You are on page 1of 8

INTERPERSONAL RELATIONS AND GROUP

PROCESSES

More "Touching" Observations:


New Insights on Men, Women, and Interpersonal Touch
Judith A. Hall and Ellen M. Veccia
Northeastern University

To describe sex differences in interpersonal touch, observation was made of 4,500 dyads in their
teens and older in public places. The principal analyses addressed the issue of asymmetry in the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

initiation of touch in mixed-sex dyads. Over all ages and initiating body parts, males touched
This document is copyrighted by the American Psychological Association or one of its allied publishers.

females (MF touch) with the same frequency that females touched males (FM touch). However,
among younger dyads, MF touch prevailed, whereas the reverse was true for older dyads. Though
there was a tendency for MF touch with the hand to exceed FM touch with the hand, significant sex
differences appeared only for "arm around," in which MF touch prevailed, and "arms linked," in
which FM touch was more common. Analyses also addressed main effects of toucher sex and
recipient sex, male-male versus female-female touch, and same- versus opposite-sex touch.

One of the cornerstones of the study of nonverbal sex differ- Two reviews of sex differences in touch have appeared (Ma-
ences is Henley's (1973) article reporting an asymmetry in the jor, 1981; Stier & Hall, 1984). The literature is not very consis-
touches exchanged between the sexes. Henley found that men tent, and these reviewers did not always concur on their conclu-
touched women more than women touched men in public sions. Also, the literature contains little objective data on the
places. The significance of this study lay not in its size or com- kinds of touches used by the sexes; exceptions are two studies of
prehensiveness, but in the provision of a coherent theoretical airport greetings and departures (Greenbaum & Rosenfeld,
account for this asymmetry as well as for many other differ- 1980; Heslin & Boss, 1980). Most investigators in observational
ences in the nonverbal behavior of men and women. This ac- studies have limited their reporting to gross frequencies; Stier
count gave central importance to male-female differences in and Hall concentrated on these studies in their meta-analysis.
status, dominance, and power and how these differences shape As far as asymmetry in mixed-sex touching is concerned,
nonverbal behavior (Henley, 1973,1977). Stier and Hall (1984) found the literature to be inconclusive
Attention to nonverbal sex differences has grown steadily because in 14 studies of adults there was, on average, little dif-
since then; reviews have been conducted involving hundreds of ference between the sexes. \et it is this question that has
descriptive studies (eg., Hall, 1979, 1984; Vrugt & Kerkstra, sparked the greatest interest. At the time of the Stier and Hall
1984). The theoretical perspective provided by Henley has been review, the literature yielded three statistically significant re-
a unifying framework for many writers. Ironically, however, in- sults, two showing more touching of females by males than of
terpersonal touch—the behavior studied in Henley's seminal males by females (Henley, 1973; Major, Schmidlin, & Williams,
article—has been relatively neglected. This neglect is likely due 1990 [first described in Major, 1981]), and one showing the
in part to the difficulties inherent in studying a behavior that is reverse (Willis, Rinck, & Dean, 1978). Since that review, Jones
both infrequent and ambiguous in meaning. An additional fac- (1986) in a participant-observation study also found more op-
tor may be premature confidence in our understanding of these posite-sex touching by women than by men for a category of
sex differences; Henley's study is often cited in such a way that touches called controlling.
one might easily assume that the asymmetrical touching she Other comparisons by sex are also of interest and have been
found between the sexes is both confirmed and unqualified
commented on in both earlier reviews (Major, 1981; Stier &
(e.g., Baron & Byrne, 1987; Schlenker, 1980; Tedeschi, Linds- Hall, 1984): main effects of initiator and recipient sex, male-
kold, & Rosenfeld, 1985; Worchel, Cooper, & Goethals, 1988).
male versus female-female touch, and same- versus mixed-sex
touching. Stier and Hall observed that a firm conclusion for
these comparisons was problematic because some investigators
We are grateful to Norman Belanger, Monica Scully, and Cindy
Morin for their help as observers, to Paul Andreasen for helpful sugges-
(e.g., Henley, 1973; Major et al., 1990) compared raw numbers of
tions regarding analysis, and to Stephen G. Harkins and Marylee C. touches rather than touches as a proportion of the number of
Taylor for their critical reading of the article. people observed. Applying a chi-square test to raw frequencies
Correspondence concerning this article should be addressed to Ju- without taking into account non-occurrences is inappropriate
dith A. Hall, Department of Psychology, 125 NI, Northeastern Univer- (Delucchi, 1983; Lewis & Burke, 1949). Such a chi-square test
sity, 360 Huntington Avenue, Boston, Massachusetts 02115. (comparing overall male versus female touch, for example)
Journal of Personality and Social Psychology 1990, m 59, ho. 6,1155-1162
Copyright 1990 by the American Psychological Association, Inc. 0022-35 H/90/S00.75

1155
1156 JUDITH A. HALL AND ELLEN M. VECCIA

must use arbitrarily chosen expected values; one might assume observers, four women and one man. Observers recorded the data in
that the same number of males and females were observed, and an on-line fashion using a hand-held compact tape recorder. The tapes
therefore one could expect equal frequencies of touch by males were subsequently transcribed onto codesheets.
and females under the null hypothesis. We consider it inadvis- To ensure that observers used the same sampling and recording
able to make such assumptions. The sampling method used in methods, each observer completed at least two training sessions as well
as a session for determining the reliability of age judgments. Reliability
the present study was chosen to avoid this problem.
checks are described later.
In our study, observers sampled 4,500 dyads in public places Once training was completed, observers conducted their sessions
and made on-line dictation of any touches they engaged in. independently. At the beginning of each observation session, the loca-
Because our chief interest was in the asymmetry question, we tion, date, and time of day were noted. During each observation ses-
included three variables—age, intention, and the body parts sion, male-male, female-female, and mixed-sex dyads were observed
involved—that had been implicated by previous studies as rele- for a 10-s period. If more than one touch occurred during the 10-s
vant to the asymmetry question. Regarding age, the strongest interval, only the first touch was included in the analysis. People who
asymmetry favoring males in the literature was found in adult were already engaged in physical contact at the start of the observation
dyads under age 30 (Henley, 1973), which raised the possibility interval were also observed for 10 s, in case they engaged in a new
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

touch. In-progress touches without a new touch during the observation


This document is copyrighted by the American Psychological Association or one of its allied publishers.

that older age groups might show a different pattern. No other


period were later analyzed separately. If a new touch occurred in a dyad
study examining asymmetry has looked at age within the adult
that was already touching, only the new touch was analyzed (and was
range. In our study, we classified subjects' ages by decades. Re- designated as a new touch). These procedures ensured that only one
garding intention, the only two significant findings of asym- touch was analyzed per dyad.
metry favoring males (Henley, 1973; Major et al., 1990) were Sampling. When observations were conducted in locations where
based exclusively on intentional touches; other studies did not people were in motion (e.g., walking through shopping areas), selection
take account of this variable. We recorded both intentional and of dyads for observation was accomplished by a time-sampling tech-
unintentional touches but limited our analyses to the inten- nique so that biases in the sample would be minimized. At the onset of
tional ones. Finally, regarding body parts, both the Henley and each observation session, the observer positioned herself or himself so
Major et al. studies recorded only touches with the hand; this that a dyad could be in view for the full 10-s observation period. Once
leaves open the question of whether touch is asymmetrical in position, the observer picked a landmark corresponding to the loca-
tion where dyads should have been at the start of the observation inter-
when other body parts are involved. We consequently took close
val. The timing device produced a beep every 10 s, which indicated
note of which body parts were involved in a touch. when the observation period would begin and end and also served as a
sampling cue. That is, once the observation of a dyad was completed,
Method the next beep signaled the observer to begin a new observation by
selecting the dyad that was now nearest to the landmark. If the dyad
Subjects being observed was blocked from view for any length of time, that dyad
was discarded from the sample. If a group of more than 2 individuals
The subjects in this study were 4,500 pairs of individuals judged to was nearest to the landmark when a beep occurred, the observer se-
be together and to be at least 14 years of age. Although subjects' race lected 2 adjacent individuals from the group and observed those indi-
was not recorded, the majority of subjects were White. People carrying viduals for the 10-s period. If no dyads were present at the landmark
children were excluded from the study. Subjects were observed in pub- when the beep occurred, the observer waited for the next beep.
lic places in the greater Boston area. Observers took special care to When observations were conducted where people were more or less
conduct their observations in an unobtrusive manner. stationary (e.g., movie lines), the observer selected persons who ap-
peared to be together, observed the dyad for 10 s, and then selected the
Apparatus next dyad that appeared to be together. In this manner, all dyads pres-
ent in the observer's field of vision were observed. As was the case with
A small timing device that produced a beep every 10 s was used to dyads in motion, if for any reason a dyad was obscured from view
time the observation interval and to select dyads for the sample. The during the 10-s interval, that dyad was eliminated from the sample.
beep was audible only through an accompanying earplug. Because the
timing device was small, it was easily hidden from view with only the
earplug showing. The data were recorded on a microcassette tape re- Variables
corder (Olympus Pearlcorder L200).
Dyad descriptor variables. The first variable recorded for each dyad
Procedure was sex composition (male-male, female-female, or mixed), followed
by age estimated by decade (teens, 20s, 30s, etc.). The activity that the
General. Observations were conducted in a total of 20 different dyad was currently engaged in was recorded as walking, standing, or
locations (both indoors and outdoors); multiple observation sessions sitting.
were conducted at some of the locations. Overall, there were 38 obser- Touch in progress. Pilot observations indicated that an appreciable
vation sessions distributed as follows. Indoor observation sites were number of sampled dyads were already touching. Observers noted this
shopping malls (4 locations on 6 occasions), movie lines (2 locations on fact so that it could be accounted for in later analyses. Even if the dyad
3 occasions), subway station (1 occasion), hotel lobby (1 occasion), and was already touching, it was observed for the full 10-s interval. If a new
the airport (1 occasion). Outdoor sites were shopping areas (e.g., down- touch occurred during this interval, it became the touch of record.
town Boston, 5 locations on 17 occasions), movie line (1 location on 2 Other variables recorded for touches in progress were toucher's sex
occasions), park (1 location on 2 occasions), parade line (1 occasion), (male, female, or unknown), whether the touch appeared to be inten-
outside of subway stations (2 locations on 2 occasions), and on the tional or unintentional, whether the touch was less or more than 3 s in
campus of Northeastern University (1 location on 2 occasions). duration, the description of the touch (e.g., female's arm around male's
The data reported here were collected over a 16-month period by five shoulders), and the apparent function of the touch (affectionate, con-
INTERPERSONAL TOUCH 1157

Table 1 ing field observations. In addition to these touches, each dyad was
•Sex and Age Composition of Dyads Observed instructed to perform an unspecified number of near misses inter-
spersed with the touches. Each dyad performed the touches and near
Sex Total misses during a single taping session during which the persons con-
versed, stood, sat, and walked around in a casual manner in a large
Age Male-male Female-female Mixed sex No. % laboratory room.
Each of the five observers was required to score the videotape by
Teens 229 235 149 613 14
376 559 873 1,808 40 providing all the information that was normally recorded during field
20s
30s 168 213 476 857 19 observations, using the on-line dictation procedure described earlier.
40s 61 77 167 305 7 Because only three different dyads were taped, the dyads' gender and
50s 27 66 98 191 4 age were eliminated from reliability analysis. Percentage agreement
60+ 9 71 76 156 3 between each of the 10 possible observer pairs was determined for the
Mixedage* 134 194 233 561 12 remaining variables as follows. For activity and touch (i.e., touch versus
Total no touch), percentage agreement was calculated across all possible
No. 1,004 1,415 2,072 4,491" observations. For toucher, duration, intention, and function, percent-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

% 22 32 46 age agreement was calculated only for those observations on which


This document is copyrighted by the American Psychological Association or one of its allied publishers.

both observers in the pair agreed that a touch had occurred.


* Mixed age refers to dyads whose ages differed by decade (e.g., 20s and The average percentage agreement was above 80% for all variables
30s). b Total does not equal 4,500 because of instances in which age except function. When function was eliminated, the average percent-
was unrecorded. The remaining dyads were 2 female-female and 7 age agreement across variables was 91%. Because reliability for func-
mixed sex. tion was lower than that of other variables (probably because it re-
quired more speculation than the other variables), it was eliminated
from further analysis.
trolling, attention getting, other). No special instructions were given to
observers regarding intention because it was generally very evident
whether a touch was intentional or not (as also reported by Major et al., Results
1990). The duration variable was not informative in later analyses and
is not discussed further. Sample Characteristics and Preliminary Analyses
For touches in progress at the onset of observation, it was not always
possible to determine which person was the toucher. For example, if a
mixed-sex dyad was observed holding hands, either member could
Table 1 contains a breakdown of the sample by age and sex
have performed the touch or touching could have been mutual (i.e., composition. The modal age was 20s; 40% of the dyads fell in
simultaneous). The toucher-sex variable was coded as unknown when this category. Twelve percent of the dyads had members whose
the toucher was unclear (this occurred very often when the dyad was ages spanned different decades. Forty-six percent of dyads were
holding hands). However, other forms of touch in progress allowed mixed sex; smaller numbers were female-female and male-
identification of the toucher. In some cases, the observer witnessed the male. The predominant activity was walking (79% of dyads);
touch while it was occurring but before the beginning of the 10-s obser- the other activities were standing (20%) or sitting (1%).
vation period (thus the touch would be called in progress). In other Eighty-five percent (3,818) of the 4,500 dyads never touched.
cases, the form of the touch itself was a guide. For example, male's arm Table 2 reveals that of the 681 dyads that did touch, 41% (278)
around female's shoulders suggests that the male had performed the did so during the 10-s observation period and the remaining
touch. Observers did not identify a toucher if there was uncertainty in
were already touching when observation began. Fourteen per-
their mind.
New touches. New touches were defined as touches that occurred
during the 10-s observation interval. For these touches, the same data
were dictated, with the addition of a mutual category for the toucher Table 2
variable. Touches Observed

Reliability Temporal
occurrence Intention Total
To determine reliability of judging subjects' age, one observer con- and
ducted an observation session in the field with each of the other four toucher's sex Intentional Unintentional No. %•
observers in turn. In this session, they judged the age, by decade, of 100
New
persons (divided into four sets of 25). The average Pearson product- Known 164 57 221 32
moment correlation between observers was .91. Unknown1" 16 41 57 8
The observed frequency of touching in the field was too low to allow In progress
efficient assessment of reliability for the other coded variables. There- Known 153 0 153 22
fore, reliability was assessed using a staged videotape (an approach to Unknown0 249 1 250 37
assessing reliability similar to that used by Jones and Yarbrough,
1985). The videotape showed two different mixed-sex dyads and one Total
No. 582 99 681
female-female dyad. The persons on the tape were psychology depart- % 86 14
ment faculty and graduate students.
The videotape was constructed by providing each dyad with a list of * Percentages do not add to 100 because of rounding. b Mutual or
10 touches that they were required to perform in an unspecified order. unrecorded. c Touch was of a form that precluded assignment of
The descriptions of these touches were taken from data obtained dur- toucher sex (e.g., holding hands).
1158 JUDITH A. HALL AND ELLEN M. VECCIA

Table3 nonhand touches, collapsing over new and in-progress touches.


Touch Asymmetry in Mixed- Sex Dyads MF touch was more common for touches with the hand (57%
MF to 43% FM); FM touch was more common for touches
MF FM
with other body parts (56% FM to 44% MF). As the table
Temporal occurrence
of touch No. % No. % x2 shows, neither of these differences achieved statistical signifi-
cance. However, an additional 2 X 2 x2 analysis asked whether
New 67 60 44 40 4.76* the MF-FM difference was significantly different for hand ver-
In progress 53 41 76 59 4.10* sus nonhand touches; data were the four frequencies listed as
Total 120 50 120 50 totals in Table 4. This test statistic was significant, x2^, N =
236) = 3.86, p < .05, indicating a significant sex difference for
Note. MF = male-to-female touch; FM = female-to-male touch. For x2 the body part that performed the touch.
tests, df= 1. All tests were two-tailed.
* p < .05. Table 5, last four columns, provides a more detailed descrip-
tion of the kinds of touches used in mixed-sex dyads. This table
collapses over new and in-progress touches. Of all the touches
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

described in Table 5, only two showed striking and significant


This document is copyrighted by the American Psychological Association or one of its allied publishers.

cent (99) of the touches were coded as unintentional. For 55% sex differences. Males preferred the "arm-around" gesture; the
(374) of the touches, it was impossible to ascertain which per- comparison between MF and FM frequencies is highly signifi-
son was the toucher. Touching showed no linear (or other evi- cant, x2(l, N= 51) = 21.35, p < .001. Females engaged much
dent) trend with respect to age; the percentages of dyads touch- more often in the "arms-linked" gesture than males did, x20,
ing in each decade varied from 11 % in the 40s to 18% in the 30s. N= 62)= 37.16, /x.OOl.1
All subsequent analyses are limited to intentional touches. To look at age as a possible moderator of these sex differ-
Furthermore, because the issue of interest is sex differences, ences, we pooled new and in-progress touches and stratified
only those touches for which the toucher could be identified are them by age of dyad. The first analysis involving age paralleled
included. As shown in Table 2, this represents 164 new and 153 Henley's (1973) cutoffs of "under 30" and "30 and over." For
in-progress touches, for a total of 317 touches. dyads under 30, MF touch prevailed (74 out of 126 touches or
59%), x20, N = 126) = 3.84, p < .05. For dyads of age 30 and over,
FM touch was more common (55 out of 90 touches or 61%),
Asymmetry in Mixed-Sex Touch
X 2 (l,A r =90)=4.44,/?<.05.
Although at the time of data collection, there seemed no The second analysis involving age was based on a finer age
theoretical significance to the distinction between in-progress breakdown, as shown in Table 6. The middle columns reveal
and new touches, data analysis revealed that opposite touch that the percentage of mixed-sex touches performed by males
asymmetries occurred for these two kinds of touch. Table 3 decreased dramatically with age, mirrored (necessarily) by a
presents these results. Differences in this table, and other analo- symmetrical increase in female touching of males. The MF
gous differences throughout this article, were tested for statisti-
cal significance using x2 with expected values set at 50% of the
total number of mixed-sex touches involved. The figure of 50% 1
Observers were not supplied with any hypotheses concerning sex
was chosen because under the null hypothesis of no asymmetry, differences. However, they might have held self-induced expectancies
one would expect to see equal frequencies of male-to-female that biased their judgments. Because 89% (55 out of 62) of the arms-
(MF) and female-to-male (FM) touching. linked gestures were already in progress (meaning that observers had
Table 3 shows that for new touches, 60% of the mixed-sex to infer which person performed the touch), we performed an addi-
touches were initiated by males and 40% by females, a signifi- tional analysis to assess the likelihood of bias. Ten photographs were
cant difference. In contrast, when the touches were already in taken of same- and mixed-sex dyads walking or standing with their
progress at the time of observation, only 41% of the mixed-sex arms linked, per request of the investigator. One of the original ob-
servers plus a convenience sample of 12 naive persons guessed which
touches were initiated by males, and 59% were initiated by fe- person was the toucher for each photo. These subjects' accuracy signifi-
males, also a significant difference. Note that when new and cantly exceeded chance, t(12) = 4.17, p < .002. Moreover, for the 8
in-progress touches are combined, as in the bottom row of Ta- mixed-sex photos (half showing MF touch, half showing FM touch),
ble 3, there is no asymmetry between the sexes. there was no tendency whatsoever for subjects to attribute touch more
Table 4 presents results separately for touches with the hand to the female than to the male. Finally, the original observer had an
versus other body parts (which were, in fact, mainly the arm). accuracy score of 100%. Altogether the form of the arms-linked ges-
Though the samples are relatively small, the percentages reveal ture apparently is such that an observer can accurately judge which
that the asymmetry favoring males as touchers in Table 3 for person performed the touch, even if the moment of touch was not
new touches is apparent for both hand and nonhand touches; witnessed. Two other touches, hand to hand and holding hands, were
for both kinds of touch, males performed about 60% of the also of potential concern when the observers were in the position of
inferring who the toucher was for an in-progress touch. As mentioned
touches. On the other hand, the asymmetry favoring females earlier, most of these touches were excluded because observers could
for in-progress touches is apparent only for nonhand touches; not make such an inference with confidence. As a result, relatively few
females initiated 62% of these touches. There is no asymmetry such touches were included in the analyses; only 12% of the analyzed
evident for in-progress touches with the hand (MF = 52%; hand-to-hand and holding-hands touches were in-progress. Therefore,
FM = 48%). no additional analysis was undertaken with regard to these two
Table 4 also compares MF and FM touching for hand and touches.
INTERPERSONAL TOUCH 1159

Table 4 hand and nonhand touches. For both kinds of touch, the pro-
Touch Asymmetry in Mixed- Sex Dyads by Body Part Used portion of touches performed by males decreased as age in-
creased, but the trend was significant only for nonhand
MF FM touches. For hand touches, the MF percentages were 62,60,52,
Body part and temporal
occurrence of touch No. % No. % x2 P and 35 for teens, 20s, 30s, and 40+, respectively, Z (linear con-
trast) = 1.42, .10 < p < .16. For nonhand touches, the MF
Hand percentages were 79, 52,47, and 6 for the corresponding four
New 42 60 28 40 2.80 <.10 age groups, Z (linear contrast) = 5.73, p < .001.
In progress 17 52 16 48 0.03 >.50
Total 59 57 44 43 2.18 <.15 These analyses demonstrated that the sex more likely to
Nonhand touch the other changes dramatically with age. This shift could
New 23 59 16 41 1.26 <.30 come about in more than one way. The relative decrease in
In progress 36 38 58 62 5.15 <.05 males' touching of females over age could be due to the males'
Total 59 44 74 56 1.69 <.20
decreasing their absolute rates of touching females, or to males
touching females at a constant rate over age but females greatly
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. MF = male-to-female touch; FM = female-to-male touch. For x2


This document is copyrighted by the American Psychological Association or one of its allied publishers.

tests, df= 1. All tests were two-tailed. Excludes 4 touches for which increasing their likelihood of touching males. To examine this,
initiating body pan was unrecorded. we calculated the rates of male and female touch in mixed-sex
dyads by dividing the frequency of male and female touches in
each age group by the number of mixed-sex dyads observed in
percentages for teens, 20s, 30s, and 40+ form a highly signifi- each age group taken from Table 1 (see last two columns of
cant linear trend as a function of age, as determined by a linear Table 6). These percentages differ from those in the middle two
contrast in proportions, Z = 4.49, p < .001 (Rosenthal & Ros- columns of Table 6 in that the denominator is dyads observed,
now, 1984; performing this contrast for the FM percentages not touches observed; because of this, the new percentages for
would yield the identical result, because the FM percentage is MF and FM touching are free to vary instead of having to add
simply 100 minus the MF percentage). Inspection of the middle to 100. These results show that the males' tendency to touch
two columns of Table 6 shows very clearly that though males are females declined with age (linear contrast in proportions, p <
the main touchers among teens—accounting for 73% of all .001) and that the females' tendency to touch males increased
mixed-sex touches in this age group, x2(l, N=22)= 4.54, p < (linear contrast in proportions, p < .05). Thus, clearly, older
.05—females are the main touchers among people over 40—ac- women touch men more than younger women do, and older
counting for 79% of touches in this age group, \\l, N=33) = men touch women less than younger men do. Analyses compa-
10.94, p < .001. The two intermediate age groups showed essen- rable to these were performed for touch rates in male-male and
tially no asymmetry (both x2 < 1.38). female-female dyads. These showed no linear age trends—that
Analogous linear contrasts were performed separately for is, for neither males touching males nor for females touching

Table 5
frequencies and Percentages of Specific Kinds of Intentional Touch
Female- Male- Female-
Male-Male Female Female Male
Kind of
touch No. % No. % No. % No. %
Hand
Hand to back* 1 5 6 10 11 9 6 5
Hand to head/neck 0 0 5 4 0
Hand to shoulder 4 21 2 3 8 7 8 7
Hand to handb 2 10 2 3 10 8 3 2
Hand to arm 2 10 19 33 15 12 21 18
Hold hands 1 5 4 7 8 7 5 4
Miscellaneous hand 0 3 5 2 2 1 1
Nonhand
Arm around0 5 26 4 7 42 35 9 8
Arms linked 0 10 17 7 6 55 46
Bump/brush" 1 5 1 2 2 2 2 2
Hug 1 5 3 5 2 2 2 2
Kiss 0 1 2 4 3 2 2
Miscellaneous nonhand 1 5 2 3 2 2 4 3
Unclassified 1 5 1 2 2 2 2 2
Total 19 58 120 120
Note. Percentages are per column. Because of rounding, percentages do not add to 100.
* Includes hand to rear end, waist. b Excludes hand holding. c Includes arm on back, around waist,
around shoulder. d Includes upper arm to upper arm.
1160 JUDITH A. HALL AND ELLEN M. VECCIA

Table 6
Frequencies and Percentages of Intentional Touch in Mixed- Sex Dyads
% of mixed-sex % of mixed-sex
Frequency touches dyads

Age of dyad MF FM MF FM MF FM
Teens 16 6 73 27 11 4
20s 58 46 56 44 7 5
30s 28 29 49 51 6 6
40+ 7 26 21 79 2 8
Note. MF = male-to-female touch; FM = female-to-male touch. Excludes mixed-age touches and touches
for which age was unrecorded.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

females were there systematic changes across ages in the likeli- tested by chi-square and was not statistically significant, x2(l,
hood of touching. This means that the trends noted in the final Ar=6,579) = 1.21,p>.20.3
two columns of Table 6 are unique to the mixed-sex context and Note that if the raw frequencies of 139 male and 178 female
do not reflect changes in the sexes' general tendencies to touch touches had been compared directly, females would have ap-
others across ages. peared to touch more than males.
Because observation was done in a variety of different loca- Sex of recipient. In her review, Major (1981) concluded that
tions, we performed an additional set of analyses to test females received more touch overall. Stier and Hall (1984)
whether the distribution of dyads' ages was different across noted a trend toward more receipt of touch by females but were
various types of locations. If age and location are related, then concerned about the problem of comparing raw frequencies
an alternative interpretation of our age effects would be that the directly. In their review, when studies using this approach were
sex asymmetry varies with the location in which people are excluded, there was no evidence of a sex difference in receipt of
observed. This seemed especially interesting because Major et touch. Because in the present study the number of MF and FM
al. (1990) found an asymmetry favoring males in "public, non- touches happened to be equal (120 in each case), the results for
intimate" locations (e.g., shopping malls) and in "recreational" receipt of touch were identical to those reported for initiation of
locations (e.g., parks) but not in "greeting and leave-taking" set- touch. Therefore, females received more intentional touch over-
tings (e.g, airports).2 all than males, but not to a statistically significant degree. As in
We classified our 20 locations using the same criteria. Unfor- the last paragraph, if we had compared the raw frequencies
tunately, we had only one greeting and leave-taking setting—an directly, we would have concluded that females receive more
airport—and it produced only five mixed-sex touches, not touch than males.
enough for a meaningful analysis. The other two types of setting Male-male versus female-female touch. Both earlier re-
had age distributions (30 and under versus 40 and over) that views concluded that female-female touch exceeded male-
were significantly different (p< .001), but in fact the difference male touch. In the present study, 2% (19 out of 1,004) of male-
was extremely small, corresponding to a Pearson correlation of male dyads engaged in touch, compared with 4% (58 out of
.06 between age and setting. In the public, nonintimate settings, 1,417) of female-female dyads. This was a significant differ-
83% of the sample was in the younger group, compared with ence, x20, N= 2,421) = 9.24 p < .002.
87% in the recreational settings. Moreover, the pattern shown Same-sex versus mixed-sex touch. Major (1981) concluded
in Table 6, wherein the proportion of touches performed by that among adults, mixed-sex touching predominated, but Stier
males decreased linearly with increasing age, was strongly evi- and Hall (1984) found the opposite. However, Stier and Hall
dent in both kinds of setting. Thus setting apparently was not a offered a number of qualifications to that conclusion. In the
confounding factor with regard to our age findings. present study, 3% (77 out of 2,421) of same-sex dyads touched,
compared with 12% (240 out of 2,079) of mixed-sex dyads. This
was a significant difference, x20, N = 4,500) = 119.49, p < .001.
Other Sex Effects
Qualitative differences. The particular kinds of touch seen in
Sex of initiator. In their review of the literature, Stier and dyads of different sex composition appear in Table 5. As noted
Hall (1984) concluded that females touch others more than
males do, disregarding recipient sex, but Major (1981) con- 2
cluded that this was true only when the recipient was female. In This analysis was suggested by an anonymous reviewer.
3
the present study, males initiated 139 intentional touches; fe- As an alternative procedure, we also denned the total number of
males observed as the number of males in mixed-sex dyads plus twice
males initiated 178. To translate these to percentages, we de- the number of male-male dyads (2,079 +1,004 +1,004 = 4,087); dou-
fined the total number of males observed as the number of bling the number of male-male dyads allows that each member had a
males observed in mixed-sex dyads plus the total number of chance to touch (though, as explained in the Method section, only the
male-male dyads (2,079 +1,004 = 3,083); an analogous calcu- first touch was recorded for a given dyad). Again, an analogous calcu-
lation was made for females. By this method, 4% of the males lation was made for females. By this method, 3% of the males touched,
touched, and 5% of the females touched. This difference was and 4% of the females touched.
INTERPERSONAL TOUCH 1161

earlier, in mixed-sex dyads, arm around was predominantly cantly more than vice versa in the under-30 group, paralleling
initiated by males and arms linked was predominantly initiated Henley (1973). In dyads over 30, the reverse was true. Linear
by females. Table 5 reveals that males seemed to prefer arm contrasts revealed that the MF asymmetry decreased steadily
around even with each other: This gesture was seen 26% of the over decades. This trend was highly significant for all mixed-sex
time in male-male dyads. In male-male dyads, hand to touches and for nonhand touches and was marginally signifi-
shoulder was of almost equal frequency (21% of touches). Simi- cant for touches with the hand.
larly, females engagedrelativelyoften in the arms linked gesture As is nearly always the case where age differences are con-
even with other females (17% of female-female touches). cerned, alternative interpretations of the age variable are possi-
Among female-female dyads, only hand to arm was more popu- ble. Because ours was a cross-sectional study we cannot know
lar (33%). whether the age differences were maturational or cohort based.
Moreover, there is undoubtedly a confounding of age with
various life-stage phenomena. Because younger people (espe-
Discussion
cially teenagers) are far less likely to be married or in enduring
The purpose of this study was to examine the generality of relationships than older people, our age effects may actually
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

asymmetry between the sexes in the use of intentional touch. reflect sex differences that vary with the nature of the relation-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Different ages of dyads (teens through 60s) and different initiat- ship between the people observed, not their age per se. Sex roles
ing body parts received special attention. Though highly mixed, may permit (even require) visible gestures of possession or be-
previous research had indicated that males may touch females ing in charge by males in less developed relationships; or per-
more than vice versa when the dyad is composed of younger haps females in such relationships touch less in an effort to
adults and when the touch is intentional and with the hand. appear noncommittal or not too forward. On the other hand,
We found that overall, each sex touched the other intention- sex roles may permit (or require) more assertive gestures, or
ally with equal frequency. This result is consistent with Stier and gestures indicating possession, by females in more developed
Hall's (1984) overall conclusion regarding touch asymmetry in relationships.
their review of the literature. However, there were dramatic In summary, our analysis of touch asymmetry revealed com-
differences when age and initiating body part were considered. plexities that preclude a simplistic understanding of this issue.
For touches with the hand, males touched females more than Contrary to the message often taken away from Henley's (1973)
vice versa. Though not statistically significant, the MF-touch seminal study, males' tendency to touch females more than vice
figure of 57% is similar to the significant figure of 60% found by versa is not a general phenomenon, at least not in public set-
Major et al. (1990) for touches among adult mixed-sex dyads. tings of the sort we studied and among pairs of people who have
For touches not initiated by the hand, females in our study selected to be together in those settings. We cannot comment,
tended to touch more (56%). Analyzing these two sets of fre- of course, on sex asymmetries in touch in other settings and
quencies together yielded the conclusion that the asymmetry among people with different kinds of relationship (for example,
differs significantly depending on whether hand or nonhand work settings, or superior-subordinate dyads). As far as we
touches are analyzed. These data therefore lend support to know, there are no observational studies to address this issue.
Henley's (1973) and Major et al.'s (1990) studies, which found Thus, interpretational issues remain complex. An analysis of
MF touches to exceed FM touches for intentional touches with the asymmetry in terms of differential dominance or status
the hand. between the sexes (Henley 1973,1977) cannot be straightfor-
Analysis of particular kinds of touch revealed that two ges- wardly applied, considering that the asymmetry progressively
tures showed significant sex differences in mixed-sex dyads, favored women as the dyads' ages increased. One could argue
but neither involved the hand. Males were more likely to put that only touches with the hand are likely to reflect dominance
their arm around a female companion, whereas females were or status (indeed, this is most likely why Henley's study concen-
more likely to put their arm through their male compan- trated on this kind of touch) and, given this, that the tendency
ion's arm. for males to touch more with the hand confirms the domi-
We found a sex difference between new and in-progress nance/status interpretation of the difference.
touches, so that MF touch predominated for new touches and We would differ with this logic, however. A touch with the
FM touch predominated for in-progress touches. Though un- hand can as easily be tentative or placating as dominating, but a
expected, this result may be due to the kinds of touches that touch with some other body part can be quite controlling or
occurred during the 10-s observation period versus those al- dominating, especially if it has a relatively long duration, as
ready in progress. In-progress touches were overwhelmingly nonhand touches tend to have. It is interesting that each sex has
long-lasting touches—specifically, arm around, arms linked, a distinctive way of holding onto the other: Males put their arm
and holding hands (few instances of holding hands were in- around the female; females put their arm through the male's
cluded in the sex difference analyses, however, because the sex arm. The difference in morphology may stem from average
of toucher for hand holding could usually not be determined for height differences or may reflect different connotations of the
in-progress touches). The arms-linked gesture, in which female touches in terms of affiliation or dominance. In any case, with-
touch was so pronounced, was particularly evident among out better data on the functions of specific touches, we believe
touches in progress. It may not be surprising, therefore, that one cannot argue that hand touches are deductively more domi-
in-progress and new touches showed different sex difference nating than other kinds.
patterns. Thus, it is not clear what different touches mean to the sexes
Age effects were notable. Males touched females signifi- and whether dominance/status can explain the sex effects.
1162 JUDITH A. HALL AND ELLEN M. VECCIA

Firm data on the relation of status to touch, irrespective of sex, Delucchi, K. L. (1983). The use and misuse of chi-square: Lewis and
would be helpful. Unfortunately, this is lacking. Hall (1984, Burke revisited. Psychological Bulletin, 94,166-176.
1987) noted that when invoking dominance/status as an expla- Goldstein, A. G, & Jeffords, J. (1981). Status and touching behavior.
nation for nonverbal sex differences (including touch), investi- Bulletin of the Psychonomic Society, 17, 79-81.
gators have often failed to produce unequivocal (or in some Greenbaum, P. E., & Rosenfeld, H. M. (1980). Varieties of touching in
cases, any) evidence of a relation of dominance/status to the greetings: Sequential structure and sex-related differences. Journal
of Nonverbal Behavior, 5,13-25,135.
nonverbal behavior in question, independent of sex. In the case
Hall, J. A. (1979). Gender, gender roles, and nonverbal communication
of touch, self-report studies indicate that people believe that skills. In R. Rosenthal (Ed.), Skill in nonverbal communication: Indi-
higher status people touch more than vice versa (see Stier and vidual differences (pp. 32-67). Cambridge, MA: Oelgeschlager,
Hall, 1984, for a review), but studies involving actual status and Gunn & Hain.
touch have not been highly supportive (Goldstein & Jeffords, Hall, J. A. (1984). Nonverbal sex differences: Communication accuracy
1981; Juni & Brannon, 1981). Evidence does indicate that the and expressive style. Baltimore: Johns Hopkins University Press.
initiation of touch increases one's relative status vis-a-vis the Hall, J. A. (1987). On explaining gender differences: The case of non-
person touched (Major & Heslin, 1982; Summerhayes & verbal communication. In P. Shaver and C. Hendrick (Eds.), Review
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Suchner, 1978). This finding, however, yields no clear predic- of personality and social psychology (Vol. 6, pp. 177-200). Newbury
tion about sex differences: One could predict either that men Park, CA: Sage.
(the higher status) would touch women more to assert their Henley, N. M. (1973). Status and sex: Some touching observations.
Bulletin of the Psychonomic Society, 2, 91-93.
position or that women (the lower status) would touch men
Henley, N. M. (1977). Body politics: Power, sex, and nonverbal commu-
more to enhance theirs. nication. Englewood Cliffs, NJ: Prentice-Hall.
Other sex differences were addressed besides the asymmetry Heslin, R, & Boss, D. (1980). Nonverbal intimacy in airport arrival and
question. Females were nonsignincantly more likely to touch departure. Personality and Social Psychology Bulletin, 6, 248-252.
others than were males. In its direction, this is consistent with Jones, S. E. (1986). Sex differences in touch communication. Western
the conclusion of Stier and Hall (1984). Jones's (1986) later Journal of Speech Communication, 50, 227-241.
study also found females to touch more overall. That and other Jones, S. E., & Yarbrough, E. (1985). A naturalistic study of the mean-
studies finding significant results for this question may have ings of touch. Communication Monographs, 52,19-56.
used more sensitive measures than we did. Observing for a Juni, S., & Brannon, R. (1981). Interpersonal touching as a function of
longer period of time would yield more touches and increase status and sex. Journal of Social Psychology, 114,135-136.
statistical power for testing sex differences. Lewis, D., & Burke, C. J. (1949). The use and misuse of the chi-square
Females were also nonsignincantly more likely to receive test. Psychological Bulletin, 46, 433-489.
touch from others than males were. Stier and Hall's (1984) re- Major, B. (1981). Gender patterns in touching behavior. In C. Mayo &
N. M. Henley (Eds.), Gender and nonverbal behavior. New York:
view found weaker evidence for this trend than was the case for Springer-Verlag.
touch initiation, especially when studies that compared raw Major, B., & Heslin, R. (1982). Perceptions of cross-sex and same-sex
frequencies were excluded. Jones (1986) also found only weak nonreciprocal touch: It is better to give than to receive. Journal of
evidence for such a trend. Nonverbal Behavior, 6,148-162.
In our study, male-male dyads touched at a significantly Major, B., Schmidlin, A. M., & Williams, L. (1990). Gender patterns in
lower rate than did female-female dyads, consistent with Stier social touch: The impact of setting and age. Journal of Personality
and Hall's (1984) review, though the difference was small in and Social Psychology, 58, 634-643.
absolute terms. In her review, Major (1981) concluded that such Rosenthal, R., & Rosnow, R. L. (1984). Essentials of behavioral re-
a difference was small, and Jones (1986) did not find any such search: Methods and data analysis. New \fork: McGraw-Hill.
difference. Schlenker, B. R. (1980). Impression management: The self-concept, so-
Finally, we found that mixed-sex dyads were much more cial identity, and interpersonal relations. Monterey, CA: Brooks/Cole.
likely to touch than same-sex dyads were, consistent with Ma- Stier, D. S., & Hall, J. A. (1984). Gender differences in touch: An empiri-
cal and theoretical review. Journal of Personality and Social Psychol-
jor's (1981) review and Jones's (1986) study. This result is sensi- ogy, 47, 440-459.
ble considering that dyads had self-selected each other, and Summerhayes, D. L, & Suchner, R. W (1978). Power implications of
many of the mixed-sex dyads were undoubtedly dating or touch in male-female relationships. Sex Roles, 4,103-110.
married. In less familiar dyads, one could expect to see less Tedeschi, J. T, Lindskold, S., & Rosenfeld, P. (1985). Introduction to
mixed-sex touch (Stier & Hall, 1984). social psychology. St. Paul: West.
In conclusion, explanations for sex differences in touch are Vrugt, A., & Kerkstra, A. (1984). Sex differences in nonverbal commu-
still obscure. Nevertheless, the present study has considerably nication. Semiotica, 50,1-41.
expanded our understanding of the phenomenon and has Willis, E N., Rinck, C. M., & Dean, L. M. (1978). Interpersonal touch
added new questions for study. We find Stier and Hall's closing among adults in cafeteria lines. Perceptual and Motor Skills, 47,
comment as appropriate now as it was in 1984: "Gender differ- 1147-1152.
ences in touch appear to be somewhat different and theoreti- Worchel, S., Cooper, J., & Goethals, G. R. (1988). Understanding social
cally more complex than has been previously stated" (p. 457). psychology (4th ed.). Chicago: Dorsey Press.

References Received August 11,1989


Baron, R. A., & Byrne, D. (1987). Social psychology: Understanding Revision received August 9,1990
human interaction (5th ed.). Needham Heights, MA: Allyn & Bacon. Accepted August 13,1990 •

You might also like