165
172

185
189
198

201

Table
3.1
4.1

4.8

4.9

4.10

4.11

LIST OF TABLES

Rater Agreement by Affect Category . . . . . .

Affective Scores for Subjects Selected for
Analysis on Plutchik's Personality Dimensions.

T-test Results for Between-Group Differences
on Affective Measures . . . . . . . . . . . .

Affective Scores for Subjects Selected for
Analysis on Birkman's Personality Dimensions .

T-test Results for Between-Group Differences
on Affective Measures . . . . . . . . . . . .

Affective Scores for Subjects Selected for
Analysis on Eysenck's Personality Dimensions .

T-test Results for Between-Group Differences
on Affective Measures . . . . . . . . . . . .

Multivariate Analysis of Variance for the
Independent Variables on the Set of Dependent
Measures 0 I O O O O O O O O O O O O O O I O I

T-test Results of Gregarious Group Differences
on Affective Measures . . . . . . . . . . . .

Affect Scores for High and Low Gregarious
Groups . . . . . . . . . . . . . . . . . . . .

T-test Results of Distrustful Group Differences

on the Affective Measures . . . . . . . . . .

Affect Scores for High and Low Distrustful
Groups . . . . . . . . . . . . . . . . . . . .

T-test Results of Aggressive Group Differences
on the Affective Measures . . . . . . . . . .

Affect Scores for High and Low Aggressive
Groups . . . . . . . . . . . . . . . . . . . .

Inter-Scale Correlations for PEPI . . . . . .

ix

Page
83

103

104

107

108

110

111

112

115

116

116

117

118

119

123

Inter-Scale Correlations for EPI . . . . . .
Inter-Scale Correlations for BM . . . . . .

Correlations of Scales from Plutchik with
Scales from Eysenck and Birkman . . . . . .

Correlations Between Scales of the EPI and
the BM 0 O O O O O O O O O C O O O O O O 0 0

Correlations Within the Set of Affective
Measures 0 O O O O I O O I O O I I O O O O 0

Correlations Between Instrument Scales and
Affective Measures . . . . . . . . . . . . .

Correlations Between Methodological Measures
and Affective Measures . . . . . . . . . . .

Page
126
127

128

130

131

134

135

Figure

3.1

3.6

3.7

LIST OF FIGURES

Percentage of rater agreement across
62 subjects 0 I O O I O O I O O O O O O O O O 82

Third rater's category where two raters were
in agreement on Interest . . . . . . . . . . 86

Third rater's category where two raters were
in agreement on Distress . . . . . . . . . . 86

Third rater's category where two raters were
in agreement on Surprise . . . . . . . . . . 86

Third rater's category where two raters were
in agreement on Anger . . . . . . . . . . . . 87

Third rater's category where two raters were
in agreement on Enjoy . . . . . . . . . . . . 87

Third rater's category where two raters were
in agreement on Disgust . . . . . . . . . . . 87

Between-group differences from independent
variables on affective measures . . . . . . . 92

Plutchik's Emotion Circle . . . . . . . . . . 122

Inter-scale relationships for PEPI . . . . . 125

xi

CHAPTER I

INTRODUCTION

While philosophers, artists, scientists, and
poets have speculated for centuries about the phenomenon
called emotion, it has only been since the late 1800's
that the study has been approached from a systematic view-
point. Scientists of the evolutionary, physiological, and
cognitive traditions have each begun to crystalize theo-
retical frameworks and compile empirical data in the area.
Others have approached emotion from a psychometric tradition,
and have worked to refine instruments that will yield data
regarding the individual's affective or emotional orienta-
tion. As new data continue to emerge, there has been
recent speculation about the use of facial emotion as a
psychodiagnostic tool.

In‘a similar manner, the face itself received
popular attention long before it became a topic of scien-
tific inquiry. The seventeenth century salons of France
for example, a popular gathering place of the literary set,
were noted for such leisure time activities as speculation
on facial type and corresponding personality characteris—

tics (Adam, 1972). In recent popular literature, books

such as Face Reading (Mar, 1975) undertake to predict per-

 

sonality characteristics by a complex system of facial
features, musculature, and bone structure. It is apparent
that the face is a rich source of information; yet it re-
mains unclear how we might systematically study and use
this information for research purposes or therapeutic in-
tervention. Only recently has empirical data begun to
emerge that provides direction as to how we might use ob-
servations of the face for these purposes.

The facial expression of emotion was introduced
by Darwin and has continued to attract researchers and
theorists toward its exploration. Only in the last twenty
years has the research and theoretical activity become suf-
ficient to produce what is now an emerging body of liter-
ature. Based largely on the work of Sylvan Tomkins (l962,
1963 , 1965, 1970) who has worked to establish a theoretical
basis for emotional experience and psychological functioning
with some speculation on facial relatedness, Paul Ekman and
Carroll Izard are currently in the forefront of research in
facial affect. Each has worked independently to establish
systems for identifying emotions through facial expressions,
to consolidate and establish major categories of facial af-
fect, to train others in the recognition and identification

of these categories, and Izard, to use these categories as

a basis for discerning psychological information about the

individual.

Theory

The three theoretical assumptions upon which this
study is based are as follows: 1) that the face provides
immediate and specific information regarding human emotions:
2) that each fundamental emotion has a characteristic fa-
cial expression recognizable as distinct from the others;
and 3) that the fundamental human emotions have a psycho-
logical relationship to the individual's personality.

That the face provides immediate and specific in-
formation regarding human emotions is in part based on evi-
dence that facial expressions are instinctive and occur re-
flexively or automatically as a part of the experiential
emotion process (Izard, 1977). While the adult may not be
as spontaneous nor as automatic in expression as is the
infant, this later inhibition is the result of learning and
experience, and does not contradict the assumption that fa-
cial expressions are instinctive. Evidence that emotions
are hereditary and bear a close resemblance to the expres-
sions of nonhuman animals (Darwin, 1910; Izard, 1971)
would seem in further support of these instinctual and re-

flexive qualities.

Studies of infants and mothers indicate that the
face is an extraordinarily important social stimulus, and
plays a vital role in the early bonding that takes place
between mother and child (Tomkins, 1962; Izard, 1971,
1977). For human beings, evidence indicates that facial
patterns as communication cues have gained independence and
in many cases priority over posture, locomotion, and envi-
ronment (Ekman, Friesen and Ellsworth, 1971; Izard, 1971).

The assumption that each fundamental emotion has
a characteristic facial expression recognizable as distinct
from the others has led researchers to investigate in one
of two ways: to assume the existence of the discrete ex-
pressions and seek to determine the cause and role of the
neuromuscular activity involved (Ekman, 1965); or to dis-
cover the degree of consensus which could be obtained in
judging discrete facial expressions (Izard, 1971). Subse-
quently, Ekman has mapped the face, and Izard has estab-
lished consensus.

In an effort to establish the cross-cultural valid-
ity of these discrete expressions, the work of Ekman and Izard
has shown that certain emotions have the same expressions
and experiential qualities in widely different cultures
and throughout the world. While it is generally agreed

that each culture has its own idiosyncratic rules for when

or under what circumstances an emotion can be appropriate-
ly displayed, and rules for how to inhibit or mask an emo-
tion, it is nevertheless apparent that when the emotion

is expressed, its corresponding facial expression is sim—
ilar across peoples. Percentages of agreement upon dis-
crete categories of facial affect range from 47 percent

to 95 percent (Ekman, 1973a; Izard, 1977).

That fundamental human emotions have a psycho-
logical relationship to the individual's personality is
based largely in the work of Tomkins and Izard. Tomkins
(1962, 1963) maintains a distinction between emotions and
drives, and the phenomenon he describes as the ideo-af-
fective organization. This ideo-affective organization
consists of a dynamic and relatively stable relationship
between an affect (emotion and drive) on the one hand,
and certain cognitive processes such as ideas and beliefs
on the other.

Tomkins assumes that emotion is not a global,
undifferentiated experience but rather that there are dis-
crete emotions which interrelate in an overall emotions
system, and that this emotions system regulates the drives
as well as other personality subsystems. In this way,
the emotions are both independent of cognition and inter-

act with it.

Izard also speaks to the interaction between
emotion and cognition, and conceptualizes it as an af-
fective-cognitive structure. This psychological organi-
zation of affect and cognition he believes is a trait-like
phenomenon that results from repeated interactions between
a particular affect or pattern of affects and a particular
set or configuration of cognitions (Izard, 1977). A com-
plex affective-cognitive structure, or an interrelated
set of structures, he posits, may constitute an affective-
cognitive orientation, or a more global personality trait,
trait complex, or disposition such as introversion.

In combination, these theoretical assumptions
suggest that the face is a reliable source of information
about human emotion, that it is possible to distinguish
which emotion is being expressed on the basis of facial
display, that these emotions have a psychological relation-
ship to the individual's personality, and that, therefore,
relationships between one's emotional experience and one's
personality traits may be studied through observation of

the facial expression of emotion.

Need

 

While there is growing evidence that an individ-
ual's emotional experience is accurately (and perhaps

most fully) expressed in his face, and while there are

years of recorded speculations about the relationship be-
tween emotional orientation and personality characteristics
there is yet very little that addresses the question of re-
latedness of facial activity to personality. To establish
this relatedness would mean that a whole range of psycho-
therapeutic possibilities and research questions currently
unavailable might become promising to pursue.

If there is evidence that the type of facial affect
one displays is related to personality dimensions, for ex-
ample, then it becomes crucial to note the client who
nearly always looks disgusted, or angry, or is smiling,
and to begin exploring this predominant affect as a diag-
nostic tool. Similarly, for the client who has a limited
range of affect display, and whose face never expresses
more than two or three of eight basic affects, it may be
possible that the observation can be used psychodiagnostically.

While research in the area of facial affect has
demonstrated relationships between personality type and
ability to identify facial affect in others, there is only
minimal research to indicate a relationship between per-
sonality dimensions for corresponding expressions of facial
affect. No study thus far has attempted to explore a vari-
ety of personality dimensions for corresponding expressions

of facial affect. Thus itwould seem that an initial study is

necessary to determine whether these questions of relat-
edness are even promising to pursue. In the event that
meaningful relatedness can be established, then it would
also seem promising to later explore the possibility of
identifying and defining basic affective personality
types from the facial affect perspective, as well as

their corresponding treatment modalities.

Purpose

The primary purpose of this study is to explore
the assumption that information about personality charac-
teristics related to emotional style is expressed in the
face. The secondary purpose of the study is to investi-
gate whether methods used in this study appear adequate
for use in subsequent research efforts in the field of
facial affect.

The study will investigate whether there are
significant differences in expressions of facial affect
as recorded on videotape and measured by trained judges
between persons with specific personality characteris-
tics as measured by pencil/peper personality instru-
ments. The specific questions that will be addressed

are: lx>persons on specified personality dimensions

vary in the types of facial affect they express; Do

persons on specified personality dimensions vary in

their frequency of change from one affect to another;

and Do persons on specified personality dimensions vary

in the range (or number of types of affects) they express.
The methodological purpose of the study will

respond to the problems of measuring, quantifying, and

of recording facial affect. The study is intended to

contribute to that literature by describing any further

refinements in techniques and procedures, and will con-

tinue to address the issue of whether current technology

permits meaningful data in this area.

Hypotheses

 

The 15 hypotheses used in this study are
designed to test the differences in expressions of facial
affect between persons of specific personality types.

This will be done by comparing groups of persons with
these specific personality traits on affective measures
designed to indicate type, range, and amount of affec-
tive change displayed. These hypotheses are theoretically
based in the literature pertaining to facial affect,
particularly in the work of Tomkins (1962, 1963, 1965 and

Izard, 1971, 1972, 1977), and exploratory in nature.

10

The following are the research hypotheses,
which are stated generally again in Chapter III, and for
which the results are reported in Chapter IV.

H1: The group mean of subjects who score high
on the Gregarious scale will be significantly higher
on the affective measure Enjoy than will be the
group mean of subjects scoring low on the Gregarious
scale, corresponding to the predominance of
gregariousness they experience.

H2: The group mean of subjects scoring high
on the Distrustful scale will be significantly
higher on the affective measure Disgust than will
be the group mean of subjects scoring low on the
Distrustful scale, corresponding to the predominance
of distrustfulness they experience.

H3: The group mean of subjects scoring high
on the Depressed scale will be significantly higher
on the affective measure Distress than will be the
group mean of subjects scoring low on the Depressed
scale, corresponding to the predominance of distress
they experience.

H4: The group mean of subjects scoring high
on the Dyscontrolled scale will be significantly
higher on the affective measure Surprise than will
be the group mean of subjects scoring low on the
Dyscontrolled scale, corresponding to the predom-
inance of dyscontrol they experience.

H5: The group mean of subjects scoring high
on the Aggressive scale will be significantly higher
on the affective measure Anger than will be the group
mean of subjects scoring low on this scale, corres—
ponding to the predominance of agression they
experience.

H6: The group mean of subjects who score high
on the Getting Along With Others scale will be signif-
icantly higher on the affective measure Range of
Affect than will be the group mean of subjects scor-
ing low on this scale, corresponding to emotionality
and a need for contact with others.

11

H7: The group mean of subjects who score high
on the Getting Alone With Others scale will be
significantly lower on the affective measure Disgust
than will be the group mean of subjects scoring low
on this scale, corresponding to the minimization of
rejection toward others.

H3: The group mean of subjects who score high
on the Getting Along With Others scale will be
significantly higher on the affective measure Enjoy
than will be the group mean of subjects scoring low
on this scale, corresponding to a need for incorpora-
tion and social bonding with others.

H9: The group mean of subjects who score high
on the Getting Along With Others scale will be
significantly higher on the affective measure Change
of Affect than will be the group mean of subjects
who score low on this scale, corresponding to
expressed emotionality and a sensitivity toward
others.

H10: The group mean of subjects who score
high on the Dominance scale will be significantly
lower on the affective measure Distress than will be
the group mean of subjects who score low on this
scale, corresponding to need for authority and
control, and a minimization of sensitivity to
distress.

H11: The group mean of subjects who score
high on the Dominance scale will be significantly
lower on the affective measure of Change of Affect
than will be the group mean of subjects who score
low on this scale, corresponding to a need for con—
trol in emotional expressiveness.

H12: The group mean of subjects who score high
on the Extraversion scale will be significantly
higher on the affective measure Enjoy than will be
the group mean of subjects who score low on this
scale, corresponding to their desire for inclusion
and social bonding.

H13: The group mean of subjects who score high
on the Extraversion scale will be significantly lower
on the affective measure Disgust than will be the
group mean of subjects who score low on this scale,
corresponding to the need to include rather than to
reject or push away from.

12

H14: The group mean of subjects who score
high on the Neuroticism scale will be significantly
higher on the affective measure Change of Affect than
will be the group mean of subjects who score low on
this scale, corresponding to emotional liability as
measured by this scale.

H15: The group mean of subjects who score high
on the Extraversion and Neuroticism scales (EN) will
be significantly higher on the affective measure
Range of Affect than will be the group mean of sub-
jects who do not score high on both of these scales
(ES, IS, IN), corresponding to emotional liability
and freedom of experiences.

Overview

In Chapter II, the literature pertinent to the
theory and research of facial affect is presented, with a
focus on the areas of emotion, the facial expression of
emotion, the scoring of facial affect, and the specific
affect categories to be used in this study. Chapter III
describes the subject sample, the instrumentation for
personality dimensions, the research procedures, the mea-
surement of facial affect, the hypotheses, the analyses,
and the experimental design. The analysis of data for
each hypothesis is presented in Chapter IV, along with
the results of supplementary analyses used for explora-
tory purposes in this study. Chapter V includes a sum-
mary of this investigation, a discussion of the findings
and the limitations, and implications for psychotherapy

and for further research in the area of facial affect and

personality dimensions.

CHAPTER II
REVIEW OF LITERATURE

This study focuses on the relationships between
personality dimensions and the expression of facial af-
fect. In order to provide the theory and supporting re-
search upon which this study is based, literature will
be reviewed in the areas of emotion, the facial expres-
sion of emotion, the scoring of facial affect, and for

the specific affect categories to be used in this study.
Emotion

Definitions and Major Theorists

 

The phenomenon we describe as human emotion has
for centuries been described, identified, researched,
and debated and has, as yet, eluded any definitive or
conclusive definition. Even the distinction between an
emotion and a feeling is a clouded one, with its own
historical and evolutional characteristics (Candland,
1977, p. 4). Despite this, however, emotion continues
to be a concept central to psychological theorists and
consistently utilized by the practicing psychotherapist.

The DSM II (as an indicator of current trends in

13

14

psychiatry) devotes an entire section to "major affective
disorders" and lists eight diagnostic categories within.
Freud, Jung, Adler, Horney, Ellis, Rogers, and Perls

each in some way incorporates the emotional experience
into his or her theoretical approach and psychotherapeu—
tic goal. And while strict behaviorists such as Wolpe
focus specifically on the cognitive aspect for treatment,
they are nevertheless striving to modify the affective
experience of fear or anxiety.

Thus usage of the term and utilization of the
concept remain consistently central to many areas of
modern psychology, despite the lack of clarity or
agreement about the phenomenon to which we refer as
"emotion." Several theorists have suggested that the
term "emotion" refers, in all probability, to a variety
of experiences and processes, yet undifferentiated from
one another (Candland, 1977; Izard, 1971; Kaywin, 1966;
Arnold, 1960) which serves to perpetuate the divergence
of thinking in the area. Additionally, however, lack
of agreement does not imply fault, but rather reflects
the complexity of the area and the fact that much re-
mains to be learned. Candland (1977) asserts that
there are yet two major disagreements: how emotion is

to be defined; and (given our models of mind and

15

behavior) what is the nature of its logical structure
(pg. 12).

With regard to these questions, three major
schools of thought have emerged over the last century
as theorists have addressed the problem from the evolu-
tional, the biological, or the cognitive perspective.
Each will be briefly summarized with regard to the ma-
jor theorists and the primary contributions to the
field of emotion.

The evolutional theorists. Of primary signifi-
cance in this area is the work of Charles Darwin, and

particularly his book entitled The Expression of Emotion

 

in Man and Animals (1910). The work was intended to
show that the ontogeny and phylogeny of emotion were
susceptible to the principle of natural selection,
though with the obvious shortcoming that there is no
historical data from which to trace an evolutionary de-
velopment. While Darwin was not the first to propose
that characteristics may change slowly over time by a
process of natural selection, he was clearly the most
influential in his observations and writings of this
phenomenon.

The impact of the evolutional approach has been

to assume a survival value for finely differentiated

16

facial expressions and bodily postures, and attempt to
determine the reason for their survival value (Candland,
1977). The work of Chevalier-Skolnikoff (1973) and of
Izard (1971, 1972, 1977) is clearly influenced by this
approach as they have sought to establish the functions
and adaptive features of the expression of emotion in
both primates and humans.

The biological theorists. While precursors of

 

William James (including Aristotle and Descartes) looked
toward the body and used physiological functioning as a
primary source for understanding and explaining emotion,
it was James (circa 1890) who brought physiological
theories to the forefront within both the lay and scien-
tific communities (Candland, 1977). The predominant
characteristics of James' approach were the emphasis on
the peripheral component of the emotional experience,
and the emphasis on Darwin‘s work which incorporated the
principle of natural selection. In the peripheral vs.
central argument, James assumed that it was the peri-
pheral body states which were initially altered, and
then searched for accompanying changes in reported emo-
tional state (James, 1890). In addition to pOpulariz-
ing and returning interest to the physiological ap—

proach in the study of emotion, James leaves us with two

17

major questidns: are there physiological correlates of
emotion; and if so, do we experience them before, after,
or during the perception.

There has been considerable research interest in
the area of physiological correlates of emotion (Black,
1970; Brady, 1970; Arnold, 1960a, 1960b, 1970; Ax,
1953), further stimulated by the fact that investigators
have consistently found some relationship between the
emotional state and the physiological state. This rela-
tionship, however, remains unspecific. It has not yet
been possible to correlate a precise emotional state, or
change from one specific emotion to another, with a mea-
surable physiological change.

The cognitive theorists. Proponents of this ap-
proach hold that psychological understanding comes from
examining consciousness, or objects as they are con-
sciously perceived. In an attempt to meld psychologi-
cal principles with newly discovered evolutional princi-
ples, or "to seek for principles and causes of mental
evolution in man" (Romanes, 1884), George Romanes de-
fined the mature mind as comprising emotion, will, and
intellect.

It was Freud (1938), however, who spoke to the

interrelatedness of these processes, and presented the

18

mind as a coherent whole, a body perceiving, apprehend-
ing, appraising, and acting as a cognitive agent. While
Freud is not best known as a theorist or researcher in
the area of emotion, his influence on the cognitive the-
orists has nevertheless had longlasting effects.

Schachter and Singer (1962) represent a contem-
porary version of the cognitive school of emotions by
defining emotion as undifferentiated arousal plus cogni-
tion. In their view the underlying physiological state
is the same for all emotions, and qualitative distinc-
tions result from cognitive appraisal or evaluation of
the situation that elicited the arousal.

More recently, however, Plutchik has developed
an argument for interaction of cognition and emotion,
and in his work "Cognitions in the Service of the Emo-
tions" (1977), he carefully develops his thesis that
cognitive functions have evolved "in the service of
needs and emotions" (pg. 190). The argument is based
on the premise that sensory stimulus is the primary emo-
tional experience, but that evaluation of the stimulus
event has led to an intricate and sophisticated system
of cognition-affect relations. Congruent with princi-
ples in his basic theory of emotions (Plutchik, 1962),

he develops this argument from an evolutionary

19

standpoint and in a context that includes all phylogen-
ic levels of animal life. Thus Plutchik presents both
emotional and cognitive processes as evolntknmrihradap-
tive as well as vital to the organism, but in a manner

which emphasizes their reciprocal nature.

The Measurement of Affective Orientations

Still another approach to the study of emotion
has been to assume a theoretical position and then at-
tempt to test it within the framework of personality
measurement. Allen and Hamsher (1974) for example dis-
tinguished three aspects of emotionality: responsive-
ness (intensity of affect), expressiveness (interper-
sonal communication of affect), and orientation (atti-

tudes toward emotion) and constructed a Test of Emotion-

 

al Styles on this basis. Other instruments such as the

 

Minnesota Multiphasic Personality Inventory have assumed
a theoretical stance pertaining not so much to what emo-
tion is, but rather how to distinguish normal from ab-
normal populations on the basis of emotion related be-
havior, and then pursued the usage of those distinctions
as a psychodiagnostic tool.

Of particular interest to this study are three
instruments which utilize very different constructs but

all of which attempt to measure affective orientation in

20

some manner: the Eysenck Personality InventQEY: the

 

Birkman Method, and Plutchik's Emotions Profile Index.

 

 

The Eysenck Personality Inventory (based on the

 

earlier Maudsley Personality Inventory) comes directly
from the work of Heyman and Wiersma, who were among the
first to use the modern approaches of measurement and
calculation (Eysenck, 1969, p. 25). Similarly, Eysenck
comes from a tradition of personality measurement rather
than emotion theory; his primary focus was to describe
behavior and classify personalities by certain hypothet-
ical traits, while working to refine the empirical ap-
proach.

At an earlier point, Heyman and Wiersma had col-
1ected data on more than 2,000 individuals using a three-
dimensional rating system composed of: emotionality,
activity, and extraversion-introversion. In a later
statistical analysis of the data, Eysenck determined
that emotionality was relatively orthogonal but that ac-
tivity and extraversion were highly correlated. Subse-
quently he eliminated activity and maintained the dimen-
sions of emotionality or neuroticism, and extraversion-
introversion, thereby incorporating elements of affective
styles into his attempts to describe personality types.

As Eysenck points out, both his method and his

21

constructs are closely related to predecessors of the
Greek and Renaissance periods, who attempted to classify
man on the basis of temperament (sanguine, melancholic,
choleric, and phlegmatic), and to Wandt who took the
fourfold division and proposed two major divisions:
strength of emotion, and speed of change of emotion.
Wandt's dimensions became emotional-non-emotional, and
changeable-unchangeable, and were based on an assumption
similar to Eysenck's: that it is desirable to reduce
the observed correlations between large numbers of
traits to a smaller number of more fundamental dimen—
sions or all-embracing types (Eysenck, 1969, p. 14).
Also from a measurement background, Birkman
(1961) undertook to develop a personality inventory
that would 1) survey personality differences using new-
ly constructed social and self-perception questionnaires,
and 2) provide data for analyzing these differences on
the basis of positive or negative social perceptions.
Working from a theoretical point of View that
assumes that behavior is a function of the meaning the
individual assigns to stimuli, and that there are two
pivotal points of positive and negative perception
around which personality organizes (Birkman, 1961, p.

5), Birkman has constructed an instrument which

22

classifies the individual on the basis of twelve perfor-
mance traits. The data from the social vs. self-percep-
tion dichotomy and the positive vs. negative social per-
ceptions generated by the instrument are used to be able
to predict interpersonal needs both in normal situations
and in situations when the individual is under stress.

As with Eysenck, Birkman enters the literature
on emotion from the standpoint of a psychometrist rath-
er than as an emotions theorist. He incorporates emo-
tional dimensions into the instrument by blending them
with behavioral and attitudinal characteristics in order
to define his basic personality traits. The self-con-
sciousness dimension, for example, is a blend of emo-
tionality, sensitivity to others, achievement orienta-
tion, and nervousness around superiors. Birkman's in—
strument, then, uses personality dimensions that are in-
terpersonal in nature and provide a collection of state-
ments about the individual's affective orientation in
combination with a variety of other statements about his
interpersonal style.

In contrast to Eysenck and Birkman, Plutchik is
an emotions theorist and offers an instrument entitled

Emotions Profile Index, which is based on his own pre-

 

viously noted theory of emotion. It is designed to

23

measure the relative importance to each individual of
what he calls prototypic patterns of emotions. These
eight prototypic patterns, so named because the behav-
iors are identifiable at all phylogenetic levels of ani-
mal life, are identified by terms such as destruction,
protection, rejection, reproduction, exploration and
orientation. These terms are intended to identify be-
havior patterns that are basic to all species in re-
sponding to stimuli, and suggest a clear evolutional
function: "prototypic patterns of behavior are 2922'
tiye and help the organism in its struggle for surviv-
al" (1965, pg. 106).

Plutchik further proposes that "all other emo-
tions are combinations of these few primary ones, just
as all colors are mixtures of a few primary colors"
1965, p. 296). In addition, the prototypic patterns have
been labeled in such a way as to emphasize the existence
of bipolarities: destruction vs. protection, or incor-
poration vs. rejection. In situations where basic reac-
tion patterns are aroused at the same time, for example
destruction (or anger) and protection (or fear), a con-
flict develops. In general, says Plutchik, "situations
which arouse Opposite action tendencies generate con-

flict, and tend to produce inhibition or immobilization"

24

(p. 106, 1966). The therapeutic implications suggested
here are to respond to the emotional conflict, either by
discerning it from the situation presented, or by iden-
tifying it on the basis of the client's profile from
Plutchik's Emotions Profile Index.

These instruments, the Test of Emotional Styles,

 

the Minnesota Multiphasic Personality Inventory, the E17

senck Personality Profile, the Birkman Method, the

 

Plutchik's Emotions Profile Index, are representative

 

of current efforts to establish personality characteris—
tics or affectively based interpersonal styles on the
basis of the individual's emotional experience or affec-
tive orientation. While some address the issue of "what
is emotion" more directly than others, all attempt to
differentiate types of people into a manageable number
of categories which describe traits or behavior related

to the emotional experience.

Facial Expression of Emotion

 

As with the phenomenon of emotion itself, the
concept that facial expressions are related to internal
emotional experience has been assumed for centuries but
elusive of systematic scientific investigation. The

arts have made use of the concept despite its lack of

25

scientific validation. Delsarte's system for training
actors relied heavily upon observations about individual
parts of the body and the moods, attitudes, and emotions
expressed by the specific position of the minutest de-
tail. To express fury or madness, for example, the ac-
tor was trained to lower the brow portion nearest the
nose, and to raise the brow near the outer edge (Steb-
bins, 1886, pg. 145). To express a neutral feeling,

the eyeball must be "calm and midway between the two
corners" (Stebbins, 1886, pg. 138).

Chekhov used his observations not to train oth-
ers but to describe them. In his short story "The Kiss,"
Chekhov describes personalities at a dinner party by giv-
ing them types of faces: a flat face, a clever face, a
well-fed face, and a face with forced smiles. Even the
central character is revealed to us in this manner, as
"one whose face seemed to say...'I am the shyest, most
modest, and most undistinguished officer in the whole
brigadel'" (Chekhov, 1965, pg. 1052).

A contemporary of Delsarte and Chekhov, it was
Charles Darwin who first proposed from the scientist's
standpoint that facial expressions were related to in-
ternal emotional states, and set about describing and

recording his observations in a manner that has stimulated

26

scientific investigation up to the present day.

Darwin and The Expression of Emotion in Man and Animals

Published subsequent to his famous Origin of the

 

Species, Darwin's The Expression of Emotion in Man and

 

Animals applied the principles of evolution and natural
selection to his notions about facial expressions of emo-
tion, suggesting that they had a purposeful function and
survival value for the organism. In this later work,
Darwin presents three principles of expression: 1) that
serviceable actions become habitual in association with
certain states of the mind (and the subsequent passage

of habits into reflex actions); 2) the principle of anti-
thesis (or development of some behaviors simply because
they are opposite to others rather than because they have
functional value of their own); and 3) the principle of
direct action of the excited nervous system on the body
(or involuntary expression and behavior) (Darwin, 1910).
Using examples from observations of children, animals,
mentally ill, blind, and members of different cultures,
Darwin identifies more than thirty emotions and expres-
sions thereof, describing in detail the elements of the
expression and theorizing about the function and surviv-

al purpose of the emotion.

27

While it was nearly a century before the tenets
expressed by Darwin became active in empirical research
(Ekman, 1973b), it is nevertheless clear that Darwin's
observations and theoretical principles have served the
function of pioneer work in the field, and have had pro-

found impact on those who followed.

Current Theory and Research in Facial Expression

of Affect

 

Proceeding from Darwin's assumptions that emotion-
al experience and facial expressions are related, Sylvan

Tomkins undertook a two-volume work entitled Affect, Im-

 

agery and Consciousness wherein he identifies and de-
scribes all major affect categories and hypothesizes in
detail regarding the development of the specific affects
as well as their psychodynamic functions. These psycho-
dynamic functions will be addressed later under'Affect
Categories to be Used in This Study" and are to date the
primary theoretical basis for assuming a relationship
between personality characteristics and expressions of
facial affect.

It is Tomkins who introduces the term affect in-
to the literature, and uses the term as synonomous with

emotion. Tomkins conceives of affect as primarily

28

facial behavior, and secondarily as bodily (outer skele-
tal and inner visceral) behavior. When we become aware
of our facial and/or visceral responses, he hypothesizes,
we are aware of our affects (Tomkins, vol. 1, 1962).

Thus Tomkins attempts the logical extension of Darwin's
work, proceeding from biological on to psychological
survival.

Both an active theorist and researcher, Izard
has recently proposed that the facial expression of an
emotion can actually determine the quality of the emo-
tional experience. How specific, mixed, conflicted, or
vague the emotional experience is, says Izard, is relat-
ed to and/or determined by not only the visceral experi-
ence, but to what extent and in what manner the emotion
is expressed in the face (1972). Thus Izard's position,
along with that of Tomkins, emerges as compatible with
Darwin's as each explores and argues the adaptive and
functional purposes of emotion and the facial expression
of emotion. Izard goes one step beyond by arguing that
the expression of the emotion interacts with and shapes
the emotional experience itself, and thatit is the facial
expression which we must study and regard as the criti-
cal feature in the process.

In this light, Izard has conducted exploratory

29

studies into the functional purposes of facial affect by
surgically sectioning the facial nerve of rhesus monkeys
(rendering them incapable of facial display or emotional
expression in the face) and observing the subsequent ef-
fects on social status, mating, and child rearing (1971).
While the findings from these studies are reported as
preliminary, the subsequent increase in unusual behav-
iors, the loss of status among ruling females, and the
social changes made within the observed monkey community
point to this direction in research as promising.

Beyond the theories and research of Tomkins and
Izard, which seek to substantiate the relationship of
internal emotional experience and facial expression of
affect as well as to address the psychological and so-
cial functions of those facial expressions, much of the
research to date has centered around the identification
and validation of facial affect categories (presented
later under "The Scoring of Facial Affect"). An out-
growth of this research in affect categories has been
to study the characteristics of how people apply the af-
fect categories or are able to use them in identifying
facial affect in others.

Zlatchin, for example, compared a group of male

medical students to male and female Haight-Ashbury

30

residents and found that those persons who were more in-
volved, steadfast, and better adjusted in terms of group
norms had significantly higher overall accuracy in rec-
ognizing specific categories of emotion than were those
who were alienated, depressed, or withdrawn (Zlatchin,
1974).

Both Schiffenbauer and Zuckerman have contrast-
ed groups of males and females in their ability to label
facial affect in same sex and opposite sex faces, and
have found significant differences 1) in the females'
ability to label more accurately (Schiffenbauer, 1976),
and 2) in the attributions that males and females assign
to faces expressing unexpectedly hnfimse (non-normative)
emotion (Zuckerman et 31., 1976).

Dougherty, Bartlett, and Izard investigated the
pattern of perceptual dysfunction in schizophrenics by
comparing their judgments of facial affect to judgments
by normals. This study produced significant findings in
l) the differential number of times specific categories
were used by the two groups, and 2) the overall accuracy
of the two groups in recognizing specific affect cate-
gories (Dougherty et. 21., 1974).

Sackheim, Gur and Saucy have approached facial

affect from the standpoint of facial symmetry and found

31

that left side composites of the face were judged to ex-
press emotions more intensely than right side composites.
Findings of this study have stimulated speculations
about using the face as a mode for examining the nature
of functional brain symmetry as well as the neurologi-
cally based organization of emotions and emotional com-
munication (Sackheim, Gur, Saucy, 1978).

Current theoretical and research efforts in this
area, then, continue to seek refinement in describing the
relationship between internal emotion states and facial
expressions, and have begun to yield empirical data

about ability to use facial affect categories as well.
Facial vs. Other Nonverbal and Verbal Expressions

To what extent the face is a reliable and accur-
ate source of information has been investigated from a
variety of approaches. The question of context, or
whether knowledge of the situation precipitating the emo-
tion is necessary in order to make a correct interpreta-
tion or judgment of the facial expression, is one ad—
dressed by both Munn (1940) and again by Turhan
(1950), Munn developed two sets of slides, one
with a context background and one with the face alone,

and found no difference in the percentage of agreement

32

among his raters for the two sets. Turhan found that
judgments of emotion based on the isolated face were con-
siderably different from judgments of the same face when
viewed in a context, and concluded that judgments of
emotions depend heavily on the perception or interpreta-
tion of the total situation in which the stimulus face

is involved.

Izard (1971) regards the argument in an evolution-
ary-phylogenetic perspective, and posits that "the impor-
tance of facial communication as compared with postural
activity in emotion, and particularly in emotion differ-
entiation and emotion communication, increases with phy-
logenetic and ontogenetic development" (1971, pg. 189).
He illustrated this by citing the 1967 work of VanHooff,
suggesting that facial displays become somewhat more in-
dependent of posture and locomotion in the anthropoid
apes compared to the rhesus monkeys. Izard summarizes
that in humans, there is mounting evidence to indicate
that facial patterns as communicative cues have gained
considerable independence from posture, locomotion, and
context (1971, pg. 192).

Ekman (1965, 1967) found that in humans, posture
indicates something of the intensity and of the global

nature of the emotion, while the facial pattern is

33

characteristic of a particular discrete emotion and may
convey additional information with regard to an emotion
mixture or combination. In later, related experiments
he has tested the prominence of facial vs. body cues by
asking subjects to try to deceive judges (Ekman and
Friesen, 1969a, 1974a). The results from these studies
indicate that if a person is trying to disguise or hide
his feelings, he will attend most carefully to his face
as the area to be most controlled. In this situation,
observers are reported to judge emotions more accurately
from the body than from the face. In addition, Ekman
has investigated the congruence between facial expres-
sion and verbal statements in an interview setting, and
found a significant and reliable relationship between
the two. Thus Ekman's work suggests the face as the more
specific conveyor of emotion, but notes that it is also
the more readily disguised communication source, sug-
gesting not so much that it must be observed in context,
but that it is useful to interpret in combination with
other body cues.

Of particular research interest has been the
question of whether facial affect is a pan-cultural or
culture specific phenomenon. From Darwin's position of

discrete emotions, each having a specific, innately

34

programmed facial pattern, the argument is pro cross-
cultural or universal expressions. The research evidence
is mixed at this point, but there is strong evidence to
suggest that basic emotions such as anger, disgust, fear,
surprise, and enjoyment are experienced and labeled sim-
ilarly across cultures, while the more subtle emotions
such as bliss, worry, and upset are more culture specif-
ic. As Harrison (1975) points out, there is not this
same kind of agreement across cultures for "emblems" or
non-facial gestures, indicating that nonverbal gestures
with arms, hands, body posture, and physical proximity
have been clearly substantiated as culture specific.
The implicit relevance of this research is that work with
nonverbally expressed emotions in culturally mixed popu-
lations must rely heavily in facial expressions as op-
posed to body cues, and will probably have greatest ef-
ficacy if restricted to the emotions previously re-
searched and established as more basic to all cultures.
The face as an indicator of emotion, then, was
introduced by Darwin and substantiated by Tomkins, Munn,
Turhan, Izard, and Ekman, both in terms of the functional
purposes of the emotion expressed, and the importance of
the emotion expressed by the face. A proliferation of

research headed by Iaard and Ekman suggests that the

35

face, in or out of context, in combination with other
body cues, in congruence with verbal behavior, and as
cross-culturally similar, is in fact not only a reli-
able and appropriate place to study affect, but perhaps
one of the most important vantage points from which to

study it.

Affective Facial Display as a Psychodiagnostic Tool

 

While viewing hours of client videotape for
another purpose, Haggard and Isaacs stumbled across a
phenomenon they later called micro-momentary expression:
a facial expression so short-lived that it seems to be
quicker than the eye (1966, pg. 154). As they further
explored the phenomenon, establishing methods for re-
cording and measuring it and searching for meaningful
correlates, some interesting speculations regarding ap-
plications to psychotherapy began to emerge.

The rapid changes of MME's seemed to occur most
often, for example, when normal defenses were function-
ing effectively (periods of general expressiveness) or
in a conflict context (when denial statements or verbal
blocking were evident as identified by incongruence be-
tween the verbal content and the facial expression).

Based on these observations, Haggard and Isaacs

36

speculated that micromomentary changes appeared to be re-
lated to intrapsychic dynamics. This speculation was
further supported by correlating the number of MME's to
the manifest content with which they were associated
during a therapy session, and finding a positive rela-
tionship between the frequency of the two (1966, pg.
162).

Haggard and Isaacs propose that micro-momentary
expressions may be indicators of ego mechanisms, and ob-
servable in psychotherapy. They speculate that MME's
might, for example, be particularly frequent during ther-
apy when newly released but not yet integrated impulses
and affects are sporadically monitored by previously es-
tablished controls. During this phase of "transitional
instability," the individual might allow split second
expressions of new affect, resulting in an MME rather
than a fully expanded expression of emotion. If that
were, in fact, the case then an increased frequency of
MME's could be an indicator of therapeutic change in
process. While many of these speculations yet lack the
empirical evidence to substantiate them, there is never-
theless sufficient evidence to suggest that they are
promising directions in which to move.

Building from this work of MME's and ego

37

mechanisms, Wilson (1976a) investigated the relationship
between repression and facial affect, hypothesizing that
there would be a positive relationship between frequency
of micro-momentary expressions and other indicators of
psychological defense Operations. "To the extent" says
Wilson "that micro-momentary expression originates as a
manifestation of self-deception, to that extent is it the
result of repression-defense Operations and fits Freud's
view of repressive Operations” (1976b).

While the study did not yield as clear cut a re-
lationship as had been hoped for, there was nevertheless
sufficient evidence to indicate that "in clinical settings,
a high rate of emission of micro-momentary expressions
suggests an hypothesis of difficulty with the expression of
anger coupled with fear or anxiety over interpersonal safe-
ty" (1976b, pg. 7). In summary, Wilson concludes that
micro-momentary expressions appear, at this point, to be a
psychological trait related to trait defensiveness, and
that the frequency of emission seems to be related to the
expression of angry feelings.

Based on Kell and Mueller' 3 observations of compacted
affective experiences (1966) , Hinds explores this phenomenon
by positing the relationship between facial expressions and
intrapsychic conflict. "Compacted affect" says Hinds "is

the inhibition Of one affect by the expression of another,

38

resulting in the compression of the affective experience
and of its display on the face (1976, pg. 1). Hinds'
theoretical position is that not only thoughts but af-
fects serve as an internal stimulus fixsanxiety, which
consequently serve to maintain the individual's neurotic
behavior through negative reinforcement. The self-de-
feating behavior may be sustained, for example, in order
to avoid the more painful affect such as fear or anxiety.
Hinds' theory is that these painful affects are experi-
enced, but the experience is compressed so as to mini-
mize the discomfort; this compression of affect leaves
the individual without a full range of affective experi-
ences, thus limiting his behaviors, alternatives, and
problem solving capacity. Hinds notes the use of facial
affect as an important diagnostic tool for the therapist
in this area and points to the potential for using facial
affective patterning "as a means for therapists to dis-
cover how affects influence and maintain neurotic behav—
ior" (1976, pg. 3).

Finally, Izard's work on emotion, and particu-
larly on emotion as it is expressed in the face, has
pointed again and again to the face as a potential psy-
chodiagnostic cue. With regard to the symmetry of facial

motion, he reports that subjects under LSD treatment show

39

a considerable decrease of symmetry quotients, and that
movements become less extensive and synchronous for per-
sons under mental stress (1971, pg. 354). Additionally
Izard reports work using the assessment of expressions
and gestures as a basis for a prognosis in behavior ther-
apy with autistic children (1971, pg. 371); work with us-
ing the identification of facial affect in others as a
psychodiagnostic cue (1971, pg. 372); and his own work
with an electromyograph (EMG) on facial muscles to de-
lineate different types of emotional tensions as well as
in training subjects to become more sensitive to their
own internal cues (1971, pg. 391).

While there are yet no readily available or wide-
ly substantiated techniques or procedures for using fa-
cial expressions diagnostically, there are nevertheless
individual and isolated evidences that this is a promis-
ing area to explore. Haggard and Isaacs, Wilson, Hinds,
and Izard all point to psychodiagnostic information po-
tentially available through the face could be useful and
is to some extent unavailable through other diagnostic
measures.

Technological issues (to be discussed in "The

Scoring of Facial Affect") are troublesome. To date,

40

the only instruments for recording facial data are video
(still photos, movies, and videotapes) and physiological
(as in the electromyograph); the equipment is cumber-
some, expensive, and yields data that is time consuming
to judge. In addition, the theoretical arguments and
experimental results are clearly at an exploratory level
in this area. Nevertheless the theoretical framework
and technological data are already providing information
useful to the trained eye of the psychotherapist, at

least for his or her own speculative purposes.

The Scoring of Facial Affect

 

General Background

 

As the problems of measuring, judging, or scoring
facial affect are reviewed, it becomes pertinent to re-
turn once again to the consideration of what an emotion
is. Since this study is focused specifically on the fa-
cial expression of emotion, considerations and defin-
itions at this point will be based on the literature in
that area rather than the literature on emotion in gen-

eral.

41

Izard offers a definition, however, that attempts a com-
parative perspective as well as a specifically useful
guideline:
Emotion is a complex concept with neurophysiolog-
ical, neuromuscular, and phenomenological aspects.
At the neurophysiological level emotion is de-
fined primarily in terms of electrochemical ac-
tivity in the nervous system, particularly in
the hypothalamus, the limbic system, and the fa-
cial trigeminal nerves....At the neuromuscular
level emotion is primarily facial activity and
facial patterning, and secondarily it is bodily
response. At the phenomenological level emotion
is essentially motivating experience and/or ex-
perience which has immediate meaning and signifi-
cance for the person (1972, pg. 185).
Emotion in the face, then, particularly as it can be de-
scribed and identified on a neuromuscular basis, will
be the context in which facial judgments are presented.
The early literature in this area
includes the names of Langfeld (1918), Rumrick (1922),
Frois-Wittmann (1930), Woodworth (1938), Schlosberg
(1941), Munn (1940), Hanawalt (1944) and Coleman (1949),
and argues issues such as whether or not emotions are
discrete and unique expressions, and if so, how they are
to be labeled or classified. Frois-Wittman worked to
identify the role of facial muscles in emotion identifi-
cation; WOodworth's contribution was to simplify the

categorization process by reducing the number of cate-

gories from 110 to six. Munn, Hanawalt and Coleman

42

focused on the reliability of judgments by researching
judgments made with and without contextual cues, and
mouth region vs. eye region judgments.

Building on the work Of these earlier research
efforts, the two most widely regarded writers at this
time are Carroll Izard and Paul Ekman. Izard (1965,
1971, 1972) has worked extensively to provide theory in
the field, but additionally has developed a categoriza-
tion system Of facial affect as well as having contrib-
uted greatly in the areas of applied research. Ekman
(in Ekman and Friesen, 1965, 1971, 1975) has been less
concerned with theory, but has made significant contri-
butions through procedures for the observing and judging

of facial behavior.

Carroll Izard

 

Izard's primary contribution to the scoring of
facial affect has been a set of research labels or affect
categories, established to a large extent on the basis
of his cross-cultural research and additionally "by the-
oretical Observations, common-sense Observations, pre-
vious descriptions from the literature, and from collabor-
ative work with Tomkins" (1971, pg. 248). The eight

categories he uses are: interest/excitement; enjoyment/

43

joy; surprise/startle; distress/anguish/ disgust/con-
tempt; anger/rage; shame/humiliation; and fear/terror.
Izard describes his initial efforts in establish-
ing these categories as attempts to establish the exis-
tence Of discrete fundamental emotions common to all man-
kind. He relied heavily on the use of emotion labeling,
showing photographs to members of four different cul-
tures and collecting data on the basis of the free-re-
sponse labels generated by the photos. The initial
study provided 244 different words or free-response 1a-
bels, which were sorted and reduces into eight categor-
ies, compatible with the same categories mentioned above.
He has subsequently used procedures of emotion
recognition, emotion labeling, and attitudes toward emo-
tions in establishing agreement within these categories
across cultures. Enjoyment/joy typically has the high-
est percentage of agreement, ranging from 71-90%; shame/
humiliation has the lowest percentage at 7.2%. The other
six categories usually have from 30 to 90% agreement,
with several marked male/female and cross cultural dif-
ferences (Izard, 1971). The development of these cate-
gories, as well as the high degree of agreement elicited
by them, provides an empirical argument for the existence

of fundamental or basic emotions as well as a firm basis

44

for continued research.

Paul Ekman

 

Ekman describes his early work in the area of
encoding and decoding of affect as having revolved
around the construction of "an Atlas of the face" to de-
pict photographically each of the universal facial ex—
pressions of emotion. He photographed models who were
instructed to move particular facial muscles listed in
the Atlas table, and separately photographed the three
areas Of the face capable of independent movement (brow/
forehead, eyes/lid and root of nose, and lower face).
Subsequently he made comparative studies with these pho-
tographs to study whether or not the Atlas was accurate,
and in addition studied videotapes, isolating and mea-
suring all muscular movements, and making independent
measurements for each of the three facial areas. This
work has provided the basis for all subsequent study of
categories of facial emotion.

A second contribution of Ekman's was the identi-
fication and description of management techniques for
the control of facial behavior. Ekman says there are
four management techniques: intensifying an emotional

expression; deintensifying it; neutralizing it; or

45

masking it by displaying another emotion instead. These
management techniques are used in combination with dis-
play rules, which are norms that dictate with whom and
in what circumstances these management techniques are to
be used. Display rules are based on static characteris-
tics of the person (age, sex, physical size); static
characteristics of the setting (ecological factors and
social definitions of the situation); transient charac-
teristics of the person (role, attitude); and transient
regularities during the course of the interaction (en-
trances, exits, periods of conversation and of listen-
ing). Display rules are established primarily on cul-
tural and personal (idiosyncratic and familial) norms.
The ability to manage the face, and the variety of pro-
cedures a person might use clearly presents a problem to
the researcher trying to judge emotion in the face. Ek-
man suggests that individual differences in knowledge Of
display rules could be explored through self-report by
subjects as a way to determine to some extent what man-
agement techniques the subject may be using (1975, 138-
39).

Ekman has, at this time (1975) established six
categories of facial affect for use in research studies:

happiness, sadness, surprise, fear, anger, and disgust.

46

These categories, says Ekman, have been found "by every
investigator in the last 30 years who sought to deter-
mine the vocabulary of emotion terms associated with fa-
cial expressions" (1975, pg. 22).

Using these six labels as the primary basis for
research, much of Ekman's earlier research focused on
judgment studies, where facial behavior was treated as
the stimulus and questions were hypothesized with respect
to the observers. From these as well as other research
experiences, he speaks to the problems of sampling and
establishing generality while recording and measuring
the face.

With regard to sampling the behavior of the face,
it is crucial that the sampling procedures be reported
says Ekman, or the related question of how 93223 the
face provides accurate information (and for what kinds
of observers, emotions, and circumstances) cannot be an-
swered (1972, pg. 41). With regard to the sampling
across persons, Ekman emphasizes the need for represen-
tative sampling in order to avoid error due to either
morphological characteristics or differing ability to
show certain emotions.

Establishing generality, for either posed or

spontaneous circumstances, must consider issues across

47

circumstances, persons, time, decoding skills. Issues
that Ekman presents, for example, are whether the find-
ings in one eliciting circumstance or setting would be
valid for another such circumstance or setting. If the
spontaneous behavior elicited is in a laboratory setting,
then the researcher must provide a discussion of the real
life events to which this lab eliciting circumstance are
relevant. Issues of generality across persons include
considerations of whether the findings are general to
most people, or just to trained persons such as actors.
He reports that generality has been severely limited
when the posers have been actors, but not when untrained
persons pose emotions (1972, pg. 21).

With regard to generality across time, the re-
searcher must consider whether his investigation has
chosen a sample from an infrequent moment when the face
showed something, or whether the facial behavior shown
in the situation provides precise information for many
points in time. Decoding considerations refer to how
readily other observers could make the same judgments.
Did the study use specially trained or gifted persons
in this area, for example, or were enlargements or
slowed motion required? If so, these elements will lim-

it the generality of who could make these observations.

48

Beyond generality, Ekman introduces another as-
pect which adds to the difficulty of making accurate
judgments: the phenomenon of blended emotions. Affect
blends are thought to occur when l) the emotion-elicit-
ing circumstance by its very nature elicits more than
one feeling, or 2) habits (common to a group, or idio-
syncratic) link the elicited emotion to another, as for
example when a second emotion is generated in response
to the initially inspired one (1972, pg. 25). The fre-
quent finding that observers disagree about which of two
affects is present "can no longer be interpreted as evi-
dence of low information in the face, but alternatively
as the consequence of presenting a multiple message
stimulus and allowing the Observer only a single mes-
sage judgment" (Ekman, 1972, pg. 25). This issue of
blended affects represents a yet unresolved problem in
the area of facial affect measurement, but is necessar-
ily one that must continue to be addressed.

As indicated in this section, the field of fa-
cial affect measurement is relatively young and faces
many unresolved problems and issues. There remains a
lot of work to be done. Nevertheless there is a history
of early research that explored the problem of where and

how to begin, and has provided a basis for later

49

refinements and continually developing new methodolo-
gies.

Izard and Ekman have contributed enormously by
establishing categories and collecting empirical data
within that framework. Despite the fact that their
numbers of categories differ, and the category labels
are slightly varied, these differences do not make the
two systems incompatible; data generated by one set of
categories is Often applicable to the other.

Ekman's continued research and writing with re-
gard to methodological procedures has been of signifi-
cant importance to the field, and has been particularly
useful in considering research procedures for this

study.
Affect Categpries to be Used in This Study

In considering what affect categories to use in
this study, literature from Tomkins, Ekman, and Izard
was reviewed as well as recent dissertation literature
from Michigan State University. From Wilson's study
(1976b) for example, it is clear that not all affects
are likely to occur during an interview setting, so that
fear, despite its high percentage of rater agreement,

is for this study not a useful category for hypothesis

50

building because it is not likely to occur in the experi-
mental setting. Inman (1976) provides additional infor-
mation on the training of judges and percentage of agree-
ment on categories used. He found (as has Izard) that
shame has a very low agreement rate, thus while shame
would be an appropriate and desirable category for tes-
ting against personality dimensions in this study, it
will not be used because of the apparent difficulty in
judging the emotion. The six affect categories to be
used are: Interest; Anger; Disgust; Surprise; Enjoyment

and Distress.

Interest

The literature on interest (Izard, 1971; Tom-
kins, 1962; Plutchik, 1962) suggests that this affect is
a central one, yet it is also one Of the most commonly
used for masking other emotions as a facial management
technique (Ekman, 1975). Tomkins states that the func-
tion "of this very general positive affect is to 'inter-

est' the human being in what is necessary and in what is

 

possible for him to be interested in" and continues by
pointing out that "to the extent to which interest is
attenuated later in life the individual thereby ceases

to develop perceptually" (1962, pg. 342). Because it is

51

a more general affect, however, the category interest
will not be treated as an affective category to be test-
ed, and consequently will not appear in the formal
statements of hypotheses.

For the purposes of this study, interest will
be considered a neutral category and will be labeled
only in the absence of an expression from one of the
five other categories. This should provide some indica-
tion of the subject's usage and/or experience of inter-

est without penalty to ratings in the other five areas.

Anger

Tomkins speaks of anger only as it can be a
learned substitute for distress. The type of innate
stimulation which activates distress, he proposes, is the
same type Of stimulation that activates anger. If the
experienced distress is unrelieved for a period of
time, "it can produce sufficient increment of stimula-
tion to innately activate anger, then this sequence may
be telescoped so that the beginning of the distress cry
becomes the learned activator Of anger" (1963, pg. 64).
Whether or not the individual then experiences bggh emo-
tions, in a chaining sequence similar to Hinds' proposal

(1976), or whether the substitution takes place so rapidly

52

that only the anger is experienced, is not clear.

Darwin refers to anger as one of the more excit-
ing emotions, particularly with regard to intensity,and
proposes that it (like other emotions) is so closely
connected to its expression that the emotion can hard-
ly exist if the body remains passive (1910, pg. 237).
The function of anger, Darwin postulates, is to prepare
oneself against perceived or anticipated danger. He
notes examples of animals bearing the canine tooth, and
humans compressing the mouth while frowning the brow,
as if in warning to the perceived enemy. While anger
may not be expressed as typically in an interview set-
ting as in other environments, it is nevertheless thought
that the subject will display it as frequently in this
setting as he would be likely to express it in everyday

interaction.

Disgust

Disgust, says Tomkins, is designed to prevent
ingestion Of noxious material or to achieve its total
rejection and regurgitation if it has been ingested.
This is as true for psychic or psychological material
as it is for the more physiologically based process.

Its function is to guard against any type of

53

incorporation or increase in intimacy with the person/
object found to be noxious.
Clearly this is a distancing technique, although
Tomkins points out that the experience of disgust can
become central to the person, as opposed to an indepen-
dently stimulated reaction:
When the experience of dngust is recurrent and
becomes central, there develops a cognitive
elaboration which organizes these experiences in-
tO a relatively unified theory. The theory sen-
sitizes the individual to contempt-relevant in-
formation and provides ready-made strategies for
coping with these paradigms (1973, pg. 250).
Disgust is usually a more passive emotional experience
than is anger, thus it is more possible for a person to
sustain the expression facially for long periods of time.
It is anticipated that disgust will be expressed fre-

quently by some subjects during the interview sessions,

and that it will be comparatively easy to measure.

Surprise

Surprise is neither a positive nor a negative
affect by its own virtue, though a value is usually
associated with it according to the stimulus event. Surprise
functions as an interrupter to ongoing activity, so that
the person may attend immediately to new information.

It functions neurologically as a circuit breaker, and

54

according to Darwin is similar to anger in that it al-
lows for preparation. It is believed that there will

be a relationship between persons who frequently experi-
ence and/or allow themselves to express surprise and
persons who find Plutchik's concept Of dyscontrol impor-

tant or central to their lives.

Enjoyment

 

Enjoyment is a social bond, says Tomkins, learned
in early infancy through interactions with the mother.
The smiling of the infant is quickly reinforced, and the
socialization process begins. Children who are reared
with a minimum of social interaction and Of smiling will
ordinarily become less social adults (Tomkins, 1972,
pg. 404). An additional feature of the smile (a primary
expression of enjoyment) is that it frees the individual
to give positive rewards or feedback without the neces-
sity of body contact; thus it is a method of social bond-
ing that has versatility. Smiling is also a frequently
used technique for facial management (Ekman, 1975),
though the choice of this expression as a mask can of

itself be a statement about the individual.

55

Distress

Distress, says Tomkins, is a natural part of the
human condition; the differences are not whether but in
how the individual experiences and expresses distress
(1973, pg. 48). He outlines four modes of distress de-
velopment, ranging from the healthy, flexible stance
where it can be expressed when experienced, to the "ice-
burg mode" characterized by denialthat becomes disrup-

‘ tive in crisis situations when the denial no longer suf-
fices. Apart from the psychodynamic implications, dis-
tress in this study will be defined and measured by ex-
pressions of sadness, worry, or the more general dis-
tress musculature. It is usually a more difficult emo-
tion to purposefully mask, and is anticipated that it

will be relatively easy to measure.

Summary

Though not conclusively defined, human emotion
has been studied from the evolutional, biological, and cog-
nitive perspecitives, as well as from the psychometric
standpoint, with each of these traditions offering its own
contribution to an overall understanding of this phenom-

enon .

56

Charles Darwin was the first widely recognized
scientist who prOposed a relationship between facial ex-
pressions and internal emotional states. From the stand-
point Of natural selection, Darwin began to identify
categories Of emotion on the basis of facial expressions,
and to speculate on the functional or survival value of
each.

Subsequent to Darwin, Sylvan Tomkins deveIOped
a theoretical framework for the emotions system as it
interplays with other physiological and psychological
mechanisms, which includes the psychodynamic functions Of
the different affects and the relatedness of facial ac-
tivity to the experience of these affects.

Based largely on the work of Darwin and Tomkins,
other researchers such as Ekman and Izard have worked to
more systematically classify categories of affect, develOp
labeling and recognition techniques for these categories,
and generate empirical data about the labeling process as
well as about the cross-cultural evidence for these funda—
mental categories. A recent development in this research
area was stimulated by Haggard and Isaacs (1966), with the
discovery that the expressions of facial affect may have
a demonstrable relationship to certain aspects Of psycho-

logical functioning, such as the presence of ego neonmfisms.

57

The scoring Of facial affect continues to be
problematic, partly due to the subjective nature of the
task, and partly due to the technological procedures, which
for the most part are cumbersome and costly. Systematic
procedures have been developed, nevertheless, to facilitate
this process. Carroll Izard has worked extensively to
continually refine fundamental categories of facial af-
fect and to validate as well as establish reliability for
the use of these categories. Paul Ekman has focused more
on a study of the face itself, identifying specific muscle
groups involved in the expression of different affects,
and proposing guidelines for the sampling, generalizing,
and accurate defining of expressions Of facial affect.

In deciding upon which facial affect categories
to use in this study, considerations Of reported rater
reliability and liklihood of occurance for each affect
category were made. The categories of interest, anger,
disgust, surprise, enjoyment, and distress were selected
to be used as the dependent measures for the study. Ad-
ditionally, however, it should be noted that interest is
considered a neutral category and does not appear in the

formal statements of hypotheses.

CHAPTER III

RESEARCH DESIGN AND METHODOLOGY

This chapter is divided into 15 sections which
cover the following areas: sample, instrumentation, se-
lection Of interview sample, interview and de-briefing
procedures, the interviewer, videotape apparatus used
during interviews, videotape editing, rater sample, rat-
er training, rater reliability, rater agreement by affect
category, research design, measurement of facial affect,
research hypotheses, statistical analyses and procedure,

and summary.

Sample

The sample for this study consisted of one hun-
dred sixty-eight undergraduate students enrolled at the
University of Texas-Austin during fall term, 1978. Ap-
proximately twenty percent Of these students were solic-
ited from Communications classes, and participated on a
strictly volunteer basis. The remaining 80 percent were
solicited from the Department of Psychology undergradu-
ate subject pool. Psychology students who signed up for
this particular research study received three hours of

experimental research participation credit toward the six

58

59

hours required by their course. The only screening cri-
terion used within this population was visibility of
face; persons who wore glasses or who had beards were
not selected for participation in the study.

Of the 168 subjects who completed the pencil/
paper instruments, 97 percent of them were between the
ages of 18 and 21. The remaining 3 percent were between
the ages of 22 and 30. Fifty-five of these subjects were
males; 107 were females.

Subjects were told that they would be partici—
pating in a study designed to investigate the relation-
ships between personality characteristics and problem
solving styles. The study was presented as a two phase
process: during phase I subjects would be asked to come
plete three pencil/paper personality inventories; phase
II would consist of a 15 minute interview which would be
videotaped, wherein the subject would be asked to talk
briefly about something that was of personal concern.

It was explained that not everyone taking the
pencil/paper instruments would be asked to participate in
the interview phase, and that subjects selected for par-
ticipation in the interview would be on the basis of the
experimenter's need for a balanced number of different

types of subjects. Subjects were also informed that

60

expressed benefits of participation in the study were
not contingent on their participation in the interview
phase; namely that all subjects were eligible for one
test interpretation of their choice as well as a writ-
ten description Of the results of the study upon its com-
pletion. All subjects were asked to fill out a Subject
Consent Form (Appendix A), as well as a form for Birkman
& Associates (Appendix A) allowing the firm to release
their test scores Unthe experimenter. Subjects were in-
structed that they could discontinue participation at
any time during the course Of the study.

After testing, and based on test scores of bal-
anced groupings, the total number of subjects selected
to participate in the interview phase of the study was
62. This group was composed of 22 males and 40 females.
Approximately 22 percent of this sub-sample was from the
Communications pool; the remaining 78 percent from the
Psychology Department pool. The racial and ethnic break-
down for this group was 57 Anglos, 1 Black, and 3 Hispa-
nics. The breakdown of academic status was 35 freshmen,

16 sophomores, 7 juniors, and 4 seniors.

Instrumentation

 

Three pencil/paper personality instruments were

61

used in the study: the Plutchik Emotions Profile Index,

 

the Birkman Method, and the Eysenck Personality Inven-

 

tory .
The Plutchik Emotions Profile Index (EPI) was

developed by Robert Plutchik and Henry Kellerman (1974)
and is based directly upon the general theory of emotion
developed by Plutchik (1962). The theory postulates
eight basic emotion dimensions: Protection (or Timid),
Destruction (or Aggressive), Reproduction (or Gregari-
ous), Reintegration (or Depressed), Incorporation (or
Trustful), Rejection (or Distrustful), Exploration (or
Controlled), and Orientation (or Dyscontrolled). The EPI
is designed to assess the relative importance of each of
these eight emotions in a person's life.

The EPI consists of 62 items, in a forced-
choice format. The test is based on 12 trait terms which
are paired in all possible combinations, and which yield,
in the final results, an "emotion circle" with a separ-
ate score for each of the eight basic emotions. The nor-
mative data presented by the manual is based upon 500 men
and 500 women. The data include college students,
housewives, nurses, motor vehicle inspectors, and public
school teachers, and represent "a broad range of individ-

uals characterized by a lack of overt pathology or

62

hospitalization" (Test Manual, pg. 2). Test-restest re-
liabilities for each scale are reported at over +.90.
Split half reliabilities for individual scales range
from +.6l to +.90.

Validity has been established through correlat-
ing individual scales to scales on the MMPI, the Edwards

Personal Preference Schedule, the Gough Adjective Check

 

List, the Barrett Impulsivity Scale, and the Clyde Mood

 

Spglg. In addition, distinctions between specific pop-
ulations (hospitalized and normal women; matched groups
of normal, neurotic, and psychotic subjects) have been
illustrated by the EPI as further data for validity as-
sessment.

Due to the specific traits Plutchik uses to de-
fine the eight basic emotion categories, only five of
the eight categories or scales are facially measurable
by current technological methods and thus usable for

this study. The scales used are:

Aggressive (primarily expressed by anger)
Gregarious (primarily expressed by joy)
Depressed (primarily expressed by sadness)
Distrustful (primarily expressed by disgust)

Dyscontrolled (primarily expressed by surprise)
Since only five of the eight scales are used, and because
of the ipsative nature of the instrument, the composite

profile cannot be compared across subjects nor used for

63

questions regarding the range of affects displayed or the
frequency of change in affect. It is possible, however,
to use the individual scales to determine whether or not
an individual's facial display matches his reported fre-
quency of experience of that emotion.

The Birkman Method (Roger W. Birkman, 1974) is

 

an instrument based largely on the dissertation work and
subsequent research Of its author. Its theoretical and
philosophical orientation is in the Maslovian concept of
self-actualization. The focus of the instrument is to
provide individuals with sufficient information about
themselves from as many aspects as possible (both person-
al-social and vocational) so that they have an increased
awareness of self as well as pertinent information upon
which to base decisions about both setting and reaching
personal-social and vocational goals.

The format of the test is divided into three
parts: perceptions of self (forced choice, 125 items);
perceptions of others (forced choice, 125 items); and
preferred interests (48 sets of quadruple groupings of
first and second choice rankings). The instrument is
computer scored. The client is provided a computer
print-out which focuses on six major areas of social

needs: ability to get along with others; the way one

64

gives and accepts directions; how one handles conflict
and competition; how one maintains emotional and physi-
cal stamina; how one organizes and plans activities; and
the manner in which one makes effective decisions. In
addition, a rank ordering of one's occupational interest
areas is provided.

Reliability measures regarding internal consis-
tency have been tested through odd-even item correlations
(using the 120 items from both Self and Most People por-
tions of the test), and have produced correlations from
.78 to .95. Test-retest procedures have also been used.
Immediate (same day) test-retest correlations are report-
ed at .69 and above. Additionally, test-retest correla-
tions of forms completed two weeks apart show more than
half of the itemrby-item correlations above £’= .60, and
for almost all of the individual items, reliabilities
were significantly different from zero.

Efforts have been made to establish criterion-
related and content validity through subject self-report
and by consultation with industrial clientele. The pri-
mary research regarding construct validity has utilized
the intercorrelations between components of the Birkman
and of specific scales from a variety of other well known

standardized instruments. Personality and interest

65

instruments used in these studies include Cattell's Six-

teen Personality Factorguestionnaire (Form A), the

 

Strong Vocational Interest Blank for Males and Females,

 

the Edwards Personal Preference Schedule, the Eysenck

 

Personality Inventory (Form A), the Minnesota Multiphas-

 

 

is Personality Inventory, and the California Psychologi-

 

 

cal Inventory. The test is normed on an industrial pop-

 

ulation ranging from senior management to hourly workers.

For the purposes of this study, two components
of the Birkman Method will be used: 1) "Getting Along With
Others" (a combination of the Self-Consciousness and
Feelings Scales), and 2) "Need for Authority" (the Domi-
nance Scale).

Persons who score high on the "Getting Along
With Others" dimension typically exhibit a high aware-
ness of others, a need for positive attention and em-
pathy from others, difficulty in self-assertion, and a
tendency toward shame over errors. Persons who score
low on this component are more generally insensitive and
aloof toward others, less reflective, and prefer more de-
tached, logical relationships.

Persons who score high on the "Dominance"
dimension tend to seek and exercise firm authority

or control. They are assertive, directive,

66

demanding, and argumentative. Those who score low pre-

fer autonomy, independence, and pleasant relationships,

are uncomfortable exerting authority, and generally non-
assertive and unable to discipline subordinates.

The Eysenck Personality Inventory (EPI) was
developed by H. J. and Sybil B. G. Eysenck (1963) . It is a revi-
sion of the earlier Maudsley Personality Ingentgry, and
has two primary scales: E (extraversion-introversion)
and N (neuroticismrstability). In addition, the EPI con-
tains a Lie Scale which was developed and adapted from
MMPI, but will be used in this study only for the purpos-
es of descriptive statistics. The test consists of 57
items in a forced choice (yes-no) format. The test is
published in both a British and an American edition. The
manual for the American edition provides percentile
norms, using American college students as the standardi-
zation group. Test-retest reliabilities range between
.80 and .97, and correlations between forms A and B
range from .75 to .91.

The test reflects Eysenck's previous research
and theory, establishing extraversion-introversion and
neuroticismrstability as two pervasive, independent di-
mensions of personality. Factorial and construct valid-

ity are spoken to in part by correlations between the

67

EPI and the MMPI, and the EPI and the California Person-

 

ality Inventory. The E factor is regarded as extraver-
sion at the high end and applies to individuals tending
to be outgoing, impulsive, and uninhibited, having many
social contacts and frequently taking part in group ac-
tivities. The low score, reflecting introversion, re-
fers to the more quiet, retiring sort of person who tends
to be introspective, fond of books rather than of people,
one who has reserve and who is distant except to inti-
mate friends. This person tends to plan ahead, and to
distrust the impulse of the moment. He does not like
excitement, takes matters of everyday living with
seriousness and likes a well-ordered mode of life

(Buros, 1972).

The high end of the neuroticism dimension is in-
dicative of emotional instability and over-reactivity.
Persons scoring high on this factor tend to be emotional-
ly over-responsive and have difficulties in returning to
a normal state after emotional experiences. Such indi-
viduals frequently complain of vague somatic upsets of
minor kinds such as headaches, digestive troubles, in-
somnia, and backaches. Such individuals are also pre-
disposed to neurotic disorders under stress. Those per-

sons with 1ow scores on the neuroticism dimension tend

68

in general to be better adjusted and more emotionally
stable. The N and E dimensions are reported as essen-

tially uncorrelated.

Selection of Interview Sample

 

The three pencil/paper personality instruments
used in the study were scored before the interview phase,
in order to determine which of the subjects from the to-
tal sample would be selected to participate in this sec-
ond part. Subjects for the interview phase were select-
ed on scores from the nine specific dimensions that were
to be the formal measures used in hypothesis testing.
Those scales are as follows:

Emotions Profile Index Aggressive

Gregarious
Depressed

Distrustful
Dyscontrolled

 

Birkman Method Getting Along With
Others
Dominance

 

Eysenck Personality E (Extraversion/Intro-

Inventory, version)

N (Neuroticism/Stabil-
ity)

 

 

In order to determine who would participate in
the interviews, groups of high and low scores were es-
tablished for each Of the Birkman and Eysenck scales by

dividing the scores at the mean and considering all

69

scores that were 1/2 standard deviation above and below
the mean on each scale. In order to establish high and
low groups on the Plutchik scale, scores above the 70th
percentile and below the 30th percentile were consid-
ered.

The next step was to consider all subjects who
fell into either a high or low group on any of the
scales, and to identify subjects who fell into the great-
est number of high/low groups, so as to minimize the to-
tal number of subjects needed for interviews. An effort
was made to include all subjects who fell at the extreme
end of any one scale so as to maximize the differentia-
tion between high and low groupings. In addition, an
effort was made to proportionally balance the number of
subjects used for any one scale, so that the two groups
(high/low) did not exceed a 1.5 to 1 ratio.

Using these criteria, a total Of 62 subjects
were selected to participate in the interview phase of
the study. This group of 62 was composed of 22 males

and 40 females, as previously described on page 60.
Interview and De—briefing Procedures

The sixty-two subjects selected for interview

participation were asked to come, by scheduled

70

appointment, to the Counseling Center where the inter-

views took place. All interviews were scheduled within
a twelve-day period, approximately 10-30 days after the
initial testing took place.

The interview setting was a room that approxi-
mates the typical office Of a counselor at the Univer-
sity of Texas Counseling-Psychological Services Center,
including several chairs, a small table, and a carpeted
floor. The control Of the video-tape equipment was man-
aged from a room adjacent to the interview room, so that
the interview could be taped without interrupting the
interview itself.

At the beginning Of the interview, the subject
was presented with a stack of index cards containing in-
dividual items from the "Personal Psychological Rela-
tions" scale of the Mooney Problem Check List (see Appen-
dix B for a complete listing of these items). The sub-
ject was asked to look at each card until he or she
found an item of current interest or concern, and then
to talk with the interviewer about that item for 15 min-
utes (see Appendix B for standardized script used by
interviewer). In the event that the subject exhausted
the topic before the 15 minutes had ended, the inter-

viewer was instructed to suggest that the subject choose

71

another topic in the same manner described above.
Immediately following the interview, the sub-
ject was met by the experimenter, who guided him to a de-
briefing room. At that time, the subject was given a
self-report form to fill out (see Appendix B), inquir-
ing about the nature of the feelings the subject has ex-
perienced during the day as well as during the interview.
After this form.was completed, the subject was given a
written description of the specific nature of the study
(Appendix B). The experimenter de-briefed the subjects
by answering any further questions, and by inquiring
about whether they had now any reservations or concerns
about having participated in the study. At the end of
the de-briefing session, the experimenter re-stated the
nature of the study, the specific use of the video-tapes,
and requested that the subject sign the video-tape re-
lease form if he or she was willing to do so. (See Ap-
pendix B for this form.) This was also the point at
which subjects who were interested could sign up to par-
ticipate in a test interpretation session and/or to have
the results Of the study mailed to them. (Subjects who
did not participate in the interview were contacted by
mail after all interviews were completed for these sign-

up procedures.)

72

The Interviewer

 

The interviewer was the same individual for all
subjects. He is a 40 year old male, employed at the time
of the study as an intern at the University of Texas
Counseling-Psychological Services Center. He was also
at that time a graduate student in the Counseling Psy-
chology doctoral program at the University of Texas.

In order to maintain a consistency of response
across subjects the interviewer was instructed to engage
with subjects in a responsive but non-confrontive manner.
He was instructed to engage with subjects by reflecting
feelings, paraphrasing content, or indicating understand-
ing in a non-verbal manner as a way to encourage the sub-
ject to continue talking. He was not, however, to make
any attempts to direct the interview or to challenge the
subject. (See Appendix B for complete instructions to
interviewer.)

The spontaneity of the interview procedure and
possibility of contamination through the interviewer's
interaction with the subject was presented as a limita-
tion of the study. In order to minimize this limitation,
the interviewer was requested to participate in prelimi-
nary practice sessions, whereby he was videotaped while

performing in the interviewer role. Subsequently he

73

reviewed the tapes with the Experimenter, and received
feedback on the extent to which he has appropriately
maintained his role proscribed by the aforementioned
guidelines. Additionally, the interviewer was asked to
record brief anecdotal information about the nature of

the interview after each subject was interviewed.
Videotape Apparatus Used During Interviews

Two Panasonic cameras were used for taping dur-
ing the interview phase of the study. The two cameras
provided for a split screen image that later facilitated
the editing of the videotapes.

One camera, a Panasonic 16-54 mm zoom lens was
situated in the corner of the interview room and focused
for close-up shots of the subject's head and shoulders.
This camera was stationery, and adjusted only at the be-
ginning Of each interview. Subjects were seated in a
high backed chaise lounge so as to minimize head and
shoulder movement.

The second camera, a Panasonic 14-70mm zoom
lens, was Operated from the adjacent equipment room and
was focused on a digital clock with minutes and tenths
of seconds. This image appeared in the upper right hand

corner of the screen, and allowed for precise editing

74

of the tape at a later time.

The two Panasonic video cameras fed into a Pana-
sonic Special Effects Generator (SEG-WS-545P) which re-
sulted in the split-screen image. The interview was re-
corded in black and white on 3/4 inch cassetts video-
tapes, using a Panasonic NV-9200 3/4 inch cassetts vid-
eotape recorder. A11 videotaping procedures were per-
formed and monitored from the equipment room adjacent to

the interview room.

Videotape Editing

 

Each subject was videotaped throughout the full
15 minutes of the interview. The videotaped interviews
were logged by subject code number. Master tapes were
then prepared for rater viewing in the following manner.

Systematic sampling of the subjects' interview
was achieved by beginning from the point at which the
subject laid the deck of problem topic cards on the ta-
ble and looked up to begin talking with the interviewer.
Beginning at this point, a 3 second sample was taken
every 20 seconds, for a total Of 30 samples per subject,
or a total of 1,860 samples for the study. This proced-
ure allowed for discrete samples of facial affect, as

well as for 10 seconds of blank tape between each sample.

75

In order to minimize rater response set, the or-
der of subjects was randomly chosen, and subject samples
were edited onto the master tape in sets of five. Be-
cause of the large number of samples collected, a total
randomization was not attempted, but rather 3-5 subjects
were randomly chosen at a time, and alternatively sam-
pled in sets of five until the total of 30 samples from
each subject had been edited onto the master tape. Rat-
ers were told not to assume that there was a chronologi-
cal order either within the five samples per set, or in
the order of the sets that they viewed for each subject.

The master tapes onto which the videotaped same
ples were edited were 3/4 inch black and white cassettes.
The equipment used for this procedure was: two Sony tel-
evision monitors; two Panasonic videotape recorders (Mod-
el #NV 9200); a video editor controlled (Model #NV-A950);
a Panasonic editor recorder (Model #NV 9500); and a Mi-

crotime time base corrector (Model #1020).

Rater Sample

 

Three raters were used in this study to judge
the videOtaped facial expressions sampled from subject
interviews. The raters were not paid for time spent dur-

ing the training program, but were paid for time spent

76

rating the tapes. All three of the raters were female,
and all between the ages of 25 and 30. At the time Of
the study, two of the raters were employed by the Uni-
versity of Texas Telephone Counseling & Referral Service,
had undergone training for that position, and were exper-
ienced crisis and telephone counselors. One of these
persons was also a graduate student in the M.S.W. School
Of Social Work program at the University of Texas, as

was the third (non-TCRS) person.

Rater Training

 

The three raters underwent a rigorous training
program which was based on programs developed by previ-
ous researchers in this area at Michigan State University
(Inman, 1976; Wilson, 1976; Bowles, 1978). The training
program focused on the six categories used in this study
(Interest, Anger, Disgust, Surprise, Enjoy, Distress)
and included diagrams, photos, slides, and videotapes
using both posed and unposed samples of facial affect.
Since there is no standardized package program for train-
ing in the recognition of moving facial affect, and
since replicability Of the study depends largely upon
the ability to reproduce rater training and measurement

procedures, the rater training program is described in

77

detail.

The program began with a brief discussion of fa-
cial affect and an overview of research on the area, par-
ticularly the research regarding the development of fa-
cial affect categories. Izard's research (1971) report-
ing the categories most commonly overlooked by males and
those most commonly overlooked by females was noted and
discussed.

The six affective categories were defined, and
synonyms common to each of these affective labels were
presented and discussed (see Appendix C). A description
of each category as well as the dynamic function of the
category was also presented and discussed (Appendix C).
A diagram of facial musculature (Appendix C) was pre-
sented and the muscle groups involved with each affect
category were illustrated and discusSed. At this time,
slides illustrating posed faces in each of these affect
categories were also presented (Appendix C).

Once the raters became familiar with the six af-
fective categories, twenty-five slides of posed facial
affect developed by Ekman (1976) were shown in groupings
of five. Each slide was projected onto the screen for
six seconds, after which raters were given a lO-second

interval during which to write their judgments of the

78

category on an Affective Data Rating Form (Appendix D).
The six-second stimulus with a 10-second interval for
rating was used to approximate the timing sequence of
the master tapes to be used in the formal rating proced-
ure.

After each set of five slides was shown, random-
ly illustrating the affect categories used in the study,
the correct answers were given. The slides were then
shown for unlimited time periods, so that questions could
be raised and discussed. After all 25 slides had been
shown and discussed, a set of 15 more slides were shown
for three seconds each at lO-second intervals. The Af-
fective Data Rating Forms were used by raters to write
their judgment for each of the 15 slides, and then col-
1ected in order to calculate rater reliability. Raters
identified the same affective label for each of the 15
slides of posed facial affect with 95.5 percent agree-
ment.

The final phase of the training program involved
showing raters segments of unposed facial expressions of
affect from a pilot videotape, edited to be identical in
format to the master tapes which raters would later be
judging for the study. The procedure for practice with

rating moving facial affect was parallel to the procedure

79

used with the slides. The raters were first shown thir-
ty samples of unposed facial expressions, in sets of
five, and requested to write their judgments on an Af-
fective Data Rating Form. After each set of five seg-
ments were viewed, correct answers were given, and rat-
ers were encouraged to compare answers as well as to
raise questions. After all thirty segments had been
viewed, rated, and discussed, raters were then shown
thirty more segments and requested to write their judg-
ments on an Affective Data Rating Form. These forms
were then collected in order to calculate rater relia-
bility. The raters identified the same affective labels
for these thirty segments of unposed facial affect with
94.4 percent agreement.

Because Of the problems unique to judging mov-
ing facial affect, the raters devised a set of rating
rules based on Bowles' training program (Bowles, 1978)
and on issues that arose while rating the pilot training
tapes. Those rules are as follows:

1) If several categories of affect are Observ-
able, record the dominant or more intense
affect category.

2) If unable to determine which affect cate-

gory was dominant, record the first dominant

80

category of affect Observed.
3) Only use the category interest if no other
category of affect is clearly observable.

These rules were briefly written and taped to the bottom
of the video monitor throughout the rating of the master
tapes. In addition, each rater had a card on a table be-
fore her listing the six affect categories throughout
the rating procedure.

Because the raters were not able to complete
the rating task in a consecutive two day period, as had
been previously anticipated, the first of the two train-
ing videotapes (which was accompanied by an answer key)
was used to establish rater reliability at the beginning
Of each rating session. Raters were asked to come in be-
fore the beginning of the session and practice with the
tape until they had established 90 percent agreement with
the answer key. Despite the fact that practice effects
began to occur by re-using the same videotape, this pro-
cedure nevertheless did serve to standardize the rater
judgment process. The rating task was completed in four

sessions, or a total of 20 hours.

Rater Reliability

 

Raters were able to choose only one affect

81

category for each sample they viewed. One thousand,
eight hundred sixty samples were viewed and judged in
this manner, resulting in a total of 5,580 affective la-
bels assigned by the three raters. Samples which had no
agreement across raters were counted as discards, thus
every affect rating used had agreement by at least two
Of the three raters. A total of 130 samples were dis-
carded, which came to an average of 2.09 discards per
subject, or a total of 2 percent discard for the total
sample.

For the purposes of this study, inter-rater re-
liability was established by percentage of agreement.
This was calculated per subject by dividing the total
number Of ratings per subject (both discards and agree-
ments) into the number of agreed upon ratings for that
subject. The mean interrater reliability across sub-
jects was 77.3 percent, which is generally high for the
task of judging moving facial affect. It is of note
that these scores varied from 60 to 90 percent, however,
indicating that reliabilities for some subjects were
much higher than for others. Figure 3.1 presents a graph

based on percentage of rater agreement per subject.

82

100
90

T t 1 80

O a
Number 70
of 60
Subjects 50
40

30
20 w
10 l ‘ l
0
60- 65- 70- 75- 80- 85- 90-
64 69 74 79 84 89 94
Percentage of Agreement

llllllllll

 

 

 

 

 

 

 

 

Figure 3.1. Percentage of rater agreement across 62
subjects.

Rater Agreement by Affect Categpry

 

In addition to varying by subject, rater agree-
ment also varied by category Of affect. Table 3.1 pre-
sents, by affect category, the percentage of times a rat-
ing was used when all three raters were in agreement, the
percentage of times a rating was used when two Of the
raters were in agreement, the percentage of single rat-
ings which became unscoreable data, and the total per-
centage of ratings that could be used as a judgment in
the category. As indicated by Table 3.1, the category
Enjoy had the highest total number Of ratings (1,956)
followed by Interest, Disgust, Distress, Anger, and fi-

nally Surprise. In the case of the first four

83

Table 3.1

Rater Agreement by Affect Category

 

 

 

Percentages of Ratings Total
1 out percentage
3 out 2 out of 3 of usable

Category Ratings of 3 of 3 (discard) ratings

 

Interest 1,352 45% 33% 22% 78%
Distress 966 25 42 33 67
Surprise 56 22 14 64 36
Anger 121 15 33 52 48
Enjoy 1,956 71 20 9 91
Disgust 1,129 42 30 28 72

 

categories, the order of unanimous agreement Was the
same: Enjoy (71 percent); Interest (45 percent); Dis-
gust (42 percent); and Distress (25 percent). Anger and
Surprise reversed order, with Anger having only 15 per-
cent unanimous agreement and Surprise 22 percent. When
looking at the percentage of usable ratings, however,

the order is once again parallel to that in the total
number of ratings column: Enjoy (91 percent); Interest
(78 percent); Disgust (72 percent); Distress (67 percent);
Anger (48 percent); and Surprise (36 percent). Thus the

categories with the highest number of ratings also had

84

the highest overall percentage of agreement by raters.

Differences between columns two and three indi-
cate that while percentages of unanimous agreement (3
out of 3) were generally higher than percentages of two-
rater agreement, this is not the case with Distress and
Anger. Each of these categories has a higher percentage
of agreement in the third column, suggesting it was less
likely that all three raters would agree on the expres-
sion of this category, and more likely that only two out
of the three would agree. In contrast, the large gap be-
tween columns two and three for Enjoy (51 percentage
points) is also Of note, indicating that raters were far
more likely to agree on this category unanimously than
on a two—rater basis.

The column of single ratings (1 out of 3) indi-
cates high percentages for Surprise and Anger. Sixty-
four percent of the ratings for Surprise were single rat-
ings and subsequently not counted as a judgment; fifty-
two percent of the ratings for Anger were single ones and
likewise were lost as data toward the subject's affect
score in this category.

Surprise, with the lowest number Of total ratings
for the study, had a relatively high percentage Of unan-

imous agreement when compared to Disgress and Anger. Yet

85

it also had the highest percentage of single ratings for
all categories indicating that it had the highest amount
of disagreement among raters.

Finally, rater agreement was studied from the
perspective of disagreement. Figures 3.2 through 3.7
are presented by affect category, and designed to illus-
trate what affect category the third rater chose when
the other two raters were in agreement on one category.

As indicated by Figure 3.2, when two raters were
in agreement on the category Interest, the third rater
was most frequently in conflict by choosing the category
Distress (91 ratings), next most frequently in conflict
with the category Disgust (69 ratings) and least fre-
quently in conflict with the categories Anger (12 rat-
ings) and Surprise (6 ratings), suggesting in this case
that it was most difficult for raters to distinguish be-
tween Interest and Distress.

As indicated by Figure 3.3, when two raters were
in agreement on the category Distress, the third rater
was most frequently in conflict by choosing the categor-
ies of Interest (83 ratings) or Disgust (80 ratings),
and least frequently in conflict with the category Sur-
prise (3 ratings), suggesting in this case that it was

difficult to distinguish between Distress and either

86

100
90

80
70
60
50
40
30
20
10

ITIIIHITI

  
   

oe

 

 

Surp Anger Dsgst Enjoy Dstrs
Figure 3.2. Third rater's category where two raters
were in agreement on Interest. (23216)

100
90

80
70
60
50

40
30
20

10 '1'?-
0 . 55;: ,,,,,,,,,, . __ .................... 315-2’ W

Surp

nnnnnnnnnnnnnnn
...................

lleTll1||

 

 

Intrst Anger Dsgst Enjoy

Figure 3.3. Third rater's category where two raters
were in agreement on Distress. (2&202)

100
90
80
70
60
50
40
30
20
10

IITIIIIIII

 

    
 

qu ui‘v-vmw.~- A AAJAJ AA.- - a. A: e e I

Intrst Anger Dsgst Enjoy Dstrs

 

Figure 3.4. Third rater's category where two raters
were in agreement on Surprise. (ps4)

100
90
80
70
60
50
4O
30
20
10

87

flllllllll

 
   
   

 

Intrst Dsgst Enjoy Dstrs Surp

Figure 3.5. Third rater's category where two raters
were in agreement on Anger. (2321)

100

C

 

  

Dstrs Surp

Figure 3.6. Third rater's category where two raters
were in agreement on Enjoy. (33195)

100

90
80
70
60
50
40

30
20

10

p..-
—
.—
h—
.—
_
n.—
—
—-
r-

 

. .:.'.-
: 5.... ....... a . .;.

Enjoy Surp

 

Figure 3.7. Third rater's category where two raters
were in agreement on Disgust. (33171)

88

Interest or Disgust.

Figure 3.4 indicates that when two raters were in
agreement on Surprise, the third rater was in conflict
by choosing either Interest (2 ratings), Anger (l rat-
ing), or Enjoy (1 rating). Because of the small number
of 2-rater judgments in this category, it is difficult
to determine whether this pattern would be representa-
tive of a rating norm in a larger sample. The Figure
does indicate, however, that the categories Interest,
Disgust, and Enjoy were the most frequently disagreed up-
on in relationship to the category Surprise.

As indicated by Figure 3.5, when two raters were
in agreement on Anger, the third rater was in conflict
by choosing Interest (10 ratings), Distress (8 ratings),
or Disgust (3 ratings). As with the previous figure il-
lustrating Surprise, the number of 2-rater judgments in
this category is so small that it is difficult to deter-
mine whether this would actually become a trend in a
larger sample. It is indicated, however, that Interest,
Distress, and Disgust were more difficult to distin-
guish from Anger than were Enjoy or Surprise.

Figure 3.6 indicates that when two raters were
in agreement on Enjoy, the third rater was in conflict

by choosing either Disgust (59 ratings) or Interest (64

89

ratings), and least frequently in conflict by choosing
Surprise (10 ratings) or Anger (5 ratings), suggesting
that as for the category Distress, it is most difficult
for raters to distinguish between Enjoy and either
Interest or Disgust.

Finally, Figure 3.7 indicates that when two rat-
ers were in agreement on Disgust, the other rater was
most frequently in conflict by choosing the category Dis-
tress (78 ratings), and least frequently in conflict by
choosing the categories Anger (7 ratings) and Surprise
(4 ratings). This is in contrast to Figure 3.3 for Dis-
tress, where both Disgust and Interest had a high number
of ratings. Distress was the single category most fre-
quently in conflict when two of the raters were in agree-
ment on Disgust.

In general, Interest, Disgust, and Distress were
the three categories most frequently conflicting in rat-
er judgments, with Enjoy the more clearly defined both
by a higher percentage of unanimous agreement (Table 3.1)
and by fewer ratings in conflict with the other categor-
ies. The only exception to this was with respect to
agreement in the category Disgust, where Enjoy was the
second most frequently chosen of the conflicting cate-

gories.

90

Anger and Surprise, the two categories with the
lowest overall frequency throughout the study, occurred
in small numbers of single ratings throughout the other
categories. Anger, which had a higher total frequency
than Surprise, also generally had more single ratings in
conflict with judgments in other categories. Each Of
these categories had more single ratings than ratings
which could be used as a judgment for a sample.

As indicated in Figures 3.4 and 3.5, the only
two categories with entirely discrete ratings were Anger
and Surprise. Neither of these categories were chosen
by the third rater where the other two raters were in
agreement on either Anger or Surprise.

While inter-rater reliability was high during
the training sessions and with the training tape used at
the beginning of each rating session (90 percent or
above), the inter-rater reliability throughout the rat-
ing task varied considerably (from 60 percent agreement
to 90 percent agreement across subjects). It should be
noted, however, that the judgment of moving facial affect
is considerably more complex and difficult than is the
judgment of either posed or unposed facial affect in
still form. The trends which have emerged regarding the

variance of rater agreement across subjects as well as

91

across affect categories will be reviewed in the Discus-
sion section in an attempt to provide still more specif-

ic guidelines for this task in the future.

Research Design

 

The research design of the study was developed
to address a comparison of groups on the dependent mea-
sures. The purpose of this design was to determine
whether subjects in the high group on any one of the in-
dependent measures differ in expression of facial affect
from subjects in the low group on that same independent
measure.

The independent measures used in the study were

the Eysenck Personality Inventory, the Plutchik Emotions

 

Profile Index, and the Birkman Method, all described pre-

 

 

viously in this chapter. In addition to using a total

of nine independent scales from these three instruments,
a combination of the Eysenck Extraversion and Neuroticism
scales were used as a tenth independent variable.

The dependent measures used in the study were
eight categories of facial affect: Interest, Disgust,
Anger, Surprise, Distress, Enjoy, Range of Affect, and
Change of Affect. The design used for hypothesis test-

ing of group differences on these affective measures is

92

graphically displayed in Figure 3.8.

 

 

81* mean score
High 2 on dependent
Group é variable
20
821 mean score
Low 2 on dependent
Group é variable
40

 

 

 

 

Figure 3.8. Between group differences from independent
variables on affective measures.

*These numbers are for illustration purposes only. While
an attempt was made to have 20 subjects in each group,
priority was given to maintaining balanced groups (see
pg. 68).

Measurement of Facial Affect

 

The measurement of facial affect was designed to
take into account the types of affect categories dis-
played, the range of affect, and the frequency of change
of affect. The six types Of affect categories used in
the study are previously researched categories of emo-
tional experience, defined by specific facial expressions
(Ekman, 1975; Izard, 1971). Scores in these categories,
Interest, Disgust, Anger, Surprise, Distress, and Enjoy,
were determined by rater judgment. An agreement by two
of the three raters constitutes a judgment in any one

particular category. A disagreement across all three

93

raters on any one sample of facial affect constitutes a
discard of that sample and becomes non-scoreable data.

A subject's score for each affect category was based on
the number of rater judgments in that category. Since
the number of samples per subject would vary by the num-
ber of discarded samples, it was determined that the fi-
nal subject's score for each category would be calculat-
ed on a percentage basis: total number of samples in
the category divided by total number of samples used for
that subject. (See Appendix D for subject scoring
sheet.)

The category Range of Affect was determined by
the number of affect categories for which the subject re-
ceived at least one judgment. The highest possible score
was six. (See Appendix D for an illustration Of this
calculation.)

The category Change of Affect was determined by
the number of times a subject changed from one judgment
of affect to another. Discards were not: counted as a
change in affect. In the case where a subject received
a judgment in the category Surprise, followed by no
judgment because of a discard, followed by a judgment in
the category Interest, the change score for the sequence

would be one. (See Appendix D for an illustration Of

94

this calculation.)

In addition to these affective measures to be
used in the formal hypothesis testing, two more measures,
Percentage of Rater Agreement and Number of Discards
Per Subject, were also used to provide further descrip-

tive data.

Research Hypotheses

 

Fifteen hypotheses were generated to empirically
test the relationship between performance on specific
scales of the three pencil/paper personality instruments
and the type, range, or frequency of change of facial
affect displayed. 4

The null hypotheses may be generally stated as
follows:

Null Hypothesis: NO difference will be found on

the affective measure between the mean of the

group scoring high and the mean of the group
scoring low on the independent variable.

 

HO: "(1 =’“2
alpha level of .05
The alternative hypotheses may be grouped into
the following two categories:
Alternative Hypothesisl: The high group's mean

score will exceed that of the low group's mean
score on the dependent affective measure.

 

95

Hla: .q 1”“ 2
or

Alternative Hypothesis : The high group's mean
score will be less thafi the low group's mean
score on the dependent affective measure.

"2.21““( 1‘“ 2

Statistical Analyses and Procedure

 

Since all hypotheses were addressed as a compar-
ison of groups on the dependent variables, and since
equality of population variance, normality of distribu-
tion, and independence between each set of high/low
grups was assumed, all hypotheses were tested by means
of a t-test. The alpha level for each hypothesis was
set at .05, which is of concern to the overall alpha
level of the study and possibility for Type I error.

Due to the exploratory nature of the study, however,
it was considered more important to use a .05 alpha lev-
el for the individual hypotheses.

Also due to the exploratory nature of the study,
analyses that would provide descriptive data beyond the
formal hypothesis testing were also used. T-tests were
used to explore the nature of group differences between
all of the dependent and independent variables in the
study. The high and low groups on each of the indepen-

dent variables were subsequently tested for differences

96

on each of the dependent variables.

For exploration in the multivariate sense, a
multivariate analysis of variance was used to examine the
set of eight dependent measures as they relate to each
of the independent measures.

Finally, an inter-correlational matrix made up
of all independent and dependent variables as well as
additional scales from Plutchik and Eysenck and the two
additional rater affect measures was explored. Areas of
focus for correlational relationships were: relation-
ships within the scales Of each personality instrument,
relationships across the set of personality instruments,
relationships across the set of affective measures, and
relationships between the personality instruments and
the affective measures. Pearson Product Moment statis-

tics were used to estimate the correlations.

Summagy

One hundred sixty-eight undergraduate students
enrolled fall term at the University of Texas-Austin vol-
unteered to participate in this study. These subjects
were administered three pencil/paper personality instru-
ments in order to determine which of them would partici-

pate in the interview phase. Cut-off points for high

97

and low groups on each of the research scales were es-
tablished. Sixty-two subjects were then selected (22
males, 40 females) on this basis to participate in in-
terviews.

Each of the 62 subjects participated in a 15 min-
ute interview by choosing items from the Personal-Social

Relations Scale of the Mooney Problem Check List and

 

discussing that item with the interviewer. Data was
collected by videotaping the interviews and then syste-
matically sampling facial expressions by editing the
tapes to 30 3-second segments per subject. A total of
1,680 samples were used for the study, and organized in
semi-random order onto master tapes. Subject samples
were grouped in sets Of five and subjects were alternat-
ed by order Of appearance on tapes.

Three female raters were trained to recognize
and accurately label six affective categories Of facial
expression: Interest, Anger, Disgust, Surprise, Dis-
tress, and Enjoy. Affective data was then calculated by
subject for: percentage of each type of facial affect
category used; percentage of change from one category to
another, and total number of categories used. Percen-
tage of rater agreement per subject was also calculated

as well as total number of discards per subject, in order

98

to provide further methodologically related data for the
study.

The study was designed to address a comparison
of groups on the affective or dependent measures. Hy-
potheses were developed to test the relationship between
performance on specific scales of the three pencil/paper
personality instruments and the type, range, or frequency
of change of facial affect displayed. T-tests were used
to empirically test these relationships, with the proba-
bility of significance set at the p==.05 level.

Due to the exploratory nature of the study, ad-
ditional tests were used to further examine the relation-
ships between the independent and dependent measures
which had not been addressed in the hypothesis testing.
Additional t-tests between all independent and dependent
variables were run, as well as a MANOVA which examined
the set of dependent measures as it relates to each of
the independent measures.

Finally, an inter-correlational matrix was used
to examine the relationships of all scores from person-
ality instruments and affective measures, with a focus
on relationships within the scales of each personality
instrument, relationships across the set of personality

instruments, relationships across the set of affective

99

measures, and relationships between the personality in-
struments and the affective measures. The analysis of

this data is presented in Chapter IV.

CHAPTER IV

ANALYSIS OF DATA

This chapter is divided into three major sections
for the purpose of reporting the results of the investiga-
tion. The first section presents the results of the
hypothesis testing, the second section presents further
analyses of group differences, and the third section

reports correlational analyses.

Results of Hypothesis Tests

 

The fifteen hypotheses used in this study can be
divided into three groups, corresponding to the three
personality inventories used as independent measures.
The first group of hypotheses (1-5) is related to the

Plutchik Emotions Profile Index; the second group (6-11)

 

is related to the Birkman Method; and the third group

 

(12-15) is related to the Eysenck Personality Inventopy.

 

Following each group of hypotheses is the empirical data
germane to the retention or rejection of the null
hypothesis.

Hypotheses related to Plutchik's Emotions

Profile Index (PEPI) propose a relationship between the

 

100

101

frequency of a specific affect expressed facially and
the percentile score of that affective area on the
subject's PEPI test profile.

Hypothesis 1

 

Null Hypothesis: There will be no significant
difference on the affective measure of Enjoy
between subjects who score above the 70th
percentile on the Gregarious scale and the sub-
jects who score below the 30th percentile on the
Gregarious scale.

 

Alternative Hypothesis: The group mean of sub-
jects scoring above the 70th percentile on the
Gregarious scale will be significantly higher on
the measure Enjoy than will be the group mean of
subjects scoring below the 30th percentile on
the Gregarious scale.

 

Hypothesis 2

 

Null Hypothesis: There will be no significant
difference on the affective measure of Disgust
between subjects who score above the 70th
percentile on the Distrustful scale and subjects,
who score below the 30th percentile on the
Distrustful scale.

 

Alternative Hypothesis: The group mean of sub-
jects scoring above the 70th percentile on the
Distrustful scale will be significantly higher
on the measure Disgust than will be the group
mean of subjects scoring below the 30th percen-
tile on the Distrustful scale.

 

Hypothesis 3

 

Null Hypothesis: There will be no significant
difference on the measure Distress between sub-
jects who score above the 70th percentile on

the Depressed scale and subjects who score below
the 30th percentile on the Depressed scale.

 

102

Alternative Hypothesis: The group mean of sub-
jects scoring above the 70th percentile on the
Depressed scale will be significantly higher on
the measure Distress than will be the group mean
of subjects scoring below the 30th percentile on
the Depressed scale.

 

Hypothesis 4

 

Null HypOthesis: There will be no significant
difference on the measure Surprise between
subjects who score above the 70th percentile
on the Dyscontrolled scale and subjects who
score below the 30th percentile on the
Dyscontrolled scale.

 

Alternative Hypothesis: The group mean of
subjects scoring above the 70th percentile on
the Dyscontrolled scale will be significantly
higher on the measure Surprise than will be
the group mean of subjects scoring below the
30th percentile on the Dyscontrolled scale.

 

Hypothesis 5

 

Null Hypothesis: There will be no significant
difference on the measure Anger between sub-
jects who score above the 70th percentile on
the Agressive scale and subjects who score
below the 30th percentile on the Aggressive
scale.

 

Alternative Hypothesis: The group mean of
subjects scoring above the 70th percentile on
the Aggressive scale will be significantly
higher on the measure Anger than will be the
group mean of subjects scoring below the 30th
percentile on the Aggressive scale.

103

Table 4.1

Affective Scores for Subjects Selected for Analysis

on Plutchik's Personality Dimensions

 

 

 

Affective
Group Measure Q E SD
Hl High Gregarious Percentage 25 37.88 23.30
of ENJOY
Low Gregarious 17 27.88 17.27
H2 High Distrustful Percentage 22 23.63 17.15
of DISGUST
Low Distrustful 23 12.52 12.27
H3 High Depressed Percentage 19 17.92 17.66
of DISTRESS
Low Depressed 28 16.19 15.44
H4 High Dyscontrol Percentage 22 .48 1.25
of SURPRISE
Low Dyscontrol 17 .62 1.38
H5 High Aggressive Percentage 17 3.51 9.08
of ANGER
Low Aggressive 1.08 2.47

27

 

104

Table 4.2

T-test Results for Between-Group Differences
on Affective Measures

 

 

Affective Signif-
Group Measure icant

Id
['0

 

H High Gregarious

l contrasted with Percentage
Low Gregarious of ENJOY 1.51 .065 No
H2 High Distrustful

contrasted with Percentage
Low Distrustful of DISGUST 2.51 .008 Yes

H3 High Depressed
contrasted with Percentage
Low Depressed of DISTRESS 0.36 .361 NO

H4 High Dyscontrol
contrasted with Percentage
Low Dyscontrol of SURPRISE -0.32 .374 No

H High Aggressive
contrasted with Percentage
Low Aggressive of ANGER 1.32 .097 NO

 

Statistical significance at the .05 level was not ob-
tained for any of the Hypotheses l, 3, 4, or 5, and there-
fore each null was not rejected. The test of Hypothesis
2, however, indicated a significant difference (with a p
Of 2.51 and a one-tailed probability of .008) between
high distrustful and low distrustful groups on the affec-
tive measure Disgust. The mean for the high distrustful
group was 23.63, and for the low distrustful group was

12.52, for a difference of 11.11 in the hypothesized

105

direction.

Hypotheses related to the Birkman Method pro-
pose a relationship between two specific personality di-
mensions (Dominance and Getting Along With Others) and
affective facial display (including specific categories
used, number of categories used, and frequency of facial
changes).

Hypothesis 6

Null Hypothesis: There will be no significant

difference on the measure Range of Affect between

subjects who score high on the Getting Along With

Others scale and subjects who score low on the
Getting Along With Others scale.

 

Alternative Hypothesis: The group mean of sub-
jects who score high on the Getting Along With
Others scale will be significantly higher on the
measure Range of Affect than will be the group
mean of subjects scoring low on this scale.

 

Hypothesis 7

 

Null Hyppthesis: There will be no significant
difference on the affective measure of Disgust
between subjects who score high on the Getting
Along With Others scale and subjects who score
low on the Getting Along With Others Scale.

 

Alternative Hypothesis: The group mean Of sub-
jects who score high on the Getting Along With

Others scale will be significantly lower on the
measure Disgust than will be the group mean of

subjects scoring low on this scale.

 

Hypothesis 8

Null Hypothesis: There will be no significant
difference on the affective measure of Enjoy be-
tween subjects who score high on the Getting
Along With Others scale and subjects who score

 

106

low on the Getting Along With Others scale.

Alternative Hypothesis: The group mean of sub-
jects who score high on the Getting Along With
Others scale will be significantly higher on the
measure Enjoy than will be the group mean of sub-
jects scoring low on this scale.

 

Hypothesis 9

 

Null Hypothesis: There will be no significant
difference on the measure Change of Affect be-
tween subjects who score high on the Getting
Along With Others scale and subjects who score
low on the Getting Along With Others scale.

 

Alternative Hypothesis: The group mean of sub-
jects who score high on the Getting Along With
Others scale will be significantly higher on the
measure Change of Affect than will be the group
mean of subjects who score low on this scale.

 

Hypothesis 10

 

Null Hypothesis: There will be no significant
difference on the affective measure of Distress
between subjects who score high on the Dominance
scale and subjects who score low on the Dominance
scale.

 

Alternative Hypothesis: The group mean of sub-
jects who score high on the Dominance scale will
be significantly lower on the measure Distress
than will be the group mean Of subjects who score
low on this scale.

 

Hypothesis ll

Null Hypothesis: There will be no significant
difference on the measure Change of Affect be-
tween subjects who score high on the Dominance
scale and subjects who score low on the Dominance
scale.

 

Alternative Hypothesis: The group mean of sub-
jects who scoreihigh on the Dominance scale will
be significantly lower on the measure Change of
Affect than will be the group mean of subjects
who score low on this scale.

 

107

Table 4.3

on Birkman's Personality Dimensions

Affective Scores for Subjects Selected for Analysis

 

 

 

Affective n i SD
Group Measure —
H6 High G.A.W.O. Number of Af- 13 4.23 0.43
fects Displayed
Low G.A.W.O. (RANGE) 15 3.73 0.79
H7 High G.A.W.O. Percentage 13 21.52 12.96
of DISGUST
Low G.A.W.O. 15 16.17 16.80
H8 High G.A.W.O. Percentage 13 35.03 19.92
of ENJOY
Low G.A.W.O. 15 42.70 24.06
H9 High G.A.W.O. Percentage Of 13 63.53 13.11
CHANGE OF Af-
Low G.A.W.O. fect 15 55.86 16.51
Hlo High Dominance Percentage 16 13.00 9.86
of DISTRESS
Low Dominance 14 10.75 9.34
Hll High Dominance Percentage of 16 61.00 18.69
CHANGE of Af-
Low Dominance fect 14 56.42 15.81

 

Statistical significance at the .05 level was not ob-

tained for Hypotheses 7-11, and therefore each null was

not rejected.

The test of Hypothesis 6, however,

indi-

cated a significant difference (with a ptof 2.00 and a

one-tailed probability of .028) in the number of affect

categories used when the high Getting Along With Others

and the low Getting Along With Others groups were

108

Table 4.4

T-test Results for Between-Group Differences

on Affective Measures

 

 

 

 

Affective Signif-
Group Measure p_ p icant
Number of
H6 High G.A.W.O. Affects
contrasted with Displayed
Low G.A.W.O. (RANGE) 2.00 .028 Yes
H7 High G.A.W.O. _
contrasted with Percentage
Low G.A.W.O. of DISGUST 0.93 .175 NO
H8 High G.A.W.O.
contrasted with Percentage
Low G.A.W.O. of ENJOY -0.91 .185 NO
H9 High G.A.W.O. Percentage
contrasted with Of CHANGE
Low G.A.W.O. of AFFECT 1.35 .095 No
Hlo High Dominance
contrasted with Percentage
Low Dominance of DISTRESS 0.64 .264 No
H11 High Dominance Percentage
contrasted with of CHANGE
Low Dominance of AFFECT 0.72 .239 No
compared. The mean of the high Getting Along With Others

group was 4.23, and the mean of the low group was 3.73,

for a difference Of .50 in the hypothesized direction.

Hypotheses related to the Eysenck Personality In-

ventory propose a relationship between the personality

characteristics Extraversion/Introversion and Neuroticism/

109

Stability, and facial affect display as measured by spe-
cific categories used, number of categories used, and fre-
quency Of facial changes.

Hypothesis 12

 

Null Hypothesis: There will be no significant
difference on the affective measure of Enjoy be-
tween subjects who score high on the Extraver-
sion scale and subjects who score low on the Ex-
traversion scale.

 

Alternative Hyppthesis: The group mean of sub-
jects who score high on the Extraversion scale

will be significantly higher on the measure En-
joy than will be the group mean of subjects who
score low on this scale.

 

Hypothesis l3

 

Null Hypothesis: There will be no significant
difference on the affective measure of Disgust
between subjects who score high on the Extraver-
sion scale and subjects who score low on the Ex-
traversion scale.

 

Alternative Hypothesis: The group mean of sub-
jects who score high on the Extraversion scale
will be significantly lower on the measure Dis-
gust than will be the group mean of subjects
who score low on this scale.

 

Hypothesis 14

 

Null Hypothesis: There will be no significant
difference on the measure Change of Affect be-
tween subjects who score high on the Neuroticism
scale and subjects who score low on the Neuroti-
cism scale.

 

Alternative Hypothesis: The group mean of sub-
jects who score high on the Neuroticism scale
will be significantly higher on the measure
Change of Affect than will be the group mean of
subjects who score low on this scale.

 

110

Hypothesis 15

Null Hypothesis: There will be no significant
difference on the measure Range of Affect between
subjects who score high on the Extraversion and
Neuroticism scales (EN) and those who do not
score high on both of these scales (ES, IS, IN).

 

Alternative Hyppthesis: The group mean of sub-
jects who score high on the Extraversion and Neu-
roticism scales (EN) will be significantly high-
er on the measure Range of Affect than will be
the group mean of subjects who do not score high

 

on both of these scales (ES,

Table 4.5

IS, IN).

Affective Scores for Subjects Selcted for Analysis
on Eysenck's Personality Dimensions

 

 

 

Affective
Group Measure pl 'i SD
H12 High Extraversion Percentage 25 40.29 24.04
Of ENJOY
Low Extraversion 26 34.88 16.04
H13 High Extraversion Percentage 25 16.56 12.96
of DISGUST
Low Extraversion 26 20.08 17.14
Hl4 High Neuroticism Percentage 27 58.22 1.87
of CHANGE
Low Neuroticism of Affect 23 60.86 .99
H15 EN Number of 16 3.75 .57
Affects
IN, IS, ES Displayed 15 4.14 .70
(RANGE)

 

Statistical significance at the .05 level was not Obtained

for Hypotheses 12-15, and therefore each null was not re-

jected. Hypothesis 15 was a contradictory finding, with

111

Table 4.6

T-test Results for Between-Group Differences
on Affective Measures

 

 

 

Affective Signif-
Group Measure 'p p_ icant
H12 High Extraversion Percentage
contrasted with of ENJOY
Low Extraversion 0.95 .174 No
H13 High Extraversion
contrasted with Percentage
Low Extraversion of DISGUST'-0.82 .207 No
Hl4 High Neuroticism Percentage
contrasted with of CHANGE
Low Neuroticism Of Affect -0.61 .273 No
H15 EN contrasted Number of
with IN, IS, Affects
ES Displayed -2.00 .027 No
(RANGE)

 

a'p of -2.00 and a one-tailed probability of p = .025.

It should be noted that since the t value is negative (in-

dicating a direction other than that which was hypothe-

sized), the degree of probability used by this study is

consequently in the wrong tail of the test, and the prob-

ability level of p .025 has no meaning for the tail in

which it suggests significance.

112

Further Analyses of Group Differences

 

Multivariate Analyses of Variance

In order to explore how the two groups of sub-
jects on each independent variable relate to the set of
eight dependent measures, ten multivariate analyses of
variance were performed. In Table 4.7 each independent
variable is presented along with the empirical data ger-
mane tO its significance test for group differences on
the set of dependent measures.

Table 4.7

Multivariate Analyses of Variance for the Independent
Variables on the Set of Dependent Measures

 

 

 

Independent Variable F p Significant
Gregarious 1.91 .091 NO
Distrustful 1.96 .079 NO
Depressed .77 .629 No
Dyscontrolled .53 .828 No
Aggressive 1.40 .227 No
Getting Along With Others .75 .648 No
Dominance 1.12 .385 No
Extraversion .98 .464 No
Neuroticism .61 .764 NO
Extraversion/Neuroticism .75 .647 No

 

the: Significance is considered at an alpha level of .05.

113

Statistical significance at the .05 level was not Obtained
for any of the independent variables when tested by a mul-

tivariate analysis of variance.

Further Analyses of Interest

While the multivariate analyses of variance were
not significant when used to relate the independent vari-
ables to the set of dependent measures, the univariate
analyses Of variance did indicate the presence of several
significant relationships beyond those mentioned in the
hypothesis tests. Univariate analyses were used to test
group differences on each of the independent variables as
they relate individually to each of the eight dependent
measures. In addition, univariate analyses were used to
provide further methodological information by testing
group differences on rater agreement and number Of dis—
cards per subject. It should be noted here that the term
significant is used with caution since it was defined at
an alpha level Of .05 and does not speak to the problems
of the overall Type I error rate.

In this section those independent variables show-
ing significant differences on the affective measures are
presented, along with the empirical data germane to the

tests for significance and the tables that display the

114

nature of the significant differences. The independent
variables presented are: Gregarious, Distrustful, and
Aggression. Tables for univariate analyses of the remain-
ing independent variables, Depressed, Dyscontrolled, Get-
ting Along With Others, Dominance, Extraversion, and Neu-
roticism, can be found in Appendix E and will be referred
to in the Discussion section, Chapter V.

NO significant differences were found among the
univariate analyses testing for group differences on the
measures of rater agreement and of discards per subject.
Tables presenting the empirical data for these tests can
be found in Appendix E and will also be referred to in
the Discussion section, Chapter V.

Table 4.8 presents the independent variable Gre-
garious along with the results of significance tests for
between-group differences on each of the affective mea-
sures. As indicated in Table 4.8, six of the eight af-
fective measures show no significant difference at the
.05 level between high and low groups on Plutchik's Gre-
garious scale. Tests on two affective measures, however,
Disgust and Change, indicate that there may be signifi-
cant differences between high and low Gregarious groups
in the percentage of disgust displayed, and in the per-

centage of affect changes the subjects displayed.

115

Table 4.8

T-test Results of Gregarious Group Differences
on the Affective Measures

 

 

 

Dependent Variable p p Significant
Percentage Of Interest 0.07 .490 No
Percentage Of Disgust -2.30 .027 Yes
Percentage of Anger -1.61 .116 No
Percentage of Distress 0.41 .682 No
Percentage of Surprise 1.24 .221 No
Percentage Of Enjoy* 1.51 .065 No
Percentage Of Change -2.20 .034 Yes
# of Affects Used (Range) -1.36 .180 No

 

*Hypothesis 1 (reported here as a one-tailed test).

Note: Significance is considered at an alpha level of
.05.

 

The mean for the high Gregarious group was 15.16,
and for the low Gregarious group 27.41 on the affective
measure Disgust, for a difference of 12.25 between groups,
at a probability level of p_= .027. The mean for the high
Gregarious group on the affective measure Change was
56.68, and for the low Gregarious group 66.82, for a dif-
ference of 10.14 between groups at a probability level of
p = .034. (See Table 4.9 for a display of these differ-

ences.)

116

Table 4.9

Affect Scores for High and Low
Gregarious Groups

 

 

 

Dependent Variable Group H i’ SD

Percentage of Disgust High Greg 25 15.16 15.85
Low Greg 17 27.41 18.41

Percentage of Change High Greg 25 56.68 15.23
Low Greg 17 66.82 13.83

 

Table 4.10 presents the independent variable Dis-
trustful along with the results of significance tests for
between-group differences on each of the affective mea-
sures.

Table 4.10

T-test Results of Distrustful Group Differences
on the Affective Measures

 

 

 

Dependent Variable p p Significant
Percentage of Interest -2.05 .047 Yes
Percentage of Disgust* 2.51 .008 Yes
Percentage of Anger 1.28 .207 No
Percentage of Distress -0.50 .621 No
Percentage Of Surprise -1.72 .093 No
Percentage of Enjoy -0.21 .832 No
Percentage of Change -0.77 .448 No
# Of Affects Used (Range) -0.21 .836 NO

 

*Hypothesis 2 (reported as a one-tailed test).
Note: Significance is considered at an alpha level of .05 .

117

As indicated in Table 4.10, six of the affective measures
show no significant difference at the .05 level between
high and low groups on Plutchik's Distrustful scale. A
significant difference is indicated between groups on the
affective measure Disgust, as presented earlier in this
chapter. In addition, the test between groups on the af-
fective measure Interest indicates that there may be a
difference in the percentage Of interest displayed by
those who score high on the Distrustful scale and those
who score low on this scale. Those who score high on Dis-
trustful have a mean interest score of 17.23, those who
score low a mean Interest score of 26.54, for a differ-
ence Of 9.31 at a probability level of p = .047. (See
Table 4.11.)

Table 4.11

Affect Scores for High and Low
Distrustful Groups

 

 

 

Dependent Variable Group 3 2 SD
Percentage of Inter- High Dstrfl 22 17.23 12.21
est Low Dstrfl 23 26.54 17.67

 

Table 4.12 presents the independent variable Ag—
gression along with the results of significance tests for
betweenégroup differences on each of the affective mea-

sures.

118

Table 4.12

T-test Results of Aggressive Group Differences
on the Affective Measures

 

 

 

Dependent Variable p p Significant
Percentage Of Interest -l.99 .053 No
Percentage Of Disgust 2.37 .022 Yes
Percentage of Anger* 1.32 .097 No
Percentage of Distress -O.10 .919 NO
Percentage of Surprise -0.51 .612 No
Percentage of Enjoy -0.81 .423 NO
Percentage of Change 0.90 .375 No
# of Affects Used (Range) 1.03 .311 No

 

*Hypothesis 5 (reported as a one-tailed test).

H933: Significance is considered at an alpha of .05.

As indicated in Table 4.12, seven of the eight affective
measures show no significant differences between high and
low groups on the Aggression scale. The test between
groups on the affective measure Disgust, however, indi-
cates that there may be a difference in the percentage of
disgust displayed between those who score high and low on
Plutchik's Aggression scale. Those who score high on Ag-
gression have a mean Disgust score of 25.75, those who

score low a mean Disgust score of 13.92, for a difference

119

of 11.83 at a probability level Of p = .022. (See Table

 

 

4.13.)
Table 4.13
Affect Scores for High and Low
Aggression Groups
Dependent Variable Group n i SD

 

Percentage of Disgust High Aggr 17 25.75 17.57
Low Aggr 27 13.92 15.17

 

It should be noted that the group differences in
this section have been explored to provide further infor-
mation about the relationships between the independent
and dependent variables in this study. While tests for
group differences did provide further evidence of signif-
icant relationships, these results should be looked at
with caution since significance was defined at an alpha
level of .05, and consequently information about the ov-

erall Type I error for these tests is inadequate.
Correlational Analyses

A 24 x 24 intercorrelational matrix was used to
further explore relationships and provide descriptive in—
formation about the variables used in this study (see

Appendix F' for complete matrix). Following are data

120

which address relationships within and between the set of
pencil/paper instruments, within the set of affective mea-
sures, and between the test instruments and the affective
measures. In addition, two methodological measures, per-
centage Of agreement among raters and percentage of dis-
carded samples per subject, were included in the analyses
and are presented at the end of this section. All data

in this section are based on scores from the 168 subjects
in the total sample, with the exception of data pertaining
to the affective measures, for which there were only 62

subjects.

Within Instrument Correlations

 

The three pencil/paper instruments used in the

study, the Plutchik Emotions Profile Index (PEPI), the

 

Eysenck Personality Inventopy (EPI), and the Birkman Meth-

 

 

pd (BM) are all standardized instruments designed to mea-
sure affective orientation. The following data include
all scales from each instrument used in this study, and
in the cases of the PEPI and the EPI, the data includes
all remaining scales of each instrument at well.

Plutchik Emotions Profile Index. The PEPI has

 

a total of nine scales, five of which were used as inde-
pendent variables in this study. The Bias scale is used

as an indicator of social desirability, though it may

121

also be an accurate description of the person. A high
score indicates a tendency to choose the more socially
desirable descriptor (shy rather than resentful, adven-
turous rather than gloomy); a low score indicates the op-
posite tendency (brooding rather than obedient, quarrel-
some rather than sociable).

The remaining eight scales are constructed so as
to be four sets of bi-polar opposites. Figure
4.1 presents these eight scales graphically on Plutchik's
"Emotion Circle," and includes the list of item descrip-
tors from the test that make up each scale. (Note that
the weights of the items per each scale are not included.)

Correlations between the scales of this instru-
ment are presented in Table 4.14. It should be noted,
however, that this is an ipsative instrument and that
consequently the reported correlations between scales will
be confounded by this factor. As indicated by Table 4.14,
the range of correlations for the instrument is -.80 to
+.79. Aggressive is the only scale which consistently
shows relationships with all other scales, indicating
that it might not be as independent a measure as the
other scales.

The bi-polar pairs Of scales have negative cor-

relations ranging from -.19 to -.60: Dyscontrol/Control

122

 

 

S‘ TR
:9 <
69
affectionate affectionate OJ-
Q; sociable sociable \S'
é obedient 00
J,” 1.
la 91°03? ;
‘q Umm3ve immune E;
O self-conscious obedient
0 THE.” “3;“
O( resentful g
(6‘ quarrelsase gloomy
O self-conscious brooding
(y '
SZRU (Leg?
STFUL ° DE?“

Figure 4.1. Plutchik's Emotion Circle

*Scales used in this study as independent
variables.

with a negative relationship at -.19; Depressead/Gregar-
ious with a negative relationship at -.33; Distrust/Trust
with a negative relationship at -.58; and Aggressive/Timid
with a negative relationship at -.60; suggesting that
Distrust/Trust, and Aggressive/Timid may have stronger

bi-polar relationships than do Dyscontrol/Control and

Depress e d/Gregarious.

123

 

mmflm Huuco eases amass

oo.H mo.o

oo.H

mono

Nv.o mh.o Hm.o
mm.o vm.ot me.ot
oo.H mm.o ha.o1

oo.H mh.o

oo.H

Hmmd

om.ol
vm.ol
om.OI
vm.ol
ov.OI

oo.H

Human
Ho.OI
no.0
hm.OI
mm.oa
Hm.ou
me.o
oo.H

mummo comma

mm.ol
NH.ot
mo.ol
hm.on
mm.OI
vv.o

no.OI

oo.H

mo.o
ma.on
om.ol
$0.0:
mH.o
Hm.o
va.ol
mm.ol
oo.H

mafia
omHHOHucOU
cases
Hsmumsua
msOwummmuu
o>flmmmumm¢
Hsmumsuumao
ommmmumma

omaaouucoomho

 

 

Hmmm How mcoflumHmquU mamomuumucH

vH.v mance

124

The three categories most generally perceived as
negative in social contexts (Depressed, Distrustful, and
Aggressive) correlate negatively to the Bias scale, indi-
cating that persons who choose the more socially desirable
descriptor items would not have high scores on these three
scales.

Figure 4.2 summarizes the relationships Of the
nine scales to one another by grouping them into categor-
ies of strong negative relationships (-.80 to -.35),
strong positive relationships (.35 to .77) and minimal
relationships (00 to 1.34).

Eysenck Personality Inventory. The EPI has a
total of three scales: Extraversion and Neuroticism were
used as independent variables in this study; the Lie
scale was used only for the purpose of eliminating sub-
jects at the high end from the interview phase of the data
collection. The extraversion scale is constructed so that
persons scoring at the high end are characterized as ex-
traverted, and at the low end, as introverted. Similar-
ly, the Neuroticism scale is constructed so that persons
scoring at the high end are characterized as neurotic, at
the low end, stable. In general, average scores on both
scales fall between the 3lst and 70th percentiles; per-

sons scoring above or below these cutting points would

JJZS

.Hamm MOM madamCOMucHOm manomuumucu .~.v musmwm

 

 

 

 

 

 

 

 

 

 

oases o>wmmoum
pmaaouu .Hsuumsuu nos .Hsmumsuu
Icoo .pmaaouusoomap .msOwummeum Imwp .pommoummp mean
moan .w>wmmmumme
stumsuu .Hsu
lumsuumwo .pmmmmum
loo .poaaouusooeap pwafiu anewummmum Houucoo
o>wmmwum
stuwsuu .mso mean Imp .stumsuumwp
nflummmum .ommmwuamp .OOHHOuucoo .OOHHouucOOmmp passe
mean .m>wmmoum
poaaouusoo .pw loo .Hsuumsuu
law» .owaaouucoomap msofiummoum leap .pmmmmuawo asuumsua
commonest .pfi mafia Houucoo .m>«mmeum
nawu .OOHHOuucoomap .stumsuu tum .Hsmumsuumwp msOwummmuo
mean
meadow» Hawumsuu peed» .Hsuumsuu
Icoo .pmaaouucoomxp swap .pwmmmummp .msOwucmwum m>wmmeuom<
mews
Omaaouucoo .pOmmoua pwswu .Hsuumsuu
Imp .pmaaouusoomxp O>flmmmummm .msOwummmum Houumsuumwn
pmaaouucoo
.pwaau .nsofiumm mean .Hsuumsuu
Imam .Hsuumsuumwp m>wmmmummc .pwaaouusoomwp ommmmummo
mean
Houusoo .stumsuu
.ms0«ummmum .m>wm
twosome .stumsuumwp pwEwu .pmmmmumoo pmHHOMuOOOmao
3m.“ 3 8.. E. 3 5.... mos Enron 8.1 omz maaom

Hashes:

mcouum

mcouum

 

126

be characterized as above or below average on the trait
under consideration.
Table 4.15 presents the inter-scale correlations
for this instrument.
Table 4.15

Inter-scale Correlations for EPI

 

 

Extraversion 1.00

Neuroticism -0.05 1.00

Lie Scale -0.24 -0.23 1.00
Extrv Neurot Lie

 

As indicated in Table 4.15, the relationship of the Ex—
traversion and Neuroticism scale shows a slight negative
correlation (-.05), which is congruent with Eysenck's
findings (Eysenck, 1968). This data seem congruent with
Eysenck's hypothesis that the two scales are essentially
uncorrelated and are orthogonal in nature.

The relationship of each of these scales to the
Lie Scale shows a slightly stronger negative correlation
(Extraversion/Lie Scale at -.24 Neuroticism/Lie Scale at
-.23), suggesting that in this sample subjects who score
high on Extraversion and Neuroticism are less prone to
"fake good" on the test. Eysenck does not provide inter-

correlational data for this scale, and suggests only that

127

a score at 10 or above shows that "faking good" is likely
to have occurred.

The Birkman Method. The BM has a total of elev-

 

en personality scales; two scales (Self-Consciousness and
Depressed) were combined for the purposes of this study to
create a research scale labeled Getting Along With Others.
The other BM scale used in the study was Dominance. Ta-
ble 4.16 presents the inter-scale correlations for these
two scales.

Table 4.16

Inter-scale Correlations for BM

 

 

Dominance _ 1.00
G.A.W.O. .58 1.00
Dom GAWO

 

As indicated in Table 4.16, the relationship between Get-
ting Along With Others and Dominance shows a positive
correlation at .58 suggesting an overlap of approximate-
ly 31 percent between these scales for this sample.

This corresponds to information reported by Birkman (1976)
indicating that Dominance and the two scales combined to
create Getting Along With Others (Self-Consciousness and

Empathy) all have positive correlations to one another.

128

Between Instrument Correlations

Correlations of scales from Plutchik's Emotions

Profile Index with scales from the Eysenck Personality

 

Profile and the Birkman Method are presented in Table

 

 

 

 

4.17.
Table 4.17
Correlations of Scales from Plutchik with
Scales from Eysenck and Birkman
Plutchik
Dyscontrolled 0.38 -0.29 -0.10 0.09 -0.13
Depressed -0.26 0.27 0.02 -0.04 0.11
Distrustful -0.08 0.39 -0.24 0.17 0.18
Aggressive -0.01 0.25 -0.16 0.13 0.09
Gregarious 0.44 -0.33 0.10 -0.23 -0.31
Trustful 0.32 -0.30 0.20 -0.14 -0.24
Timid -0.40 -0.11 0.23 -0.06 0.14
Controlled -0.42 0.03 0.03 0.02 0.20
Bias 0.14 -0.43 0.18 -0.17 -0.24
Extra- Neuro- Lie Domin- G.A.W.
versn ticsm ance O.
\ y:/ \; 11
Eysenck Birkman

 

As indicated by Table 4.17, the range of correlations be-
tween the Plutchik scalesarfl.those of Eysenck and Birkman

is -.43 to +.43. Again it should be noted that the

 

 

129

Plutchik instrument is an ipsative one, so that correla-
tions between its scales and the scales of the other two
instruments may be confounded by this factor.

Of these between-instrument correlations, only
four correlations indicate relationships at i.40 or more.
Gregarious and Extraversion show a positive correlation
at .44; both Timid and Controlled show negative correla-
tions to Extraversion at -.40 and -.42, respectively.

The Plutchik Bias scale shows a negative correlation to
the Eysenck Neuroticism scale (-.43) indicating an inverse
correlation between those who score high on Neuroticism
and those who choose the more socially acceptable descrip-
tor items in describing themselves.

Neither Of the Birkman scales shows evidence Of
strong relationships with any of the Plutchik scales.

It is of interest to this study, however, that the Birk-
man scale Dominance indicates a positive relationship
with Distrustful and Aggressive (.17 and .13, respective-
ly) and a negative relationship with Trustful and Gregar-
ious (-.14 and -.23, respectively). It is also of inter-
est that the Birkman Getting Along With Others Scale
shows a negative relationship with Gregarious and Trust-
ful (-.31 and -.24, respectively).

Correlations between the scales of the Eysenck

130

Personality Inventory and the Birkman Method are present-

 

 

ed in Table 4.18.
Table 4.18

Correlations Between Scales of the Eysenck Personality
Inventory and the Birkman Method

 

 

 

 

 

Eysenck
Extraversion 0.05 -0.19
Neuroticism 0.10 0.36
Lie -0.17 -0.20
Domie G.A.W.O.
nance
Birkman

 

As indicated by Table 4.18, the range of correlations be-
tween the scales of Eysenck and Birkman is -.20 to +.36.
All relationships between the scales of these two instru-
ments are slight. Both Birkman scales have mild negative
correlations to the Eysenck Lie scale. The two scales
differentiate slightly, however, in their relationships
to the Extraversion and Neuroticism scales. While Domi-
nance shows virtually no relationship to Extraversion or
Neuroticism, suggesting that it measures characteristics
quite independent of these two scales, Getting Along With
Others does suggest a slight overlap, with a mild posi-
tive relationship with Neuroticism (.36) and a mild nega-

tive relationship with Extraversion (-.19).

131

Correlations Within the Affective Measures
As previously noted, the data for these correla-
tions is based only on the 62 subjects participating in
the interview phase of the study. Table 4.19 presents
the correlations within the eight affective measures used
in this study.
Table 4.19

Correlations Within the Set of Affective Measures

 

 

INTEREST 1.00

DISGUST 0.35 1.00

ANGER 0.09 0.17 1.00

DISTRESS 0.47 0.33 0.13 1.00

SURPRISE 0.30 0.19 -0.04 0.17 1.00

ENJOY 0.41 0.41 0.13 0.34 0.16 1.00

CHANGE 0.75 0.67 0.28 0.68 0.26 0.69 1.00

RANGE 0.73 0.67 0.31 0.69 0.35 0.75 0.96 1.00
INTER-DIS- ANGER DIS- SUR- ENJOY CHANGE RANGE
EST GUST TRESS PRISE

 

As indicated by Table

4.19, all relationships within the

set of affective measures are positive ones with the ex-

ception of Surprise and Anger, which show a negative cor-

relation at -.04.

As might be expected, due to their low

frequency of occurrence during the interviews, Surprise

and Anger in general show the lowest correlations to one

132

another and to all the other scales within the set of
affective measures.

The four strongest relationships are within the
categories of Change, Range, Interest, and Enjoy. The
high positive correlation between Change and Range (.96)
suggests that those who change from one category of af-
fect to another most frequently are also most likely to
display a wider range of affect categories. The high
positive relationships between both Enjoy and Interest
with Range (.75 and .73, respectively) suggest that those
who most frequently use Enjoy and/or Interest are likely
to display a wider range of affect categories. Finally,
the high positive relationship between Interest and
Change (.75) suggests that those who use Interest most
frequently are also most likely to change from one affect
category to another the most frequently.

With regard to potential patterns of usage among
affect categories, correlations within the four most fre-
quently used affects (Distress, Interest, Enjoy, and Dis-
gust) do show some differential combinations. The high-
est correlates Of Distress and Interest are one another,
at a .47 correlation. The highest correlate for Disgust
is Enjoy, at a .41 correlation. The highest correlates

for Enjoy are Disgust and Interest, each at a .41

133

correlation. Thus Distress/Interest and Disgust/Enjoy

seem to be the two most consistent sets of combinations.

Correlations Between Test Instruments and Affective
Measures ‘

Correlations between scales of the pencil/paper
instruments and the eight affective measures are present-
ed in Table 4.20.

As Table 4.20 indicates, the range of correla-
tions between the affective measures and scales of the
pencil/paper instruments is -.29 to +.27, suggesting only
mild relationships between any of these dependent and in-
dependent variables. The affect measure Disgust shows the
largest number of relationships at 1.20 or above, and
also has the highest correlational values, with a nega-
tive relationship with Trustful (-.29), a positive rela-
tionship with Distrustful (.27), and a negative relation-
ship with Gregarious (-.25).

As previously noted, the data for correlations
between the independent variables and the affective mea-
sures are based on scores from 168 subjects on the pen-
cil/paper instruments and scores from 62 subjects on the
affective measures. Since some of the earlier tests
(based only on subjects who score high or low on the in-

dependent measures) do show significant relationships, the

134

 

 

 

mmHmm mmmme 9mm
moz<m moz<m0 Moozm Imam ImHQ mmwz¢ EmaomHD ImmBzH mmmmBO mBHZ
no.0! 00.0 50.0! no.0 m0.o No.0 H0.o! v0.0! uzoqd UZHBBMU
$0.0! m0.0! 0H.o! NH.0! 0H.o! NH.0 no.0 Ho.o! muzfizHEOQ
hH.o! 0H.0I no.0! No.0 va.0! HN.0! om.0! 00.0 m<Hm
$H.o ha.o 00.0 50.0 $0.0 $0.0 HH.0 mH.o omqqomezou
$0.0! No.0! 00.0 00.0 HH.0I 0H.0! MH.0! 00.0 DHEHB
mm.o! mm.o! 0H.0! no.0! HH.0I NN.0! 0N.0! $0.0! ADMBmDmB
NN.0! mm.ol HH.0! $0.0 00.0! NN.o! mN.0! 00.0! MDOHmdommo
mH.0 mH.o $0.0 Ho.o NH.o wd.o HN.0 No.0! m>HmmmmOU¢
$H.0 va.o 00.0 00.0! 0H.o ma.o hN.0 no.0 ADmBmDMBmHQ
$0.0 $0.0 no.0 $0.0 00.0 mH.o No.0 no.0! Qmmmmmmma
00.0 Ho.0! 00.0 no.0! no.0 00.0! 00.0! no.0! Omnqomszoomwo
50.0! 00.0! No.0! Ho.0 $0.0! $0.0! $0.0! H0.o quum mHA
v0.0 mo.o $0.0 $0.0! 00.0 00.0 $0.0 $0.0! ZmHUHBOMsz
$H.o! hH.0! $0.0! va.0! $0.0! 0H.0I NH.0! 00.0! ZOHmmm><mem
meDmmw—L Q>HHUQHW< U36 WQHMUW HCQEDHUWCH G003H0m MCOHUMHGHHOU

o~.¢ OHQOB

135

generally weak relationships indicated by Table 4.20 sug-
gest that there may be a curvilinear relationship between
the independent and dependent variables. That is, rela-
tionships between test scores and affective measures ap-
pear more clearly differentiated when only high and low
groups are compared on the affective measures, and not
when the full range of test scores is compared on these

measures .

Correlations Between Methodological Measures and

Affective Measures

Rater scores on two measures, percentage of
agreement per subject and number Of discards per subject,
were averaged and included in the correlational matrix.
Table 4.21 presents correlations between these scores and
subject scores in each of the affect categories.

Table 4.21

Correlations Between Methodological Measures
and Affective Measures

 

 

Percentage
of Agree-
ment 0.76 0.65 0.22 0.63 0.29 0.82 0.93 0.96
# Of Dis-
cards 0.45 0.62 0.15 0.56 0.25 0.58 0.75 0.75

INT- DIS- ANGER DIS- SUR- ENJOY CHANGERANGE
EREST GUST TRESS PRISE

 

136

As indicated by Table 4.21, correlations between the per-
centages of rater agreement per subject and subjects'
scores on affective measures all show positive relation-
ships. Subjects who displayed high percentages Of Enjoy
show the strongest relationship to high percentages of
rater agreement (.82) followed by subjects who displayed
high percentages Of Interest (.76), of Disgust (.65), of
Distress (.63), of Surprise (.29) and of Anger (.22).
These relationships are similar to those presented in
Table 3.1 (Rater Agreement by Affect Category) where the
category with the highest rater agreement was Enjoy, fol-
lowed by Interest, Disgust, and Distress. Surprise and
Anger switch order here, as the category Anger had a
higher percentage of rater agreement than did Surprise,
while subjects displaying a high percentage Of the af-
fect Surprise had a higher correlational relationship to

rater agreement than did those displaying Anger.

Summary

The fifteen hypotheses used in the study were
tested to determine group differences on the affective
measures. The nine independent variables on which sub-

jects were divided into high and low groups were Gregar-

ious, Distrustful, Depressed, Dyscontrolled, Aggressive,

137

Getting Along With Others, Dominance, Extraversion, and
Neuroticism. The affective measures used were Enjoy,
Disgust, Distress, Surprise, Anger, Interest, Range of
Affect, and Change Of Affect. T-tests were used to test
group differences, with the alpha level set at p‘(.05.
Significant differences were found between high and low
Distrustful groups on the affective measure of Disgust
(p=.008); and between high and low groups Of Getting
Along With Others on the affective measure of Range of
Affect (p=.028).

A multivariate analysis Of variance was per-
formed in order to examine how the two groups of subjects
on each independent variable relate to the set of eight
dependent measures. Significance was considered at an
alpha level Of .05 for this test. No significant dif-
ferences were found for any of the independent variables
on the set of eight affective measures.

Additional t-tests were used to further explore
the relationships between independent and dependent vari-
ables not addressed in the hypotheses. In addition,
t-tests were used to test group differences on two meth-
odological measures, rater agreement and number of dis-
cards per subject. NO significant differences were

found between groups on the methodological measures.

138

Additional differences were found, however, between high
and low Gregarious groups on the affective measure DiSe
gust and on the affective measure Change of Affect;
between high and low Distrustful groups on the affective
measure Interest; and between high and low Aggressive
groups on the affective measure Disgust. Significance
was considered at an alpha level of .05 for these tests;
evidence of significant differences, however, was re-
garded cautiously since there is inadequate information
about the overall Type I error rate for this group of
univariate analyses.

Correlational analyses were used to provide
descriptive information about relationships within and
between the set Of pencil/paper instruments used as the
independent variables, within the set of affective meas-
ures used as the dependent variables, and between the
test instruments and the affective measures.

Scales from Plutchik's Emotions Profile Index
were found to range in correlational values from -.80 to
+.79. Scales comprising bi-polar pairs showed differ-
ential strengths of negative relationships, ranging from
-.19 to -.60. It was noted that correlations within the
scales of this test may be confounded by the ipsative

nature of the instrument.

139

Scales from the Hysenck Personality Inventory
were found to range in correlational value from -.24 to
-.05, with the two major scales (Extraversion and Neuro-
ticism) at -.05. This relationship seems congruent with
Eysenck's hypothesis that the two scales are orthogonal
in nature.

The two scales from the Hirkman Method used in

 

the correlational analyses show a positive relationship
at .58, suggesting a moderate overlap between these two
dimensions.

Correlations between the instruments range from
-.43 to +.43 for Plutchik and Eysenck, from -.31 to —.20
for Plutchik and Birkman, and from -.20 to +.36 for
Eysenck and Birkman. Relationships that were of interest
to the study were noted and will be discussed further in
Chapter V.

Correlations within the set of eight affective
measures ranged from -.04 to +.96. The strongest cor-
relational relationships were found between the measures
of Change and Range (.96), Enjoy and Range (.75), Enjoy
and Interest (.73), and Interest and Change (.75). The
possible implications of these relationships will, again,

be discussed in Chapter V.

140

Correlations between test instruments and af-
fective measures ranged from -.29 to +.29, suggesting
only mild relationships between any Of the independent
and dependent variables. Since the earlier t-tests
(using only high and low scores from the independent vari-
ables) indicated stronger relationships than did the cor—
relations (which included all scores from the independent
variables) it was noted that there may be a curvilinear
relationship between the sets of independent and depen-

dent variables.

CHAPTER V

SUMMARY, CONCLUSIONS, AND DISCUSSION

In this chapter, the study is summarized, con-
clusions based on data analysis and methodology are ex-
plored, and a discussion of suggestions for future re-

search and implications for therapy is presented.

Summary

The primary purpose of this study was to inves-
tigate whether there are significant differences in ex-
pressions Of facial affect as recorded on videotape and
measured by trained judges between persons with specific
personality characteristics as measured by pencil/paper
personality instruments. The specific questions addres-
sed were: DO persons on specific personality dimensions
vary in the types of facial affect they express; Do per-
sons on specific personality dimensions vary in their
frequency of change from one affect to another; and DO
persons on specific personality dimensions vary in the

range (or number of types of affects) they express.

The secondary purpose of the study was a metho-

dological one, and responded to the problems of measuring,

141

142

quantifying, and recording facial affect. The study was
intended to contribute to the methodological literature by
describing any further refinements in techniques and pro-
cedures, and by continuing to address whether current
technology permits meaningful data in this area.

The assumption that personality characteristics
related to emotional style are discernable in the face
by specific expressions Of facial affect was the basis
for this research. The potential use Of such information
for subsequent research efforts and therapeutic inter-
vention provided impetus for the study.

A review of the literature was conducted in
four related areas: Emotion, Facial Expression of Emotion,
The Scoring of Facial Affect, and Specific Affects to be
Used in This Study. The summary of Emotion noted that
while this phenomenon has been studied from the evolu—
tional, biological, and cognitive perspectives, as well
as from the psychometric standpoint, it has not yet been
successfully delimited nor conclusively defined. Darwin,
as the first widely recognized scientist to prOpose a re-
lationship between facial expressions and internal emo-
tional states, began to identify categories of emotion on
the basis of facial eXpressions, and to speculate on the

functional or survival value Of each.

143

Subsequent to Darwin, Sylvan Tomkins develOped
a theoretical framework for the emotions system as it
interplays with the other physiological and psychological
mechanisms, and includes the psychodynamic functions of
the different affects and the relatedness of facial ac-
tivity to the experience Of these affects.

Based largely on the work Of Darwin and Tomkins,
other researchers such as Ekman and Izard have worked to
more systematically classify categories Of affect, develop
labeling and recognition techniques for these categories,
and generate empirical data about the labeling process as
well as about the cross—cultural evidence for these funda-
mental categories. A recent develOpment in this research
area was stimulated by Haggard and Isaacs, with the dis-
covery that the expressions of facial affect may have a
demonstrable relationship to certain aspects of psycho-
logical functioning, such as the presence Of ego mechanismh

The scoring of facial affect continues to be
problematic, partly due to the subjective nature of the
task, and partly due to the technological procedures, which
are for the most part cumbersome and costly. Systematic
procedures have been developed, nevertheless, to facilitate
this process, and were used where applicable as guidelines

for research procedures in this study.

.144

In deciding upon which facial affect categories
to use in this study, considerations of reported rater
reliability and likeLflmod Of occurance for each affect
category was made. The categories Of interest, anger,
disgust, surprise, enjoyment, and distress were selected
to be used as the dependent measures for the study. It
was noted, however, that interest was used as a neutral

category and did not appear in the formal statements of

hypotheses.

Subjects were solicited from the undergraduate
population Of the University of Texas-Austin. One hun-
dred sixty-eight subjects were administered the battery
Of pencil/paper research instruments used in the study.
Sixty-two were subsequently selected to participate in the
interview, on the basis of their high or low scores across
the nine personality dimensions used.

The three pencil/paper instruments used to assess
:personality characteristics were the Plutchik Emotions PrOw
_£ile Index, the Eysenck Personality Inventory, and the
Birkman Method. Subjects were divided into two groups
(high.and low) based on their scores on one of the per-
sonality dimensions at a time.

The interviewer was the same individual for all

145

subjects, and was instructed to engage in a responsive but
non-confrontive manner. Subjects were asked to select an

item of personal interest from the Mooney Problem Check

 

Hisp's "Personal Psychological Relations" scale, and then
to talk with the interviewer about that topic for 15
minutes.

The interviews were videotaped, and a systematic
sampling procedure was used to prepare the data for rating.
The last three seconds of each 20 second interval of the
interview was recorded as a discrete segment onto a master
tape, resulting in a total Of 1,860 samples for the study.
Samples from individual subjects were grouped in sets of
five, and arranged in semi-random order on the master
tapes so as to minimize response set on the part of the
raters. Raters subsequently scored 30 individual samples
of facial affect for each subject.

Three females were trained as raters for the
study. The training was an intensive five hour program
based on techniques developed by other researchers in the
area at Michigan State University (Inman, 1976; Wilson,
1976; Bowles, 1978). The training included the use Of
both slides and videotapes, with both posed and unposed
samples of facial affect. The six categories used in the

study (Interest, Anger, Disgust, Surprise, Enjoyment, and

146

Distress) were clearly defined, and criterion characteris-
tics labeled. Two additional measures, Change of Affect
and Range of Affect, were compiled on the basis of cal-
culating number of changes from one category to another
and number of categories used/from the rater scoring sheets.

A judgement for any one of the six categories
was defined as agreement between at least two of the three
raters. In the case where there was no agreement for a
specific sample, the sample was considered unscoreable
data. Judges did not begin the rating task until a .8
percentage of agreement had been established.

All hypotheses were stated so as to test the
relationship between performance on Specific personality
dimensions and the type, range, or frequency of change of
facial affect. Hypotheses were tested by a comparison of
groups on the affective measures, using t—tests, with an
alpha level set at .05. Supplementary analyses from the
multivariate, univariate, and correlational aspects were
also used for further exploration of differences and re-

lationships.

Results

The fifteen hypotheses used in the study were

designed to determine group differences on the affective

147

measures. The nine independent variables on which sub-
jects were divided into high and low groups were Gregar-
ious, Distrustful, Depressed, Dyscontrolled, Aggressive,
Getting Along With Others, Dominance, Extraversion, and
Neuroticism. The affective measures used were Enjoy, Dis-
tress, Disgust, Surprise, Anger, Interest, Range of Affect
and Change of Affect. All hypotheses were tested at an
alpha level of .05.

The null hypothesis was rejected for two of the 15
hypotheses. A significant difference was found between
high and low Distrustful groups on the affective measure
of Disgust (p;.008) indicating that persons who score high
on Distrustful tend to display more disgust than do per-
sons who score low on this scale. A significant difference
was also found between high and low groups of Getting
Along With Others on the affective measure Range of Af-
fect (ps.028), indicating that persons who score high on
Getting Along With Others tend to display more types of
affect than do persons scoring low on this scale.

While the hypothesized differences for Distrust-
ful and Getting Along With Others were significant and
congruent with Tomkins (1962, 1963), the lack of support
for the remaining hypotheses may, tentatively, indicate
a lack of empirical support for Tomkins' suggested rela-
tionships between facial affect and personality character-

istics. This lack of support might have been more a

148

measurement than a theoretical problem, since the variance
within groups indicated that means between high and low
were not as independent as would have been desired; never-
theless, the lack of significance for predicted relation-
ships and the evidence of significance for non-predicted
relationships (discussed under Findings) remains trouble-

some .

Discussion

 

The discussion section will be presented in two
parts: Findings and Methodology. The Findings section
will present conclusions and limitations related to the
hypothesized relationships between facial affect and per-
sonality dimensions, and to the supplementary analyses per-
tinent to this issue. The Methodology section will pre-
sent conclusions and limitations related to the use of
raters and affect categories in the study.

Findings

Significant findings of hypothesized differences
were limited, but additional t-tests did show evidence of
further differences between high and low groups on the
independent variables. The independent variable Dyscon-
trol was the single personality dimension which failed to

show any evidence of group differences on any one of the

 

 

 

‘ ML

149

affective measures.

The independent variables Gregarious, Distrust-
ful, Aggressive, and Getting Along With Others all showed
evidence of significant differences between their respec-
tive high and low groups on at least one affective
measure at p < .05. The high Gregarious group showed
less Disgust (p=0.27) and less Change of Affect (p=.034)
than did the low Gregarious group. The high Distrustful
group showed less Interest (p=.047) than did the low
Distrustful group. The high Aggressive group showed more
Disgust (p=.022) than did the low Aggressive group. The
high Getting Along With Others group showed greater Range
of Affect (p=.028 as reported under Hypothesis 6) than
did the low group.

In addition, Gregarious and Aggressive did show
group differences between means in hypothesized direc-
tions, but the high within-group variances may have
obscured significance. The high Gregarious group had a
higher mean than the low Gregarious group on the measure
Enjoy, at p=.065; the high Aggressive group had a higher
mean than the low Aggressive group on the measure Anger,
at p=.097 (see Appendix E for these tables).

The four remaining independent variables, Depres-
sed,Dominance, Extraversion, and Neuroticism, all showed

evidence of group differences, though not always in the

150

hypothesized direction, nor on the affective measure ad-
dressed in the hypothesis testing. Depressed showed no
evidence of high/low group differences on the hypothesized
measure of Distress, but did show group differences on
the measures of Interest, Anger, Range of Affect, and Per-
centage of Rater Agreement, at probability levels of .164
or less.

Dominance showed evidence of high/low group differ-
ences on the hypothesized measures of Distress and Change
of Affect, but in the direction opposite of that hypothe-
sized, at probability levels of .264 and .239, respectively

Extraversion showed evidence of high/low group dif-
ferences on the measures of Enjoy and Disgust, at proba-
bility levels of .174 and .175, respectively, and in the
hypothesized directions. In addition, there was evidence
for group differences on the measures Range of Affect,
Change of Affect, and Anger at probability levels ranging
from .063 to .118.

Finally, Neuroticism did not show evidence of
group differences on the hypothesized measure Change of
Affect, but did show evidence of group differences on
the measure Interest at p=.080.

The probability level of these later tests is such

that they must all be viewed with caution; nevertheless

151

there appears to be evidence beyond the two hypotheses
showing significant test results that high and low groups
on these personality dimensions do differ in their expres-
sions of facial affect when measured by the categories in
this study. While this evidence for further differences
is encouraging, it is nonetheless difficult to interpret.
Evidence of further differences between these groups on
the affective measures mentioned in some cases appears to
contradict theoretical relationships prOposed by Tomkins
(1962, 1963), and in many cases appears to be without
theoretical underpinnings and is consequently left unex-
plained.

Aside from the basic question of whether there is
sufficient theory upon which to base predictable differ-
ences, a second question is whether a broader population
containing more extreme personality scores might be needed
in order to find the differences postulated by this study.
Noting that there was evidence of a curvilinear relationship
between the independent and dependent variables, it is pos-
sible that while the population in this study did range
from high to low on the independent measures, it was too
homogeneous or too limited in its extremes to allow evi-
dence for clearer differences between groups.

In addition, the subjects used for high and low

152

groups on the independent measures overlapped, so that
one subject was used in from two to five of these high/
low subgroups, thereby decreasing the independence of
affective measures between these subgroups. The limita-
tions of the subject pool,then, both in terms of extreme
scores and in terms of overlap, may have contributed to the
lack of significant group differences, despite the fact that
group differences were found on some of the measures.

A third question is whether the differences pre-
dictedcxlthe affective measures between these personality
types are observable in the interview setting that was
used. While this setting was purposefully designed to be
a neutral stimulus so that natural interaction patterns
might emerge, it was limited to a 15 minute time frame,
and might have elicited only those patterns which are pre-
sent in the initial phase of an interaction between two
strangers. It is possible that samples of facial affect
over a much longer period of time would reflect an increase
of self-disclosure such as that described by Jourard (1964),
and would be different from the introductory behaviors
that were measured by this study.

Finally, there is the question of whether the af-
fective measures used as dependent variables were the ap-

propriate ones for which to test these differences. Since

153

these behavioral measures are considerably more specific
than the more global trait measures to which they were com-
pared, it may be that the information they yield is too
limited to establish significance on these grosser person-
ality measures. A more appropriate procedure might have
been to have established a single, primary mode of affect
for each subject which could then have been tested across
personality dimensions. A second possibility would have
been to collapse the specific affect categories into
grosser measures such as happy and unhappy, or into the
more general depressed/non-depressed categories used by
Ekman and Friesen (1974b) in their study of depression in
a clinical pOpulation.

In conclusion, while it was possible to establish
empirical support for two of the hypotheses, and to estab-
lish evidence of differences or of significant differences
between personality traits on the affective measures, it
remains unclear whether the lack of support for the re-
maining hypothesized differences is a result of inadequate
theory or of the limitations of the study. The following
section addresses further limitations in considering the
methodological aspects of the study.

Methodology

 

While rater agreement was generally high for the

154

task of rating moving facial affect, it was less than op-
timal for considerations of accuracy and replicability.

In comparison to Izard's study reporting percentage of
agreement among American females for labeling affect cate-
gories, the inter—rater reliability per affect category for
the raters in this study was generally five to fifteen
percentage points higher for each category (Izard, 1971,
pg. 272). For the categories of Anger and Surprise, how-
ever, agreement was considerably lower than than reported
by Izard (Anger, 48 vs. 70 percent; Surprise, 36 vs. 89
percent). Low percentages of rater agreement for Surprise
and Anger in this study were confounded by the fact that
the two affects appeared so infrequently they had few to-
tal ratings on which to establish a percentage of agree-
ment. A second consideration, however, is that of the six
types of affects used, Anger and Surprise are typically the
most intense but also the most fleeting of these. Thus
their transient nature not only makes them more difficult
to sample, but also more difficult to judge.

It is also of note that Izard's percentages of
agreement are based on judgments of photos or slides, thus
while an adequately high percentage of agreement has been
established for these categories on the basis of still sam-

ples, agreement based on this study suggests that these

155

categories might be much less reliable measures in the
judgment of moving facial affect.

Of the four categories used most frequently by
subjects (Enjoy, Interest, Disgust, and Distress), Dis-
gust and Distress had the lowest agreements and seemed the
most difficult to differentiate from one another (indi-
cated by Figures 3.2 through 3.7 showing rater disagree-
ment). It is difficult to determine whether the problem
was confusion about the distinctions between the two
categories, whether both appeared in the same 3-second
sample and there was disagreement about which to label as
predominant, or whether they were experienced simultaneously
by the subject and thus expressed asaablended affect.
Since Enjoy is also a frequent component of blended
affect, however (Ekman and Friesen, 1975),and since the
distinctions between Enjoy and either Distress or Dis-
gust seemed much clearer (Figurei3.6h it.seems more
likely that the problem was in adequately distinguishing
them from one another, rather than that they occurred as
blends or consistently together within the same samples.

The subjectivity of the rating task remains a per-
plexing issue. Informal observations suggested that it
was difficult for raters to maintainaaconsistent standard

for whether an expression of affect was sufficiently strong

156

to be labeled as such. While raters maintained agreement
with one another, they were more demanding of a strong,
clear affect display at the beginning of a rating session
in order to consider it different from neutral; toward the
end of a rating session they were more likely to accept
subtle expressions as sufficiently different from neutral
to be judged as one of the other categories.

Some subjects were more difficult to rate than
others, as evidenced by Figure 3.1, which indicates a range
of 30 percentage points of agreement across subjects.
There is also evidence to suggest that percentage of rater
agreement may vary with personality type in some cases.
The supplementary univariate analyses, for example, show
rater agreement as higher for those scoring low on Getting
Along With Others and on Dominance (ps.078; ps.077, respec-
tively) than for those who scored high on these measures.
(See Appendix E for Group Differences on Percentage of
Rater Agreement.)

A final problem was with the affect measures them-
selves, in that raters commented on differences between
some subjects which were not captured by the affect cate-
gories used in this study. On an informal basis, for ex-
ample, raters agreed that some faces were distinctly more

labile than others, and that this dimension was not always

157

reflected in the Change of Affect measure. Additionally,
they agreed that types of smiles between subjects were
qualitatively different (anxious smiles; rigid, mask—like
smiles; relaxed, casual smiles) and were not discriminated
for by the one measure of Enjoy.

In conclusion, for the task of rating moving
facial affect it was possible to establish normatively high
rater agreement for categories of Enjoy, Interest, Distress,
and Disgust, but not for Anger and Surprise. While this
might be reflected as a limitation of the study, in that
the neutral setting did not elicit these affects as fre-
quently, or in that the methodology did not provide for
prOportional sampling of these categories, it may also im-
ply that the reliability for these affects is not suffic-
iently high to use with moving facial affect.

In addition, it was difficult to distinguish be-
tween Distress and Disgust, suggesting that these two
categories may yield less reliable data and might profit
from the development of more clearly differential criteria
regarding their respective qualities.

While subjectivity was controlled for at the be-
ginning of each rating session by re-viewing the training
tape, there were differences in rater standards betweentme

beginning and end of a session, suggesting that this must

158

still be considered a limitation of the methodology as well
as of the study.

Finally, there were observable differences between
subjects that were not scoreable by existing measures,
which did not appear to influence the reliability of the
measures used, but did result in..a loss of data that
might have been pertinent to this particular study.

A notable deficit in the scoring of moving facial
affect is the lack of a standardized, packaged training
program such as Ekman's "Facial Affect Scoring Technique"
(Ekman, Friesen and Tomkins, 1971), which was developed
for use with still samples. While standards of reliability
established by percentage of rater agreement were viewed
as adequate for the purposes of this study, the replica-
bility is severely limited by the subjective and idiosyn-
cratic nature of the training program, and of the process
itself. Repeated use of this methodology with different
samples, however, might allow for the develOpment of a
standardized rater training program for use with videotape

or moving affect.

Implications

 

Suggestions for Further Research

 

The field of facial affect is a relatively new

159

one, and the possibilities for relating expressions of
facial affect to personality dimensions or using them as
psychodiagnostic information is even more recent (Haggard.
and Isaacs, 1966; Wilson, 1976a; Izard, 1971; Ekman and
Friesen, 1974b). While this study shows only limited em-
pirical support for these efforts, it nevertheless does
not contradict these possibilities. Of primary importance
is a tighter, more comprehensive theory to provide direc-
tion in this area. Continued research efforts can provide
more stimulus for theoretical models, as well as empirical
support or lack thereof for existing theory.

In order to work within the current state-of-the-
art perspective, more descriptive studies should be done in
order to factor out which variables can be most profitably
considered. Measurement, research conditions, and popu-
lation parameters must continue to be investigated in or-
der to determine which procedures may be most useful.

In order to overcome the possible effects of homo-
geneity within the sample, the current study should be
replicated with a larger sample, and with a sample substan-
tially more diverse on the abnormality/normality continuum
with respect to each of the independent measures used.

Finally, considerations of facial affect measure-

ment based on this study suggest that at this stage, it may

 

 

 

160

be more profitable to establish a primary mode of affect
for each subject, which can be compared across independent
measures, rather than to consider a multiple set of
affects in comparison to a diverse set of personality
dimensions, with an insufficient amount of theory to
direct hypothesized differences. Secondarily, it may be
more appropriate to use grosser measures such as micro-
momentary expressions or happy/unhappy dimensions than it
is to use the more discriminate categories of facial
affect to relate to the more global personality measures
such as those used by this study. It is of utmost
importance that methodological considerations continue to

form part of the basis for future research in this area.

Implications for Therapy

 

Due to the exploratory nature of this study and
the aforementioned limitations, results must be regarded
with caution, and particularly those which were not
formally hypothesized. There is evidence to suggest,
nonetheless, that the clinician who observes facial
affect might begin to speculate on the amount and type of
information available through these facial expressions.
Based on this study, the clinician might begin to observe

whether, indeed, a high frequency of the affect Disgust

161

signals personality characteristics of high Distrustful-
ness or Aggression, or of low Gregariousness; or whether
a wide range of affect signals a personality character-

istic of internal sensitivity, such as that described by
the Getting Along With Others scale.

Hopefully, for those who use nonverbal informa-
tion in the clinical setting, this research has pin-
pointed one particular area Of nonverbal behavior in a
manner that will stimulate further thought and observa-
tion, as well as having explicated it in such a way that
the limited amount of information we do now have in this

area can become useful.

 

- 4.. su.1...!...c.4..u.. .l.l.. . . f . . 4.... .. .. . -1... .

A . ..
. 1:! f1 71%).}; l. . . 4 . .l‘

 

APPENDIX A

SUBJECT CONSENT FORMS FOR

RESEARCH PARTICIPATION

162

 

 

1153

SHORT CONSENT FORM FOR RESEARCH STUDY

1, , hereby give my consent to
participate in research study on ,
the general plan of which has been explained to me including anticipated bene-

 

fits, risks, and potential complications.

I fully understand as it has been explained to me that by notice given
to the undersigned principal investigator that I may withdraw from this re-
search project anytime that I may elect to do so.

 

 

Parent's/Guardian's Signature in the Participant's Signature
case of a minor (under 18 yrs of age)

**********

I hereby certify that I have given to the above individua1(s) an explana-
tion of the contemplated study and its risks and potential complications.

 

 

Date Signature of Principal Investigator

 

 

164

H

 

0.53253

.nomomusm noumomou you

manomem smexu.m on» one mauow omega cm: 0» co.umwwsmouo no :Omuom damn n.5u ouwuosuse

.>m00u we >n pouoaeeco on maqu convex scexuwm one mo nwm>qmsm uwonu uo madame» onu

 

Ou uuomou ou ..0:H .moum«00mm¢ a scaxuwm caduceusm >nmuon H

fizmzmmmud ZOH9<AHU~BK<Q AdZOHmmthmm

 

 

 

 

om<

.8 O0 8 0033...? o:

.26. 35:330.. 85. xom

30:: 8a 30> c2632. .82an”. 8a»... .25 80.62.. .0083.an he. so 5332 so. 08: 2. =0: guesses. 9.0.6.3.. 2....

 

"50...? 3.. £9. .—

02 D 3> D ”80.2. 8.2.5.330 00502 550.0 05 082an". :2. 26..

 

see 3...? 80.8..

 

”on. 00.3.? seeds—ob £50092 no. .—

 

.850 D 35800 D 05.8560 D .5003... no— D

A28 b.0260 8:35.330 55...: mean—assoc 3.. 083%

 

 

 

8...... 00.. 388..
”3.3.0:... 388..
.5 23m .96 .925
uoasis—5.
0.00.: .3... .3
02.82 2. :8 =9. ”oEez

 

050:3 boar—.3: GEOF—m

 

6:5

hmmzm 20....(5mcuz.

 

 

 

 

.2... corpus. zSszm ure.

 

Obi .>wt

APPENDIX B

INTERVIEW MATERIALS

165

 

 

 

16.
17.
18.
19.
20.

71.
72.
73.
74.
75.

126.
127.
128.
129.
130.

181.
182.
183.
184.
185.

236.
237.
238.
239.
240.

291.
292.
293.
294.
295.

166

Items of the Personal Psychological Relations

Scale of the Mooney Problem Check List

Being timid or shy

Being too easily embarrassed

Being ill at east with other people
Having no close friends in college
Missing someone back home

Wanting a more pleasing personality
Losing friends

Wanting to be more popular

Being left out of things

Having feelings of extreme loneliness

Feelings too easily hurt
Being talked about

Being watched by other people
Worrying how I impress people
Feeling inferior

Being too envious or jealous

Being stubborn or obstinate

Getting into arguments

Speaking or acting without thinking
Sometimes acting childish or immature

Disliking someone

Being disliked by someone

Feeling that no one understands me
Having no one to tell my troubles to
Finding it hard to talk about my troubles

Too self-centered

Hurting other people's feelings
Avoiding someone I don't like
Too easily led by other people
Lacking leadership ability

167

VERBAL INSTRUCTIONS TO SUBJECTS DURING INTERVIEW

(As subject enters the interview room) Hi, my
name is . (Wait for subject to respond with
his/her name; if s/he doesn't, ask "What's your name?")
(Motion for subject to sit in appropriate chair while ex-
plaining.) This is the interview part of the research
study. We'll be talking for about 15 minutes about
something that is of concern to you. I'd like you to
take this stack Of cards (hand subject the index cards)
and just turn them.over, one by one, until you see a top-
ic that is of concern or interest to you. Stop with the
first topic you find that you can talk about; then just
let me know what the topic is as you begin to talk about
it.

Interviewer instructions, as previously stated in "Pro-
cedures" are to indicate understanding by nonverbal ac-
knowledgement, or to verbally reflect or briefly para-
phrase the content that the subject is presenting. The
interviewer will not challenge, nor lead the subject in

a specific direction, but rather engage in a more passive
style that simply encourages the subject to continue

talking.

If the subject exhausts the topic before the 15 minutes
are us, s/he may look at the cards again, select another
topic, and begin again. In the event that a subject ap-
pears particularly anxious and unable to explore any cho-
sen topic at length, the interviewer may respond to the
anxiety in a reflective manner, and encourage the sub-
ject to stay with a topic a little longer by considering
different aspects of the issue.

1158

Code #
Ageyyv Male Female

Fresh Soph Jr Sr

SELF-REPORT FORM

1) Check any of the following adjectives that apply to how you have been
feeling in general today:

ANGRY CONTENT DISAPPOINTED upszr suapaxszn
HAPPY IRRITATED ANXIOUS SAD INTEREst WORRIED
DISGUSTED SCARED FRUSTRATED CYNICAL EMBARRASSED

2) Here any of these feelings particularly strong for you during the )5 minute
interview? If so, which one(s)?

3) How are you feeling now?

4) Please check the level of comfort you felt in talking with the interview person:

L l J l 1 a
I j I . T V V
very mostly slightly slightly mostly very
comfortable comfortable comfortable uncomfortable uncomfortable uncomfortable

5) Here you aware of any emotions that you experienced during the interview,
but tried not to let the interview person know about? If so, please list
those emotions here:

Thanks again for your cooperation. Your participation has been greatly
appreciated.

Fran Stott,Staff Member Augustine Baron, Jr.. Psy.D.
Counseling Center, Chairman, Research Committee
and Principal Investigator Counseling Center

169

AUDIO/VISUAL RELEASE FORM

I understand that my signature below gives the experimenter
permission to keep and use the audio/visual tape recording made
during my interview, and to take selected samples from it in the
preparation of a master tape that will be used to rate facial
affect.

I understand that the primary purpose of these tapes will be
for the collection of dissertation data, and that their primary
use will be for observation and scoring by trained judges par-
ticipating in the study. I also understand that segments of
the tapes may on occasion be used for teaching purposes or
presentations to professional groups.

I understand that the confidentiality of the material will
be protected, and that in no way will my name or other personal-
ly identifying information be linked with the audio or video
taped material. I also understand, however, that inasmuch as
my facial features will be a part of the recording, there is a
possibility that I may be recognized by those to whom the re—
corded material is played, and for this I do not hold the
experimenter or her associates reSponsible.

SIGNATURE

 

DATE

170

EXPLANATION OF THE STUDY

There is a good deal of recent literature to
suggest that one's emotions or emotional experiences can
be studied on the basis of facial expressions. The type
of emotions that we express on our faces, the emotions
that we try to hide or cover up, and even the emotions
that never appear can all be clues to the emotions we
are experiencing, or to our emotional experience.

This project has undertaken to study the facial
expression of emotion in three ways: by the types of
emotions that are expressed facially, by the range of
types that are expressed, and by the frequency of change
from one emotion to another. This information about the
facial expressions will be correlated to information
about particular personality characteristics as measured
by the three instruments completed earlier in the study,
in order to examine possible relationships between
personality characteristics and facial affect.

The interview setting was designed so that
participants would be talking about a problem or personal
concern, and so that the facial expressions recorded
would to some extent provide information about the emo-
tions the individual normally experiences in a problem
solving setting. This information, in combination with

the other test instruments, should provide new and useful

171

information regarding the manner in which students cope
with problems. It is hoped that this information can be
useful in designing programs and interventions that will
be helpful to students in the future.

We would prefer that you not discuss the focus on
facial affect with other participants until after all
interviews have been completed. Our experience has been
that people become very self-conscious when aware that
their faces are being studied, thus this information
could distort or alter the facial expressions we are
videotaping. Our preference is that participants be
aware of the more general nature of the study--the issue
of problem solving styles--until they are de-briefed
following this interview.

Again we thank you for your cooperation and
participation in this study. Should you wish to receive
a report on the outcome of the research, please write

your name and address on the attached sign-up sheet.

Fran Stott, Principal Investigator
Counseling Center

University of Texas, Austin

APPENDIX C

RATER TRAINING MATERIALS

172

I73

CROSS-CULTURAL SIMILARITIES IN FREE-RESPONSE LABELING
0F EXPRESSIONS (Izard)
N= 268: 89 Americans, 62 British, 67 French, 50 Greeks

A_Priori definitions of emotion categories are centered, followed by
correct transcultural free-response labels in columns.

 

INTEREST-EXCITEMENT
concentrating, attending, attracted, curious

attentive fervor (b) questioning
concentration inquisitive reflection (a)
concern (a) interest religious fervor (b)
contemplation (a) ovservation (b) seriousness
curiosity pensive somber reflection
deliberating pondering thoughtfulness
excitement (b) puzzlement wonder

expectation (a)

ENJOYMENT-JOY
glad, merry, delighted, joyful

amusement gratitude (b) playful

bliss (a) happiness pleasantness

clowning humor (a) pleasure

contentment jovial rapture

delight (a) joy satisfaction

ecstasy laugh sees something pleasant
elation merry self-satisfaction (a)
enjoyment (b) mystical ecstasy serenity (a)

gaiety optimism smile

glee (a)

SURPRISE-STARTLE
sudden reaction to something unexpected, astonished

amazed pleasant astonishment surprise
amused surprise (b) pleasant surprise (b) joyful surprise
astonishment shock startle (a

fearful astonishment

 

a; most commonly used by females
b most commonly used by males

174

DISGUST-CONTEMPT
sneering, scornful, disdainful, revulsion

aversion (b) dislike scorn
contempt distaste skepticism
cynical insolence (b) smirk (a)
derision mockery (a) smug (b)
disapproval repugnance (a) sneer
disdain repulsion superiority
disgust sarcasm
ANGER-RAGE

angry, hostile, furious, enraged
aggressive furious revenge (a)
anger fury spite
bitterness (b) mad vengeful (a)
enmity (a) rage vexation (b)

ferocity (a)

DISTRESS-ANGUISH
sad, unhappy, feels like crying

about to cry (a) grief sad

anguish hurt sorrow

bad news loneliness suffering (a)
crying melancholy (b) troubled (a)
dejected misery (a) uneasiness
dejection not going well unhappy
depression pain (a) unloved (a)
despair pathetic upset (a)
disappointment pity (b) worry
distress

 

From Izard, Carroll, 1971.

A.

B.

C.

175

DESCRIPTION OF FACIAL AFFECT CATEGORIES

INTEREST

(l)

(5)

Description: Interest is usually aroused when one spontaneously
attends to the environment.

Duration: Most people can maintain this affect over long periods
of time.

Function: The function of Interest is taking in information about
the environment.

Synonyms: Curious, concentrating, attending, absorbed, involved,
attracted, intense, fascinated.

Facial Components: Eyebrows level but may be slightly raised or
lowered. Eyes open and usually fixated. Lips may be parted and
jaw dropped slightly.

ENJOYMENT

(l)

(2)

(3)

(5)

(1)

Description: Enjoyment is most people's favorite affect--
stimulated either by the onset of pleasant stimuli or the
cessation of unpleasant/painful stimuli.

Duration: The duration of this emotion varies with intensity.
Mild enjoyment can be maintained for extended periods of time
while ecstasy is maintained for shorter periods.

Function: The function of Enjoyment includes both an internal
component (self reinforcement) and an external component
(social bonding).

Synonyms: Happy, glad, merry, joyful, cheerful, blissful,
jubilant, gay, elated, ecstatic, gleeful, jovial.

Facial components: Eyebrows level or slightly lowered. Eyes
bright or may be partially closed; eyes often have wrinkles
("crows' feet") in the outer corners. Corners of the mouth
lifted back and up (exaggerated with laughing). Teeth often
partially exposed, upper lip tensed. Nasolabial folds (run-
ning from the nose to the outer edge of the mouth) are
evident.

SURPRISE

Description: Surprise is triggered by the unexpected. If one
anticipates an event, then it is impossible to be startled or

D.

E.

176

surprised. Dramatic examples of stimuli that trigger the
surprise/startle response include an unexpected gunshot, or
a flashbulb going off close at hand.

(2) Duration: Surprise is distinguished by its very brief dura—
tion. It is sudden in its onset and fades quickly.

(3) Function: The function of Surprise is the clearing of the
sensory systems for assimilation of new information.

(4) Synonyms: Astonished, amazed, shocked, flabbergasted.

(5) Facial Components: The brows are raised so that they are
curved and high, the forehead wrinkles horizontally. The
eyes are widened so that the white is visible above and some—
times below the pupil. The jaw drops so that the lips are
loosely parted.

DISGUST

(1) Description: Disgust is aroused by a disagreeable sensory
experience: a "yeechhy" taste, a bad smell, an unpleasant
sight, or a repulsive feeling object. More complex stimuli,
such as ideas, things, or people can arouse disgust as well.
There is an aloofness and a distancing or "put-down" com-
ponent to the emotion Disgust; a sense of getting away from
or getting rid of.

(2) Duration: In mild to moderate forms, Disgust may be experienced
for extended periods of time. However as disgust increases in
intensity, so does the probability of nausea and vomiting, thus
people often try to "leave the scene" before the feeling be-
comes this intense.

(3) Function: The function of Disgust is expulsion, a getting rid
of or getting away from the disagreeable object or experience.

(4) Synonyms: Scornful, disdainful, skeptical, condemning, critical,
arrogant, sarcastic, spiteful, revolted, indignant.

(5) Facial Components: The brow is lowered; one or both cheeks are
raised; the nose is sometimes wrinkled. One or both sides of
the upper lip are raised and protrude slightly; teeth are
sometimes exposed. At times the tongue is slightly extended.

ANGER

(l) Description: Anger is likely to be triggered by frustration,
physical threat, psychological hurt, violation of values, and

F.

177 -3-
failure to fulfill expectations.

(2) Duration: Anger can be of short or long duration depending on
circumstances and the reaction received from the environment.
Duration does not vary with intensity; one may be mildly angry
for a short or a long period of time and one may also be in-
tensely angry for a short or a long period of time.

(3) Function: The function of Anger is to mobilize the individual's
resources for confrontation with a disturbing element in the
environment. Further, Anger serves to arouse fear in others--
fear of loss of control on the part of the angry individual--
creating distance.

(4) Synonyms: Irritated, hostile, mad, hateful, aggressive, annoyed,
cranky, cross, disagreeable, furious, resentful.

(5) Facial Components: Brows are lowered and drawn together, creating
vertical lines in the forehead. Eyes have a hard stare and a
bulging appearance. Lips are either firmly pressed together
with corners down or drawn back in a squarish shape, baring the
teeth.

DISTRESS

(1) Description: Distress is stimulated by loss: loss of a loved one,
loss of an opportunity, loss of self-esteem, loss of health, loss
of security are all experiences which stimulate feelings of distress.
One can also experience distress in anticipation of a loss.

(2) Duration: Distress is often a prolonged feeling, lasting more
than a few minutes, often hours, sometimes days, and occasionally
a year or more.

(3) Function: Distress communicates deficiency to others in the environ-
ment and attracts help-givers.

(4) Synonyms: Sad, unhappy, miserable, hurt, dejected, depressed,
despondent, dismal, low, grieved, suffering, worried.

(5) Facial Components: Inner corners of the eyebrows are drawn up,
vertical wrinkles in the forehead often appear. The skin below
' the forehead is triangulated with the inner corner up. The corners
of the mouth are down and the lower lip may be trembling.

 

From Inman, David, 1976.

 

179

 

Surprise*

 

*Taken from Ekman and Friesen slide series (1976).

180

 

Distress*

 

*Taken from Ekman and Friesen slide series (1976).

181

 

Disgust*

 

*Taken from Ekman and Friesen slide series (1976).

182

.4w’.~..w.....;‘ ,

.a- "“

 

Anger*

 

*Taken from Ekman and Friesen slide series (1976).

183

 

Enjoyment*

 

*Taken from Ekman and Friesen slide series (1976).

184

«1.4.,

Tm.“

49"
'md“"‘ a» ,1:
w
34

3, ,‘x
>

 

Interest*

 

*Taken from Ekman and Friesen slide series (1976).

APPENDIX D

RATER AND SCORING FORMS USED IN

MEASUREMENT OF FACIAL AFFECT

185

186

 

all

 

 

 

 

Atlanta

 

6011‘

 

 

 

 

1.2.3.41...

 

Coda”

 

 

 

 

 

Lasts.

 

Codi“

 

 

 

 

 

 

 

M1

 

 

 

 

 

Cola."

 

 

 

 

 

0041.11

 

 

 

 

 

Com!

 

 

 

 

 

1.11%.»).

 

Mt

 

 

 

 

1.1.14.5

 

 

Cecilat

 

 

 

 

L2.3.d..$.

 

 

Coda.”

 

 

 

 

 

1.134s

 

Codi?!

 

 

 

 

 

1.2.3.4.:

187

 

 

 

Femiooaw

$5.230
nrmdhns

63.2».
Lnoona

.33 mt:

#7.
7.
“7.
Z.
i.

”7.
J.

 

 

 

 

 

 

 

# Dam

a< ' a.
W'

11 4451A—

45:“
404.344!
44
dgoLu-qu
#chfio'muA-IJ.____

 

188

Ill/TE Les‘r
D15 6 031'
SULPBI$€

a W
4 DI.

1‘ 4A5.“—
14 Dim...
41W

“W

11% olefin
fiafigom‘awwl. §
t%6h~tv‘é(“ can; i

905‘]; 201.14 I‘lfi‘.uo".‘v~? 9] 6/6

 

APPENDIX E

SUPPLEMENTARY UNIVARIATE ANALYSES

189

 

190

T-test Results of Depressed Group Differences

on the Affective Measures

 

 

 

Dependent Variable E 2 Significant
Percentage of Interest -1.45 .154 No
Percentage of Disgust 0.12 .907 No
Percentage of Anger 1.42 .164 No
Percentage of Distress* 0.36 .361 No
Percentage of Surprise 0.86 .395 No
Percentage of Enjoy 0.20 .840 No
Percentage of Change 0.74 .466 No
# of Affects Used (Range) 1.53 .133 No

 

*Hypothesis 3 (reported as a one tailed test).

191

T-test Results of Dyscontrol Group Differences

on the Affective Measures

 

 

 

Dependent Variable E 2 Significant
Percentage of Interest -0.25 .805 No
Percentage of Disgust 0.12 .902 No
Percentage of Anger -0.86 .394 No
Percentage of Distress 0.49 .627 No
Percentage of Surprise* -0.32 .374 No
Percentage of Enjoy 0.13 .900 No
Percentage of Change 0.63 .533 No
# of Affects Used (Range) 1.18 .246 No

 

*Hypothesis 4 (reported as

a one-tailed test).

192

T-test Results of Getting Along With Others Group

Differences on the Affective Measures

 

 

 

Dependent Variable ‘E p_ Significant
Percentage of Interest -0.19 .853 No
Percentage of Disgust* 0.93 .175 No
Percentage of Anger -0.10 .921 No
Percentage of Distress 0.52 .607 No
Percentage of Surprise 1.08 .291 No
Percentage of Enjoy* -0.91 .185 No
Percentage of Change* 1.35 .095 No
# of Affects Used (Range)* 2.00 .028 Yes

 

*Hypotheses 6-9 (reported as one tailed tests).

 

193

T-test Results of Dominance Group Differences

on the Affective Measures

 

 

 

Dependent Variable ‘t 2, Significant
Percentage of Interest 0.55 .589 No
Percentage of Disgust 1.09 .284 No
Percentage of Anger 0.98 .335 No
Percentage of Distress* 0.64 .264 No
Percentage of Surprise -1.61 .118 No
Percentage of Enjoy -1.91 .066 No
Percentage of Change* 0.72 .239 No
# of Affects Used (Range) 0.03 .974 No

 

*Hypotheses 10-11 (reported as one tailed tests).

194

 

T-test Results of Extraversion Group Differences

on the Affective Measures

 

 

 

 

Dependent Variable E 2 Significant
Percentage of Interest -0.11 .911 No
Percentage of Disgust* -0.82 .207 No
Percentage of Anger -1.70 .096 No
Percentage of Distress 0.51 .615 No
Percentage of Surprise -1.31 .196 No
Percentage of Enjoy* 0.95 .174 No
Percentage of Change -1.59 .118 No
# of Affects Used (Range) -1.90 .063 No

 

*Hypotheses 12-13 (reported as one tailed tests).

 

 

1.95

T-test Results of Neuroticism Group Differences

on the Affective Measures

 

 

 

Dependent Variable ‘E 2 Significant
Percentage of Interest -l.79 .080 No
Percentage of Disgust 1.06 .296 No
Percentage of Anger -0.07 .943 No
Percentage of Distress 0.31 .760 No
Percentage of Surprise -0.88 .386 No
Percentage of Enjoy 0.49 .627 No
Percentage of Change* -0.61 .273 No
# of Affects Used (Range) -0.84 .403 No

 

*Hypothesis 14 (reported as a one tailed test).

 

196

T-test Results of Group Differences

on Percentage of Rater Agreement

 

 

 

 

_ Signif-
Group ‘3 x SD ‘3 p, icant
22222;. 223“ 22 32:22 3:22 0-44 -663 No
Ziggoti- High 27 76.81 7.58 _.40 .688 No
Low 23 77.69 7.78
Siigari' 222“ 23 32:22 2:22 -68 ~499 No
3:3: :3 3:32 3:3: N.
2:31: 2: 33:22 3:32 N.
2:25:25 High 22 76.36 8.81 _.24 .812 No
Low 17 77.00 7.42
222263‘ 222h 23 32:22 2:22 --83 .412 No
GA”° 222“ 22 32:22 2:32 -1-84 .073 No
D°minan°e 222“ 22 22:22 2:22 -l-84 ~077 No

 

T-test Results of Group Differences on

Number of Rater Discards

197

 

 

 

 

Signif-

Group E x SD ‘3 E icant
22222Sn 222h 22 2:22 2222 -03 ~974 No
2222°ti' 222h 22 2222 2232 ~21 -836 No
Egigar— High 25 2.64 1.72 -1.40 .171 No

Low 17 3.41 1.80

2222““St' 222“ 3; 3:23 1:32 -o.51 .609 No
......... 222“ 2: 2:23 2:3: ... .... N.
222222; 222“ 22 2:22 2:22 -06 -949 No
2:258- 2:3: :3 3:2: 2:32 N.
:23“ 22 2:2: 1:3: N.
2222; 222“ 22 2:22 2:22 1-04 .307 No

 

 

APPENDIX F

24 X 24 INTERCORRELATIONAL MATRIX

198

1199

mmwm .coAAouucou .cAEAe .Aswumoua .msoAummouu .o>Aomoucm¢ .Aomumzugon .oomnoummc .UOAAouucoomhc "Homo xAnuusAm

coma Mom omcm mczu mam muom Dunc hand ammo HucH 0349 son

oc.A cc. m5. mh.
oc.A on. no.
oc.A on.

oc.A

om.
No.

mp.
co.
oc.A

mm.
mm.

mm.
on.
cA.
oc.A

omcmm .omcmnu .honcm .umAumuom .mmouuon .uom:< .umamnAo .umououau "mundane: o>Auoowu<

om.
nm.

no.
no.
on.
NA.
oc.A

mA.
nN.

An.
on.
nA.
vo.1
nA.
oc.A

No.
co.

no.
no.
Av.
cA.
nn.
hA.
oc.A

me.
oh.

nb.
m5.
Av.
on.
hv.
mo.
mm.
oc.A

pumomwo .ucoaomumc "moucmooz noummo

p

mumcuo 29A: mcoA¢ chuuoo .oocchEoo "vogue: cmEqumo

Ho.- oo.u mH.-
so.- po.s SH.-
no.1 mo.. AH.-
co. mc.- cA.-
ho.- cA.- so.u
.mo. ~H.- No.
me. on.. «H.-
Na. NH. H~.-
Ho.- No. n~.-
vo.- Ho.n co.
oc.A mm. v~.-
oc.A AH.-
oc.A

n

oAmom 0AA .EMAOAu0uooz .coAmuo>muuxm "Ham xocomxmm

«can ucoo oefia
so. no.1
ma. co.u
SH. oc.i
AM. No.1
mo. cc.
ho. Ho.
mo. HH.-
mo. cA.-
Ha. ma.u
ma. mo.
cm. «A.
«o. oo.1
me. «e.
oc.A mm.

oc.A

umHB

hA.1
oA.1

nN.1
nN.1
cA.1
no.1
AA.1
NN.I
o~.1
no.1

vN.1
VA.1

on.
vn.1
nm.
oc.A

mono hood uon ammo coho 0AA

cA.1
oA.1

NN.1
nN.1
AA.1
mo.

no.1
NN.1
m~.1
no.1

An.1
nN.1

Ac.
mv.1
5A.!
or.
oc.A

cA. AA.
AA. cA.
mA. cA.
mA. VA.
mo. mo.
Ac. no.1
AA. cA.
cA. oA.
AN. hm.
no.1 no.
cc. cA.
nA. FA.
oo.1 Ao.1
on.1 no.
cc.1 hn.1
vo.1 cm.1
ov.1 Am.1
oc.A av.
oc.A

ho. Ac.1 no.1
Ao. No.1 no.1
mo. co. no.1
mo. Ac.1 vo.1
no. cc. no.1
mo. no.1 Ac.
vo. no. no.1
mA. no.1 oo.1
no. no.1 no.1
no.1 no.1 Ao.
AA. nA.1 cN.1
vc.1 co. hA.1
mm.1 co. cA.
NA.1 mA.1 no.
no.1 om.1 nN.
hn.1 oc.1 on.
nn.1 mA. cA.
vv. AN. cA.1
no.1 «A.1 vm.1
oc.A mn.1 No.
oc.A cA.1
oc.A

usoz uuxm

no.
No.

vc.
no.
mo.
no.1
no.
vo.
cc.
no.1

on.
cA.

no.1
no.
AA.1
cn.1
nn.1
mm.
on.
mm.
mm.1

n~.1
oc.A

ma.u coma
~H.i oum<
oa.u omam
FH.- menu
oo.u flaw
cA.- dusm
oo.u name
«A.1 umc<
~H.- ammo
ma.u ouocH
mH.- 0340
mo. oeoo
«a. mafia
«v.1 ucoo
94.- cans
Na. anus
«e. mono
Ho.- name
no.1 yawn
o~.u ammo
an house
¢~.- and
mo.. usmz
oc.A auoxm

 

BIBLIOGRAPHY

200

 

BIBLIOGRAPHY

Abbott, R. "Improving the validity of affective self-
report measures through constructing personality
scales unconfounded with social desirability: A
study of the personality research form." Educa-

tional and Psychological Measurement. 1975, 35,
361-370.

Adam, A. Grandeur and Illusion; French Literature and
Society 1600-1715. London: Weidenfeld and
Nicolson, 1972.

Allen, J. "Correlates of emotional styles." Journal of

Consulting and Clincial Psychology. 1976, 44(4),
679 . _

Allen, J. & Hamsher, J.H. "The development and validation
of a test of emotional styles." Journal of Con-
sulting and Clinical Psychology. 1974, 42(S),
663-668.

 

Argyle, Michael. "The laws of looking." Human Nature.
January, 1978, 32-40.

 

Arnold, M. Emotion and Personality: Neurological and
Physiological Aspects. New York: Columbia Univer-
sity Press, 1960a.

Arnold, M. Emotion and Personality: Psychological Aspects.
New York: Columbia University Press, 1960b.

Arnold, M. (Ed.) Feelings and Emotions - The Loyola Sym-
posium. New York: Academic Press, 1970.

Ax, A.F. "The physiological differentiation between fear
and anger in humans." Psychosomatic Medicine.
1953, 15(5), 433-442.

 

Birkman & Associates, Inc. The Birkman Method: An Inter-

pretative Handbook for Participants. Unpublished
manual, 1974*

 

 

 

*Materials available from Birkman & Associates,
Inc., 3637 West Alabama, Houston, Texas 77024.

201

 

202

Birkman & Associates, Inc. The Birkman Method: Reliabili-
ties and Validities for Business and Industry. Un-
published report, January 15{T1975.*

 

 

Birkman, R.W. Development of a Personality Test Using
Social and Self-PerceptiOn Inventories. Unpub-
Iisheleh.D. Dissertation, University of Texas-
Austin, 1961.*

 

Black, P. (Ed.) Physiological Correlates of Emotion. New
York: Academic Press, 1970.

 

 

Bowles, J.M. Learnedgelplessness: Personality Correlates
and Affective Re5ponses. UnpublishediP .D. Disser-
tation, Michigan State University, 1978.

 

Brady, J.V. "Endoctrine and autonomic correlates of
emotions." in Black, P. (Ed.), Physiological
Correlates of Emotion. New York: Academic Press,
1970.

 

 

Buros, O.K. (Ed.) The Seventh Mental Measurements Year-
book. Highland Park, New Jersey: The Gryphon
Press, 1972.2

 

Candland, D.K. "The persistent problems of emotion." in
Candland, D.K. gt 31., Emotion. Monterey, Cali-
fornia: Brooks/Cole Publishing Co., 1977.

Chekhov, A. "The Kiss." in Mack, M. (Ed.), World Master-
pieces. New York: W.W. Norton and Company, 1965.

 

Chevalier-Skolnikoff, S. "Facial expressions of emotion
in nonhuman primates." in Ekman, P. (Ed.),
Darwin and Facial Expression: A Century of Re-
search in Review. New York? Academic Press, 1973.

 

 

Clynes, M. Sentics: The Touch of Emotions. Garden City,
New York: Anchor Press7Doubleday, 1977.

 

Coleman, J.C. "Facial expressions of emotion." Psycho—
logical Monographs. 1949, 63 (1, whole No. 296).

 

Darwin, C. The Expression of the Emotions in Man and
Animals. New York: D. Appleton and Co., 1910.
(originally published in 1872.)

 

203

Diagnostic and Statistical Manual of Mental Disorders.
Washington, D.C.: American Psychiatric Assoc., 1968.

Dougherty, F., Bartlett, E., & Izard, C. "Responses of
schizophrenics to expressions of fundamental emo-
tions." Journal of Clinical Psychology. 1974,
'5g(3), 243-246.

 

Ekman, P. "Body position, facial expression, and verbal
behavior during interviews." Journal of Abnormal
and Social Psychology. 1964, 55, 295-301.

 

 

Ekman, P. "Differential communication of affect by head
and body cues." Journal of Personality and Social
Psychology. 1965, 2f5), 725-735,

 

 

Ekman, P. "Universals and cultural differences in facial
expressions of emotion." in Cole, J.D. (Ed.),
Nebraska Symposium on Motivation, 1971. Lincoln:
Universify of Nebraska Press, 1972.

 

Ekman, P. "Universal facial expressions in emotion."
Studia Psychologica, 1973a, 55(2), 140-147.

 

Ekman, P. Darwin and Facial Expression. New York:
Academic Press, Inc., 1973b.

 

Ekman, P. & Friesen, W. "Head and body cues in the judg-
ment of emotion: a reformulation." Perceptual
and Motor Skills. 1967, 24, 711-724.

 

 

Ekman, P. & Friesen, W. "Nonberbal leakage and clues to
deception." Psychiatry. 1969a, 55(1), 88-106.

 

Ekman, P. & Friesen, W. "The repetoire of nonverbal be-
havior: Categories, origins, usage, and coding."
Semiotica. 1969b, 5, 49-98.

 

Ekman, P. & Friesen, W. "Constants across cultures in the
face and emotion." Journal of Personality and
Social Psychology. 1971, 11(2), 124-129.

 

 

Ekman, P. & Friesen, W. "Detecting deception from the
body or face." Journal of Personality and Social
Psychology. 1974a, 25(3), 288-298.

 

 

204

Ekman, P. & Friesen, W. "Nonverbal behavior and psycho-
pathology." in Friedman & Katz (Eds.), The Psy-
chology oleepression. Washington, D.C.: V.H.
Winston & Sons, 1974b.

 

Ekman, P. & Friesen, W. Unmasking_the Page: A Guide to
Recognizing Emotions from Facial Expressions.
Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1975.

 

 

Ekman, P. & Friesen, W. "Pictures of Facial Affect"
(slide series). Palo Alto, California: Consulting
Psychologists Press, Inc., 1976.

Ekman, P., Friesen, W. & Tomkins, S. "Facial affect
scoring technique: A first validity study."
Semiotica. 1971, 5! 37-58.

 

Ekman, P., Friesen, W. & Ellsworth, P. Emotion in the
Human Face. New York: Pergamon Press, 1972.

 

Ekman, P. & Oster, H. "Facial Expressions of Emotion."
Annual Review of Psychology. 1979, 55, 527-554.

 

 

Ekman, P., Sorenson, E.R. & Friesen, W. "Pan-cultural
elements in facial displays of emotion." Science.
1969, 164, 86-88.

Eysenck, H.J. & Eysenck, 8.8. Manual for the Eysenck
Personality_Inventory. San Diego: Educational and
Industrial Testing Service, 1968.

 

 

Eysenck, H.J. & Eysenck, S.B. Personality Structure and
Measurement. San Diego: Robert Knapp, Publisher,
1969.

 

 

Finalyson, G. "Effects of exposure to and re-appraisal of
facial expressions depicting affect.” Dissertation
Abstracts International. 1974, 55(10-B), 5188.

 

 

Forrest, D.W. & Lee, S.G. "Mechanisms of defense and
readiness in perception and recall." Psychological
Monographs. 1962, 15(4), (No. 523).

 

Freud, Sigmund. The Basic Writings of Sigmund Freud. New
York: Random House, 1938.

 

 

205

Frois-Wittmann, J. 'The judgment of facial expression."
Journal of Experimental Psychology. 1930, 55,
I13-151.

 

Gordon, Jesse, "Interpersonal predictions of repressors
and sensitizers." Journal of Personality. 1957,
35, 686-698.

 

Haggard, E. & Isaacs, K. "Micromomentary facial expres-
sions as indicators of ego mechanisms in psycho-
therapy.” in Gottschalk, L. & Auerbach, A. (Eds.)
Methods of Research in Psychotherapy. New York:
Appleton Century Croft, 1966, 154-165.

 

Hanawalt, N.G. "The role of the upper and lower parts of
the face as a basis for judging expressions."
Journal of General Psychology. 1944, 55, 23-36.

 

Harrison, R. & Crouch, W. "Nonverbal communication:
theory and research." in Hanneman, G. & McEwen,
W. (Eds.), Communication and Behavior. New York:
Addison-Wesley, 1975.

 

Harrison, R. Controversy On the Face. Kenilworth, N.J.:
Schering Corporation, 1977.

 

Harvey, N. The Feelings of Man: Their Nature, Function,
and Interpretation. Baltimore: WarwiCk & York,
Inc., 1914.

 

 

Hill, D.J. The Elements of Psychology: A Textbook. New
York: Sheldon, 1888.

 

Hinds, W.C. "Compacted affect: a therapeutic concept."
Paper presented at 84th Annual Conference of the
American Psychological Association, Washington,
D.C., September 7, 1976.

Inman, D.J. "Investigation of affective facial expressions
through slow motion and normal speed videotape
techniques." Unpublished Ph.D. Dissertation,
Michigan State University, 1976.

Izard, C. The Face of Emotion. New York: Appleton-
Century-Crofts, Meredith Corporation, 1971.

 

 

206

Izard, C. Patterns of Emotions: A New Analysis of Anxiety
and Depres§ion. New York: Academic Press, 1972.

 

Izard, C. Human Emotions. New York: Plenum Press, 1977.

 

James. W. The Principles of Psychology. New York: Holt,
1890.

Jourard, S. The Transparent Self. New York: Van Nostrand
Reinhoid Company, 1964.

 

Jung, C.W. Analytical Psychology: Its Theory and Practice.
New York: Random House, 1968.

Kaywin, L. "Notes on the psychoanalytic theory of affect."
Psychoanalytic Review. 1966, 55, 112-118.

 

Keen, Ernest. "Emotion in personality theory." in
Candland, D.K. 55 al., Emotion. Monterey, Califor-
nia: Brooks/Cole PEEIisEing'Company, 1977.

Kell, B.L. & Mueller, W.J. Impact and Change: A Study
of Counseling Relationships. New York: Appleton-
Century Crofts, I966.

 

Laird, J. "Emotion and touch: interesting observations
and a discordant theory." Contemporary Psychology;
1977, 55, 570-571.

Langfeld, H.S. "The judgment of emotions from facial ex-
pressions." Journal of Abnormal and Social Psy-
chology. 1918, 55, 172-184.

Lazarus, R.S. "Emotions and adaptation: conceptual and
empirical relations." in Arnold, W. (Ed.),
Nebraska Symposium on Motivation. Lincoln: Uni-
versity of Nebraska Press, 1968.

Lazarus, R., Averill, J. & Opton, E. "Psychological ap-
proaches to the study of emotion." in Arnold, M
(Ed.), Feelings and Emotions: The Loyola Sympos-
igm, New York: Academic Press, 1970.

Leathers, D.G. Nonverbal Commupication Systems. Boston:
Allyn & Bacon, Inc., 1976.

 

 

207

Mar, T. Face Reading. New York: The New American Library
1975.

 

 

Mooney, R. Mooney Problem Check List. New York: The Psy-
chological Corporation, 1950. '

Munn, N.L. "The effect of knowledge of the situation upon
judgment of emotion from facial expression."
Journal of Abnormal and Social Psychology. 1940,
55, 324-338.

 

Plutchik, R. "The Multifactor-analytic theory of emotion."
Journal of Psychology. 1960, 22x 153-171.

 

Plutchik, R. The Emotions: Facts, Theories and A New
Model. New York: Random House, 1962.

Plutchik, R. "What is an emotion?" Journal of Psychology.
1965, 55, 295-303.

 

Plutchik, R. "Emotions as adaptive reactions: Implications
for therapy." Psychoanalytic Review. 1966, 55,
106-110.

Plutchik, R. "Emotions, evolution and adaptive processes."
in Arnold, M. (Ed.), Feelings and Emotions: The
Loyola Symposium. New York: Academic Press, 1970.

 

Plutchik, R. "Individual and breed differences in approach
and withdrawal in dogs." Behavior. 1971, 55,
302-311.

Plutchik, R. Foundations of Experimental Research. New
York: Harper & Row, 1974.

Plutchik, R. "Cognitions in the service of emotions." in
Candland, D. 55 55., Emotion. Monterey, Califor-
nia: Brooks/Cole Publishing Company, 1977.

Reymert, M. (Ed.) Feelings and Emotions: The Wittenbepg
Symposium. Worcester, Mass: Clark University
Press, 1928.

 

Rockberger, H. "The role of an eclectic affect theory
in multiple therapy." Psychoanalytic Review.
1966, 55, 120-128.

208

Romanes, G. Mental Evolution in Animals. New York:
Appleton, 1884.

 

Sackeim, H., Gur, R. & Saucy, M. "Emotions are eXpressed
more intensely on the left side of the face."
Science. 1978, 202, 434-436.

Schachter, S. & Singer, J. "Cognitive, social, and phy-
siological determinants of emotional state."
Psychological Review. 1962, 55, 379-399.

 

Schiffenbauer, A. "Sex role stereotypes and the spon-
taneous attribution of emotion." Journal of
Research in Personality. 1976, 55(2), I37-145.

 

Schlosberg, H. "A scale for the judgment of facial ex-
pressions." Journal of Experimental Psychology.
1941, 55, 497-510.

 

Schlosberg, H. "Three dimensions of emotion." The Psy-
chological Review. 1954, 55(2), 81-88.

 

Sloman, C. "Micromomentary facial expressions and the
actor: an investigation.” Empirical Research in
the Theatre. 1972, 5(1), 52-60.

 

 

Stebbins, G. Delsarte System of Dramatic Exppession.
New Yotk: Edgar S. Werner, 1886.

 

Strongman, K. The Psychology of Emotion. London: John
Wiley & Sons, 1973.

 

Tomkins, S. Affect, Imagery, and Consciousness: V01 I,
The Positive Affects. New York: Springer Pub-
lishing Company, 1962.

 

 

Tomkins, S. Affect, Imagegy, and Consciousness:Vol II,
The Negative Affects. New York: Springer Pub-
Iishing Company, 1963.

 

 

Tomkins, S. "Affect as the primary motivational system."
in Arnold, M. (Ed.), Feelings and Emotions: The
Loycla Symposium. New York: Academic Press, 1970.

 

Tomkins, S & Izard, C. (Eds.) Affect, Cognition, and
Personality: Empirical Studies. New York: Sprin-
ger Publishing Sompnay, 1965.

 

 

209

Turhan, M. 'Uber die deutung des gesichtsausdrucks."
Psychologic Beitrage. 1960, 5.

 

Wilmot, T. "The effects of sensitizing clients in therapy
to the analysis of facial expressions." Disser-
tation Abstracts International. 1973, 55(2-A), 574

 

Wilson, F.R. Micro-Momentary Facial Expressions and
Psychological Defenses. Unpublished Ph.D. Disser-
tation, Mighigan State University, 1976a.

 

 

Wilson, F.R. "Repression and affect: a facial cue to de-
fense operations." Paper presented at the 84th
Annual Meeting of the American Psychological Assoc-
iation, Washington, D.C., September 7, 1976b.

Winer, B.J. Statistical Principles in Experimental Design.
New York: McGraw-Hill, 1962.

 

Woodworth, R.S. Experimental Psychology. New York: Holt,
1938.

 

Young, P. Emotion in Man and Animal. New York: John Wiley
& Sons, Inc., 1943.

 

Zerba, M. Labeling and Recognitign of Izard's Facial
Expressions by Three Age Groups. Unpublished Ph.D.
Dissertation, Michigan State University, 1977.

 

 

Zlatchin, C. "Accuracy and inaccuracy in the recognition
of facial affects and the relationship to perso-
nality." Dissertation Abstracts International.
1974, 55(12-8, Part I), 6228.

 

Zuckerman, M., DeFrank, R., Hall, J & Rosenthal, R.
"Encoding and decoding of spontaneous and posed
facial expressions." Journal of Personality and
Social Psychology. 1976, 5545), 966-977.

"‘2222222222‘2272