This article was downloaded by: [University of Arizona]

On: 04 January 2013, At: 03:43

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

The Journal of Social

Psychology
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/vsoc20

The Development of
Interpersonal Relationships:
Social Penetration Processes
Dalmas A. Taylor

University of Delaware, USA

Version of record first published: 01 Jul 2010.

To cite this article: Dalmas A. Taylor (1968): The Development of Interpersonal

Relationships: Social Penetration Processes, The Journal of Social Psychology, 75:1,
79-90
To link to this article: http://dx.doi.org/10.1080/00224545.1968.9712476

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/
terms-and-conditions
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any
representation that the contents will be complete or accurate or up to
date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not
be liable for any loss, actions, claims, proceedings, demand, or costs or

Downloaded by [University of Arizona] at 03:43 04 January 2013

damages whatsoever or howsoever caused arising directly or indirectly in

connection with or arising out of the use of this material.

The Journal o f Social Psychology, 1968, 76, 79-90.

T H E D E V E L O P M E N T OF INTERPERSONAL R E L A T I O N S H I P S :
SOCIAL PENETRATION PROCESSES*
University of Delaware2

Downloaded by [University of Arizona] at 03:43 04 January 2013

DALMAS
A. TAYLOR~

A. INTRODUCTION
T h e theory of social penetration (1, 2, 4, 16, 17) provides a framework for
describing the development of interpersonal relationships. Social penetration
refers to the reciprocal behaviors that occur between individuals in the development of an interpersonal relationship. These behaviors include exchange of
information (e.g., attitudes, values, biographical-demographic, and personal
data), exchange of expressions of positive and negative a f f e c t , and mutual activities (e.g., sports, dating, studying, etc.). Interpersonal development is
thought to proceed along two related dimensions, brendth of penetration, or the
amount of interaction, information exchange, etc., per unit time (e.g., the
amount of communication per week), and depth of penetration or the degree of
intimacy of a typical interaction or exchange. Related studies on social penetration have demonstrated that interpersonal exchange processes are modified by
situational factors, such as socially restricted environments (2), and compatibility or reward-cost characteristics (3, 4). T h e present study is concerned
with two aspects of interpersonal development : ( a ) changes in amount of intimate and nonintimate exchanges over time, and ( b ) the role of individual differences in social penetration processes. T h e interactions of personality and
level of intimacy with time are of critical interest here.

1. Self-Disclosure
An element of interaction which has recently received considerable attention
is self-disclosure. I n some early, nonexperimental work, Lewin ( 11) speculated

* Received

in the Editorial Office, Provincetown, Massachusetts, on February 10,

1967. Copyright, 1968, by T h e Journal Press.

1 From Bureau of Medicine and Surgery, Navy Department, Research T a s k

MF022.01.03-1002, Subtask 1. T h e opinions and statements contained herein a r e the
private ones of the writer and are not to be construed as official or reflecting the views
of the Navy Department or the Naval Service at large.
2 T h i s report is based on a dissertation submitted to the University of Delaware in
partial fulfillment of the requirements for the Ph.D. degree. T h e author is deeply grateful to Dr. Irwin Altman, Dr. William Haythorn, and Dr. Dean Pruitt, who served
jointly a s the mentors of that undertaking.
T h e author is currently a Postdoctoral Resident Research Associate, Naval Medical
Research Institute, Bethesda, Maryland, under the auspices of the National Academy of
Sciences, National Reeeatch Council.

79

Downloaded by [University of Arizona] at 03:43 04 January 2013

80

JOURNAL OF SOCIAL PSYCHOLOGY

on differences between Germans and Americans, indicating that the typical

American is quite willing to reveal a large portion of himself to others, including casual acquaintances, whereas the German is more reluctant about engaging in social intercourse with others. Using a German and English version
of a self-disclosure questionnaire, Plog ( 14) reported differences in self-disclosure patterns between German and American college students of comparable
backgrounds which empirically verified Lewins theorizing. Jourard ( 7 ) also
found similar differences in a comparison of British and American college
students.
Recently, self-disclosure among American Ss has shown relationships with
such biographical characteristics as age and sex (6, 10) ,cultural and subcultural
characteristics (7, 10, 14) , and personality properties measured by Rorschach
scores (8). Jourard (7) , Jourard and Lasakow ( 10) , and Rickers-Ovsiankina
( 15) have reported use and development of self-disclosure questionnaires which
have adequate reliability. All these studies, however, are correlational and are
not concerned with the development of interpersonal relationships.

2. Hypotheses
T h e following hypotheses are heuristic and directed toward an initial theory, rather than being derived from theory, and are addressed to changes over
time in the breadth of interpersonal exchange at different levels of intimacy.
A fairly obvious expectation is that interpersonal exchanges increase over
time. However, a more interesting question deals with the possibly differential
rates of increase between frequencies of intimate and nonintimate exchanges
over time. A main focus of this study was to determine whether or not there
are differential rates of increase over time in intimate and nonintimate levels of
mutual activities and information exchanges between roommates, It is predicted that dyads composed of two high-revealing individuals (to best friend
as target person) will exhibit a greater breadth of penetration (variety of activities engaged in and information revealed) than low-revealers. If self-revealing or self-disclosure is an individual characteristic generalizable across target
persons, this is a fairly obvious prediction, but of necessity must be demonstrated in order to examine a more important question. T h a t is, in what way
will high- and low-revealer dyads differ over time and across levels of intimacy
with regard to activities engaged in with a roommate and information disclosed
about the self? Finally, high-revealer dyads are expected to exhibit a more
rapid growth in breadth of exchange than low-revealer dyads, especially in intimate areas.

DALMAS A. TAYLOR

81

B. METHOD

Downloaded by [University of Arizona] at 03:43 04 January 2013

1. Questionnaire Measures
T w o questionnaires were used to measure the mutus. activities an information exchange involved in social penetration, A 30-item roommate questionnaire, developed by the author, was administered a t weeks 1, 3, 6, 9, and 13 to
each S with instructions to indicate whether or not they had engaged in the activity indicated in the item during the preceding three weeks. Typical items
from this questionnaire are : Have you borrowed or loaned money to your
roommate? Have you invited your roommate to your home? T h i s was a
measure of the behavioral activity aspect of social penetration. Ss also completed a self-disclosure questionnaire ( 9 ) which requested them to indicate
whether or not they had revealed information about themselves to their roommates in each of 40 content areas (e.g., religion, family, etc.).
Items on both questionnaires were previously scaled for intimacy by undergraduates from a general psychology class, using the Thurstone procedure of
equal appearing intervals ( 18). Both questionnaires were administered a t
weeks 1, 3, 6, 9, and 13 of the fall semester. T h e self-disclosure questionnaire
was also administered during Freshman Orientation Week.
I n an effort to assess biasing factors resulting from repeated administration
of the questionnaires to the same Ss, several control groups were used. T h e
control groups were randomly drawn from the total initial pool of high- and
low-revealer dyads and were tested a t designated control points. Six high-revealer and five low-revealer dyads were tested a t weeks 6, 9, and 13 only; and
six high-revealer and two low-revealer dyads were tested a t week 13 only.

2. Subjects

8s were male freshman students at the University of Delaware. T h e selfdisclosure questionnaire (9, 10) was administered to 695 Ss to identify roommates who were either both high or both low on self-disclosure (for best friend
as target person). Split-half reliability of this instrument to several targets is
.94 (10). Only pairs in which both members were strangers before coming to
the university, and in which both members were either high- or low-revealers,
were selected. Ss were told that they would be participating in research related
to Navy future weapons systems. Further, we were interested in how men go
about getting to know each other. Only two Ss refused to participate. All vol4 Data from related research indicate that disclosure to best friend an a target person m
i independent of disclosure to casual acquaintance as a target perion (r = .18).

82

J O U R N A L OF SOCIAL PSYCHOLOGY

unteers were informed that they would be expected to answer questionnaires

several times during the semester.
Sixty-one pairs (122 Ss) met these criteria. Thirty-two dyads were composed of high-revealers and 29 of low-revealers. Fifteen high- and 15 low-revealer pairs were randomly selected to fill the main experimental conditions.
T h e remainder of the high and low pairs were used as controls to check on the
effects of repeated measures.

Downloaded by [University of Arizona] at 03:43 04 January 2013

3. Design

A three-dimensional factorial design with repeated measures was used. T h e

major independent variables were time (weeks of interaction over the college
semester), intimacy level, and predisposition to reveal oneself to others, as
measured by a pre-experimental self-disclosure questionnaire in response to best
friend as target person. Dependent variables were the number of mutual roommate activities and amount of self-disclosure to roommate, measured by questionnaire responses. Each questionnaire was scored in two ways: ( a ) total
number of items and ( b ) number of items at each of four levels of topical intimacy. T h e amount of esteem each S had for his roommate was obtained at
each session. This was measured through the use of 24 bipolar adjectives as a
semantic differential scale (12).
C. RESULTS
Analyses of the data are summarized in Table 1 and are centered around
the three independent variables. T h e major dependent variables were mutual
activities reported and amount of information revealed.

1. Activities
Figure 1 compares the mean number of activities engaged in by high-revealer
dyads with the mean number of activities engaged in by low-revealer dyads
over the 13-week interval. From the figure it can be seen, as expected, that
both high- and low-revealer dyads increased in amount of mutual activities reported over time. Of greater importance is the finding that, regardless of number of weeks, more reported social activities occurred at nonintimate than at
intimate levels (see Table 2). While the rank ordering of intimacy levels was
not as consistent as expected, more mutual activities were reported at Intimacy
Level I than at more intimate levels (11, 111, and I V ) . T h e intimacy x time
interaction (Table 1 ) for activities failed to reach an acceptable level of significance ( F = 1.45, p < .15). Thus, while there was a gradual increase over
time in mutual activities reported at each level of intimacy, the amount of in-

83

DALMAS A. TAYLOR

TABLE 1
ANALYSES
OF VARIANCEOF NUMBEROF MUTUAL
ACTIVITIES
REPORTED
AND AMOUNT
OF SELF-DISCLOSURE
TO ROOMMATE
Source of
variation

df

Activities
Mean
square

Information revealed
Mean
dj
square
F

Downloaded by [University of Arizona] at 03:43 04 January 2013

Between dyads

A (personality)
Dyads w groups
Within dyads
B (time)
AB
B x dyads w groups
C (intimacy)
AC
C x dyads w groups
BC
ABC
BC x dyads w groups
Total

1
18

287.30
52.78

5.44.

18.00
1.12
2.60
293.68
21.80
8.64
1.74
1.16
1.20

6.90..

72
3
3
54
12
12
216
399

1
18

5
5
90
3
3
54
15
15
270
479

3.9 6***
2.52+
1.45

.
. .-

169.22
28.84
78.20
1.82
1.74
451.67
12.33
5.84
5.10
.74
.77

6.55'
44.86***
1.05
77.31***
2.11
6.62***

p < .05.
** p < .01.
*** p < .001.
f p <.lo.
crease was not significantly different for nonintimate and intimate areas. Inspection of Figure 1 and mean scores in Table 2 indicates that high-revealer
dyads engaged in significantly more activities with each other than low-revealer
dyads. I t has already been shown in Figure 1 that both high- and low-revealer
dyads reported more mutual activities over time and that high-revealer dyads
reported a greater number of activities than low-revealer dyads at all points in
time. However, of special interest is the fact that the discrepancy between
these two groups was constant over time, as reflected in a nonsignificant personality x time interaction (Table 1).
Even though the personality x intimacy interaction only reached a marginal
level of significance ( p < .lo), the means are in the predicted direction. T h a t
is, high-revealer dyads reported more activities than low-revealer dyads at all
levels of intimacy, but more so at the more intimate levels (Table 2).
~

2. Information Revealed
Amount of breadth of disclosure of personal information by high- and lowrevealer dyads is summarized in Table 1 and shown in Figure 2.
As hypothesized, greater amounts of disclosure by both high- and low-revealer dyads occurred in later weeks of their relationship than in earlier ones.
T h e prediction that high-revealer dyads would exhibit greater breadth of pen-

a4

JOURNAL OF SOCIAL PSYCHOLOGY

-----.--

High Revealers
Q
W

24.0

Downloaded by [University of Arizona] at 03:43 04 January 2013

I-

g 20.0
a
W

m 16.0

I3
-

./
.---.
0

LOW Revealers
,,c.-90

12.0

0 ,

-0
/-

IV

cl

8.0

a
2

4.0

5
I

13

Weeks
FIGURE 1
MEANNUMBER
OF ACTIVITIES
REPORTED
FOR HIGH- AND LOW-RBVEALER
DYADS

etration (more disclosure) than low-revealer dyads at all levels of intimacy

was only partially substantiated (Table 3). These data also support the expectation that a greater amount of information exchange would occur a t nonintimate than at intimate levels of disclosure. A more important hypothesis, however, was concerned with developmental changes in intimate and nonintimate
disclosures. Inspection of means associated with these conditions indicates that
there is little initial difference in intimacy of disclosure. Over time, however,
there is a rapid increase in nonintimate disclosure oersus a slow and gradual
increase in intimate disclosures (see Figure 3). T h e results of a Newman-

85

DALMAS A. TAYLOR

TABLE 2
ACTIVITYMEANSFOR HIGH- AND LOW-REVEALER
DYADS

AT

FOURLEVELSO F

INTIMACY

Levels of intimacy
Nonintimate
I

Catenorv
High-revealer dyads
Low-revealer dyads
Total

6.44b
5.74c

12.18

I1

I11

Intimate
IV

Total

2.68
1.40
4.08

6.38
3.50
9.88

5.68.
3.76
9.44

21.18
14.40
35.58

Significantly larger than mean for low-revealer dyads a t same level of intimacy

Downloaded by [University of Arizona] at 03:43 04 January 2013

( p < .OS) by Newman-Keuls test.

b Significantly larger than mean a t Level I1 ( p

Newman-Keuls test.
C Significantly larger than means at Level I1 ( p
in the same condition by Newman-Keuls test.

< .01)
< .01)

in the same condition by

and Level I11

(p

< .OS)

Keuls test (19, p. 309) indicate that there were more significant mean differences between nonintimate and intimate levels of disclosure at later weeks than
a t earlier ones. T h e family of curves in Figure 3 lends empirical validation to
Altman and Haythorns (2) conceptualization of a wedge-like development
of interaction in interpersonal relations.
As was the case with activities, the rate of development for high-revealer
dyads was not significantly different from that of the low-revealer dyads (see
Figures 1 and 2). Although high-revealer dyads reported greater amounts of
exchange than low-revealer dyads at all points in time, both groups showed
equal rates of development.
It was also predicted that high-revealer dyads would report a greater proportion of intimate relative to nonintimate exchanges than low-revealer dyads.
Since the personality X time interaction (Table 1) was not significant, this
hypothesis was not confirmed.
T h e second-order interaction (personality X time x intimacy) for activiTABLE 3
MEANINFORMATION REVEALED
FOR HIGH- AND LOW-REVEALER
AT FOURLEVELS
OF INTIMACY

DYADS

Levels of intimacy
Nonintimate
I

Category
High-revealer dyads
Low-revealer dyads

6.58alb
4.52C

11

111

Intimate
IV

3.66
2.53

2.10
1.56

1.65
.63

Total
13.99
9.24

Significantly larger than means for low-revealer dyads a t intimacy Levels 11,
111, and IV ( p
.01) by Newman-Keuls test.
b Significantly larger than mean8 at Levels 11, 111, and IV ( p
.01) in the name
condition by Newman-Keuls test.
c Significantly larger than means a t Levels I11 and IV ( p
.01) in the same condition by Newman-Keuls test.

<

<

<

86

J O U R N A L OF SOCTAL PSYCHOLOGY

40.0

High Revealers
32.0

28.0

n
Downloaded by [University of Arizona] at 03:43 04 January 2013

24.0
W

>
W
a 20.0
z

r
2
2
0:
8

16.0

12.0

z
-

0.0

/.---

Low Re vea le r s

--3

,a//

/
/

a /

I
a
4.0

0 1

13

Weeks
FIGURE 2
AMOUNTOF DISCLOSURE
MR HIGH-AND LOW-REVEALER
DYAD8

OVER

TIME

ties and self-disclosure was not significant ; thus the consistently higher breadth
of penetration of high- versus low-revealer dyads a t the different levels of intimacy remained constant over time.

3. Control Analyses
Analysis of the esteem measure showed no differences between high- and
low-revealer dyads. Esteem for roommate changed over time (F = 6.77, p <
.001), however, and in the direction of less esteem over time. Since interpersonal exchange increased over time, it seems that esteem decreased with in-

8i

DALMAS A. TAYLOR

10.0
9

Level I

Downloaded by [University of Arizona] at 03:43 04 January 2013

7 .o

6 .O

5 .O
4 .O

.-*.

3 .O

,*/

2 .o

*
Level I l l
Level IV

A*\.

I .o

13

Weeks
FIGURE 3
AMOUNT
OF DISCLOSURE
FOR DIFFERENT
LEVELSOF TOPICAL
INTIMACY

creasing interaction. Further support for this position is evidenced in a trend

toward greater decrement of esteem for high-revealers than for low-revealers.
This trend accompanied the more rapid increase in interpersonal exchange for
high-revealers. Perhaps the old maxim familiarity breeds contempt is applicable here.
I n an effort to account for or rule out other variables which might influence
the development of interpersonal relations] information on each 8s emotional
and social adjustment as measured by the Opinion, Attitude, and Interest Sur-

Downloaded by [University of Arizona] at 03:43 04 January 2013

88

JOURNAL OF SOCIAL PSYCHOLOGY

vey ( 5 ) 18 (measured by The School and College Ability Tests) and age
was obtained from the university counseling office. Mean scores on these factors were computed for high- and low-revealer dyads and evaluated by t tests.
No differences were significant,
I n 13 out of 16 times, the control groups had somewhat higher means (more
activities and information revealed) than the experimental groups. This appears to rule out any biases resulting from repeated testing that may have operated to elevate disclosure and activity scores artificially. If anything, the effect was in the direction contrary to hypotheses, making the results a more
conservative test of hypotheses. Thus, it seems that means of the experimental
groups, if anything, were suppressed by constant re-exposure to the same
questionnaire.

D. DISCUSSION
This study was concerned with some broad aspects of a theory of social penetration. As expected, reported mutual activities and self-disclosure between
roommates increased significantly over time, and a greater number of these exchanges were classified as nonintimate as opposed to intimate. High-revealer
dyads reported significantly more exchanges (both activities and disclosures)
than low-revealer dyads. For reported activities, the difference was more pronounced in intimate areas.
These data are in agreement with those from other studies designed to evaluate aspects of the social penetration theoretical framework. Altman and Haythorn ( 2 ) found that socially isolated pairs revealed more to each other than
did nonisolated pairs and especially in nonintimate areas. Frankfurt (4) and
Colson (3) obtained similar findings in different situations.
T h e conceptual framework of social penetration recognizes the importance
of reward-cost characteristics in the development of interpersonal relationships.
T h i s feature is implied in the present study and explicit in other studies developed in the context of social penetration theory. T h a t is, if the level of exchange in a relationship is sustained or increases over time, it can be assumed
that both persons are profiting from the interaction. Jourard (9) has suggested
that openness, within limits, can be mutually rewarding in interpersonal relations. I n regard to roommates, Newcomb theorizes that
person-to-person
interaction tends, actuarially speaking, to be rewarding-especially when one
has recently been introduced into a community of strangers (4, p. 210). T h e
decrement of mutual esteem scores by both high- and low-revealer dyads over
time, with a concomitant increase in exchange, is an apparent contradiction
to the reward-cost formulation. Newcomb (13) explains a similar decrement

. . .

DALMAS A. TAYLOR

89

in roommate attractions by suggesting that initial estimates of attraction are

autistic. T h a t is, because of a need or desire to be compatible in a roommate
situation, individuals tend initially to overestimate the favorableness of the relationship. These overestimations are later modified by reality, causing a decrease in estimates of attraction. While we have no data on this, it appears to
be a reasonable and testable assumption. Laboratory experiments are needed to
answer this and other questions derived from social penetration.

Downloaded by [University of Arizona] at 03:43 04 January 2013

E. SUMMARY
T h i s study was concerned with the development of interpersonal relations
and with differences in the dynamics of growth as a function of personality.
Thirty college roommate pairs were studied over the course of a semester regarding various aspects of their interpersonal relations. Fifteen dyads were
composed of members who were high-revealers of personal information to their
best friend and 15 were composed of low-revealers. Additional pairs were
used as controls. T h e major dependent variables were reports of mutual activities engaged in between roommates and amount of information revealed to
roommate. Ss responded to questionnaires covering these areas several times
over a 13-week interval. Mutual activities and self-disclosure both increased
over time, and nonintimate or superficial exchanges of activities and information about the self occurred to a greater extent than intimate ones. Furthermore, dyads composed of high-revealers engaged in a significantly greater
amount of exchange than dyads composed of low-revealers. T h i s latter difference was greater in intimate areas of exchange than in nonintimate ones. T h e
results are interpreted as offering support for a general theoretical framework
of social penetration processes.

REFERENCES
1. ALTMAN,
I. Social interaction: A theoretical framework and behavior classification
system. Mimeo, Bethesda, Maryland, 1963.
2. ALTMAN,
I., & HAYTHORN,
W . W . Interpersonal exchange in isolation. Sociometry,
1965, a8, 411-426.
3. COLSON,
W . C. Self-disclosure a8 a function of social approval. Unpublished paper, Howard University, Washington, D. C., 1965.
4. FRANKFURT,
L. P. T h e role of some individual and interpersonal factors on the
acquaintance process. Unpublished Doctoral dissertation, The American University, Washington, D. C., 1965.
5. FRICKE,
B. G . Opinion, Attitude and Interest Survey. Ann Arbor, Mich.: Oaie
Testing Program, 1955-1963.
6. JOURARD,
S. M. Age trends in self-disclosure. Merrill-Palmer Quart. Bebao. ed
Deoel.. 1961.. 7.. 191-197.
7.
Self-disclosure patterns in British and American college females. J . SOC.
64, 315-320.
P ~ y c h o l . 1961,
,

-.

90
8.

JOURNAL OF SOCIAL PSYCHOLOGY

,

Self-disclosure and Rorschach productivity. Percept. W Motor Skills,

1961, 13, 232.

9.
10.

11.
12.

Downloaded by [University of Arizona] at 03:43 04 January 2013

13.
14.
15.
16.

17.

18.
19.

T h e Transparent Self-Openness, Effectiveness, and Health. Princeton,

N. J.: Van Nostrand, 1964.
JOURARD, S. M., & LASAKOW,
P. Some factors in self-disclosure. J. Abn. W SOC.
PJychol., 1958, 66, 91-98.
LEWIN,K. Some social-psychological differences between the United States and
Germany. Charac. W Personal., 1936, 4, 265-293.
MYERS,A. T e a m competition, success and the adjustment of group members. J .
Abn. W SOC.Psychol., 1962, 68, 325-332.
NEWCOMB,
T. M. T h e Acquaintance Process, New York: Holt, Rinehart & Winston, 1961.
P a , S. C. T h e disclosure of self in the United States and Germany. J . SOC.Plychol., 1965, 66, 193-205.
RICKERS-OVSIANKINA,
M. A. Social accessibility in three age groups. PJychol. Rep.,
1956, 2, 283-284.
TAYLOR,
D. A. T h e effects of social reinforcement and self-disclosure patterns on
interpersonal behavior. Unpublished paper, University of Delaware, Newark,
Delaware, 1964.
Some aspects of the development of interpersonal relationships: Social
penetration processes. Office of Naval Research Technical Report Number 1,
Contr. Number Nonr-2285( 0 4 ) , Bethesda, Maryland, May, 1965.
THURSTONE,
L. L., & CHAVE,E. J. T h e Measurement of Attitude. Chicago, Ill.:
Univ. Chicago Press, 1929.
WINER,B. J. Statistical Principles in Experimental Design. New York: McGrawHill, 1962.

Naval Medical Research Institute

National Naval Medical Center
Bethesda, Maryland 20014