EYE-CONTACT PHENOMENA RELATED TO CHOICE OF
PARTNER AND AGGRESSIVE PASSIVE ROLE PLAYING
JOSEPH F. KERSEY
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The writer would like to thank the individual members of his
committee: Drs. Madeline Carey-Ramuy, Richard Anderson, Vernon Van
de Riet, and Herschel Elliott. Especial thanks to his chairman,
Dr. Marvin E. Shaw, for his understanding as well as his more technical
Further acknowledgment to Misses Frances I!aemmerlie, Jean Russo,
Jane Harper, and Janis Vincent; to Messrs. Gil Betz, Tom Bowman, Barry
master, and Joe Mathis without whose aid as experimental confederates
the mechanics of this dissertation could not have been accomplished.
Finally, an expression of gratitude to Allan Lind whose aid made
the data collection temporally feasible.
TALL OF CONTENTS
List of Tables
E:xpcriment T .o:
methodd and Results
tlethod and Results
Si o tIrapl
LIST OF TABLES
Table 1 Choice of Future Partner Frequencies
Obtained from Same-Sex (Male or Female)
Panel Conditions 21
Table 2 Chi-square Summary Table for Same-Sex
Panel Conditions 22
Table 3 Choice of Future Partner Frequencies Ob-
tained from Mixed-Sex (Male and Female)
Panel Conditions 23
Table 4 Chi-squares Obtained from Analyses of Six
Intervening Mixed-Sex Panel Conditions 24
Table 5 Mean Millimeters of Event Recorded In-
scription Measuring Time Subjects Made
Eye-contact Available to Target Person 25
Table 6 Analysis of Variance Summary Table for
Millimeters of Inscription Measuring
Time Subjects Made Eye-Contact Available
to Target Person 26
Table 7 Mean Millimeters of Event Recorded In-
scription Measuring Time Subjects Made
Eye-Contact Available to Target Person 27
Table 8 Analysis of Variance Summary Table for
Millimeters of Inscription Measuring
Time Subjects Made Eye-Contact Available
to Target Person 28
A definitive provision of -o-ial performance is that it takes
place within the confines of the interrelated behavior of two or more
individuals whose behavior is not solely determined by the wishes of a
single person. That che idiosyncrjLic behavior of the individuals
can be interrelated supplies the bar-, for, or basis of, social per-
formance. As Argyle and Kendon (l'07) pointed out, . /individ-
uals/ must be able to agree as co what the encounter is about, who is
dominant, who is submissive, they rnus igrce upon the level of inti-
macy; and there must be coordinatii,'' in terms of emotionality and
the patterning of actions in time," (p.. 59).
It is readily apparent that t-*e above adjustments may come about
only by the transmission of signal or cues (information) between the con
fronting individuals. How is this ci,-ns-iission accomplished? The
answers range from the ob"ious--che. tell each other--to the abstruse
(Goffman, 1959, 1963). One method Of transmitting this information
which long has been popular wich pocr.s and writers las been the
"message of the eyes." Or,stated in the accepted academic terminol-
ogy, that of eye-contacc.
A great deal of effort has be..-e' devoted to the demonstration of
eye-c")ntact as a behavioral variable within social p.-rformance. Simmel
(1921) has stated that one's desire for union with another determines
whether a one-way or mutual glance is used in establishing communica-
tion. Tomkin's (1963) discussion of taboos of mutual looking suggests
that the emotional involvement implicit in a mutual glance (particu-
larly of a sexual nature) will, under certain circumstances, inhibit--
or encourage--the desire to become involved. This inhibition effect
may be found within many animal studies concerned with dominance and
aggression (Devore, 1965). Exline and his associates (Exline, 1963;
Exline & Eldridge, 1965; Exline & Winters, 1965; Exline, Gray & Schuette,
1965) have lent generous support to the notion that the amount of eye-
contact (attempted and accomplished) is related to the orientation a
person holds toward his situation or companionss. )Exline and Winter
(1965) note, ". one observes another's visual behavior and infers.
we believe, the degree and affective sign of the other's involvement
in a momentary interpersonal relationship . or learns from the
behavior of the other's eyes something of the other's desire, willing-
ness, or ability to relate emotionally to another" (p. 320).
SIt would se.'m, then, that enough thought ani evidence exists to
consider eye-contact as a meaningful variable within social perform-
ance. However, a distinction in non-verbal acts such as eye-contact
has been made by Ekman (1965). His system classifies an act as an
indicator or a communicator. Essentially, the indicator focuses
attention upon the sender and the communicator focuses upon the re-
ceiver. Perhaps the terminology of information sending and receiving
is more descriptive. The previously mentioned research has focused
on the receiving or interpretation of eye-contact and as yet fully
failed to ask the question of whether subjects will use eye-contact
to elicit effects chac hav.c b.:en established prev'.io'is ly by manipulat-
ing the amount of eye-contact. l.iac, chen, have some normal pacttrns
of eve-contact found to be? 1..hen engaged in normal conversation two
people will engage in mutual eye-contact on an intermittent basis.
The percentage of che time chat cwo people will share this Trutual ?lanct:
,.ill usua llv fall between 5 and 30 percent of the tim- (Argyle, [967;
E:.line & Winters, 1965). This percentage of eye-contact is established
by each individual looking non-mutually at the other between 30 and 60
percent of the total time In an unpublished study Cathc.rt, Ker.ey,
Breed, Garcia, Rubin, and Wells (1968) found consistent individual
patterns of mutual eye-conr.act percentages well above 70) percent.
Argyle (1967) has also noti.i the close interrelation between eye-
contact and the pattern of speech taking place. People seem to look
more uhen li;teniig than 1..-il transmitting information T.hei e.;laice,
or "looking beha.,ior" tends to be longer and their a'- ay-glances or
"avoidance behavior," arc shorter in duration. When a person is pre-
paring to speak, he uill glance away from the person :.'ho is the object
of his message; at che end of sentences or natural break.s I,--- glances
briefl-, at the other person, and at the te r-ination of hi, comur-unica-
Lion he uill give a longer gaze. The listener will at the same time
be giving glances that are much longer (Kendon, 1965).
Along with the information obtained demonstrating the existence
of general eye-contact norms, great variation between iLdi.vi.duals in
the amount of looking was found. Argyle and Dean (1965"' have demon-
strated that women engage in more eye-contact than do men, particular-
ly when calkirng to other women. Exline and Winters (1965) also found
same-sex pairs to engage in more eye-contact than cross-sex pairs. In
an attempted replication of these data, Catlcart, et al. (1968) could
not substantiate the findings. Exline and Winters (1965) also found
that women will look more at the other while speaking if they like
him, while men look more when listening if the other is liked.
People high in the need for affiliation look more, but only in
a cooperative situation. If the situation is competitive, the subject
high in dominative needs tend to look more. These effects are height-
ened among women. This has been used to explain women Looking more,
i. e., women are higher in affiliation and lower in dominance, so that
a friendly expetimrntal setting elicits more looking (E:line, 1963).
The essence of the information aLL'rnacivois are ,summarized by
Argyle (1967) as follows:
1. A wants to initiate iiieraction with B.
2. A shows B the attitude or emotions which he feels
3. E wishes to show A that he accepts--or rejects--A's
4. At the end of each speech A looks at to signal
he may speak.
5. A wishes to show B he is more believable, more
confident, and more favorable in his attitude
6. If A gives B glances rather longer than normal, he
indicates he is more concerned with B as a person
than with chi- i lsuj at hand.
While the above possibilities are essentially concerned with
the sending of inforna t ion, it i equally, necessary, for a communicator
to get feedback on the other r. is':zn's response, i e. atti nd to com-
municator acts. Person A may i.ish to know simply, if he still has B's
attention, or if B has understood and/or agrees, or, perhaps most
importantly, if he still has permission to continue the interaction.
lIhile an extensive effort I:I been made to catalog the norms of
eye-contact data, quite a bit of r-st-atch has been completed concern-
ing the motivational aspect of ey..-contact.
Arg;yle (1967) used che informai(ion seeking function to hypoth-
esize the nature of the underlying motivation of the avoidance bhchav-
ior at the outset of a communicajtiuin He felt that the speaker does
not wish to be distracted by extra inputs of information while tie is
planning and organizing his message. Several scudies are available
as support for this theory of eye-contact motivation; Kendon (1967) stati-s
that A ". .iill tend to look more 'hen he is engaging in fluent,
well-rehearsed material than when engaging in hesitant and unfiuent
passages." Exline (1963) has found that the difficulty of the material,
per se, correlated with looking. That is, more difficult and complex
material elicits less looking behavior on the part of the sender.
Argyle also offered the hypothesis that the avoidance is really an
effort to reduce anxiety as an a'tcrnative to his information theory
of motivation. (Argyle & Dean, 1965; Stass and Willis, 1967).
While short periods of looking may simply be information processes,
longer periods signify a heightened interest in the other person (Argyle,
1967), either in an affiliative, sexual, or aggrc:..sive/competitive
sense. Looking should he correlated positively with these motives
(Argyle, 1967). For example, Exline and Winters (1965) have found
high affiliators to engage in more eye-contact when in the presence
of a person whom they like. However, there is evidence that being
aware of being observed is disturbing (Liang, 1960).
Paralleling the research devoted to the information and
anxiety hypotheses is work directed toward suggesting eye-contact to
be a component of "intimacy" (Argyle & Dean, 1965). Exline, Gray,
and Schuette (1965) found fewei personal questions to be asked of
an interviewer when the interview, gazed continuously Kendon
(1967) found a negative relation between smiling and eye--conjtact.
If A smiles, then B smiles, thus raising the intimacy in the con-
versation. At this point of higher intimacy the amount of eye-contact
decreased. Argyle and Dean (1.965) tested the. hypothesis that distance
and eye-contact would also be negatively related. Their results add
evidence that eye-contact decreases with intimacy--in this case oper-
atiaialized by assuming physical closeness to be intimacy.
It is obvious that considerable effort has been directed toward
the variable of eye-contact within social performance; however, sev-.
eral criticisms may be levied against the preceding research. One
aspect is that the paradigms selected,almost without exception,assume
the subject to be using eye-contt.ct in connection wich a pre-seleccted
variable. That is, a receiver indicates he is interpreting a condition
which contains increased cyc-contact as a condition that contains
aggression; therefore, the experimenter ascribes aggression to the
sender. It seems, however, that no one has asked subjects to be ag-
gressive and checked to sec if indeed the subject as a sender uses the
variable is predicted. Of notable cxccption to this cri cic sm is an
atItempt tL examine the motivational aspect of eye-contact in uhicl the
behavioral state of the subjecr, i.e., anxiety, is manipulated (Stanley
& Martin, 1967). Even this study, however, fails to get at the ques-
tion of hoi.' the subject interprets his efforts at info: motion sending.
By utilizing the method of role playing, perhaps two questions mna be
answered: 1) Do subjects as senders recognize the social value of
eye-contact as information sending, and 2) if subjects do recognize
the variable as useful, will they uie it in thc manner predicted by
the behavioral scientist from his : ata gatheLed from rcei'.ers'
Another seeming failure in the attempt to demonstrate the eyc-
contact phenomenon has been the choice of explr:rimental designs. Only
one study (Stass & Willis, 1967) has attempted a true within-subject
design, the remainder have been casL into a b'jteen-subjects; analysis.
That is to say, rather than attempting tu find differi.n ial effects of
several eye-contact conditions to the sane individual, experimenters
have relied upon assigning one condition oF eye-contact to a group and
then comparing it. results to another group uhich has received a dif-
ferent condition. Since it is readily demonstrated that baseline rates
between individuals is a highly significant factor (Cathcart, et al.,
1969) between groups seems to be tenuous--or at least a less powerful
demonstration of the viability of the variable of eye-contact.
Although less severe, several additional criticisms exist. Vir-
tually all the data collected is from dyadic discussion or interview
situations. That is to say, the effect has generally been studied
only among the smallest of groups. Furthermore, the available cross-
sex data is conflicting and quite scarce.
The following series of two experiments were designed t.z attempt
to correct the first two criticisms. The manipulations were: 1) To
ask subjects to "play roles," i.e., attempt to send a specified type
of information, and determine if the subjects use the variable as
previously hypothesized, and 2) cast the subject in such an experi-
mental setting as to expose him to several conditions of eye-contact
rather than a single condition. The experiments will alos accept to
extend specific previous findings, e.g., to groups largest than dyads,
and to acr.enmt to simply replicate findings of previou- e:.:perim nts
in order to more fully demonstrate their -xisternce.
Experiment One: MethoJ and Pesults
The 176 male and 176 female subjects involved in the experiment
were, for the most part, from the University of Florida's introductory
psychology course subject pool. Most subjects were fulfilling intro-
ductory psychology course requirements for experimental participation,
however some were recruited by asking passing studenLt to "help out."
With only a few exceptions, the passersby responded affirmatively to
A posted announcement indicated an experiment Lhat offered
enough credit to meet the student's entire obligation, but that a
prerequisite of an interview by a committee was necessary. The subjects
were informed that perhaps one-half of those interviewed would be
selected to continue with the second half--this information was given
after the completion of the collection of the data from the interview.
In reality, only 32 subjects were selected to continue in the second
Upon reporting to the experimental room, the subject found a
panel of three persons waiting for him. The panel was comprised of
either all males, all females, or a mixture. Thc three specific indi-
viduals comprising the panel were one combination of all the possible
combinations of four male and four female confederates of the expcri-
nenter. The compositions were counterbalanced so that each confcder-
ate was a member of differently constituted panels an equal number of
times. The members of the panel had been previously instructed to
conform to one of three eye-contact availability conditions: constant
availability, spontaneous availability, or non-availability. The por-
trayal of the three conditions was also rotated among panel members
in a counterbalanced manner.
Following a brief introduction to the panel the experimenter
left the interview room. The subject found himself seated facing
the panel which was in turn seated about a small table some 10 to 12
feet away. The seating order of the panel was haphazardly changed
with each change of subject. A 4-minute discussion of the sjbject's
concepts, attitudes, and goals in the field of psychology and experi-
mentation in psychology took place. During this interview, the panel
members attempted to maintain a neutral presentation cf their questions
--each asking three of the same nine questions (see Appendi:-: A). The
choice of which three co ask was also changed in a haphazard manner,
i. e., each panel member asked any three questions that he had not
asked in the iirrnedi3ately previous interview. The order of the ques-
tions was established only by the subject's responses, i.e., if the
response of a subject led naturally to another question it was aoked.
Upon completion of the discussion, the experimenter returned and
asked the subject to follow him outside the interview room. Once
the subject was away from the panel he was asked to simply identify,
the member of the panel he preferred co ,.ork with, or for, in the re-
mainder of the experiment. If the subject questioned the .xperimente r
about the feasibility of the choice, the experimenter acknowledged the
difficulty, but stated ic was to te. a variable in the next e:..periment
and was necessary. This satisfied all questioning subjects enough so
chat they made the choice requested. After the subject identified the
memnber, he was informed r.hat notification of those selected would be
posted on the experiment notice bulletin board.
If one makes the simple assuimptioi, that people F .i l choose to
work with those .whom they like: or find moie attractive, then previously
reported data indicating that dLcgee of eye-contact makes a person
attracLive should predict the ch',ice; of the subjects. Winei and
Hehrabian (no date) had tw.o female subjects~ interviewed by a male con-
federate who looked more at one of the subjects. The most looked at
subject found the interviewer to be more positive in value. Kersey
(1968) in a re-analysis of the Cathcart, et al. data found that when
the effect of individuals was statistically controlled,a significant
interaction existed between liking, eye-contact, availability condi-
tion, and sex of subject. In essence, the females did notdislike being
stared at, at least,did not devalue those who made eye-contact con-
stantly available, but did dev.alue those who ignored their attempts at
eye-contact. The male subjects did nct dislike the condition where the
confederate made no eye-contact available--that is, the males did not
devalue those who gave no eve-contact while they did devalue males who
made it constantly available.
Based on the two studies above, the following speLific predic-
tions were made:
Prediction One: The female subjects will choose members
of th.e panel in thle folluo'ing manner:
spontaneous eye-contact availability
more often than or with the same fre-
quency as constant eye-contact avail-
ability, with no eye-contact availabil-
ity chosen significantly less often.
Prediction Two: Tile male subjects will chcose memb..rs
of the panel in the following manner.
spontaneous y-i-contac' availability
more ofteri than or with the s.-me fre-
quency as no eye-contact availability,
with constant eye-contact availability
chosen significantly less often.
Because of the conflicting data on cross-sex iruaLiors combined
with the lack of data on triads, no hypotheses were advanced on the
six intervening male/female combination panels. Rather, these data were
to be viewed as exp]orator,' investigation within the area of eye-
The frequencies with which subjects chose panel members for future
partners in a second experiment constituted the data. The
data Jcre subjected to a series of chi-square tests. The data
obtained from the all-male and all-female panels were cast in a chree-
way contingency table wi..di sex of subject, sex of panel, and eve- .
contact availability as the classification factors. The choice fre-
quency from each of the six possible combinations found in the mi:ed
panel situation, i. e., a male playing each of the three eye-contact
availability conditions xith Lto females playing the other, and a female
playing each of the three eye-contact a'.ailability conditions with two
males playing the others, were subjected to separate tists of ind pend-
ence in 2 x 3 contingency tables (Li, 1964).
The obtained frequencies of choice of future partner from same
sex panels, i.e., all male or all female, tor each cell are presented
in Table 1. The expected frequencies are in parentheses. The resulting
chi-square analysis ,of the sam--se:- panels indicated significant effects
as shouTn ir. Table 2.
One rote of explanation is necessary concerning the statistical
analysis. Sex of subject effect and sex of target effect were forced
to zero by the choice of the a priori assignments of expected cell
frequencies. That is, all between-subjects data is not testable. This
may be demonstrated if the male subject x all-male panel cell is exam-
ined. Since each of the subjects in that cell (n = 18) must make a
choice among the three conditions in che cell, and the a priori
expectancies call for a frequency of 18 in chat cell, it is readily
seen that che two numbers will exactly coincide giving a chi-square of
zero. The practical consequences of the situation are as follows:
1) the loss of one degree of freedom for each of the tw, bt-.'een-
subjects effect as well as a loss of another degree of freedom for
University of Florida's introductory psychology course subjCct pool.
The 16 male and 16 female subjects were fulfilling introductory psy-
chology course requirements for experimental participation.
Upon reporting to the experimental waiting room, the subject
found another person, the confederate of the experimenter, already
waiting there. The experimenter immediately appeared and asked if the
two were there for experiment number 191. Upon receiving affirmative
answers from the subjects, they were led to the experimental setting:
a room containing a small table (approximately 32 inches wide), two
chairs, and a mirror. The experimenter then gave the following directions
and explanations: "Would you (pointing to the confederate) please be
seated there (pointing in a casual manner to the chair between the
table and the mirror). And you, there (indicating in a manner that
appeared to be a random choice for the subjecr to be sealed in the
chair facing the mirror). I'm exploring the .-onditions surrounding
dyadic interaction; that is, interaction between two people. I would
like to explain two aspects of this experiment. First, the mirror be-
hind you is obviously a one-way screen. It is placed there so that my
presence is minimized during this experiment. After all, I am investi-
gating dads and not three people, or what's called triads. Secondly,
because I shall be asking each of you to participate in this experi-
ment in a slightly different manner, I shall present the remaining
instructions for this experience as written te.t rather than verbally.
Do you understand? 0. k., here are your further instructions."
their interaction, and 2) all of the testable variances being forced to
the within-subject condition, i.e., eye-contact availability conditions,
and its interactions.
The two significant chi-squares found were for eye-contact avail-
ability condition (X = 7.6, df = 2, p<.025) and for eve-contact avail-
ability conditions interacting with the sex: of the subject
(X = 7.6, df = 2, p<.025). This, then, is support that eye-contact
availability conditions lead to a differential choice of future part-
ners, (However, the effect seems to be different depending upon the
sex of the person receiving the eye-contact; a male does not choose
another male who looks at him continually or a female who avoids look-
ing at him and females rarely choose anyone who avoids looking at
The data from the intervening mixed-sex panel conditions dis-
played in Table 3 appeal to agree with the above. That is, eye-contact
availability conditions found to be aversive in the same-sex panel
conditions continue to be chosen less often. Table 4 compares the
chi-squares found in the six mixed-so:, conditions. The results
indicate that as conditions establish morc numerous aversive cells,
choices are forced to the remaining cells, thus yielding chi-squares
with smaller probability levels.
Experiment Two: Mechod and Pesulcs
The 32 subjects involved in the experiment were a sample from
the 252 subjects involved in Experiment One. All were from the
The confederate received no particular instructions. He was
handed t. pewritten pages of te::t of obviously longer length than the
subject's instructions, and, on the next occasion, a handwritten note.
This was to make the subject unsure if the confederate were receiving
the same set or type of instructions. The subject's instructions
were one of two sets:
Set of Instructions Number One:--This part of the experiment re-
quests that ycu discuss Mormons (Church of the Latter Day Saints).* Any-
thing that you know, feel, or believe about the Mormons may be discussed.
In other words, anything you feel to be appropriate to the discussion
may be brought forth, but please do attempt to play the "role" of an
AGGRESSIVE and VITAL debater. You may define the "role" as you see
fit. PLEASE DO NOT BEGIN THE DISCUSSION UNTIL YOU HEAR A SMALL TAP
ON THE ONE-WAY MIRROR.
Set of Instructions Number Two:--This part of the experiment re-
quests that you discuss Mormons (Church of the Latter Day Saints).
Anything that you know, feel, or believe about the Mormons may be dis-
cussed. In other words, anything you feel to be appropriate to the
discussion may be brought forth, but please do attempt to keep the
discussion only on Mormons. However, I further wish you to attetIpt
to play the "role" of a PASSIVE and RETIRING debater. You ma' define
the "role" as you see fit. PLEASE DO NOT BEGIN THE DISCUSSION UNtTIL
YOU HEAR A SMALL TAP ON THE ONE-WAY MIRROR.
After the subject and confederate read the in; tructions the e:-
perimenter took the instructions and retired to behind the one-way
*'The subject of "Mormons" was chosen because the experim-enter's
previous use of the subject had established it as a neutral topic for
the majority of University of Florida students.
screen. He then capped on the one-way mirror to signal subjects to be-
gin the discussion. A ten-point Esterline-Angus Event Recorder was
used to record the amount of eye-contact availability the subject used
in his interpretation of his assigned role. The recorder was wired in
such a manner that the experimenter controlled the movements of one
stylus with a telegraph key. Deflections of the stylus recorded eye-
At the end of a 4-minute period, the experimenter returned to the
experimental room, said, "Fine, now please read these." The subject was
handed the second set of instructions and the confederate received an-
other copy of what appeared to be different instructions. When che sub-
ject had read the instructions, the e:pe rimenter took them and once
again retired behind the screen for another 4-minute period.
At the end of the second "scagirg." the subject and confederate
were thanked for their assistance. At this time the two were asked if
they would consent to do a favor for the next subject. It was explained
that one of the pair due to participate next on the schedule had failed
to arrive, and while it was a bit irregular, perhaps one of them would
consent to do the experiment again. 'he confederate refused because of
an extremely important prior commitment. After gaining the subject's
acceptance (only two refused), another confederate of the opposite sex
from the first confederate was ushered into the experiment room.
The procedure was then repeated, the experiment terminated, and
the subject debriefed.
Since the level of looking has been assumed to be related to the
subject's orientation toward the others, this experiment was set up
to give the subject a specific orientation toward another and find out
if he did, indeed, use eye-contact to differentially sc;id information.
Of specific iinteresc in this experirr.ent was Atg.g le's (1967) assertion
chat aggression should correlate positively with eye-contact. Cross-
sex data once again were offered [n a r.normative manner.
Based on Kendon's (1967) findings that
the subject looks more when he is fluent
and rehearsed, the subject should look
more during his second presentation.
Based on Argyle (1967) findings, it is
predicted thac women will engage in more
eye-contact than men, particularly with
Based on ExJine and Winters' (1965)
findings, it is predicted chat the same
sex-Fairs will engage in noLe eye-con-
tact ch.n cross-sex pairs.
Argyle's (1967) assertiur. predicts that
male subjects will stare more vhen play-
ing rth aggressive role. Since no fe-
male or cross-sex data affErd a pre-dic-
tion, it should be assum-d co also
hold true for females a;id cross-se:
situations as well.
The data, consisting of the millimeters of inscription on cvcnt-
recorder tape, were subjected to a 2 x 2 x 2 (sex of confederate x sex
of subject x passive/aggressive condition) repeated measures design.
The repeated measures were over two factors--pass ive/aggressive condi-
tion and sex of confederate. This is referred to by Wirter (1962) as
Case-i type. The order of presentation was randomized for sex of con-
federate and randomized within sex of confederate for passive/aggres-
Table 5 shows the means for each of the experimental conditions.
No effort was made to convert the millimeters of inscription on the
event-recorder tape to increments of rime since this amounts only to
multiplying the data by a constant and would in no way change tle sta-
tistical presentation. Tibic 6 presecnts a sumwar, of the analysis of.
variance applied co the data. Se. of subject is a very strong effect
(F = 258.6, df = 1/30, p<.001) with female subjects presenting far more
eye-contact availability than male subjects. The eye-contact availabil-
ity used by subjects in tht two role playing conditions, i.e., aggressive
or passive, is significantly different (F = 17.3. df = 1/30, p<,01).
The subjects use less eye-contact when role playing the passive condi-
tion. And, finally, the amount of eye-contact availability used in the
role conditions is found to interact significantly with the sex of the
target person (F = 6.7, df = 1/15, p<.05).
The data may be recombined in such a manner as to demonstrate
any effect found over trials. That is, by ignoring the sex of the
target, which in the actual experiment was counterbalanced, one obtains
the measure of eye-contact during the first and second role playings of
the aggressive and passive conditions. Table 7 shows the means of the
cells obtained from the recombination of the data and Table 8 presents
a summary of the separate analysis performed.
The prediction that women will look more across all conditions estab-
lished from Ex]ine and Winters' (1965) data is firmly substantiated.
However, neither Kendon's (1967) prediction that more fluency will
allow more eye-contact, nor Exline and Winters' (1965) findings that
same-sex pairs engage in more eye-contact than cross-sex pairs finds
support from the analysis of the recombined data. Nor is Argyle's
(1967) assertion that aggression is positively related to eye-coiitact
sustained by the analysis of the data in their original form. PuL,
since there is a significant interaction between eye-Lortact availabil-
ity and sex of target, the hypothesis does find some support.
Choice of Future Partner Frequencies Obtained
from Samc-Se:. (Mal or Female) Panel Condicions
All Male Panel
ConscanL Spontaneous I-ev'er
8 t6) 8 (6)
7 (6) 2 (6)
All Female Vanel
Constant Spontarnecus lever
7 (6) 10 (6) 1 (6)
11 (6) 5 (o0 2 (6)
Noce: The expected cell freq'uncies shown in FarenLhesCS are a
priori assignments based on the assumptLon of independence of ciondi-
Si on s
Chi-square Sunrary Table for Same-Sex Panel Conditions
Sex of subject (SS)
Sex of target (ST
SS x ST interaction
SS x EC interaction
ST x EC interaction
SS x ST x EC interaction
Choice of Future Partnecr Frequencies Obtained
from Mixed-Sc-x (Male and Female) Panel Condicions
Panel Condition Hale Subject Female Subjecc
Constant Male 2 (5) 6 (5)
lI.ver lale 7 (5) 3 (5)
Spontaneous Female 6 (5) 6 (5)
Spontaneous Male 4 (5) (5)
Heve.r Male 5 (5) C? .5)
Constanc Female 6 (5) 6 (5)
Sponcaneous Male (5) 6 ()
Constant Male 2 (5) 7 (5)
Lever Female 4 (5) L 5)
Constant Femalp 7 (5) 7 (5)
Never Female 2 (5) 3 (5)
Spontaneous Male: 6 (3) 5 (5)
Spontaneous Female 11 (5) 6 (5)
Never FemalL 3 (5) 2 (5)
Constant Male 1 (5) 7 (5)
Spontaneous Female 5 (5) 6 (5)
Constant Female 6 (5) b (5'
lever Male 4 (5) (S
lioce: Expccced call frequencies established a priori on the
assumption of independence of conditions.
Obtained from Analyses of Six
Mixed-Sex Panel Conditions
Cons tant Male
Mean Millimeters nf r e'.'en Recorded Inscription
Measuring Time Subject- H-ide Eye-Contact Available
to r1rget Person
Aggre si.e Pass ive
237. 9 228.50
Analysis of Variance Summary Table for Millimeters of Inscription
Measuring Time Subjects Made Eye-Contact Available to Target Person
Sex of Subject (SS)
Subject Within Groups
Sex of Target (ST)
SS :,: ST
ST : Subject Within
Groups (error B)
Role Condition (C)
C : Subject Within
Groups (error C)
ST x C
SS x ST : C
ST x C x Subject
flean i11 ilimeters of ['.'-nt Recorded Inscription
feasu r ng Ii iC Su btj-c Ls -l.aJ ce Eye-Con tac: t Av'.ailabl
to ?"Tai.- c P-erson
A grUss L P3ss i.'
230. 243. 7
Analysis of Variance Summary Table for lillimeters of Inscription
Ileasuring Time Subjects Made Lye-Concact Available co Target Person
Sex of Subject (SS)
Subject Within Groups
i th in Subjec ts
SS x T
T x Subject Iichin
Groups (error T)
Role Condition (C)
SS x C
C x Subject ltithin
Groups (error C)
I x C
SS x T x C
T x C x Subject
(error T x C)
riS F Probnbi lit',
51,5S6.S 15 343.9
The results of E:-:pctiment One Seetn straighcfori-iard and cleat.
First and foremost, they demonstrate the effect of eye-contact in pro-
ducing differential behavior within an individual. To those who had
faith in the bet.,een-gtoups dara perhaps chis finding may seem unimpot-
cant; however, to the degree chat tlh effectiveness of scientific vari-
ables rests upon demonstration in varying situations ard consent, this
finding supports the belief that ,'-e-coirnact may be viewed as a rcien-
cifically useful phenomenon. In this s,:nse, E::pieiment One has duti-
fully contributed. Tie degree of consistency w:ith w..hich individuals in
this experiment avoid the choice i.f certain t,'pes of partners in prefer-
ence to other types seems to indicate tle existence of generally known
and accepted norms.
The conditions that are avoided are somewhat different for the
cuo se:-:es. For males a "looking" male seems to be undesirable. If
the animal daca concerning dominance m.ay be extended to humans ac this
point, it wouldl d seem reasonable to argue chat this is merely one male
resisting the attempted dominance by another male. What makes this
ad hoc explanation somewhat tenuous is the data showing an opposite
effect by a "looking" female. Sin-e she is readily accepted--irdeed,
sought ouL--by the male, why is she not seen as attempting to dominate?
Does the male wish to be dominated by the female? It does not seem
likely. Perhaps it is merely that the male does not perceive an attempt
by a physically weaker female to dominate us plausible and therefore not
as threatening. Another finding reported in the literature is that
females just use more eye-contact than males, so it may merely be a
case of a "looking" female meeting a cultural expectation while the
"looking" male is not. If this be the case, one may explain both occur-
rances by noting that the unexpected is threatening. It must be stated,
however, that invoking cultural expectations as an explanation says
absolutely nothing about the genILsis of the expectations.
In passing, it is easiest to explain the "looking" female choice
of the male in terms of sexual concoctions. Tomkins (1963) used this
to explain his finding that female. did not look at males as juch as
at other females, although this p rtic,,Ler finding has not been repli-
cated in several attempts by t'- p rimeater and some of his immediate
associates. However, it does rot se,,n implausible that sexual impli-
cations may be found in eye-cont c*- situations. This is different from
saying they always are.
The psychology of individual il .fferco cs offers one possible
option fur the female cultural tra;.r ct locking. She is found to be
more field dependent in her behavior; that is, she leans upon the en-
vironmental milieu to guide her actions uore than the male, it is per-
haps for this reason that the reu-iltr of Experiment One show the female
subjects to avoid those--male or fct ale --who do not offer there. eye-
contact. The explanation of expectation must be twisted a bit to fit
this situation, but if the female xpccts the male to look at her for
sexual reasons and another female to look at her for information-
gathering reasons, then a nonlooking rale or female is not meeting
her expect cation.
Finally, C:-perimenti Onc, cl.- rl sheo -- that prev.iousu S reported
data in dvadic situations !old- LIue at least for a troup of four.
This, of course, was an in tLr.i sit'3ction and it is. possible chat a
discussion group could piodice *- r.. of il;fferc-in nature if suffi-
cient competition were to be i cr duced. Thi. couul particu early affect
the cross-sex crr, i cion.. .'hich in this .-per ijmenr. followed rhe same-
Experiment Two so.m3.; t be less straigi, ttorward in tr. results.
Primarily the diffictultcy li. i 'n Ihe hI'potc'.,e-L: concerning the amount
of eye-concact that will 1,c i, .-,I when pliyinp the nggretsive role.
Argyle's prediction of pr?'-i .- ielatic i.nshLt' I ~:ceen aggression and
eye-contaci is not suppcr- te-l. 7T faLL, r.l results of the sta-istical
test of the d.3at for thli- r-i i ect of rol: ccndition- would indicate
the rev.ers. L. e., thcli LC i-.o b.t- e i .?7-gre l Lto aid eye-cuntacc
However, in c:-.aminIn'. '--: c. 1i1 ri.an li-;ted ir Toble 1 t1.t i.s
possible to s-e that the d.nr L. i:ied fow male t.-rgatc. are in the
direction predicted. It is po Ibl that ihe Lrium.ir'idous difference
found in the results obtained frou- che teriales is overriding and hid-
ing any possible effects co 1-- flounr' .n the results obtained from the
males. This situation is r-.fic t.-ced in tle significance of sex of
target x role condition. 1 .e r: perimenrti believes the significant
interaction to be more truly ref' fec ive of the situation than the
main effect. -Wnile thi:; juL.st.i;ies to some degree the belief that Argyle',
contention is supported b,, I- i,;ate target data, it necessitaces asking
whv it does not occur with fI.-, lc;s as targ cts.
The first place to look for an explanation of the result is in
the experiment itself. Did some aspect of the experiment "force" the
subjects into changed behavior? An examination of the role instruc-
tions offers one possible explanation. Did the subjects read the
aggressive and active role demands as telling them to be hostile and/or
negative? If so, then the results of the experiment agree remarkably
well with the results of the first experiment. That is to say, if one
does not care to hb chosen (with its implication of not being liked), then
what better way than to "look" at a male and not look at a female? This
explanation fits the data well with the exception of the female-male
cell in which the females choose to look more at the male. The males
did not find this undesirable.
Alternatively, the passive and retiring instructions nig t have
cued the subject to "sit and listen." This behavior ha.3 Leen dLmon-
strated as positively related to more looking behavior. This does not
seem likely since neither males and females followed this pattern of
behavior when corfLonted with a maJe target.
It is exceedingly difficult tc staue which of the two, or if
either of the two phenomena occurred. This is due to the failure on
the part of the experimenter to include a control group in the design.
In an effort to remedy the failure, a post-experimental attempt was
made to find if the instructions were seen differently. This investi-
gation consisted of asking 20 subjects to write three-sentence para-
graphs explaining the behavior they thought was appropriate in a debate be-
tween two strangers. Next they were asked what behavior was indicated
by either the one or the other set of instructions. That is, 10 subjects
wrote a paragraph about .: debatCe nd Lhen a paragraph abou-t the passive
instructions, and 10 ocher suhj,-,.. ts wrote a paragraph abouc a debate and
then a parragraph about the aggrr.isLv instructions. Threc judges (grad-
uate students in clinical psychle o. l '. ) were unable to detect a signifi-
cant difference in either coimp.ar--.on at the two sirCu.acions.
If no element of the e:
factory interpretation of the rc ;i'-L the next place to look is at the
subjects as subjects. Does th,: -ubj'.ect actually kno%. enough about his
behavior to successfully "play" i ole? The introductory te:
chology are replete with one ty[: ..f teaching mechanism: the initiate
is offered a list of true-falsei a .-. tions concerning cultural truisms
which he anrcers to the best u. I -- la',aanii's kno- led ,e He L, chh-n
astonished to find that the behai;.n 1 1 scientist offers ev'.ide'i rt that
his pr-c-cnceptions--those that. *'. body juct knows!--are false. This
is supposed to make the studenL malleabl to the new I r-ov. ied',e he
is about to encounter. Perhaps. Li :n the role behavior re lusted of the
subjects in Experiment Two wais -ri -.- :mple of the behavioral scLence/
layman dispori ,, Tlat is, pe'rhb..is eo-,rie do not e., evye-concac' as
related to aggression. However. mn". folk-sayings, e.g., "'tate the
man dovrm," exist that attest to ch- layman seeing a positive' 2 correla-
tion between che two variables. In.' ed, if ques cioned, tl-e layman
would probably predict as the ec:<.,timenter and Argy,,le did, This particu-
lar questionnaire has yet to Le -.:.cn to a sample of subj-cts.
In line with the present d.. :-ission is i distinction that has
been made concerning the differe,..c between "role playing" and "role
enactment." The essence of the di..tinction is chat a subject will
offer different results if no ego investment is made in the situation.
In other words, the role player does not have anything "on the line."
Just why this would affect only half of the experiment is not very
clear, and therefore open to some skepticism as an explanation.
Finally, let us look at thu results of Experiment Two and the
subject's actions as if they do '"uly reflect the culture and its
norms, and try to isolate potentially effectual variables. Tf the
experiment consisted of allowing some choice of behavior to the sub-
ject, it would surprise very few people if little aggressiveness was
overtly directed toward females. Our cultural stereotypes of male
and female are dichotomized by nir ii things but very few so tr..ngl.y,
as th-- one demanding r "man" to ;. s LLong, cormn: t iti\.e, aggressive,
and dominating, and a truly "flomi iii woman" to be soft, duinure, and
ie ldi-g. Our norils also d-emnanu .n'-ist nt bch:i'.ioi to-'.,1 L h(m. T.,
attest to this, one m;y, refer to jan' .)f the -cu ient studies i alyzing
the difficulty of the emancipate' women in resolving conflicts of role.
Therefore, aggression is not oil', -.riissible iih males 1ut o'JL cul-
ture d,-m:.nds it. Con.'ersely cu" c' LLure does not re,2i;i ', : ept.
aggression it, relations with wo-.rn This, then, indli i:atsl th experi-
wiental results are reflecting socLal norms.
If the above relationship L s rentedd, then two possiLe.- sources
of motivation may be cited: 1) tl- di.scomfort from the dis:;onance
(Fescti igc r, 196 ) suffered fro,, tlhe t-:havior d-emanded kb., the instructionso -
being in conflict with thl cu lrtual norms, or 21 discomfort rrom guilt
felt by the subject from the suppos-_d discomfort, or mental pain, in-
flicted upon the target person. In both cases, the increased anxiety
or discomfort should dec rcas- .-,y -:ontact (Stanley & MlrnrLin, 1968).
A choice among the possibLe interpret stations requires a modi fi-
cation of e:.:perimental design, 5.., introducing controls, as well as
experimental technique, e. g., u;ing a method other than role playing
to elicit aggression or pas-i'i.ty. These would have 'r, b._ accomplished
in further work on the eye-co iL a t .-ariablet .T-e thenr .r not this is
reflecting society's admoni.ion to not be ov.erly aggr Li,.e toward
females or some inner disc, mfor'. about a possible transgiession also
remains to be tested.
(1) Do you find experimental pr rticipation requirements to be a
(If answer is yes or no, respond with, "Would you please ex-
(2) Is general psychological research--that you are aware of--useful?
(If answer is yes or no, respond with, "Would you please explain.")
(3) Is general psychological research--that you are aware of--valid?
(If answer is yes or no, respond with, "Would you please explain.")
(4) What experiments have you participated in previously?
(5) Why did you take Psychology 201 (or 300 if applicable)?
(6) Is psychology what you thought it to be?
(7) If you weren't required tr~ participate in experiments, what would
motivate you to parti(cite?
(8) What is most exciting ':o you in psychology?
(9) What is most disappointii'g to you in psychology?
Argy, le, 1. The psychology of irlt.-rFpersonal beIha..ior. London: Penguin
Books, Ltd., 1967.
/ Argyl N. and Dean, J. Evc-con ta,- L, distance and affiliation. Sociom-
e t rP 1965, 2., 289-30.4.
Argyle, 1. and [endon, A. The *::periomental anal: sis of social perform-
ance, in Cerkouitz (ed ), ..d .-inrces in ::e perimental social ps.'chol-
o0y III. lew York: Academic Press, 1967, 55-98.
Cathcart, E., Kersey J., Breed. G., Rubin, J., Garcia, J., and Wells, S.
Attitude change as a function of eye-contact. Unpublished paper:
Uni'.ersity of Florida, 1':8.
De'.'ore, I. (ed.) Primate behav.ior- field studies of monkeys and apes.
liet,: York: Holt, Rineh-i r. ,. 'Uinscon, 1965.
,,-Cknan, P. A methodological discussion of nonverbal behavior. Journal of
P-.-cholog.', 1957, 43, 141-14q.
E>:line, P. V. E:-:plorations in the process of person perception: 'isual
integration in relation to competition, sex and need for affilia-
tion. Journal of personality, !,63, i-f 1-20.
Exline, P. V., and Eldridge. Uirpublished. Reported in S. Tomkins and
C. Issard (eds.) Affect, condition and personality. New York:
Exline, R. V. and Iinters, L. C. Affecti've relations and mutual glances
in dyads. In S. Tomkirns .,,rJ C. I?.zard (eds.) Affect. cogniLion
and personal i,. fIew York : Springer, 1965, 319-351.
S-'Exlinc- e, R. V., Gray, D., and Schl:ette, D. Visual beha'.ior in a dyad
as affected by interview. ciunt-.nt and sex:: of respondent. Journal
of Personalit' and Social Pv'.'cholo.y, 1965, 1, 201-209.
Fesfinger, L. Conflict, decis -,n, nid dissonance. Stanford: Stanford
Goffman, E. Presentation of zclf- in i-.er',day life. 1Ne,. York: Double-
-3 7 -
Goffman, E. Behavior in public places. New York: Free Press, 1963.
Kersey, J. Eye-contact as an independent variable in the measurement
of person perception and attitude change. Unpublished paper:
University of Florida, 1968.
Kendon, A. The experimental analysis of social performance. In
Berkowitz (ed.) Advances in experimental social psychology III.
New York: Academic Press, 1967, 55-98.
Li, J. C. R. Statistical Inference I. Ann Arbor: Edwards Brothers,
Liang, R. D. The self and others. London: Travistock, 1960.
Simmel, G. Sociology of the senses: Visual interaction. In R. Park
and E. Burgess (eds.) Introduction to the science of sociology.
Chicago: Chicago Press, 1921.
Stanley,G. and Martin, D. S. Eye-contact and the recall of material
involving competitive and noncompetitive associations. Psycho-
nomic Science, 1968, 13, 337-340.
Stass, T. W., and Willis, F. N., Jr. Eye contact, pupil dilation, and
personal preference. Psychonomic Science, 1967, 7, 375-376.
Tomkins, S. S. Affect, imagery, consciousness, Vol. II. The negative
Affects. New York: Springer, 1963.
Winer, B. J. Statistical principles in experimental design. New York:
Winter, B. J., and Mehrabian. Reported in M. Argyle and A. Kendon, The
Experimental analysis of social performance. In L. Berkowitz
(ed.) Advances in experimental social psychology III. New York:
Academic Press, 1967, 55-98.
Joseph Francis Kersey t'as ho,.r, Jarudry 18, 1939, at Laki.l.ind,
Florida. In June, 1957, he .',as graduated from Edgew.ate:r High School,
Orlando, Florida. In April, 1960, he entered the United States Army
and served in Europe as a member of tie Intelligence Corps until
1963. In December, 1964, he recr-ived the degree of Bachelor of Arts
from the University of Florida. In i965. he enrolled in rhc Graduate
School of the University of Florida anr. mintriculated for a las~ter of
Arts and Doctor of Philosophy degre-e, in, psychology.
This dissertation was prepared under che direction of che chair-
man of the candidate's supervisory cunmiti Le and has be.n approved by
all members of that commiCtee. IL was submitted to the Dean of the
College of Arts and Sciences and to che Graduace Council, and was
approved as partial fulfillment ot the requirements for the degree of
Doctor of Philosophy.
Dean, Colleg of Arts/and Sciences
Dean, Graduate School
Supervisory Cormmittee -