“*vvv—

-'

 

 

. SIMILARITY AS A BASIS FOR _ -
INTERFERSONAL ATTRACTION AMONG
GROUPS IN LONG-TERM ISOLATION

Thesis for the Degree of M. A.
MlCHIGAN STATE UNIVERSITY
ROBIN RICHARD VALLACHER ‘

       
 
  

    

I. 19] 2
a
- I
. .
l I . -

.-’. n ‘ ‘ . - . .

I. . . : ' . v

'a- , .

n ' . ' V
. ‘ ‘ ’ ' ' ’ .
a I -
... ‘ '
"v‘ ‘ ‘ .A. .
. .. .

.- . r o - ' . . '5 n : . -
.. . ' ' - -‘ . l ' I ‘ --." ’

.1. . ‘ I

" a - " I cr'. -.' .

-.. , . - ,»

r7 - " . 4 .a _ - - . .QO -
_._ . ' - . . - I’I . -

J. ’ ‘- " ‘7‘ 3' :r’ — . .

‘ ‘ . . . . ° . _ ‘ " '

’ . . -"' ' . ‘

. . . _ .. n f} o " - ' ~ ‘ . ‘ -
Ifo . " ‘ . l .., .‘\' ‘, 1"- J .’ - ' '5 . '
:r- ‘ ’ " ’ z' ‘i c’l ' 'vr- l;- ‘ ' ,

-' v . . . . . . > .., '. . . ,
(>— , . . - . ~.-l.0 h.” I’A ‘ ' - . ' . -—
’4 , . . 1’ , ' . ' - _.‘ f ’4 0 3
.. . . ' . o '

->< . ' - ,
I._ 0 I - ' ' .

F '. ’ ‘ I. ‘ ' ' ' ’t ' _

‘ . . u . - .v < ' - ,
f" . ‘ J ‘ l t 3].. ,r. . I v 4'

.2” _ . 14,- -' ‘

o I " '
‘ ‘ . _ .‘. -. ‘ u ‘_
a ‘ ._ ovu.‘ 1' 7'1' .

,6. . r w 0 '. _r I! . ‘ _’. - r , '0 ‘ 1", V“ ”—{ - - 4 .

h ‘ . . O..v> ,.-., 714-: .I'. ” . - r '- ‘ "-

‘J'o ' v I. ‘ ' " . ' ' ‘ In ' " ’ . . I-"'/ ' l‘ . ‘ . "~ -.:-‘.

9‘. . - ', - .I . -.-:'-J W" - —. 'I--'°""" """ ' -' r--Hi 'M
I.’ o — on I l ' P ' C ' . ‘ v ‘ 'tl ’ 2’ ' . - T—.; "'4' l"""" "" :f’fl': '.’.‘ "I if...“ V”. ‘ t: ‘i'

’v ,. , Z“... . ‘,'.. ~ . _ y‘_.._-JIJ"IOI’JO".’ ' .', 3 , A 1’31"”

'f' ‘ . v "‘ ." ." '." Ir." I ' ' J " I‘ '5’ p E: . - J" ‘

- '~ " ' .' .-.-:’ -.. . n 1- """ " r~a.';;'!:"1¥ ' "
v' ' - ,. - .t - ,r;

’ O , , t . .; ,r . 6'! O

.. ... "(’3,’ ---mflw7

910 '9 , I '- ... 3-... . 1f- ‘ ) ' .." (VI"_!{'

A

.~ . o I . ,' ' ‘1. ’ ‘ ;V'>II.J . :14 I " ' 'V'f" ':
. .,.,. ‘ . . ,..v -. -p I ' ‘ " ._ n
'v‘. n - ' ‘ " l ' You! - at.1( I?
. ' ‘ _ l I 4-v I “'F’ ‘ r.
. . ' 33K. ‘ " . .4, r " 24"" " ',' vylf “WC
.. 4 n- v .

o .a ' t c o ' .
o . . ., . ' ’ .

. , ' o A II.- ' ' ' ‘. I aft-1“,
. - . , . ’. . .' 0 , .1 ‘l _ .c’oJ .. ’1‘;- 0" "(I . . I). :I’ 2.153.]. ',I : :'O... :14" 4’

'. A - " . ., ' ‘ ' I‘ 'I‘
.q _ .I I - q . ' , ' -‘l . ' v.hf’ '-
n ‘ A ‘ A . - ‘Il ‘ - .‘.
- . '! .-_‘.';':.a-‘ I’A‘ ' '1’-

ABSTRACT
SIMILARITY AS A BASIS FOR INTERPERSONAL

ATTRACTION AMONG GROUPS IN
LONG-TERM ISOLATION

BY

Robin Richard Vallacher

The present study investigated the role of simi-
larity in friendship formation. Research in the area of
interpersonal attraction and friendship formation is
marked by conflict in theory and inconsistency in empir-
ical findings. Particularly confusing is the role of sim-
ilarity in attracting individuals to one another. This
confused state of affairs is in part due to the limited
generalizability of the settings and subject samples em—
ployed in many of the studies dealing with this issue.
The present study avoided such difficulties by examining
natural groups in a field setting. Specifically, groups
of men who lived and worked together for approximately
one year in an isolated environment (Antarctica) were
examined to see if any consistent relationships between

1

Robin Richard Vallacher

friends could be found. Correlational analyses were used
to assess these relationships.

Contrary to prediction, friends in Antarctica were
found to be similar on few personality dimensions, and
only slightly similar in background characteristics.
While in Antarctica, however, friends did tend to develop
similar attitudes regarding their station group experi—
ence. On the other hand, as predicted, pairs of individ—
uals who avoided each other as friends were significantly
different on various dimensions and tended to deve10p
conflicting attitudes while in Antarctica. It appears,
therefore, that in a stressful task-oriented situation
it is easier to predict incompatibility than compatibility
on the basis of various traits and characteristics of the
individuals involved.

It was also predicted that depending on how sal—
ient a given dimension is for an individual, he should
be attracted to a larger or smaller range of peOple on
that dimension. The results of the analyses of this
hypothesis were generally nonsignificant, although the
results for several personality measures were in the pre-
dicted direction. Thus, at least to some extent,

2

Robin Richard Vallacher

individuals who were "high" on a personality dimension
tended, as a group, to have as friends individuals who
were relatively homogeneous on that dimension, i.e., they
were similar to each other.

Several aSpects of the setting involved in this
study were discussed as possible reasons why more signif—
icant results were not obtained. Some of these (task—
orientation, prescribed membership, temporary suspension
of common psychological needs) were held as more tenable
than others (dissatisfaction with self, dissimilarity as
stimulation in a monotonous environment).

{Various theoretical and methodological problems
with the published research in the area of friendship for-
mation and interpersonal attraction were discussed. Some
of these (single trait approach, unknown relationship be-
tween test responses and behavior, use of correlational
analyses) were evident in this study, while others (con—
trol of temporal factors, examination of nonlinear rela-

tionships) were not.

SIMILARITY AS A BASIS FOR INTERPERSONAL
ATTRACTION AMONG GROUPS IN

LONG-TERM ISOLATION

BY

Robin Richard Vallacher

A THESIS

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

MASTER OF ARTS

Department of Psychology

1972

To the members of my karass, one and all:

ACKNOWLEDGEMENTS

This thesis would have been impossible without the
invaluable assistance given me by members of the Navy Med-
ical NeurOpsychiatric Research Unit in San Diego, Cali-
fornia. In addition to making the data for this thesis
available to me, they assisted a great deal in the analyses
and provided me with relevant Unit reports and publications.
I especially wish to thank Dr. E. K. Eric Gunderson, Head
of the Operational Psychiatry Division, who made the data
and assistance available to me; Mary Paul, who wrote the
necessary computer programs; and George E. Seymour, a good
friend as well as a competent research associate, who pro—
vided much assistance and acted as the liaison between the
Unit and myself at all stages in the deve10pment of this
thesis.

I would also like to thank the members of my
thesis committee, Dr. Lawrence A. Messé, Dr. Gary E.
Stollak, and Dr. William D. Crano, for their assistance

and patience during the writing of this thesis. I would

iii

like to give special thanks to Dr. Messé, the committee
chairman, who made helpful comments on the early drafts.
This research was supported by the Bureau of Med—
icine and Surgery, Navy Department, under Research Work
Unit MFlZ.524.00l—9003D. Opinions eXpressed are those of
the author and are not to be construed as necessarily re—
flecting the official view or endorsement of the Depart-

ment of the Navy.

iv

TABLE OF CONTENTS

Page

LIST OF TABLES. . . . . . . . . . . . . . . . . . . vii

INTRODUCTION. . . . . . . . . . . . . . . . . . . . 1

Theoretical Support for Similarity-Attraction
Hypothesis. . . . . . . . . . . . . . . . . . 3

Theoretical Support for Complementarity-
Attraction Hypothesis . . . . . . . . . . . . 5

Methodological Approaches to Similarity-

Attraction Hypothesis . . . . . . . . . . . . 8
Review of Similarity-Attraction Literature. . . 9
Setting . . . . . . . . . . . . . . . . . . . . l4
Hypotheses. . . . . . . . . . . . . . . . . . . 19

METHOD. . . . . . . . . . . . . . . . . . . . . . . 26
Subjects. . . . . . . . . . . . . . . . . . . . 26
Procedure and Variables . . . . . . . . . . . . 27
Analysis. . . . . . . . . . . . . . . . . . . . 34

RESUDTS . . . . . . . . . . . . . . . . . . . . . . 4O
Prediction Variables. . . . . . . . . . . . . . 40
Station variables . . . . . . . . . . . . . . . 51

DISCUSSION. . . . . . . . . . . . . . . . . . . . . 59

TABLE OF CONTENTS (Cont.)

Page

SUMMARY . . . . . . . . . . . . . . . . . . . . . . 80

REFERENCES. . . . . . . . . . . . . . . . . . . . . 82

vi

Table

10.

LIST OF TABLES

Page
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T—TESTS FOR PERSONAL HISTORY VARIABLES. . . 41
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR NEED SCALES . . . . . . . . . . 43
RESULTS OF F—TESTS FOR NEED SCALES. . . . . . 44
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR ATTITUDE SCALES . . . . . . . . 46
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR SELF SCALES . . . . . . . . . . 47
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T—TESTS FOR FIRO-B SCALES . . . . . . . . . 49
RESULTS OF F-TESTS FOR FIRO—B SCALES. . . . . 50
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR EARLY AND LATE STATION SCALES . 52
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR EARLY AND LATE.SYMPTOM SCALES . 55
AVERAGE CORRELATIONS, STANDARD ERRORS, AND
T-TESTS FOR SUPERVISOR AND PEER SCALES. . . 56

vii

The purpose of this thesis is twofold: l) to re—
view theoretical approaches and research findings regarding
interpersonal attraction and friendship formation; and
2) to relate these to the problem of interpersonal attrac-
tion in groups in long-term isolation. More specifically,
groups of men who "wintered-over" at United States Antarc-
tic scientific stations were examined on a wide variety of
measures to see if there are any consistent relationships
between friend pairs. In addition to providing a test of
several issues in interpersonal attraction (e.g., simi—
larity vs. complementarity, the role of temporal factors,
etc.), the present study was designed to identify factors
which contribute to the prediction of compatibility under
unique and often stressful circumstances. This latter
aspect has increasing relevance as man more and more must
adapt to new and continually changing conditions such as
increasing loss of "personal space."

Perhaps the most parsimonious model relevant to
the prediction of interpersonal attraction involves the
notion of similarity along various dimensions between

1

individuals who are attracted to one another. No one, of
course, maintains that similarity is the sole determinant
of interpersonal attraction. Other important factors in-
clude such things as prOpinquity (e.g., Festinger, Schac-
ter, and Back, 1950), the reinforcing prOperties of the
situation (e.g., Lott and Lott, 1960), the temporal length
of the relationship (e.g., Rosenfeld and Jackson, 1965),
characteristics like boastfulness and self—depreciation
(Pepitone, 1964), the respective status of each individual
(e.g., Jones, 1964), and perceived liking for oneself on
the part of the other person (e.g., Backman and Secord,
1959). However, these other factors are not as potentially
useful in choosing the composition of groups as is the
similarity principle. Those factors that are related to
the situational context (e.g., reinforcing properties of
the situation) give no clue as to the possible influence
of personality variables, while those factors that do re-
late to personality either are not of a predictive nature
(e.g., perceived liking for oneself) or are contingent on
the situational context (e.g., the reapective status of

each individual).

Theoretical Support for

Similarity-Attraction
Hypothesis

Several theories are often advanced to support the
idea that similarity is conducive to positive interper-
sonal attraction. Some of these appear to be more rele-
vant to the problem at hand than others. Similarity be-
tween friends would seem to follow naturally from the var—
ious consistency theories. Heider's balance theory (1958),
for example, holds that peOple strive to make their senti—
ment relationships harmonious with their perception of the
unit relationship existent between objects. According to
Heider, separate entities which are similar tend to be
perceived as belonging together (having a unit relation—
ship). Thus, positive unit formation (perceived similar-
ity) shouLd induce a harmonious sentiment relationship
(liking). However, this line of reasoning is not partic-
ularly convincing since it is abstract and non-intuitive,
that is, it doesn't consider attraction from the point of
view of a person's needs, feelings, etc.

A much more intuitive notion is that similarity
facilitates attraction because it allows for the predic-

tion of another's future behavior. If a person appears

sociable and says he likes to be included in group activ—
ities, we may reasonably eXpect that he will engage in
group activities at every Opportunity. If we ourselves
are also high in need for affiliation, we may like another
who we know to be similarly high on that need—-not for
reasons of cognitive consistency but simply because we
anticipate many rewarding interactions, engaging in af-
filiative behavior with someone who also enjoys engaging
in affiliative behavior.

Another possible reason, related to the one de—
scribed above, involves the notion of perceived liking
for oneself on the part of the other person. It has been
shown that people tend to like those who like them. It
may be the case that when we learn that another is similar
to us, we assume that he is likely to like us, and thus we
in turn like him.

Perhaps the theoretical approach concerning simi-
larity and attraction that is the most directly relevant
to the conditions in the present study is Festinger's
theory of social comparison (1954). The theory states
that the individual has a drive to evaluate his Opinions
and abilities, and that the opinions and abilities of

Others will be used as a basis of comparison when no

objective criteria for evaluation exists. Persons who are
similar to oneself provide more precise comparisons than
do those who are dissimilar, and the similarity functions
as reinforcement. Because Of the greater amount Of rein-
forcement provided, similar persons are more attractive
than dissimilar ones. Although this theory is more con-
cerned with the evaluation Of one's Opinions and abilities
through comparison with others, it has been extended to
the evaluation Of one’s personality (Griffitt, 1966).
Given the unique circumstances in which the Antarctic
volunteer finds himself, and hence the lack of Objective
criteria by which to judge his feelings, attitudes, be-
havior, etc., it seems plausible that something like so-

cial comparison would Operate in friendship formation.

Theoretical Suppgrt for
Complementarity-Attraction

Hypothesis

In contradiction to the folk adage that "birds Of
a feather flock together" is the equally pOpular idea that
“Opposites attract." A leading prOponent of this point Of
view is Winch (Winch gg_§1., 1954), who developed a theory

Of "complementary needs." Although his theory and research

were primarily concerned with married and engaged couples,
he felt that the basic idea could be extended to inter-
personal attraction in general.

There are two general reasons why persons who
differ in need structure should be attracted to each other.
First of all, each member Of the dyad finds interaction
mutually or reciprocally rewarding because his needs are
expressed in behavior that is rewarding to the other
person. For example, a person with strong nurturance
needs behaves in a protective, nurturant manner toward
another person who has strong needs tO be dependent. In
this way, each individual satisfies the needs Of the other
person, and is in turn satisfied.

A second general reason is that persons are at-
tracted to others who have characteristics to which they
once aspired, but were prevented by circumstances from
developing. Instead, they have modeled themselves after
the image of a person with the Opposite traits, but they
find themselves attracted to individual who possess the
once-coveted traits because of feelings of wistful admira—
tion. This second idea concerning need complementarity
and interpersonal attraction reflects the "need completion

principle" advanced by Cattell and Nesselroade (1967).

Considering the context of the present investiga-
tion-~groups Of men isolated for a long period Of time in
a monotonous environment--it might reasonably be expected
that individuals would be attracted to dissimilar rather
than similar others, at least on certain dimensions.
Dissimilar others can provide varied stimulation, such as
different points of View and insights with life, new in—
terests, unpredictability. Hence some degree Of excitement
may be generated by interacting with these persons. Con-
versely, it may be that individuals would prefer to inter-
act with dissimilar others, except that they fear they
would not be as well liked by dissimilar others as they
are by similar others. Walster and walster (1963) found
that individuals who were confident that everyone they
came in contact with would like them were likely to asso—
ciate with dissimilar others. On the other hand, they
found that when it was especially important to be liked,
or when one was unsure of his likability, an individual
was likely to "play it safe" and to associate with sim-

ilar peOple.

Methodological Approaches to
Similarity-Attraction Hypothesis

Just as there are several theoretical approaches
relevant to the role of similarity in interpersonal attrac-
tion, different methods of investigation are employed by
various investigators. One approach involves comparing an
individual's test responses on some measure with the test
reSponses of his friends. The scores of the pairs are
then correlated and often compared with similar correla—
tions for random pairs within the same population or with
pairs Of mutually antagonistic or mutually indifferent
subjects. In another general approach, the measure or
measures are Obtained, and then previously unacquainted
subjects are placed in a situation requiring some degree
of interaction. Thus, similar and dissimilar pairs or
groups are created, and their interpersonal reSponses are
assessed following the interaction. Yet another method
has been employed by Byrne and his colleagues (e.g., Byrne,
Griffitt, and Stefaniak, 1967). This method involves
first Obtaining the test responses of a subject and then
presenting him with the test responses of a "stranger" and
asking the subject toevaluate the stranger on various

dimensions. The test responses of the "stranger" are

actually created by the experimenter to be similar to the
test responses of the subject on from 0 to 100% of the

items.

Review of Similarity-
Attraction Literature

Despite the amount of research that has been con-
ducted regarding the relative effects of similarity and
complementarity on interpersonal attraction, the issue is
far from settled. First of all, there are problems in-
herent in the methods described above which confuse the
interpretation of the results found in many studies (some
Of these problems are discussed below). Also, whether
similarity or complementarity is a better predictor of
friendship is no doubt dependent in part on the situa-
tional context (e.g., task versus social-emotional orien-
tation) as well as on the kinds of variables being con-
sidered.

Generally, similarity on such characteristics as
age (Glick and Landau, 1950; Hollingshead, 1951), race
(Golden, 1954; Morton, 1941), religious affiliation (Hol-
lingshead, 1950; Thomas, 1951), ethnic origin (Bossard,

1939), economic status (Byrne, Clore, and Worchel, 1966),

10

education (Landis, 1945), and previous marital status
(Bowerman, 1953) has been shown to have a positive effect
on compatibility.

Attitudinal similarity has also been found to be
positively associated with interpersonal attraction. New-
comb (1961), for example, in a study of friendship forma-
tion among male university students, found support for the
general hypothesis that given adequate Opportunity for
individuals to become familiar with each other's attitudes,
attraction is predictable from actual attitudinal agree—
ment. Other studies, Of a more experimental nature, have
also supported the proposition that attitude similarity
is a determinant Of interpersonal attraction. Byrne and
Nelson (1965), for example, showed that attraction is a
positive linear function of the prOportion of similar to
dissimilar attitudes rather than the absolute number of
similar attitudes.

With regard to personality variables, however, the
results have been less consistent. Evidence has been
found to provide support for the similarity hypothesis
(Banta and Hetherington, 1963; Izard, 1960a, 1960b;

Maisonneuve, 1954; Mehlman, 1962; Miller, Campbell, Twedt,

11

and O'Connell, 1966; Murstein, 1961), for the complemen-
tarity hypothesis (Cohen, 1956; Kerckhoff and Davis, 1962;
Rychlak, 1965; Winch, Ktsanes, and Ktsanes, 1954), and for
some combination of the two (Becker, 1964; Secord and
Backman, 1964).

Other findings give only partial support to the
similarity hypothesis in that similarity is positively
associated with attraction only under limited conditions,
or only in Specific groups or with respect to only a few
variables (Bonney, 1946; Bowerman and Day, 1956; Katz,
Glucksberg, and Krauss, 1960; McLaughlin, 1971; Posovac,
1971; Rosenfeld and Jackson, 1965; Van Dyke, 1940). Izard
(1963), for example, in a replication Of an earlier study,
found a relationship between similarity and sociometric
choice among college freshmen, but not among college
seniors.

Finally, a number of investigators have simply
found no relationship between personality similarity and
attraction (Corsini, 1956; Gordon, 1957; Hoffman, 1958;
Hoffman and Maeir, 1966; Katz, Cohen, and Castiglione,
1963; Pintner, Forlano, and Freedman, 1937; Reilly, Cum-

mins, and Stefic, 1960; Thorpe, 1955).

12

Since many Of the studies in this area have used
as subjects married or engaged couples, the applicability
Of such findings tO interpersonal attraction in general is
limited. Those studies that have dealt with interpersonal
attraction more generally have, for the most part, involved
short-term laboratory groups or presented subjects with
personality profiles Of alleged strangers. The generali-
zability of such studies is also limited since they prob-
ably do not make salient important characteristics which
could be a potential source of conflict in many "real life"
situations, particularly over an extended period Of time.
Several studies have shown, for instance, that temporal
factors have an influence on the relationship between sim-
ilarity and attraction (Newcomb, 1961; Rosenfeld and Jack-
son, 1965). In addition, subjects in laboratory studies
generally represent a rather homogeneous pOpulation, i.e.,
college sephomores. Again, this limits the generaliza-
bility of such studies. The sample and setting involved
in the present study, heterogeneous groups Of men isolated
for an extended period Of time, did not encounter the par—
ticular limits on generalizability mentioned above.

A different type of problem encountered in studies

Of this nature involves the explanation for any Observed

l3

relationship between similarity and attraction. That is,
since most of these studies involve correlational analyses,
there are several possible explanations for any Observed
relationship between similarity and attraction. The pOp-
ular eXplanation is that peOple select their friends on
the basis of similar characteristics. Another possibility
is that friends become similar due to their association.
This possibility seems more likely with regard tO atti-
tudes than with regard to personality characteristics.
Finally, it may be that individuals run into similar
peOple for reasons that have nothing to do with personal
preferences. The importance of propinquity as a deter-
minant of interpersonal attraction is well established
(e.g., Festinger §§_§l,, 1950), and it may be that peOple
tend to be thrown together with those who possess similar
characteristics. This might be particularly true in cases
where a particular type Of personality is drawn to a cer-
tain type Of occupation.

For most Of the variables used in the present
study, any relationship between similarity and friendship
would most likely be the result of individuals choosing
their friends on the basis Of similarity. This is so be—

cause scores for most variables were Obtained from the

14

subjects before their deployment to Antarctica. Scores
for a few variables (attitudes, emotional symptoms, and
performance criteria), however, were Obtained while sub-
jects were living together at the duty station, so that
for these variables there is the possibility that any Ob-
served relationship between similarity and friendship is
due to friends becoming similar as a result Of their asso—
ciation. The third possibility (similarity as an artifact
Of prOpinquity or occupational similarity) is possible for
all Of the variables, but not likely, since the groups are
isolated for several months and thus there is much Oppor-
tunity for group members.to interact with and get to know
one another intimately. Thus, for most of the variables
examined in the present study, the interpretation Of any
Observed relationship between similarity and friendship

is not as ambiguous as it was in most studies in this area.

Setting

Since the International GeOphysical Year of 1957-
1958, the United States has maintained several year-round
stations on the Antarctic continent to implement the Ant—

arctic Research Program supported by the National Science

15

Foundation and the United States Navy. At the Antarctic
stations, civilian scientists and technicians collect re-
search data, while Navy personnel provide necessary logis-
tical support. Groups varying in size from 15 to 40 men
live and work together in close association for approxi—
mately one year. For a period of approximately eight
months, all stations are completely isolated from each
other and the outside world, except for intermittent radio
communication. There is no possible way for members to
leave the station nor for help to reach them during the
isolation period.

In addition to the extreme isolation of the Ant—
arctic stations, the physical setting is undoubtedly the
most rugged environment inhabited by man. Temperatures
below —100 degrees Fahrenheit have been recorded, and winds
Of more than 100 miles per hour may prevail. During the
Antarctic summer months (when sunlight is nearly contin-
uous), construction, repair, and storage tasks must be
performed at every station in addition to the collection
Of scientific data. It is essential for these tasks to
be accomplished if the group is to survive the savage on-
slaught Of the Antarctic winter. Once the darkness of

the winter sets in, station members must limit themselves

16

almost exclusively to indoor activities. For the remainder
of the long winter, the men live and work together in vir-
tual confinement with no possibility Of evacuation or Of
additional supplies.

Men are initially selected for Antarctic assign—
ments on the basis of competence in an occupational Spe-
cialty. However, all applicants are subjected to thorough
physical and psychiatric screening examinations. Volun—
teers tend to be youthful; participants range in age from
20 to 45, but most are under 35. The typical member of
an Antarctic wintering group is likely to be totally in-
experienced with respect to cold weather and isolation.

Since each station must be a completely self-
sustaining community for many months, a variety Of scien-
tific, technical, and military occupations are represented.
The Navy personnel work in occupational specialties in-
volving support and maintenance, while scientific research
is conducted by civilians in areas such as glaciology,
ionospheric physics, and meterorology. Because of the
wide variety Of occupational Specialties, Antarctic groups
represent very heterogeneous pOpulations. Occupational
group differences (largely Navy-civilian differences) have

been demonstrated in areas such as motivation and personal

17

values (Gunderson and Nelson, 1966), personality traits
(Gunderson and Mahan, 1966), job satisfaction and perform-
ance (Doll and Gunderson, 1969), and in the predictive
validity of screening measures (Doll, Gunderson, and Ryman,
1969).

While the frontier-like existence at the Antarctic
station lends itself to a friendly and informal atmOSphere,
the close quarters and sameness of the physical and social
environments tend to highlight personal and cultural idio-
syncrasies during the long winter. Morale and work are
affected to some degree by the interpersonal frictions
that inevitably arise. Thus, psychosocial factors become
important, and it is here that the greatest complexities
exist, and knowledge is least adequate.

In 1961 a program of psychological studies involv-
ing groups wintering-over in Antarctica was instituted at
the Navy Medical Neuropsychiatric Research Unit (NMNRU) in
San Diego, California. This research is sponsored by the
Bureau Of Medicine and Surgery and is under the direction
of Dr. E. K. Eric Gunderson. These studies have been pri-
marily concerned with the validity Of selection criteria,
the measurement of job performance, and the determination

Of related correlates Of adaptation in individual and

18

group behaviors which might influence task accomplishment.
The NMNRU reports have been largely based upon data Ob-
tained in absentia from questionnaires, supervisor evalu—
ations, and peer ratings administered by the station
Officer-in-charge two or three times during the winter.
(Problems in the collection and limitations of this type
of data were outlined by Gunderson and Nelson, 1965.)
Gunderson (1966) has prepared a "Deep Freeze" bibliography
with listings of NMNRU reports from 1962 to 1966. Wilkins
(1967) reviewed Antarctic psychological studies in the
context of group behavior in long-term isolation.

The basis of interpersonal attraction in Antarc-
tica has not been previously examined p§£_§§, although
Nelson (1964) has looked at the sociometric structures
that develop during the winter. These small wintering-
Over Antarctic station groups seemed to provide a unique
Opportunity to examine the basis Of interpersonal attrac-
tion. During the isolated winter months group members
have little work to occupy their time, and, consequently,
they have much time available to cultivate interpersonal
relationships. A long period of time is involved, so the
effects of temporal factors can be studied through repeated

measures Of many Of the variables. A wide variety of data

19

is available on all Of the individuals involved, including
such things as personal history, personality measures,
attitudes, measures Of adaptation, satisfaction, and job
performance while in Antarctica, and various sociometric
measures. The sample is large (N = 204) and heterogeneous,
and thus somewhat more representative Of the general pOpu-
lation than, say, college SOphomores. Since a field or
natural laboratory setting is involved, problems with
demand characteristics, evaluation apprehension, etc.,

are negligible. Finally, as mentioned earlier, problems
with the interpretation Of any Observed relationship be-

tween similarity and friendship are not involved.

Hypotheses

Personal history yapiablpp, Interpersonal attrac-
tion has been found fairly consistently to be facilitated
by similarity on various background characteristics (see
p. 9). Newcomb (1961), however, found that the importance
Of similarity on certain "quasi-demographic" character-
istics in friendship formation diminished over time.
Therefore, since friendship choices are ascertained after

months in isolation, during which time no doubt other

20

traits assume prepotence, rather low level but positive
correlations were expected between friends.

Pppponality variables. As pointed out earlier
(p. 10), the empirical findings regarding personality
similarity (or complementarity) and friendship are in—
consistent and conflicting. No doubt this state of af-
fairs is due in part to the relative neutrality Of the
settings used in such studies, i.e., certain psychological
traits are not made salient. Given the setting Of the
present study, however, it is unlikely that an individ-
ual's psychological characteristics would not become
highly manifest.

As Secord and Backman (1964) have pointed out,
whether similarity or complementarity is a better predic—
tor Of attraction depends on the particular variable in
question. Thus, with regard to certain needs examined in
the present study (autonomy, achievement, and orderliness),
similarity between friends seems a reasonable expectation.
For instance, highly autonomous peOple should not be at—
tracted to persons low on autonomy who need leadership
and attention from others, but rather to individuals like

themselves who are not demanding Of others. On the other

21

hand, complementarity between friends seems a more reason-
able prediction with regard tO need for nurturance, which
was also examined in the present study. That is, it was
expected that a highly nurturant individual would not be
attracted to another individual who similarly "needs" to
give attention to others, but rather to an individual who
has a "need" for receiving attention from others.

Certain scales measuring self-perceptions were
employed in the present study. Generally, research has
shown that friends tend to perceive themselves in much
the same way (e.g., Griffitt, 1966), although the findings
are less than perfectly consistent (e.g., Miller 22.913}
1966). Thus, given the results Of most past studies,
similarity between friends was expected for the scales
used in the present research.

Additional scales (FIRO-B; Schutz, 1958), which
purportedly measure an individual's orientation to inter-
personal relations, were examined in the present study.
Similarity was expected between friends on those scales
measuring the individual's affection and inclusion ten-
dencies; that is, sociable and warm peOple should be drawn
to each other. On the other hand, complementarity was

predicted between friends on the scales assessing control

22

tendencies, since, for example, individuals who want to
be controlled should not be attracted to someone who has
similar motivations but rather to someone who will provide
control.

Attitudes. The theoretical orientations and em-
pirical findings are fairly consistent regarding attitude
or opinion similarity and interpersonal attraction, as
was pointed out earlier (p. lfim Therefore, with respect
to the attitudes examined in this study (some of which
were measured prior to Antarctic duty and some of which
were measured while at the duty station), similarity was
expected between friends.

Emotional symptoms. While at the Antarctic sta-

 

tion, several stress symptoms were assessed at two time
periods (early and late). Since research has shown that
there is similarity between friends regarding their emo-
tional states (Zimbardo and Formica, 1963), similarity was
predicted for friends in their symptomatology. Further-
more, this similarity was eXpected to increase from the
early to the late administration, as the volunteers' reac—
tions to their environment become more manifest in stress

symptoms.

23

Performance evaluations. It has been shown that
performance evaluations Of friends tend tO be similar
(Deutsch and Solomon, 1959). Furthermore, friends tend
to have reputational similarity (Miller pp 31,, 1966).
Therefore, in the present study, friends were expected to
be seen by both supervisors and peers as having similar
standing on various task and social—emotional dimensions.

In addition to assessing the degree of similarity
or complementarity between friends, tests were also per—
formed tO assess the influence of each of several person-
ality dimensions on the tendency for subjects as a group
to choose as friends persons who are similar or dissimilar
to each other on the trait in question. It is an everyday
observation that some people seem to relate effectively
with a wide variety Of peOple, while others seem to relate
only to a very narrow range of personalities. It seems
plausible that these differences are related to differences
in certain personality traits. For instance, it may be
that individuals who are highly nurturant relate only to
highly succorant individuals, while individuals for whom
nurturance is less salient can relate well to individuals
with widely differing degrees Of succorance. Thus, it was

hypothesized that individuals who scored "high" on a given

24

personality dimension would, as a group, have as friends
individuals whose scores on the same personality dimension
would be relatively homogeneous, that is, the within group
variance of scores would be small. On the other hand,
individuals who scored "low" on a personality dimension
were expected as a group to have friends whose scores on
that dimension were relatively heterogeneous.

Additional variables. The sample employed in this
study was split into two groups: Navy and civilian per-
sonnel. However, since the sample was split primarily for
methodological reasons no hypotheses are presented to
account for possible Navy-civilian differences in observed
relationships. Also, as a basis Of comparison for rela-
tionships Observed between friends, the analyses were per-
formed on a group Of "mutual isolates," that is, pairs Of
individuals who did not choose each other as friends.
Although no hypotheses are presented for this group, it
should be noted that it is not a control group but rather
a group composed of pairs Of individuals who chose to
avoid each other rather than to become friends. In many
situations, ignoring or not choosing a person could merely

mean a lack of sufficient contact for the development Of a

25

crystallized attitude. Given the circumstances Of the
present study, however, it is unlikely that the individ-
uals at a station had not formed definite attitudes about
every other station member. Furthermore, Borgatta (1968),
in a discussion of sociometric literature, concluded that
persons who are unchosen when the positive choice form of
the sociometric question is used tend to be the ones who
are rejected in the negative form of the question. Thus,
relationships Opposite to those hypothesized to occur
between friends were eXpected to occur between "mutual

isolates."

MET HOD

 

The subjects for the present investigation were
204 male volunteers who were assigned to U.S. Antarctic
stations for a period Of one year during 1964—1969. Sixty-
One percent of the subjects were Navy personnel; the re—
maining 39%.were civilian scientists, technicians, and
engineers. Only subjects from the two largest stations
(Byrd and South Pole) were included in the analysis. The
size of these stations ranged from 18 to 30 men.

The median age for both the Navy and civilian
personnel was 27 years, with a range for the total sample
Of 18 to 43 years. Approximately half of the men were
married. In terms of personal background, the Navy and
civilian personnel represented different populations. The
civilian personnel had higher levels of formal education
(66% were college graduates) than did the Navy personnel
(59%.were high school graduates). The civilians more fre—
quently came from urban communities and were raised in

26

27

smaller families Of higher socio—economic status than were
the Navy personnel.

In addition to background differences, the Navy
and civilian personnel have been shown to differ along
various other dimensions. As mentioned in the Introduc-
tion (p. 16), there are Navy-civilian differences in the
areas Of motivation and personal values, personality
traits, job satisfaction and performance, and in the pre—
dictive validity Of their scores on the screening measures.

Because of these cultural and psychological dif-
ferences between the Navy and civilian personnel, the
sample was split into these two sub—groups for purposes
Of analysis. That is, since they come from different
backgrounds and have different values, etc., they may have
different criteria for friend selection, which, when con—
sidering the two groups together, could mask or wash out
strong relationships that might exist in each group sep-

arately.

Procedupe and Variablgp

As mentioned earlier, all applicants undergo a

rigorous screening program several months prior to their

28

deployment to Antarctica. In addition to a physical exam-
ination and separate interviews with a psychiatrist and a
psychologist, each applicant is administered several ques-
tionnaires and inventories. Of these, the following were

included in the present investigation.

(1) Personal History Questionnaire

This questionnaire Obtained basic background data
pertaining to the applicant's developmental history. The
particular items included in the analysis were level Of
education, church attendance, number Of siblings, boyhood

town size, father's education, and mother's education.

(2) Opinion Survey

This survey was developed eSpecially for the Ant-
arctic screening program. Factor analysis was employed to
identify highly inter—correlated clusters of items which
appeared to represent meaningful psychological concepts.
Four of the test scales measured common psychological needs
(Need Scales): Achievement, Autonomy, Nurturance, and
ngerliness. The content Of these four scales are gener-
ally similar tO those of the corresponding Edwards Per-
sonal Preference Schedule scales (1958), although the

format of the items was entirely different. Statements

29

were presented as single items to be evaluated on a six—
pOint scale ("strongly agree" to "strongly disagree") and
response values were summed for items composing each scale.
Also derived through factor analysis were five
scales which measured attitudes toward the Antarctic as-
signment (Attitude Scales): Motivation, Usefulness,
Boredom, Confidence in organization, and Confidence in

medical care. As with the Need scales, statements were

 

presented as single items to be evaluated on a six-point
scale Of agree-disagree, and response values were summed
for items composing each scale.

Finally, four scales represented personality self—
descriptions based upon clusters Of intercorrelated traits
(Self Scales). The scales and the traits composing them
are as follows: Decisive (Decisive, Obedient, Handy,
Alert, Orderly, and Self—improving); Excitable (Excitable,
Argumentative, Hostile, Suspicious, and Impulsive); Blppp
(Blunt, Stubborn, Rough in manner, and Hard); and Absent-
minded (Absentminded, Awkward, and Slow). Each of these
traits was presented as an adjective to be evaluated by
the volunteer as being descriptive of himself along a six-
point scale. Response values were summed for all traits

composing a scale.

30

(3) FIRO—B (Fundamental
Interpersonal Relations
Orientation-Behavior)

The only published test used consistently in Ant-
arctic research has been the FIRO-B questionnaire (Schutz,
1958). According to Schutz's "Postulate of Interpersonal
Needs": "(a) Every individual has three interpersonal
needs: inclusion, control, and affection. (b) Inclusion,
control, and affection constitute a sufficient set Of
areas Of interpersonal behavior for the prediction and

explanation of interpersonal phenomena [1958, p. 13]."

He defined these three needs as follows:

Inclusion: ". . . the need to establish and maintain a
satisfactory relation with peOple with respect to interac-

tion and association." (1958, p. 18.)

Control: ". . . the need to establish and maintain a
satisfactory relation with respect to control and power."

(1958, p. 18.)

Affection: ". . . the need to establish and maintain a
satisfactory relation with people with respect to love

and affection." (1958, p. 20.)

31

The FIRO-B measures two complementary aspects (expressing
and wanting) of these three needs. There are thus six
scales, each of which consists of nine items, and the

scales are cumulative (Guttman) in type:

Ipplpsion Exppessed (IE): participating in group activ—
ities;

Inclusion Wanted (IE): wanting to be included in group
activities;

Control Expressed (QE): controlling others, exercising
dominance or leadership;

Control Wanted (CW): wanting direction or regulation from
others;

Affpption Exprepsed (pg): being affectionate with others;

Affection Wanted (AW): wanting affection from others.

Scores on each of the six scales can range from 0 (low
need) to 9 (high need).

In addition to the above measures (ppediction
variables), which were Obtained prior to Antarctic duty,
various measures were Obtained on each subject while he
was at his duty station (station variablgp). Of these
measures, the following were examined in the present

study. All of the following measures except the Supervisor

32

evaluations (6) and Peer evaluations (7) had two adminis—
trations: one month after the beginning of the winter
isolation period (March) and one month before the winter

isolation period ended (September).

(4) Station Opinion Survey

This questionnaire tapped the station member‘s
perceptions of his Antarctic eXperience and his impres-
sions Of how well the station functioned. Each station
member rated 34 statements along a six—point scale of
agree-disagree. Factor analysis was used to derive six

scales (Station Scales): Egalitarian atmosphere, Compat—

 

ibility, Motivation, Accomplishment, Usefulness, and Job
dissatisfaction. Response values were summed for all items

composing a scale.

(5) Common Complaints

Each of ten "stress" symptoms was rated by the
station member along a four-point scale according to how
much he was troubled by it while at the station. Four
scales were derived by factor analysis (Symptom Scalgp):
Depression, Insomnia, Hostility, and Anxiety. ReSponse

values were summed for all items composing a scale.

33

(6) Supervisor Evaluations

Both the military supervisor and the civilian
supervisor at a station independently rated each station
member on ten dimensions, using an eight—point scale for
each dimension. The military and civilian evaluations
on each dimension were then averaged for each subject.
On the basis Of content, four scales were derived and
used in the present study (Supervisor Scales). The scales
and the items of which they are composed are as follows:
Tapkfpgrformance (Industriousness, Motivation, Proficiency);
Emotional ptabilipy (Emotional control, Acceptance of
authority); Social compatibility (Likability, Cheerfulness,
Considerate Of others); and Leadership (Leadership abil-
ity). Response values were summed for items composing a
scale. Individuals at each station were ranked and as-
signed T-scores (§'= 50, S.D. = 10) in terms Of average
supervisor ratings on each of the four behavior dimensions.
T-scores were utilized in order to control for station

size and differences in supervisor's rating distributions.

(7) Peer Evaluations
On each Of ten items, each station member nominated

from one to five men at the station whom he felt stood out

34

on the dimension tapped by the particular item. Four Of
these items correspond to the four Supervisor scales de-
scribed above and were examined in the analysis. Thus,

there were four scales (Peer Scales): Task performance,

 

Emotional stability, Social compatibility, and Leadership.
As with the Supervisor scales, the subjects at each sta-
tion were ranked and assigned T-scores in terms Of the
number of peer nominations on each of the four behavior
dimensions in order to control for station size. As men-
tioned earlier, peer nominations were Obtained only once
during the Antarctic assignment: one month before the end
of the winter isolation period (September).

One of the items from Peer Evaluations was con-
sidered separately in the present analysis. On this item,
each station member nominated those men at the station
whom he had found to be his closest friends while at the

station (Closest friend).

Analysis

For each of the variables described above, Product—
Moment correlations were computed for each station between

"closest friends" (as determined by the Closest friend

35

item). That is, for each variable there were two columns
Of figures: in one column were the scores on that vari-
able Of all individuals at a station; in the other column
were the correSponding scores of the "closest friend"
(first choice on the Closest friend item) of each indi-
vidual in the first column. Correlations were then com-
puted between these columns of scores separately for each
station on each variable.

The Fischer Z transformation was then used to
transform each of the station correlations on each var—
iable. These z—scores were weighted by Ni (the number of
pairs of "closest friends" at station i) and an average
z-score was Obtained across stations on each variable,
which was then transformed back into a correlation and
the mean correlations were tested against the null hypoth-
esis that the population correlation was zero by means of
t-tests.

As a basis Of comparison, a similar analysis was
performed for "mutual isolates," i.e., pairs Of individ—
uals at a station who did not choose one another on the
Closest friend item. This comparison group was derived
in the following way. First, all individuals at a station

who were not chosen at all by individual A on the Closest

36

friend item were looked at to see if they in turn did not

choose individual A on the Closest friend item. Since

 

there were Often several such pairings between individual
A and his mutual non-choices, only one such pairing was
chosen randomly for individual A. This procedure was
carried out for every individual at a station, so that
every individual was paired with a "mutual isolate."
Correlations were then computed between these pairs of
"mutual isolates" and tested against the null hypothesis
that the pOpulation correlation was zero in the manner
described above for "closest friends."

In addition to the correlations obtained for each
variable individually, correlations were computed between
the following pairs of FIRO-B scales for both the "closest
friends" and "mutual isolates" groups: Inclusion EXpressed
(IE) vs. Inclusion Wanted (IW); Inclusion Wanted (IW) vs.
Inclusion EXpressed (IE); Control Expressed (CE) vs. Con-
trol Wanted (CW); Control Wanted (CW) vs. Control Ex—
pressed (CE); Affection Expressed (AE) vs. Affection
wanted (AW); Affection Wanted (AW) vs. Affection EXpressed
(AE). According to Schutz (1958), ideal reciprocal com—
patibility can be achieved by a dyad only if for each need

area the expressed behavior of each member is equal to the

37

wanted behavior of the other. This reciprocal condition
parallels that which Winch pp_al. (1954) refers to as
"type II complementariness," where a positive correlation
exists between the need of one member and the complemen-
tary need of the other. Thus, these correlations tested
the notion of need complementarity between "closest
friends" and "mutual isolates."

The correlational analysis described above for the
total sample was also performed separately for the Navy
sample. The rationale for splitting the sample was given
earlier (see Subjects, pp. 26-27). Thus, both "closest
friends" and "mutual isolates” groups were derived for
the Navy personnel. In a few cases, Navy personnel chose
civilian scientists as "closest friends"; such cases were
included in the analysis. However, the "mutual isolates"
group was composed entirely of Navy subjects. Originally
it was planned to perform this same analysis on the civil-
ian sample, but because of the low number of civilians at
each station (average N = 5) it was felt that the result—
ing correlations from separate stations would be tOO un-
stable.

In addition to the correlational analyses, further

analyses were employed for certain Of the personality

38

variables (Need scales and FIRO-B scales). The assumption
underlying these analyses (see pp. 23-24) was that if an
individual is ”high" on some characteristic (e.g., need
for Nurturance), he should be highly sensitive to that
characteristic in others and hence should choose friends
on that basis. On the other hand, a person "low" on that
characteristic should be less discriminate in his choice
of friends along that dimension since it has less salience
for him. Thus, F-tests were employed to test the general
hypothesis that for a group of individuals "high" on some
characteristic there should be greater homogeneity on

that characteristic in their "closest friend" choices

than there should be in the "closest friend" choices Of

a group Of individuals "low" on that characteristic.

These tests were performed in the following way. First,

a frequency distribution Of scores was Obtained for each
variable in question (e.g., need for Nurturance). Those
persons whose Nurturance scores were in the upper one-
third of the distribution were classified as "high." Sim—
ilarly, those persons whose Nurturance scores were in the
lower one-third of the distribution were classified as
"lows." Next, the "closest friends" chosen by the "highs"

were determined (high chosen), as were the "closest friends"

 

39

chosen by the "lows" (19w chosen). The variance Of Nur-
turance scores in the high chosen group was then compared
to the variance of Nurturance scores in the lgw chosen
group by means of an F-test that was the ratio of the
respective mean squares Of the two groups of scores. The
F-tests were performed separately for the Navy "closest
friends" group and the civilian "closest friends" group
on each of the Need scales and the FIRO-B scales.

For both the correlational and the F-test analyses
the .05 level (two-tailed) was used to define statistical

significance.

RESULTS

Prediction variables

 

Table 1 presents the results of the correlational
analyses for the Personal History variables. It appears
that "closest friends" in the total sample were signifi-
cantly similar on church attendance, father's education,
and mother's education. None Of the variables were signif-
icant for "closest friends" in the Navy sample. "Mutual
isolates" in the total sample differed significantly on
church attendance and boyhood town size, while "mutual
isolates" in the Navy sample differed significantly on
church attendance.

Thus, "closest friends" tended to be similar on a
few aspects of their personal history, while “mutual iso-
lates" tended to Be different on certain aspects Of their
personal history. An exception to this pattern, however,
is the significant similarity between "mutual isolates" in
the total sample on amount Of formal education. There was
no relationship between "closest friends" on this dimen-

sion. Apparently, individuals who are somewhat similar

40

41

~66. v m*.«

 

 

 

Ho. v mac
mo. v.@«
o o o o o o e o o o o o HMO-Humoflflwm
NNH HI Nmo Nmo I mHm H NHH HBH whN HI NmO mOH I «NHm N N50 mmH m.H0£UQz
o o o o o O o o. e e o . o. HHOflgmoavm
th I mmo who I Ofim H mmo NwH mom I mmo mNO I «fitmNN fi #50 NHM m.H0£¥Mh
o o e o o o o I o o o o o ”NI-Hm
com H- mmo cos . mum H Hes mew «mew m moo mma . mmm H mas ems asap eooasom
0 II o o ' o I o a II o I o e o o o mmfiflfl m
*va N mOH mmm ®m¢ NNH mmo mflm H hmo HvH th hmo hNO HO Hmnfiflz
. . . occupsmuus
asOmn MI Mbo NBN I mHm. mmo. Nmo. «hmv.NI Nmo. mmH.I smfih.N omo. mmH.
stages
om@.I mOH. Mbo.l mmv.HI GMH. fiON.I amhm.N Vmo. MGN. mvm. mOH.. mmo. GOHHMUDVH
.H .H .H .H
u m up u m m. u m up u m .m
; fiend;
=mmumaomH assusa= gmpsmfium umomOHO: smoumHOmw Hs5u58= gmpcmfinm ummmoau: .

 

 

 

HQfiZflm H>¢z

 

mam2¢m A<909

.mflflMflHm¢> NmOEMHm AdzOmmmm mom mBmNBIB 92E .mmommm DMflQZfiHm .mZOHBGAmmmOO m04mm>¢

H mnmdfi

42

in educational background tend to avoid one another in
this type of environment, at least to a certain extent.

In Table 2 are shown the results of the correla-
tional analyses for the Need scales. Within the total
sample there were no significant correlations, positive or
negative, on any of these scales between either "closest
friends" or "mutual isolates.” Within the Navy sample,
however, ”closest friends" were significantly different
(complementary) on need for Achievement. There were no
significant correlations on any of the scales between "mu—
tual isolates" in the Navy sample, although there was a
marginally significant trend toward complementarity on
need for Orderliness.

The results of the F—tests on the Need scales for
the Navy and civilian "closest friends" group are shown in
Table 3. Although only three statistically significant F-
ratios were Obtained, in all cases the variance (MS) Of the
low phosen group was greater than the variance (MS) of the
high chosen group. Thus, the hypothesis was generally con-
firmed; that is, a group composed Of individuals with high
need tended to choose friends who were more homogeneous
than were the choices of persons with low need on that di-

mension. The significant results were Obtained on (a) need

43

.mo. v ma

 

 

mom.au moo. mma.n moo.n mod. vao.u moo. mso. ooo. me~.n ems. omo.u mmmaaaumonouc
msm.au mod. mvH.u mom.- was. moo.u oo~.au moo. sea.» Hoo.n ono. ooo.u mocmusuuszuc
oom.n sad. AHH.I mm».- was. NHH.- moo. Has. moo. oso. mad. moo. ssoaousanc
oom.u was. moa.n «vo¢.Nu ooa. mom.u qmm.au moo. sma.u smo.u moo. omo.n usmsmpmasuanc
.H .H H .H
u m m. u m .m u m .m u m .m

 

:mmusaomw Hmsusfi:

 

=monmnum umwmoaog

=mmumaomw assume:

 

Hum
=mpsmwnm pmomOHO= mans. >

 

mamZflm M>¢Z

 

WAfiZGm AS808

 

 

mmA¢Um Qmfiz.m0fl mBmmBIB 92¢ .mmommm Qm¢dz<fim .mZOHB¢ANmm00 m9¢mm2fl

N mflmda

44

for Autonomy for civilians, (b) need for Orderliness for
Navy personnel, and (c) need for Orderliness for civilian
personnel.1 Thus, of the four needs measured, Autonomy
proved to be salient in friend selection for the civilians,
while Orderliness was salient in friend selection for both

Navy and civilian personnel.

TABLE 3

RESULTS OF F-TESTS FOR NEED SCALES

 

 

 

 

 

High Low
Variable Sample Chosen Chosen F
N MS N MS
n A hi V e t Navy 22 17.87 22 24.95 1.40
C e em “ Civilian 12 11.52 12 10.27 1.12
n_Autonom Navy 22 17.19 22 18.50 1.08
y Civilian 12 10.27 12 30.39 2.96*
n_Nurturance Navy 22 12.46 22 18.81 1.52
Civilian 12 11.27 12 16.15 1.43
n_0rderliness Navy 22 8.86 22 19.47 2.20*
Civilian 12 8.88 12 33.12 3.73**

 

*p < .05
**p < .025

 

1It is interesting to note that in none of these

three cases did the means Of the high chosen group appear

to be significantly different from the correSponding means
Of the low chosen group: Autonomy (civilian)--high chosen
2': 31.42, low chosen 2': 30.25; Orderliness (Navy)--high

chosen §'= 17.00, low chosen 2': 15.05; Orderliness

 

 

45

Table 4 shows the results of the correlational
analyses for the Attitude scales. In both the total
sample and the Navy sample there were no significant cor—
relations between "closest friends." There were signifi-
cant correlations, however, between "mutual isolates" in
both the total and Navy samples. In the total sample,
"mutual isolates" tended to be significantly different
on Confidence in organization and Confidence in medical
care. "Mutual isolates" in the Navy sample tended to be

significantly different on Boredom, as well as on Confi-

 

dence in organization. Thus, it appears as though it is
easier to predict mutual avoidance than it is friendship
formation on the basis of certain attitudes related to
anticipated group experience.

The results of the correlational analyses for the
Self scales are shown in Table 5. There were no signifi-
cant correlations, either positive Or negative, for any
of the scales. Apparently, self-descriptions bear no
relationship to either friendship formation or mutual

avoidance in this type of environment.

 

(civilian)--high chosen 2': 16.17, low chosen §'= 16.67.
However, since the F—tests were significant the assumption
Of homogeneity of variance was clearly violated, thereby
making t-tests inapprOpriate. Thus, statistical compari-
sons between the means were not made.

46

.mo. v m».

 

 

 

omm.| ONH. mvo.n mvm.n and. «no.1 mvo.l HOH. moo.| Nev. omo. mvo. soHum>Huoz
G 6 O O O O 0 ' O 0' O O 0 ”H60 Hmolflcmz
moo an who NNH I How HOH and «How N ovo mma vma I moa vao 1 ca .msoo
sowusNHsmmHo
«hom.mn Ono. ooa.a oom.a wed. mmH. avao.mn who. OOH.1 mme.a mmo. mma. as .
. msoo
«mov.mu moo. oma.| mvo.| oma. $00.: mam.u moo. 060.1 oma. maa. mHO. Sopmuom
mmo.n «me. 600.: oom.u oma. ovo.l mmm.a Hmo. OOH. wom.H moo. Oma. mmosasmomo
H H H H
u m H u m H u m .m u m .m
mHOOHHm>

=mmumaomw Hooves:

 

=monoanm unmmoflo=

=mmumaomfl Hmsussa

 

=nocwaum unmaoao.

 

mqgm gflz

mgm WQDBHFHAN mom mammalfi

 

mflgm 93.08

O24 .9856 9335 52035580 mommEE

v mflmfir

47

 

mmH.I

 

 

vvm. NOH. mmo. mHH. mHo.I hNH.H Ono. mho. OMH.I mOH. wHo.I vowaflfilucmmnm
vow.l @mo. mmo.l hHw.I NVH. mmo.l mhv.l vvo. HNO.I NNN. vHH. mNo. uGDHm
mum.HI mmo. VHH.I 0mm.HI vOH. mMH.I NMh.I omo. moo.l mvm.l omo. mNo.I anmuHOxm
mmn.l Hmo. mmo.l mVM.HI 00H. QHN.I vmh.l vmo. moo.l Nmo.I mmH. moo.l mbflmfloma
H H H H
u m H u n .W H w .W p m .m
387.83,

:mmumHomH assassa

 

:mUGGHHM DammOHU:

amwumaOmH Heapssa

 

:MUGQHHM “mTMOHU:

 

WAQZQm w>4z

 

mqmzdm A4808

mmAdUm mnmm mom mBmMBIB 02¢ ammommm GM¢DZ¢Hm .mZOHBflAmmmOU WUflMWPG

m mam¢9

48

The results of the correlational analyses on
and between the FIRO—B scales are similarly disappoint-
ing (see Table 6). In both the total and Navy samples
there were no significant correlations between "closest
friends" on any of the scales or between complementary
scales. There were, however, significant negative cor-
relations between ”mutual isolates" On Inclusion Ex-
pressed (IE) in both the total sample and in the Navy
sample. Thus it appears that peOple who avoided each
other in this situation differed in their affiliation
tendencies.

Like the correlational analyses, the F—tests
for the FIRO—B scales yielded statistically nonsignifi-
cant results. As can be seen in Table 7, which summar-
izes these comparisons, the only significant F-ratio was
on Inclusion Wanted (IW) for the Navy sample. This F-
ratio, contrary to the general hypothesis, indicates
that as a group Navy personnel who scored high on IW
tended to choose as "closest friends" a group of indi—

viduals who were less homogeneous on that dimension than

49

 

 

 

 

=mODMaOma amausfia

=noawanw ummmoaog

:WOUMHOmH HMHHDE:

 

:mosoaHm HmomOaO:

 

HAAZGm M>¢z

madam mIOmHm mom menses oza .mmOMmm Omaozsem .szHefimmmOO moamman

 

m HAM¢H

NAQZflm AQBQE

.HG. v mas
.mo. v at
omm. moo. oao. moo.II, «oak moa.I oom.a «co. moo. omm. maa. omo. mt .m> 3m
ooo.aI amo. omo.I ova. aoa. omo. oov.I moo. vmo.I aov.I oaa. omo.I 3m .m> m4
omm.I amo. «oo.I Nam.al oma. Noa.I vmm.I ooo. ooo.I woe. moo. mmo. m0 .m> so
ova. omo. mao. vom.I Naa. omo.I aoo.a moo. oaa. aoo.aI moo. amo.I 30 .m> mu
emo.I ooo. ooo.I omo.a aaa. mma. mo~.I moo. omo.I ama.a moo. omo. Ma .m> 3H
mmo. moo. ovo. mao. voa. moo. avm.I omo. amo.I mmo.I Nmo. moo.I 3H .m> NH
o m.aI Nmo. Nva.I omo. m a. ao. mmo.I moo. mmo.I mum. «ma. mvo. Azmv pounds
a a m 8368.?
l O O I- O O O O O O O 0' Aav cmmmwug
oom mmo amo ooo mma moo «mo ooo omo mmo I moa voo soauoommc
. . .- .- . .. .- . . . . .- one .4353
mmm I Oaa moa mam oaa omo mmm omo omo I amo I omo ooo aOHusou
m. o o. a. o . oa. o o. ma.a ooo. amo. a . oo. «0. Amuv Commmmem
mm a o mv vv m v N «v v o aoHusoo
O O O O O C O O O O O O AgHv @853
ooo Noa moo moo oaa ooo mom ooa omo aoo I omo ooo I scamsaosa
. . . . . . _ . . . . . . :5 Banana
«amo NI moo mma I mam ama aaa «aomm MI mmo ooa I aom mmo mmo seamsaosa
H H H H
u. m m. u m m u m m u m m
wanvaum>

iiiillllllllll

 

TABLE 7

50

RESULTS OF F-TESTS FOR FIRO-B SCALES

 

 

 

 

High Low
Variable Sample Chosen Chosen F
N ms N ms

Inclusion Navy 24 4.26 24 3.99 1.07

EXpressed (IE) Civilian 12 4.63 12 4.45 1.04

. Navy 20 14.75 20 .79 18.74*

Inclu51on

wanted (IW) Civilian 12 15.30 12 8.99 1.70
Control Navy 23 3.80 23 3.42 1.11

Expressed (CE) Civilian 12 10.39 12 4.81 2.16
Control Navy 24 4.39 24 4.51 1.03

wanted (CW) Civilian 12 3.66 12 6.45 1.76
Affection Navy 24 2.34 24 3.39 1.45

ExPressed (AB) Civilian 12 2.45 12 2.64 1.08
Affection Navy 24 5.52 24 5.56 1.01

wanted (AW) Civilian 12 5.30 12 3.79 1.40

 

*p < .005.

51

did persons who scored low on IW.2 The most Obvious in-
terpretation of this finding is that individuals who have
a high need for affiliation (i.e., high on IW) want to be
accepted by everyone, even by those for whom affiliation
or inclusion is less important. However, it is not imme-
diately clear why this should be so for only the Navy per-
sonnel, and not for the civilians as well. Perhaps the
greater autonomy for civilians in their work roles made
inclusion a less salient dimension than it was for the
Navy personnel, whose work roles required much interper-

sonal contact.

Station Variables

 

In Table 8 are shown the results of the correla-
tional analyses for the Station scales. For "closest
friends" in the total sample, significant similarity was

demonstrated on early Motivation, late Accomplishment,

 

It is interesting to note that the mean of the
high chosen group (3.7) appeared to be significantly
greater than the mean Of the low chosen group (.55), al-
though significance tests were not performed (see foot-
note l). Apparently a similarity effect was Operating in
friend choice for individuals low on IW; the very large
variance in the scores Of friends chosen by "high" indi-
viduals masked this effect, however, so that a nonsignifi-
cant correlation was Obtained on this variable (see Table 6).

52

.aoo. v m«««

 

 

 

.Ho. v on.
.mo. v m«

. . . u I . . . . . . . . soauoom
moa HI ooo ooa I amo a moa Hon ooo ooo omo IaIeaa m ooo ovm Inaumnoao nos
mv~.I ooo. moo.I oom.a «ma. Hon. ooo.I ooo. ooo. oma.a aoa. oaa. mnmcasmomo

«mo~.~I moo. ooa.I aoo~.~ ooa. ooo. ooo. ooo. aoo. «Iaom~.v ooo. omo. ucmsaoaaosoooa
omm.I ooo. aoo.I aoo.aI oaa. aoo.I ooo.a moa. ooa. ~oo.a ooa. oaa. coaum>auos
aoo.aI ooo. ooo.I omv.a ooa. mma. mam.I ooo. omo. omo.a moa. ooa. Huaaanaumasoo

. . . . . . . . . . . . 3.8.8....

«ooo «I omo ooa I ooo I flea aaa I amm I ooo omo com a oaa ova anaumuaamom
1388

. . . . . . . . . . . . coauomm
mos I «ma aoo I omo moo moo mam oaa ooa ooo a ooo moa Imaumomao ooo
omo. moo. ooo. amm. aoo. omo. aoa. ooo. ooo. Hom.I ooo. omo.I mmoaasmmno
oom.I «ma. mao.I «ma. aaa. mao. ooo. aoa. ooa. mam.a oaa. ova. unassmaaasoooa

«Hom.oI ooo. oaa.I omv.aI ooo. oaH.I ooa.a ooa. oaa. «Ioao.m aoo. oom. coaum>auoz
aoo.aI ooa. ova.I omo. moa. oma. «aom.oI aoo. ova. ova.a ooo. oaa. muaaanaumasoo

. . . . . . . . . . . . mnoamnosum
ooa a ooo aoo «as ooo omo Ivoo o ooo ooa ooo I moo ooo I amanmuaamoo

Amaummo
H .H .H .H
u m H u m H u m H u m H
mannaum>

:moumaoma amauss:

 

.moamanm ummnoao=

amoumaOma assuza:

 

zmoamanm uowmoao=

 

mam2<w M>4z

 

m mnmfla

mnmz<m AflBOB

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIoIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

mmAdUm ZOHB¢Hm mfidn 92¢ wam¢m mom mammBIB 92¢ .mmemmm QMfiQZdBm .mZOHfidflmmmOO mwdmmkfl

 

53

and late Job dissatisfaction. For the Navy sample the

only significant similarity demonstrated between "closest
friends" was on late Accomplishment. The "mutual isolates"
in the total sample were significantly different on early
Compatibility, but significantly similar on early Egali-
tarian atmOSphere. "Mutual isolates" in the Navy sample
were significantly different on early Motivation, late
Egalitarian atmOSphere, and late Accomplishment.

Thus, similarity was demonstrated between "closest
friends" on certain attitudes, while with one exception
(early Egalitarian atmOSphere for "mutual isolates" in
the total sample) differences were demonstrated between
"mutual isolates" on certain attitudes. It is interesting
to note that for none Of the groups did the same scales
demonstrate significance both early and late. Apparently
different attitudes were salient in these two time periods.
Generally, the late correlations were higher than the
early correlations for "closest friends" in both samples,
while no early—late differences are apparent for the
"mutual isolates" groups. It appears, therefore, that
"closest friends" became somewhat more similar in their

attitudes as the winter isolation progressed.

54

The results Of the correlational analyses for the
Symptom scales are presented in Table 9. There were no
significant correlations, positive or negative, between
"closest friends" in the total sample. In the Navy sample
there was a significant difference between "closest friends"
on late Depression. There were no significant correlations,
positive or negative, between "mutual isolates" in either
the total sample or in the Navy sample. Apparently,
friends in this type of environment do not necessarily
develOp similar symptoms, nor do people avoid those with
similar symptoms.

Table 10 presents the results of the correlational
analyses for the Supervisor and Peer scales. Between
"closest friends" in the total sample there was significant

similarity on Peer Emotional stability, while "closest

 

friends" in the Navy sample were significantly different
on Supervisor Leadership. Thus, for the sample as a whole,
"closest friends" were seen to some extent by their peers
as being similar in their emotional stability. Within the
Navy sample, "closest friends" tended to be seen by super—
visors as differing in their leadership. vPerhaps leader-

ship among Navy personnel (to some extent) reflects need

55

.mo. v my

 

 

 

 

voa.aI ooo. aoa.I mo~.aI oma. oma.I moo.I ooo. ooo.I oaa.oI ooo. ooo.I oumaxco
omo.I ooa. ooo.I ov~.aI oaa. oma.I ooo. ooo. amo. ooa.I moo. oao.I Huaaaunom
moo. omo. omo. mov.I ama. aoo.I oe~.a ooo. moa. mao. aoo. moo. maosonsH
ova.a moo. ooa. «omo.mI moa. mo~.I ovm.a ooo. ooa. omo.I moo. amo.I aoaonoumwo
Aoumqv
omo.aI moa. ooa.I moa.a aoa. ooa. ooa.aI ooo. maa.I mom. ooo. omo. Humaxso
oaa.I voa. ooo.I amo.I ama. ooa.I ooa. ooo. mao. oo~.I ooa. omo.I Huaaaumom
moo.aI ooo. ama.I mom.aI ooa. mma.I oaa.aI ooo. ooo.I ooo. ooo. ooo. maesonaa
avo.I aoa. moo.I ooa.aI oaa. ooa.I mo~.I ooa. omo.I ooo. ooo. ooo. coamomumoo
Amaummo
.H H H .H
u m m. u o mp u m .H u. n .H
mannaum>

amoumaOma amsusE:

 

amosoaHm HommOaO:

:monaoma aMOHOS:

.moamana unmnoau=

 

mam2<m W>¢z

 

mamZGm 44809

mma¢0m Edamzwm afida Qz< NQMdfl MOE mfimmfila n24 ammommm Qmmflzmfim .mZOHadammmOO m0¢mm>d

m wands

56

.ao. v 944
.mo. v 9*

 

ooo.l
woo.al

«amNH.MI

«aHmo.MI

mmo.l
«www.mI

Nam.l

moH.NI

amo.
oaa.

ooo.

aoo.

Qua.
moa.

mma.

Naa.

Noo.l
oaa.l

NVN.I

MNN.I

ooo.l
mmN.I

moo.l

m¢N.I

mom.al
mmm.al

oaa.

mmm.aI

amm>.Nl

moo.I

omv.l

omv.l

aaa.
oaa.

wNa.

omo.

moa.
moa.

Nva.

oma.

ooa.l
mma.l

mac.

mmH.I

omN.I
voa.l

Noo.l

Noo.l

oom.I
ava.l

omm.l

mom.l

Nwm.
mmm.al

ooN.I

voo.I

ooo.
moo.

ooo.

moo.

Naa.
mmo.

mmo.

mmo.

mNo.I
aao.I

omo.l

ooo.l

mmo.

.moa.l

mmo.l

voo.l

amm.l
0mm.a

«omo.m

oom.a

NmN.al
mma.l

mmm..

mum.l

NHa.
mmo.

Noa.

mwo.

omo.
oma.

aoa.

maa.

mvo.l
mva.

mmN.

mma.

vma.l
omo.l

mmo.

moo.l

maanoomoa
Huaaanaumosoo

amaoom
suaaanmuo

amsoauosm
oosmEHouHom

Home

Amoamom Hommv

masmHmomoq
muaaaoaummaoo
amaoom
Huaaanmuo
amsOauOSm
mosmaHOMHom
xmma
Ammamom
Homa>Homdmv

 

u

H
m

H

 

u

H
m

H

u

H
m

H

u

H
m

H

 

amoumaOma amsusfi=

=MGC6HHM HOQMOHO:

amoumaoma assusaa

 

=noamaum uoomoaoa

 

9992€m w>¢z

 

9992dm 99909

oanmanm>

m99¢0m 9999 92¢.mOmH>Mflmbm 909 mammBIB 92¢ .mMQmmm 9m¢9z¢8m .mZOHBm9mmmOO mwdePG

OH 99999

57

for Dominance and "closest friends" are complementary on
that dimension.

For the "mutual isolates" in the total sample,
there were no significant correlations on any of the Super-
visor or Peer scales. Within the Navy sample, however,

"mutual isolates" were significantly different on Super-

 

visor Social compatibility, Peer Task performance, and

Peer Emotional stability. Within the Navy sample, there-
fore, peOple who avoided each other as friends had a
stronger relation on these dimensions (negative) than did
peOple who chose one another as friends. Apparently, then,
in this type Of environment individuals tend to avoid those
who are different on task and socio—emotional dimensions,
but do not necessarily form friendships with those who

are similar.

On balance, the results Of the correlational anal-
yses are disappointing in that few significant correlations
were found between "closest friends," particularly on the
prediction variables. In fact, the greater number of
correlations (mostly negative) on the prediction variables
for "mutual isolates" suggests that perhaps avoidance is

easier to predict in this type of environment than is

58

positive attraction. On the other hand, the results of
the F-tests indicate that personality measures (Need and
FIRO-B scales) do have some predictive power for friend-

ship formation but not in the usual linear sense.

DISCUSSION

Generally speaking, the results Of the present
study were disappointing. While similarity was demon-
strated between "closest friends" on certain attitudes,
personality similarity (and complementarity) proved to be
a poor predictor of friendship choice. Within the total
sample, similarity was demonstrated between "closest
friends" on church attendance, father's education, mother's
education, and in their early feelings Of motivation and
late feelings Of accomplishment and job dissatisfaction
while in Antarctica. In addition, "closest friends" were
perceived by their peers in Antarctica as being similar
in their emotional stability. The pattern Of results for
"closest friends" in the Navy sample generally resembled
that Of the total sample, although fewer results reached
statistical significance. In fact, the only significant
similarity demonstrated between "closest friends" in the
Navy sample was in their late feelings of accomplishment
while in Antarctica. Moreover, unlike the results for the

I

59

60

total sample, complementarity was demonstrated between
"closest friends" on three variables for the Navy sample:
need for Achievement, feelings Of depression while in
Antarctica, and supervisor perceptions of leadership
ability.

Since no complementary relationships were Obtained
for "closest friends" in the total sample, the complemen-
tary relationships demonstrated between "closest friends"
in the Navy sample suggest certain Navy—civilian differ-
ences in the criteria used in the selection of friends.

It seems plausible that these differences are in large
part related to differences between these two samples in
their prescribed roles while in Antarctica. The civilians
at these stations are there primarily for purposes of in-
dividual research, so a high degree of autonomy is pos—
sible and, if desired, these persons can avoid interper-
sonal contact. The Navy personnel, on the other hand, are
responsible for station maintenance, which requires much
coordinated effort on their part, and hence a great deal
Of interpersonal contact. In addition, the Navy personnel
perform jobs that are more demanding and less intrinsic-
ally rewarding than the relatively "soft" duties performed

by the civilians. Thus, different factors are likely to

61

become manifest and assume importance in friend selection
for these two groups. In addition, established Navy-
civilian differences in background and personality char-
acteristics (see p. 16) may also contribute to differences
between the two groups with respect to the criteria used
in friend selection. Unfortunately, Navy-civilian dif-
ferences in the criteria for friend selection can only be
inferred since, as pointed out earlier (p. 37), the small
number of civilians at each station made separate corre-
lational analysis for the civilian sample impractical.
Thus, differences between the total and Navy samples are
seen as merely suggestive of possible differences between
the Navy and civilian samples.

It is interesting to note the relatively large
number Of significant relationships that were Obtained
between "mutual isolates." As expected, these relation—
ships were generally Opposite tO those Observed between
"closest friends," that is, relationships tended to re-
flect complementarity rather than similarity. Within the
total sample, "mutual isolates" were found to be comple-
mentary with regard to church attendance, boyhood town

size, their confidence in the organization and medical

care prior to their Antarctic deployment, their reported

62

participation in group activities (Inclusion EXpressed--
FIRO-B), and in their early perceptions of group compati-
bility while in Antarctica. On the other hand, similarity
was obtained for level Of education, and in their early
perceptions of egalitarian atmosphere in Antarctica. It
is not clear why similarity should be obtained on these
two variables, although the relatively low level of sig-
nificance (p < .05) suggests that these results may have
been due to chance.

As with the results for "closest friends," the
pattern Of results were generally similar for both the
total and Navy samples Of "mutual isolates." Complemen-
tarity was demonstrated between "mutual isolates" in the
Navy sample with regard to frequency of church attendance,
number Of siblings, their anticipated boredom and confi-
dence in the organization prior to Antarctic deployment,
their reported participation in group activities (Inclu-
sion EXpressed——FIRO-B), and their early motivation and
late perceptions Of egalitarian atmOSphere and accomplish-
ment in Antarctica. In addition, in Antarctica they were
seen by their supervisors as different in their social
compatibility, and by their peers as different in their

tas performance and emotional stability. Unlike "mutual

63

isolates" in the total sample, they did not demonstrate
significant similarity on any Of the variables.

Thus, the hypothesis of complementarity between
mutually indifferent or antagonistic individuals in this
setting was in general supported. In fact, more signifi—
cant relationships were Obtained for "mutual isolates"
(19) than for "closest friends" (11). It appears, there—
fore, that mutual indifference or antagonism bears an
equal or somewhat stronger relationship to the variables
in this study than does friendship. Furthermore, only
four of the prediction variables yielded a significant
relationship between "closest friends," while for the
"mutual isolates" eight of the prediction variables yielded
significant relationships. This suggests that, at least
in the setting of this study, mutual antagonism or indif—
ference is easier to predict than attraction. Perhaps
even stronger relationships on many of the other variables
would have been Obtained if the analysis had been done on
pairs of individuals who were more clearly antagonistic.
Unfortunately, an unambiguous measure of antagonism (e.g.,
"who don't you like?") was not Obtained from the station

members.

64

In addition to the hypothesized relationships
between similarity or complementarity and friendship, it
was eXpected that the salience of certain personality
traits would relate to the homogeneity or heterogeneity
of chosen friends on the trait in question. However, the
F—tests which were employed to test this hypothesis yielded
only four significant relationships of this nature. Al—
though the F—ratios were generally nonsignificant, it was
noted earlier (p. 42) that for the Need scales, the dif-
ferences in variances were all in the predicted direction.
Thus, with regard to need for Autonomy and need for Order—
liness, and to some extent with regard to the other needs
(Achievement and Nurturance), "high" individuals as a
group tended to choose as friends individuals who were
relatively homogeneous on the same need. In addition to
theoretical interest, a finding of this nature has Obvious
practical application to the selection Of individuals for
group composition, particularly in the type Of setting
investigated in this study. That is, in a situation such
as Antarctica where individuals must live and work to-
gether in virtual isolation from the remainder of society,

it is important that individuals are chosen who can

65

establish a compatible relationship with most, if not all,
of the group members.

Perhaps the main conclusion that can be drawn on
the basis Of the results of the present investigation is
that similarity (or complementarity) is not as strong a
_predictor Of friendship formation in a situation of this
type as originally thought. Certain questions, however,
remain unanswered. Why were generally "nonsignificant"
results Obtained? Would a different analysis have been
more apprOpriate? Can these results be generalized to
interpersonal attraction in general? Considering the
published research in this area, can any firm conclusions
be made about the determinants of friendship? What meth—
odological changes are needed in future research in this
area?

First of all, certain aSpects of the situation may
have accounted for the relative unimportance Of similarity
and complementarity. Life at the Antarctic station is
very task-oriented; physical survival is dependent upon
the performance Of various jobs. Thus, group members may
have been drawn to those who were most competent at per-
forming their work role, and it has been shown that simi—

larity affects interpersonal judgments along

66

social-emotional dimensions, but not along task-oriented
dimensions (McLaughlin, 1971). Similarly, Hoffman and
Maeir (1966) found that task-oriented groups composed of
individuals with either similar or dissimilar personality
profiles did not differ in their attractiveness for their
members. Most Of the positive correlations between simi—
larity and attraction have been found in groups which
were principally social in nature, such as rooming-house
residents (Newcomb, 1961), or work associates who social-
ized together outside Of the work setting, but who were
not necessarily members of the same work group (Rosenfeld
and Jackson, 1959). In these situations the principal
demand placed on the relationship is merely one of getting
along with one another and, hOpefully, enjoying the rela-
tionship. Under such circumstances, similarity of person-
ality could provide the basis for enjoying the same kinds
of activities, sharing similar feelings about various
events, and offering sympathy when things go wrong. The
extreme salience of task performance in the present study,
however, may have overwhelmed the importance Of person-
ality similarity in friendship.

Another possible factor specific to the Antarctic

setting that may have influenced the Obtained findings

67

involves the motivation of the typical Antarctic volunteer.
Law (1960) concluded that "escape" is a motivating factor
for these individuals. Certainly this type of situation
is a haven for the technically competent individual who

is deficient in social skills. Thus, to the extent that
these individuals volunteer for Antarctic duty in order to
"escape" their personalities, they should not be attracted
to other station members who are similar to them in per-
sonality. This interpretation would be more plausible

if similarity had been demonstrated between "mutual iso—
lates." However, the results suggest that similar indi—
viduals did not necessarily avoid each other, so the above
interpretation is not well supported.

Perhaps the most important factor involving the
setting of the present study was the prescribed membership
of the station groups. Although there is considerable
Opportunity for interaction between all individuals at
the Antarctic stations, interaction between Navy and
civilian personnel is relatively infrequent, presumably
because they have vastly different roles to perform (al—
though personality and cultural differences may also be a
factor). Thus, ”functional prOpinquity" (Festinger pp al.,

1950) may have put restrictions on friendship to a

68

considerable degree. In most studies finding a positive
association between similarity and attraction, the groups
were more able to regulate the extent of their relations
with others or to associate more with certain members than
others (e.g., Rosenfeld and Jackson, 1959). In unstruc—
tured social situations, people with similar personalities
may become attracted to each other as they find their
attitudes, interests, and reactions to shared experiences
to be compatible. Friendship formation in a group whose
membership is prescribed by the organization, may, on the
other hand, be a function of factors more directly related
to the goals of the group.

In addition to the specific factors related to the
Antarctic setting, there are factors related to conditions
of isolation in general that may in part account for the
lack Of a consistent relationship between similarity and
attraction. As pointed out in the Introduction to this
study (p. 7), it could be argued that individuals re-
stricted to a monotonous unchanging environment should
be attracted to others who are dissimilar rather than
similar, since they would provide a source of varied
stimulation. However, the results indicate that while

"closest friends" were not similar on many dimensions,

69

neither were they significantly dissimilar on many dimen-
sions. This interpretation, therefore, does not appear
to be well supported.

Another, more plausible, interpretation is that
this type of environmental deprivation amounts to the
temporary suSpension Of basic human needs such as affec-
tion, security, and significance. Thus, to the extent
that the various personality measures tapped such needs,
similarity (or dissimilarity) on these dimensions would
not necessarily be related to friendship. The results
support this interpretation. In fact, the only person-
ality measure that had a relationship (negative) with
friendship choice was need for Achievement among the Navy
sample. This particular relationship on need for Achieve—
ment probably reflects the increased salience of job per—
formance in Antarctica, although it is not intuitively
clear why friends should be dissimilar on this dimension.
Perhaps need for Achievement also reflects need for Domi-
nance, at least within the Navy sample, in which case
complementarity would be expected.

Besides the factors related to isolation that may
in part account for the Obtained results, there are more

general reasons why a consistent relationship between

70

similarity and friendship was not demonstrated. For one
thing, there probably are large and consistent individual
differences in the extent to which similarity or comple-
mentarity affects attraction. For example, a person with
high self—esteem may be attracted to similar others, while
a person with low self—esteem may shun similar others and
be attracted instead to those who have traits which he
lacks and/or to which he aspires. Support for this idea
is given by Posavac (1971), who found that similarity was
a prime factor in interpersonal attraction for certain
personality types, while complementarity or some other
factor was a major determinant of attraction for other
personality types.

Furthermore, temporal factors have been shown to
influence attraction. Newcomb (1961), for example, found
that the effect of attitudinal similarity on attraction
increased over time. At the same time, he also found that
similarity in certain "quasi—demographic" variables re—
lated to interpersonal attraction early in acquaintance,
but not later. He reasoned that early eXpectations about
interpersonal agreement may have been formed on the basis
of similarity Of initially observable prOperties of the

persons, but that these eXpectations subsequently were

71

disconfirmed when previously covert attitudes became ex—
pressed. This interpretation is consistent with the pro—
posal that the behaviors exhibited by persons in the ini—
tial stages of interaction tend to be facades (Goffman,
1959) or at least nonrepresentative samples Of the be—
haviors they display under more familiar circumstances
(Thibaut and Kelley, 1959).

In addition to the possibility that overt behavior
changes over time, changes in social motivation may also
occur. It is likely that increased acceptance by one's
peer grOUp leads to a reduction in insecurity, thereby
changing the criteria by which social choices are made.
Research on early acquaintance has demonstrated that ex—
pectations and fears of rejection are positively related
to the choice Of similar persons (Rosenfled, 1964). Fur—
thermore, even though highly similar persons are chosen
because of eXpected reciprocation, less similar persons
tend to be viewed as more attractive (Walster and Walster,
1963). Thus, social choices in later acquaintance may be
based more upon the potential for pleasurable interaction
than upon anxiety reduction.

Thus, it may be that early in acquaintance station

members were attracted to one another, in part, on the

72

basis of similarity in personality or social behavior.

But as individuals got to know one another better, facades
were drOpped and social motivation changed, so that
friendships were formed on other bases, such as task
performance or attitudinal agreement on topics of mutual
concern. Unfortunately, there are no data regarding
friendship choice early in the winter with which to test
this interpretation. However, the results suggest that
there was greater attitudinal agreement between "closest
friends" late in the winter than early in the winter.

In addition to the situational and theoretical
reasons outlined above, there are methodological reasons
why more "significant" findings were not Obtained. These
problems are not limited to the analysis employed in the
present study, but prevail throughout the literature in
this area.

First Of all, there is the general inadequacy of
the single trait approach. In studies of this type an
unknown number of unidentified variables are potentially
Operative, but are not consistently controlled. Further,
personality similarity itself is defined in terms Of one
or at best a small subgroup Of personality variables so

that similarity along all other dimensions is also not

73

controlled. The situation, therefore, is one in which the
effect of a very limited number of independent variables on
attraction is determined in a context where a large number
Of uncontrolled independent variables are Operating.
Furthermore, even when studies examine a large num—
ber of independent variables (as in this study), they are
usually examined in isolation, that is, separate correla-
tions are computed for each variable with the criterion.
This implies the rather naive assumption that traits are
additive. A much more reasonable position is that traits

are interactive. Only if the variable under investigation

 

were of sufficient strength to override all other indepen-
dent variables, Or if a sufficient number of the other in-
dependent variables happened to convary with it, or if the
other variables were accidentally controlled through ran—
domization, would the hypothesized relationship be ob—
served. Since the Obtained findings in this area lack
consistency, these special circumstances do not appear to
occur regularly. For example, a person high in need for
affiliation and low in anxiety is likely to be attracted
to entirely different people than is a person high on both
affiliation need and anxiety. Or, as another example, a

person high in need for dominance and low in need for

74

affiliation may interact with very few people and for
limited selfish purposes. On the other hand, a person
similarly high in dominance but also high in affiliation
may be attracted to many different peOple and act as a
pOpular leader rather than a ruthless manipulator. The
above examples show how complex the situation can be con-
sidering only two variables at two defined levels; in the
real world literally hundreds of variables may be inter—
acting in a much more complex fashion.

Therefore, rather than looking at traits individ-
ually, the patterning of traits should be the independent
variables in research of this nature. Many existing tests
(e.g., MMPI, Edwards Personal Preference Schedule) yield
personality profiles which could be employed in such re-
search. Hoffman and Maeir (1966), for instance, used
personality profiles based on responses to the Guilford—
Zimmerman Temperament Survey tO form homogeneous and
heterogeneous groups.

Another methodological problem involves the rela-
tionship between scores on a personality measure and be—
havior in an interpersonal situation. Personality dimen—
sions influence attraction only to the extent that such

dimensions have an effect on behavior in a given situation;

75

it is the behavioral stimuli to which subjects respond
and not to the hypothesized personality dimension. But
the relationship between personality test scores and be—
havior in various everyday life encounters or even behav-
ior in a controlled laboratory situation is for the most
part unknown. However, it seems reasonable that a one-
to—one relationship does not hold. It may be that test
responses, because of social desirability, idealized self—
image, etc., either do not correSpond to behavior at all,
or correSpond to the behaviors exhibited early in an in-
teraction, or between individuals who are less well—.
acquainted than are friends. If that is the case, it is
not surprising that friends do not have "similar" person-
alities as measured by test responses. Friends may indeed
be very similar in their behavior, motivations, and per—
Spective toward life, but analyses that are based on tests
that tap superficial behavior or no behavior at all will
not demonstrate this.

One way to overcome this problem is to have actual
behavioral measures of the subjects' personalities. For
instance, separate judges could rate an individual on
various dimensions as he is Observed in interaction with

others. However, in addition to being highly impractical—-

76

even for a small number of subjects--this method is less
than satisfactory, since the behavior exhibited by a sub-
ject in that situation may not reflect his behavior in
different settings or with other interactants. Another
possibility is to employ projective type tests to assess
the subject's personality. Such tests supposedly reveal
unconscious determinants of personality and hence are un-
likely to be distorted by the subject's desire to appear
"normal."

Another methodological weakness of studies in this
area is the pervasive use of correlational analyses. It
seems rather naive to assume that the relationship between
similarity and attraction should be a linear one. There
is no reason to suppose, for example, that a person
slightly below average in need for dominance should choose
as friends those slightly below (or above) average in need
for dominance. He has other needs that are more salient
and which should determine friendship choice. The F-tests
employed in this study were an attempt to examine possible
relationships that were free of the assumption of linear-
ity in correlational analyses. It is an everyday Observa—
tion that certain individuals can effectively (or super—

ficially) relate only to a narrow range of peOple, while

77

others seem capable and/Or willing to build relationships
with practically everyone. This aSpect of friendship
formation seems just as important as asking in what ways
are all friends similar, but has never been examined em-
pirically.

Another possible alternative to correlational
analyses would be to compute eta coefficients for friend
pairs. Eta does not yield insight into the Specific na-
ture of a relationship, however, and therefore has less
predictive value than the product-moment correlation. Yet
another possibility is to compute correlations separately
for different points on the distribution Of subject scores
on the variable in question. Thus, it may be that indi-
viduals low on some dimension choose similar people as
friends, while individuals high on the same dimension
choose dissimilar (complementary) people as friends. The
total correlation, meanwhile, would show nO relationship.
The above possibilities are not by any means exhaustive.
Certainly other analyses are also possible; the point is
that results Obtained from traditional correlational anal—
yses can be misleading since only linear relationships are

examined.

78

Still another problem with studies in this area
is the confusion between "interpersonal attraction" and
"friendship." It is one thing to view another person as
attractive and quite another thing to build a friendship
with that person. While many peOple may be seen by an
individual as attractive, most peOple have a relatively
small number of close friends. The qualities we admire
in a person may not be the qualities we would desire in
a person with whom we confide, confess our weaknesses,
etc. As pointed out earlier (pp. 70—71), at different
steps in the acquaintance process we may exhibit different
behaviors or have different social motivations. Suppo—
sedly, friends are at a fairly advanced point in the
acquaintance process, so that facades to some extent are
drOpped and persons at this stage interact for different
reasons than do those who are less well acquainted. Lab-
oratory studies of interpersonal attraction, however, look
at individuals with no prior acquaintance (e.g., Walster
and Walster, 1963), or ask subjects to evaluate supposed

"strangers" whom they never will meet (e.g., Byrne pp al.,

 

1967). Thus, conclusions regarding similarity and attrac-

tion from such studies may not be generalizable to

79

real-life interaction past the first encounter, let alone
to something as unique as friendship.

This particular weakness was not evident in this
study since the subjects were isolated together for sev-
eral months. However, it may be that really close rela—
tionships were not formed and that "closest friends" were
individuals who were merely more compatible with each
other than with anyone else at thestation. This seems
plausible since there were relatively few mutual or re-
ciprocated friend choices; only 40 percent of the indi-
viduals chosen first on the Closest friend item recipro-
cated in their first choice (although almost all individ—
uals chosen first reciprocated on later choices). Thus,
the results of the present study may reflect interpersonal

attraction in this type Of situation rather than friend-

ship per pp,

SUMMARY

Contrary to prediction, "closest friends" in Ant-
arctica were found to be similar on few personality dimen—
sions, and only slightly similar in background character—
istics. While in Antarctica, however, "closest friends"
did tend to develOp similar attitudes regarding their
experience at the station. On the other hand, as pre-
dicted, "mutual isolates" (pairs of individuals who were
not attracted to one another) were significantly different
on various dimensions and tended to develOpIconflicting
attitudes while in Antarctica. It appears, therefore,
that in a stressful task-oriented situation it is easier
to predict incompatibility than compatibility on the basis
Of various traits and characteristics of the individuals
involved.

It was also predicted that depending on how sal—
ient a given dimension is for an individual, he should be
attracted to a larger or smaller range of peOple on that
dimension. The results of the analyses of this hypothesis

8O

81

were generally nonsignificant, although the Obtained F-
ratios for several personality measures (Need scales) were
in the predicted direction.

Several aSpects of the setting involved in this
study were discussed as possible reasons why more signif-
icant results were not obtained. Some Of these (task-
orientation, prescribed membership, temporary suspension
of common psychOlOgical needs) were held as more tenable
than others (dissatisfaction with self, dissimilarity as
stimulation in a monotonous environment).

Various theoretical and methodological problems
with the published research in the area of friendship for-
mation and interpersonal attraction were discussed. Some
Of these (single trait approach, unknown relationship be-
tween test responses and behavior, use of correlational
analyses) were evident in this study, while others (con—
trOl of temporal factors, examination of nonlinear rela-

tionships) were not.

REFERENCES

REFERENCES

Backman, C. W., and Secord, P. F. The effect of perceived
liking on interpersonal attraction. Human Rela—
tions, 1959, ng 379-384.

 

Banta, T. J., and Hetherington, M. Relations between needs
Of friends and fiances. Journal of Abnormal and

Becker, G. The complementary—needs hypothesis, authori-
tarianism, dominance, and other Edwards Personal
Preference Schedule scores. Journal Of Person—
ality, 1964, 32, 45—46.

 

Bonney, M. E. A sociometric study of the relationship Of
some factors to mutual friendships on the elemen-
tary, secondary, and college levels. Sociometpy,
1946, 2, 21-47.

 

Borgatta, E. F. Sociometry. In Sills, D. (Ed.), Interna—
tional EncyclOpedia of the Social Sciences, Vol.
15. The MacMillan Company and The Free Press:
1968, 53-57.

Bossard, J. H. S. Nationality and nativity as factors in
marriage. American Sociological Review, 1939, 4,
792—798.

Bowerman, C. E. Assortive mating by previous marital
status: Seattle, 1939—1946. American Sociolog—
ical Review, 1953, 18, 170-177.

Bowerman, C. E., and Day, B. R. A test of the theory Of
complementary needs as applied to couples during
courtship. American Sociological Review, 1956,
21, 602—605.

82

83

Byrne, D., Clore, G. L., Jr., and Worchel, P. Effect of
economic similarity-dissimilarity on interpersonal
attraction. Journal of Personality and Social
Psychology, 1966, 4, 220-224.

 

Byrne, D., Griffitt, W., and Stefaniak, D. Attraction and
similarity of personality characteristics. Jour-
nal of Personality and Social Psychology, 1967, 5,
82-90.

Byrne, D., and Nelson, D. Attraction as a linear function
of prOportion Of positive reinforcements. Journal
of Personality and Social Psychology, 1965, 1,
659-663.

Cattell, R. B., and Nesselroade, J. R. Likeness and com-
pleteness theories examined by 16 personality
factor measures on stably and unstably married
couples. Journal of Personality and Social Psy-
chology, 1967, 1, 351-361.

Coates, T., and Mazur, S. Personality characteristics and
interpersonal attraction. Psychology, 1969, p, 2—9.

Cohen, A. R. EXperimental effects of ego-defense prefer—
ence on interpersonal relations. Journal of Ab—
normal and Social Psychology, 1956, p2, l9-27.

Corsini, R. J. Understanding and similarity in marriage.
Journal Of Abnormal and Social Psychology, 1956,
.52. 327-332.

Deutsch, M., and Solomon, L. Reactions to evaluations by
others as influenced by self-evaluations. Socio-
metry, 1959, 22, 93-112.

Doll, R. E., and Gunderson, E. K. E. Occupational group
as a moderator Of the job satisfaction—job per-
formance relationship. Journal of Applied Psye
chology, 1969, 5;, 359—361.

84

Doll, R. E., Gunderson, E. K. E., and Ryman, D. H. Rela-
tive predictability of occupational groups and
performance criteria in an extreme environment.
Journal of Clinical Psychology, 1969, 4;, 399-402.

Edwards, A. L. Manual for the Edwards Personal Preference.
Schedule (revised edition). New York: Psycholog—
ical Corp., 1959.

Festinger, L. A theory of social comparison processes.
Human Relations, 1954, 1, 117—140.

 

Festinger, L., Schachter, S., and Back, K. Socialgpres-
sures in informal groups: A study of human factors
in housing. New York: Harper, 1950.

 

Glick, P. C., and Landau, E. Age as a factor in marriage.
American Sociological Review, 1950, 4;, 517-529.

Goffman, E. The presentation Of self in everyday life.
Garden City, N.Y.: Doubleday, 1959.

Golden, J. Patterns of Negro-white intermarriage. Amer-
ican Sociological Review, 1954, 12, 144-147.

Gordon, J. E. Interpersonal predictions of repressors and
sensitizers. Journal of Personality, 1957, 2;,
686-698.

Griffitt, W. B. Interpersonal attraction as a function Of
self-concept and personality similarity-
dissimilarity. Journal of Personalityyand Social
Psychology, 1966, 4, 581-584.

Gunderson, E. K. E. Adaptation to extreme environments:
Prediction of performance. U.S. Navy Medical
NeurOpsychiatric Research Unit Report NO. 66—17,
April, 1966.

Gunderson, E. K. E., and Mahan, J. L. Cultural and psy-
chological differences among occupational groups.
Journal of Psychology, 1966, 4;, 287-304.

 

85

Gunderson, E. K. E., and Nelson, P. D. Personality differ—
ences among Navy occupational groups. Personnel
and Guidance Journal, 1966, 44, 956-961.

 

Gunderson, E. K. E., and Nelson, P. D. Individual perform-
ance measures for extremely isolated groups. U.S.
Navy Medical NeurOpsychiatric Research Unit Report
NO. 65-6, February, 1965.

Heider, F. The psychology of interpersonal relations.
Wiley, 1958.

Hoffman, L. R. Similarity of personality: A basis for
interpersonal attraction? Sociometry, 1958, 4;,
300-308.

HOffman, L. R., and Maeir, N. R. F. An eXperimental re—
examination Of the similarity-attraction hypoth-
esis. Journal of Personality and Social Psychol—
pgy, 1966, 4, 145-152.

Hollingshead, A. B. Cultural factors in the selection of
marriage mates. American Sociological Review,
1950, lg, 619-627.

 

Hollingshead, A. B. Age relationships and marriage.
American Sociological Review, 1951, 44, 492-499.

Izard, C. E. Personality similarity and friendship.
Journal Of Abnormal and Social Psychology, 1960,
91. 47-51 (a).

Izard, C. E. Personality similarity, positive affect, and
interpersonal attraction. Journal Of Abnormal and
Social Psychology, 1960, pi, 484-485 (b).

Izard, C. E. Personality similarity and friendship: A
follow-up study. Journal of Abnormal and Social
Psychology, 1963, 44, 598-600.

Jones, E. E. Ingratiation: Aysocial psychological anal-
ysis. New York: Appleton-Century-Crofts, 1964.

86

Katz, I., Cohen, M., and Castiglione, L. Effect Of one
type of need complementarity on marriage partners'
conformity to one another's judgments. Journal
of Abnormal and Social Psychology, 1963, pg, 8-14.

Katz, I., Glucksberg, S., and Krauss, R. Need-satisfaction
and Edwards PPS scores in married couples. Journal
of Consultipg Psychology, 1960, 44, 203-208.

Kerckhoff, A. C., and Davis, K. E. Value consensus and
need complementarity in mate selection. American
Sociological Review, 1962, 21, 295—303.

 

Landis, P. H., and Day, K. H. Education as a factor in
mate selection. American Sociological Review,
1945. $9; 558-560.

Law, P. Personality problems in Antarctica. Medical
Journal Of Australia, 1960, 41, 273-282.

 

Lott, B. E., and Lott, A. J. The formation of positive
attitudes toward group members. Journal of Ab-
normal and Social Psychology, 1960, pl, 297-300.

 

 

Maisonneuve, J. A contribution to the sociometry Of mu-
tual choices. Sociometry, 1954, 41, 33-46.

McLaughlin, B. Effects Of similarity and likableness on
attraction and recall. Journal Of Personality and
Social Psychology, 1971, 49, 65-69.

 

 

Mehlman, B. Similarity in friendships. Journal of Social
Psychology, 1962, p1, 195-202.

Merton, R. K. Intermarriage and the social structure.
Psychiatry, 1941, 4, 371-374.

Miller, N., Campbell, D. T., Twedt, H., and O'Connell,
E. J. Similarity, contrast, and complementarity
in friendship choice. Journal of Personality and
Social Psyghology, 1966, ;, 3-12.

87

Murstein, B. I. The complementary need hypothesis in
newlyweds and middle-aged couples. Journal of
Abnormal and Social Psycholggy, 1961, 4;, 94—197.

 

Nelson, P. D. Structural change in small isolated groups.
U.S. Navy Medical NeurOpsychiatric Research Unit
Report NO. 64-24, 1964.

Newcomb, T. M. The acqpaintanceyprocess. New York: Holt,
Rinehart, and Winston, 1961.

Pepitone, A. Attraction and hostility. New York: Ather-
ton, 1964.

 

Pintner, R., Forlano, G., and Freedman, H. Personality
and attitudinal similarity among classroom friends.
Journal of Applied Psychology, 1937, 4;, 48-65.

Posavac, E. J. Dimensions of trait preferences and per-
sonality type. Journal of Personalipy and Social
Psychology, 1971, 42, 274-281.

 

Reilly, M. S. A., Commins, W. D., and Stefic, E. C. The
complementarity Of personality needs in friendship
choice. Journal of Abnormal and Social Psychology,
1960, pl, 292-294. '

Rosenfeld, H. M. Social choice conceived as a level of
aSpiration. Journal of Abnormal and Social Psy-
chology, 1964, pg, 491-499.

Rosenfeld, H. M., and Jackson, J. Effect of similarity of
personality on interpersonal attraction. American
Psychologist, 1959, i4, 366-367.

Rosenfeld, H. M., and Jackson, J. Temporal mediation Of
the similarity-attraction hypothesis. Journal of
Personality, 1964, 4;, 649-656.

 

 

Rychlak, J. F. The similarity, compatibility, or incom-
patibility Of needs in interpersonal selection.
Journal of Personality and Social Psychology,
1965, 4, 334-340. '

88

Secord, P. F., and Backman, C. W. Interpersonal congru-
ency, perceived similarity, and friendship.
Sociometpy, 1964, 41, 115—127.

Schutz, W. C. FIRO-B: A three—dimensional theory of
interpersonal behavior. New York: Rinehart and
CO., 1958.

Thibaut, J. W., and Kelley, H. H. The social psychology
Of groups. New York: Wiley, 1959.

Thomas, J. L. The factor Of religion in the selection of
marriage mates. American Sociological Review,
1951, lg, 487-491.

Thorpe, J. G. A study Of some factors in friendship forma-
tion. Sociometry, 1955, 48, 207-214.

 

Van Dyke, V. E. Personality traits and friendship forma-
tion in adolescent girls. Journal Of Social Psy-
chology, 1940, 4;, 291—303.

Walster, E., and Walster, G. W. Effect Of eXpecting to
be liked on choice of associates. Journal of
Abnormal and Social Psychology, 1963, 91, 402-404.

Wilkins, W. L. Group behavior in.long-term isolation.
In M. H. Appley and R. Trumbull (Eds.), Psycho—
logical Stress. New York: Appleton—Century-
Crofts, 1967.

 

Winch, R. F., Ktsanes, J., and Ktsanes, V. The theory of
complementary needs in mate selection: An anal-
ytic and descriptive study. American Sociological
Review, 1954, 49, 241—249.

Zimbardo, P., and Formica, R. Emotional comparison and
self-esteem as determinants Of affiliation.
Journal of Personality, 1963, 4;, 141-162.

MICHIGAN STATE UNIV. LIBRARIES

I III II II lllllH ll

 

31293100287121