SOCIAL NETWORKS, SUPPORT

CLIQUES, AND KINSHIP

R. I. M. Dunbar
U n i v e r s i t y o f L i v e r p o o l
M. Spoors
U n i v e r s i t y C o l l e g e L o n d o n
Data on the number of adul t s t hat an i ndi vi dual contacts at least once
a mont h i n a set of British popul at i ons yi el d est i mat es of net wor k
sizes that correspond closely to t hose of the typical "sympat hy group"
size i n humans. Men and women do not differ i n t hei r total net wor k
size, but women have more femal es and more ki n i n t hei r net wor ks
t han men do. Ki n account for a si gni f i cant l y hi gher pr opor t i on of net -
work member s t han woul d be expected by chance. The numbe r of ki n
i n the net wor k increases i n pr opor t i on to the size of the fami l y; as a
result, peopl e from large fami l i es have proport i onat el y fewer non- ki n
i n t hei r net works, suggest i ng t hat there is ei t her a t i me const rai nt or a
cogni t i ve const rai nt on net wor k size. A smal l i nner cl i que of the net -
work funct i ons as a support group from whom an i ndi vi dual is par-
ticularly l i kel y to seek advice or assi st ance i n t i me of need. Ki n do not
account for a si gni fi cant l y hi gher pr opor t i on of the support cl i que t han
they do for the wi der net work of regul ar social contacts for ei t her men
or women, but each sex exhi bi t s a st rong preference for member s of
their own sex.
KEY WORDS: Net works; Ki nshi p; Sex di fferences; Fami l y size;
Support group.
Received August 22, 1994; accepted January 23, 1995.
Addr es s all correspondence to R. I. M . Dunbar , De p a r t me n t o f Psychology, U n i v e r s i t y o f Li v e r -
pool, P. O. Box 147, Liverpool L69 3 B X , England.
Copyright 9 1995 by Walter de Gruyter, Inc. New York
Human Nature, Vol. 6, No. 3, pp. 273-290. 1045-6767/95/$1.00 + .10
273
274 Human Nature, Vol. 6, No. 3, 1995
There has been considerable interest in both the sociological and the
anthropological literatures regarding the people with whom individuals
interact. The current view emphasizes the fact that an individual lies at
the focal point of a number of partially overlapping social networks,
each of which is oriented towards a different social context or purpose
(Milardo 1988). Thus, an individual may have a number of different sets
of friends based on work, leisure activities (such as an interest group or
sports club), church, the extended family, and so on. Moreover, the pat-
tern of relationships may change over an individual' s lifetime (Larson
and Bradney 1988).
It is clear, however, that there are limits to the number of contacts
that an individual can maintain over a given period of time. "Small
world" experiments in which individuals are asked to send messages
to distant parts of the world via a chain of contacts rooted in their cir-
cle of personal acquaintances suggest that the number of people on
whom any one individual can call for such favors is limited to
between 130 and 250 (Killworth et al. 1984). This estimate of the size
of an individual' s social world is on the same order as that estimated
from the size of the modern human neocortex based on a relationship
between neocortex size and group size derived from primates (Dun-
bar 1993).
Many of these individuals, however, are likely to be acquaintances
rather than intimate friends. It is widely recognized that this inner cir-
cle of more intense relationships plays a crucial role in mediating the
individual' s interactions with (and place within) the local communi t y
(Bott 1971; Milardo, ed. 1988; Mitchell 1969; Young and Willmott 1957).
Estimates of the size of this inner circle (the so-called sympat hy group)
have yielded values on the order of 10-12 individuals (Buys and
Larsen 1979). In this case, the estimates were obtained by asking indi-
viduals to list all the people whose death they woul d find personally
devastating, but they probably correspond to those people with whom
an individual keeps in regular contact. Net work sizes estimated from
frequency of contact have yielded values ranging from seven in U.S.
college freshmen (Hays and Oxley 1986) to 16.6 in young women
(McCannell 1988) and around 20 in married couples (Rands 1988). Dif-
ferences in both the criteria used to define network membership and
the stages of the life cycle at which networks are sampled are largely
responsible for the differences found between studies. Nonetheless, it is
clear that all these values tend to converge on the same group size
(10-15 individuals).
One reason for our interest in the size of these groupings derives from
the suggestion that language may have evolved to allow the exchange
of social information in order to facilitate the integration of relatively
Kinship and Network Size 275
large social groups (Dunbar 1993). There is some evidence to suggest
that the constraints in this respect derive not so much from the ability
to monitor all individuals in the communi t y but rather from the need to
monitor the activities and doings of one' s key social allies (principally,
presumably, one' s family and i mmedi at e friends) (Kudo et al. 1995). One
aim of this study, then, was to try to determine the size and composi-
tion of the inner circle (or network) that might constitute the focus of an
individual' s social interest.
Kinship is known to play an important role in both human social rela-
tionships and the structure of human groups in traditional as well as
modern postindustrial societies (see, for example, Firth 1956; Hughes
1988; Keesing 1975). Hames (1979) has shown that Ye' kwana villagers of
Venezuela interact more often with individuals who are closely related
to them, while Bert6 (1988) found that among the horticultural K'ekchi"
of Central America the availability of a network of kin is an i mport ant
determinant of the amount of land an individual can cultivate. The
extent to which kinship is a consideration in the creation of social net-
works in industrial societies remains unclear, however, even t hough
interview-based studies suggest that considerable weight is placed on
kinship in contemporary European societies (Bott 1971; Young and Will-
mott 1957).
We examine here the role that kinship plays in determining the com-
position of an individual' s social network in a modern European soci-
ety. Our main concern is with the circle of friends and relations wi t h
whom an individual maintains regular contact. Given that we can iden-
tify such a group, we can then ask what role kinship plays in determin-
ing its composition. In addition, we examine the size and composition
of the inner clique of intimates (the support clique) that individuals
would normally approach for advice or assistance when in difficulty. We
might expect kinship to be an important factor in the selection of sup-
port clique members because the opport uni t y for reciprocal altruism is
likely to be much less in situations of advice and/ or help than it is with
respect to social interaction.
Studies of social networks have t ended to focus either on quantitative
analyses of network size and structure (Burt 1982; Coleman 1964; Knoke
and Kuklinski 1982) or on more descriptive studies of individuals' net-
works and their role in facilitating social life (Bott 1971; Fischer 1982;
studies in Mitchell, ed. 1969 and Milardo, ed. 1988). In general, the more
quantitative studies have typically concerned themselves wi t h large-
scale structures at the societal level, often with a focus on organizations
rather than individuals (e.g., business and political networks) and the
functional roles that exist within organizations of this kind. In contrast,
studies of personal and support networks have t ended to be based on
276 Human Nature, Vol. 6, No. 3, 1995
in-depth interviews with a handful of individuals, with the focus on
how individuals relate to their immediate social circles. Our aim here is
to provide a preliminary assessment of the size of an individual' s net-
work of intimates and the extent to which kin contribute to it.
M E T H O D S
A questionnaire was designed which asked respondents to list the first
names of individuals whom they (a) lived with, ( b ) contacted wi t h vary-
ing degrees of frequency (termed their n e t w o r k ) , and (c) relied on for
advice and/ or help at the personal level (termed the s u p p o r t c l i q u e ) , as well
as ( d ) the size of their extended biological family (defined as all individu-
als related to the subject by r > 0.125, assuming full paternity certainty).
Contacts were defined as social exchanges involving face-to-face, letter, or
telephone interaction. Respondents were specifically asked not to include
business and professional contacts, unless the individuals concerned were
deemed to be personal friends. In responding to ( a ) - ( c ) , subjects were
required to distinguish between relatives and nonrelatives, with the crite-
rion for a relative being limited to full cousins or more closely related indi-
viduals. Subjects were asked to list the first names of individuals they
contacted daily, twice weekly, weekly, and at least once a month, as well
as those individuals they contacted regularly but less than once a month.
Considerable effort has been put into questionnaire design duri ng the
past decade (see Milardo 1988, and references therein). The burden of
this work has been to suggest that questionnaires that take more than a
few minutes to complete and have too many instructions and/ or more
than 10-12 name-eticiting questions tend to result in loss of concentra-
tion. The questionnaire was therefore designed to contain just four sets
of questions, each accompanied by a series of boxes to be filled in. The
questionnaire was tested on students and refined until it required no
more than 5 minutes to complete.
The questionnaire design used a recall procedure rather than asking
individuals to list all those whom they actually contacted during a set
period of time from receipt of the questionnaire. This approach was
largely chosen for speed and convenience. Although recall procedures
run a risk that respondents will overlook contacts they have made (see
Bernard et al. 1982, 1984), the relatively short time depth used in the pres-
ent case should tend to minimize this source of inaccuracy. In contrast,
prospective questionnaires greatly increase the load on the subject
because they require respondents to keep a daily tally of whom t hey
have contacted; as a result, under-reporting by those who lead especially
busy lives tends to increase. In addition, they t end to underestimate the
Kinship and Network Size 277
actual network size because individuals may not always be able to con-
tact all their normal interactees duri ng a particular sample period (e.g.,
because they are away or other unusual circumstances intervene). In
order to try to circumvent some of these problems, subjects were asked
to say whom they normally contacted duri ng a given time period. It was
felt that this would reduce both errors of omission and the number of
trivial contacts listed (i.e., those casual contacts who were not members
of the respondent' s circle of friends).
Questionnaires were distributed at five different locations within Eng-
land and Scotland. These respondents were not chosen to be a represen-
tative sample in any conventional sense, but simply to provide a broad
sample of subjects of different age, background, and geographical loca-
tion. In each case, a single assistant was responsible for handi ng out and
collecting completed questionnaires. Each assistant was instructed to tell
respondents only that the purpose of the questionnaire was to find out
about people' s social contacts. They were, however, allowed to help with
the filling in of questionnaires if requested to do so. Questionnaires were
distributed only to subjects between 18 and 65 years of age who were
members of a golf club or staff at a hospital in Lincolnshire, staff at an
empl oyment consultancy in Aberdeen, employees at a farm machinery
factory in Doncaster, and personal contacts within the London area. With
just a few exceptions, only one respondent was sampled per household.
Individuals over 65 years of age were excluded from the sample because
they are known to have reduced network sizes owing to greater vulnera-
bility to infirmity as well as deaths among lifelong friends (Bliezner 1988;
Brown 1981). Similarly, children and younger teenagers were excluded
because their networks are known to be atypical in both composition and
stability (Foot et al_ 1980; Levinger and Levinger 1986; Thorne 1986).
A total of 155 questionnaires were returned from the 170 given out.
The resulting response rate of 91% is high by normal questionnaire stan-
dards and can largely be attributed to the fact that in most cases the sub-
jects were themselves known to the assistants distributing the
questionnaires. Personal loyalty is thus likely to have been an important
factor influencing the completion of questionnaires. Of the returned
questionnaires, 54 were excluded from the analysis: the majority had
been incompletely (or in a few cases, incorrectly) filled in (mostly fail-
ure to identify contacts by sex), but 11 were spoiled in transit or other-
wise unreadable and 8 contained too many comments and queries to be
considered reliable. So far as we could tell, incomplete and spoiled ques-
tionnaires were not biased in favor of any particular category of subject
by sex, age, or domestic status. The remaining 101 subjects consisted of
34 men and 67 women. Of these, 24.8% (5 men and 20 women) were
single, 50.5% (18 men and 33 women) lived with a partner but did not
278 Human Nature, Vol. 6, No. 3, 1995
ha ve d e p e n d e n t of f s pr i ng l i vi ng wi t h t hem, 21.8% (9 me n a nd 13
wome n) l i ved wi t h a pa r t ne r a nd d e p e n d e n t chi l dr en, a nd t wo wo me n
l i ved al one wi t h d e p e n d e n t chi l dr en.
Ou r c onc e r n was t o i dent i f y t he set of p r i ma r y associ at es ( " f r i e nds "
in t he nor ma l me a ni ng of t he t er m) t hat an i ndi vi dua l has. Thi s is t he
set of pe opl e wh o s e act i vi t i es a nd l i fe hi s t or i es ar e of suf f i ci ent i nt er est
t o an i ndi vi dua l f or t hat pe r s on t o be wi l l i ng t o ma ke s ome ef f or t t o
keep up t o dat e. Rat her t ha n ask r e s p o n d e n t s t o s peci f y t hes e i ndi vi du-
als us i ng t hei r o wn cri t eri a, we pr e f e r r e d t o us e a mor e obj ect i ve cri t e-
r i on t hat coul d be appl i ed u n i f o r ml y acr oss t he ent i r e s ampl e. The
n u mb e r of cont act s was bi modal , wi t h pe a ks i n t he we e kl y a nd mo n t h l y
f r e que nc y cat egor i es. We t her ef or e us e d t he f r e que nc y wi t h whi c h i ndi -
vi dual s we r e cont act ed on at l east a mo n t h l y basi s as t he cr i t er i on f or
i ncl usi on i n a subj ect ' s net wor k. We t ook t he vi e w t hat i ndi vi dua l s wh o
cont act ed each ot her less of t en t ha n onc e a mo n t h we r e unl i kel y t o
r e ma i n up t o dat e wi t h each ot he r ' s mo r e i nt i mat e exper i ences , a nd t hus
fell out s i de t he s cope of t he pr es ent concer n.
Al l st at i st i cal t est s ar e t wo- t ai l ed. Dat a we r e l og- t r a ns f or me d i n o r d e r
t o nor ma l i z e val ues f or all pa r a me t r i c st at i st i cal anal yses.
RESULTS
Network Si ze and Composi t i on
Fi gur e 1 s hows t he di s t r i but i on of n e t wo r k si zes in t he s ampl e. The
me a n ne t wor k si ze f or all subj ect s was I I . 6 ( r ange 0- 30, sd = 5.64; N =
101). The di s t r i but i on exhi bi t s s ome d e g r e e of bi modai i t y, wi t h pe a ks at
ne t wor k si zes of 6 a nd 11-12, s ugges t i ng t hat it ma y be pos s i bl e t o di s-
t i ngui s h be t we e n mor e- a nd l ess- soci abl e i ndi vi dual s . Bef or e we can
saf el y d r a w t hi s concl usi on, howe ve r , we ne e d t o check t hat t he
bi moda l i t y is not due t o c onf oundi ng var i abl es s uch as g e n d e r or
domes t i c ci r cumst ances.
Ther e was a sl i ght , but nons i gni f i cant , g e n d e r di f f er ence i n n e t wo r k
size: wo me n a ve r a ge d a ne t wor k of 12.4 i ndi vi dua l s ( N = 67) c o mp a r e d
wi t h 10.9 (N = 34) f or me n ( Ma nn- Whi t ne y t est , z = -1. 006, P -- 0.315).
Thi s is l ar gel y a c ons e que nc e of t he f act t hat t he di s t r i but i on of n e t wo r k
si zes f or wo me n was mor e s ke we d ( l onger t ai l t o r i ght ) t han t hat f or t he
men. Moda l n e t wo r k si ze was v e r y si mi l ar f or t he t wo sexes.
As mi ght be ant i ci pat ed, t he dome s t i c st at us of subj ect s di d ha ve
s ome i nf l uence on t hei r ne t wor k size. Si ngl e subj ect s ha d a me a n net -
wo r k si ze of 15.4 ( r ange 7-30, sd = 6.20; N = 25), c o mp a r e d wi t h me a n s
of 11.1 ( r ange 0-25, sd = 5.55; N = 51) f or c oupl e s wi t h o u t chi l dr en at
12-
10-
8 -
0
:::1 6 -
0 "
r i I i I I I I L I l I I I I I
0 1 2 3 4 6 7 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9
Kinship and Network Size 279
Network size
Figure 1. Distribution of network sizes.
home and 8.7 (range 0-19, sd = 4.16; N = 24) for couples with dependent
children (including the two single mothers). However, the variance
within each category was considerable, and the differences bet ween
them were not significant (Mann-Whitney tests: singles vs couples with-
out children, z = 0.99, P = 0.322; couples wi t hout children vs couples
with children, z = 1.48, P = 0.139). For all three distributions, however,
the same pattern is evident: 40.0% of singles had net works of size 6-12
compared with 66.7% of couples both with and wi t hout dependent chil-
dren. The differences are largely in the lengths of the tails on either side
of the modal values. Singles had a t runcat ed l ower range, whereas cou-
ples with children had a truncated upper range, while couples wi t h no
dependent children had a more even distribution.
We examined gender differences in the composition of the net work
using a MANOVA with respondent ' s gender as the i ndependent vari-
able and the proportions of contacts that were male (as opposed to
female) and kin (as opposed to non-kin) as the dependent variables. (For
these purposes, kin were defined as individuals related to the subject by
r > 0.125.) This examination revealed that the t wo sexes differed signifi-
cantly in terms of the ratio of male to female contacts (mean percent of
280 Huma n Nat ure, Vol. 6, No. 3, 1995
n e t wo r k me mb e r s t hat we r e mal e: 67. 7% f or me n a n d 30. 8% f or wo me n ;
F1,94 = 47.75, P < 0.001), n u mb e r of f e ma l e ki n c ont a c t e d ( me a n s of 1.88
f or me n a nd 3.06 f or wo me n : F1,94 = 9.64, P = 0.002), n u mb e r of ma l e
non- ki n cont act ed ( me a ns of 5.62 f or me n a n d 1.75 f or wo me n : F1,94 =
52.81, P < 0.001), a nd n u mb e r of f e ma l e n o n - k i n c ont a c t e d ( me a n s of
1.65 f or me n a nd 5.51 f or wo me n : F1,94 = 57.97, P < 0.001), b u t not i n
t e r ms of t he n u mb e r of ma l e ki n c ont a c t e d ( me a ns of 1.74 f or me n a n d
2.06 f or wo me n : F1,94 = 0.14, P = 0.705). I n s u mma r y , wo me n ha d a
l ar ger n u mb e r of f e ma l e f r i ends a n d r el at i ves i n t hei r n e t wo r k s , wh e r e a s
me n h a d a l ar ger n u mb e r of ma l e f r i ends , wi t h ki n b e i n g s i gni f i cant l y
l ess i mp o r t a n t f or me n t ha n f or wo me n .
The r e we r e no di f f er ences i n t he n u mb e r of n o n - k i n c ont a c t e d
mo n t h l y ( me a ns of 7.26 f or me n a n d 7.25 f or wo me n ) , or i n t he p r o p o r -
t i on of non- ki n cont act ed at l east once a mo n t h wh o we r e c ont a c t e d at
l east once a we e k ( me a ns of 39.3% f or me n a n d 43. 9% f or wo me n : Ma n n
Wh i t n e y t est , z = 0.99, P = 0.327). Ho we v e r , t her e wa s a s i gni f i cant di f -
f er ence i n t he p r o p o r t i o n of all mo n t h l y non- ki n ma l e cont act s t hat we r e
cont act ed at l east we e k l y ( me a ns of 81. 4% f or me n a n d 22. 8% f or
wo me n : Ma n n - Wh i t n e y t est , z = --4.78, P < 0.001).
The r e wa s no di f f er ence b e t we e n t he s exes i n t he p r o p o r t i o n of t hei r
e x t e n d e d f ami l i es ( def i ned as t he t ot al n u mb e r of l i vi ng i n d i v i d u a l s
r el at ed t o t he subj ect b y r > 0.125) wh o we r e c ont a c t e d mo n t h l y ( me a n s
of 29. 9% out of an a v e r a g e f a mi l y of 12.1 me mb e r s f or me n a nd 36. 3%
out of an a v e r a g e f a mi l y of 14.1 f or wo me n : Ma n n - Wh i t n e y t est , z =
0.18, P = 0.857). Al t h o u g h me n di d not c ont a c t mo r e ma l e ki n t ha n
wo me n di d i n a bs ol ut e t e r ms , t he y di d c ont a c t a hi ghe r p r o p o r t i o n of
t he ma l e s i n t hei r e x t e n d e d f ami l i es t h a n wo me n di d ( me a n s of 48. 8%
of 3.6 ma l e ki n f or me n vs 40. 2% of 5.1 ma l e ki n f or wo me n ; Ma n n -
Wh i t n e y t est , z = - 2. 67, P = 0.008).
Over al l , ki n a c c ount e d f or 37. 5% of t he n e t wo r k , a f i gur e t hat is
a l mos t cer t ai nl y s i gni f i cant l y hi ghe r t ha n wo u l d be e xpe c t e d if p e o p l e
chos e t hei r n e t wo r k me mb e r s at r a n d o m ei t her f r o m t he l ocal p o p u l a -
t i on as a whol e or f r o m t he s u b s a mp l e of t hat p o p u l a t i o n wh o m t he y
k n o w b y si ght . Unf or t unat el y, we c a n n o t t est t he s i gni f i cance of t hi s
be c a us e we do not h a v e a n y a p p r o p r i a t e va l ue s t o us e f or t he nul l
hypot he s i s . Nonet hel es s , ki n ar e l i kel y t o a c c ount f or a r e l a t i ve l y s ma l l
p r o p o r t i o n of all t he i ndi vi dua l s t hat a n y one p e r s o n k n o ws . I f we t a ke
t he l owe r mo r e c ons e r va t i ve f i gur e of 150 a c qua i nt a nc e s f r o m t he " s ma l l
wo r l d " e x p e r i me n t s ( Ki l l wor t h et al. 1984) a n d t he a v e r a g e e x t e n d e d
f a mi l y si ze obt a i ne d i n t hi s s t u d y of 12.1 f or me n a nd 14.1 f or wo me n
(see above) , we wo u l d expect onl y a b o u t 8.1% a n d 9. 4%, r es pect i vel y, of
n e t wo r k me mb e r s t o be ki n if c hos e n at r a n d o m. On t hi s basi s, t he
o b s e r v e d p r o p o r t i o n is cl ear l y s i gni f i cant l y bi a s e d i n f a v o r of ki n f or
Kinship and Network Size 281
bot h sexes: t he p r o p o r t i o n of ki n i n t he n e t wo r k is gr e a t e r t h a n t he
e xpe c t e d va l ue f or 32/ 34 me n (;(2 = 26.47, df = 1, P < 0.001) a n d f or
63/ 66 wo me n ( one wo ma n wi t h a n e t wo r k si ze of 0 wa s e xc l ude d) (X 2
= 54.55, df = 1, P < 0.001).
Network Si ze and Kin Group Si ze
I f t he n u mb e r of i ndi vi dua l s wh o can be ma i n t a i n e d i n a cl ose soci al
n e t wo r k is l i mi t ed ei t her b y t he t i me a va i l a bl e f or i nt e r a c t i on (e.g., Du n -
ba r 1992) or b y cons t r ai nt s i mp o s e d b y t he pr oc e s s i ng c a pa c i t y of t he
cogni t i ve ma c h i n e r y (e.g., Du n b a r 1993), t he n we mi g h t e xpe c t t he r e t o
be an i nve r s e r el at i ons hi p b e t we e n t ot al n e t wo r k si ze a n d t he si ze of t he
fami l y. I n ot he r wor ds , i ndi vi dua l s wh o h a v e l ar ge e x t e n d e d f ami l i es
ma y be mo r e l i kel y t o conf i ne t hei r soci al cont act s t o me mb e r s of t hei r
f ami l y ci rcl e t ha n ar e t hos e i ndi vi dua l s wi t h f e we r cl ose r el at i ves t o
c hoos e f r om.
Fi gur e 2 s ugge s t s t hat t her e is a we a k n e g a t i v e r e l a t i ons hi p b e t we e n
t he n u mb e r s of ki n a nd non- ki n c ont a c t e d at l east mo n t h l y ( Pe a r s on' s r
= -0. 138, t99 = 1.37, P > 0.05). On e l i kel y r e a s on wh y t he c or r e l a t i on is
25
20
15'
10'
Number of non- ki n
9 0 0 0
00 O 9
OO
0 0 0 0 0
0 0 0 0 9
0 0 0 0 9 9
0 0 0 0 0 9
0 0 9 00
9 00
O 0 9 9 9 9 9 @
@ Q
0 I _k J I I 9 I
0 5 10 15 20 25 30
Number of kin
Figure 2. Number of non-kin contacted at least once a mont h pl ot t ed agai nst
the number of kin contacted at least once a mont h.
282 Human Nature, Vol. 6, No. 3, 1995
poor is t he l ar ge n u mb e r of i ndi vi dual s i n t he l owe r l eft q u a d r a n t ( who
cont act onl y a smal l n u mb e r of bot h ki n a nd non- ki n) . In ot he r wor ds ,
it ma y be t hat f or r el at i vel y asoci al i ndi vi dua l s wh o s e n e t wo r k si ze is
wel l be l ow t he cogni t i ve l i mi t , t he n u mb e r of ki n doe s not r est r i ct t he
n u mb e r of non- ki n cont act ed. If al l subj ect s wh o cont act ed f e we r t ha n
10 i ndi vi dual s ar e excl uded, t he n t he r e is a hi ghl y s i gni f i cant ne ga t i ve
cor r el at i on be t we e n numbe r s of ki n a n d non- ki n cont act ed (r = - 0. 554,
t58 = -5. 06, P < 0.001).
One r eas on f or t hi s s eems t o be t hat i ndi vi dua l s f r om l ar ge f ami l i es
t end t o cont act mor e ki n ( Fi gur e 3: Pe a r s on' s r = 0.397, t99 = 4.30, P <
0.001). In cont r ast , t he n u mb e r of non- ki n cont act ed is not s i gni f i cant l y
r el at ed t o t he si ze of t he f ami l y ( Fi gur e 4: r = - 0. 032, t80 = - 0. 32, P > 0.05;
f or subj ect s wi t h ne t wor ks l ar ger t ha n ni ne me mbe r s : r = - 0. 096, t58 =
-0. 74, P > 0.05); r at her , it ma y be r el at ed mor e cl osel y t o i ndi vi dual s '
r es pect i ve oppor t uni t i e s f or i nt er act i on out s i de t he fami l y.
Thes e r esul t s s ugges t t hat pe opl e pl ace a p r e mi u m on ma i nt a i ni ng
f ami l y cont act s a nd onl y e xt e nd t hei r n e t wo r k of cont act s b e y o n d t he
f ami l y if t hey ha ve s par e capaci t y i n t hei r t ot al n e t wo r k si ze onc e t hei r
ke y f ami l y cont act s ha ve be e n e xha us t e d. Thi s s eems t o be i n d e p e n d e n t
30
25
20
15
10
Number of kin cont act ed
O - . . - 9 9 _ , , ,
0 10 20 30 40
Family si ze
5O
Figure 3. Number of kin contacted at least once a mont h plotted against total
family size.
Kinship and Network Size 283
Number of non-ki n cont act ed
25
20
15
lOq
5
9 9 9 9 9
10 20 30 40
Family size
0
0 5 0
Figure 4. Number of non-kin contacted at least once a mont h plotted against
total family size.
of t he di s t r i but i on of degr ees of ki ns hi p wi t hi n t he f ami l y (i.e., ki ns hi p
densi t y) : n u mb e r of ki n cont act ed is not r el at ed to subj ect ' s me a n de gr e e
of r el at edness, rmean, to all t he me mb e r s of h i s / h e r e x t e n d e d f a mi l y
( Pear s on' s r = 0.092, F1,99 = 0.47, P = 0.496). In ot he r wor ds , f ami l i es wi t h
a hi gher pr opor t i on of cl osel y r el at ed i ndi vi dua l s (e.g., si bl i ngs) do not
s how a ny t e n d e n c y t o i nt er act mor e f r e que nt l y wi t h each ot he r t ha n
t hose wi t h a l owe r pr opor t i on (e.g., f e we r si bl i ngs, mor e cousi ns) . Not e ,
however , t hat i ndi vi dual s do not neces s ar i l y i nt er act wi t h all t he me m-
ber s of t hei r e xt e nde d family. On aver age, me n c ont a c t e d onl y 30.0% of
t he me mb e r s of t hei r e xt e nde d f ami l y at l east once a mont h, whi l e
wo me n cont act ed 36.6% of t hei r f ami l y me mbe r s .
Kinship and the Support Clique
The me a n n u mb e r of i ndi vi dua l s f r om wh o m s u p p o r t wo u l d be
s ought wa s 4.72 ( r ange 0-14, s d = 2.95; N = 101). The r e wa s no di f f er -
ence i n t he si zes of t he s u p p o r t cl i ques of me n a nd wo me n ( me a ns of
4.47 f or 34 me n a nd 4.85 f or 67 wo me n ; Ma n n - Wh i t n e y t est , z = - 1. 18,
284 Huma n Nature, Vol. 6, No. 3, 1995
16-
14-
12-
lO-
e~
6-
0 1 7 8 9 10 11 12 13 14
Cl i que size
I Men [ - - ] Women
Figure 5. Distribution of support clique sizes,
P = 0.236). Fi gur e 5 s ugge s t s t hat , as wi t h t he n e t wo r k si ze, t he di s t r i b-
ut i on is b i mo d a l wi t h p e a k s at 2- 3 a n d 5. Di s a g g r e g a t i o n of t he da t a f or
me n a nd wo me n yi e l ds a s i mi l ar pi ct ur e, b u t wi t h s l i ght l y of f set pe a ks :
me n at 2 a nd 5, wo me n at 3 a n d 7. As wi t h t he di s t r i but i on of n e t wo r k
si zes, t he p e a k s in t he si ze of t he s u p p o r t cl i que ma y c o r r e s p o n d t o con-
t r ast s b e t we e n mor e - a nd l es s - s oci abl e i ndi vi dua l s . So me e v i d e n c e t o
s u p p o r t t hi s s ugge s t i on c ome s f r o m t he f act t hat t he si ze of t he s u p p o r t
cl i que is l i near l y r el at ed t o t ot al n e t wo r k s i ze ( Fi gur e 6: Pe a r s o n ' s r =
0.427, t97 = 4.651, P < 0.001) a nd r e pr e s e nt s a n a v e r a g e of 39.8% of t he
i ndi vi dua l ' s t ot al cont act ne t wor k. The r e s e e m t o be no c o n s p i c u o u s di f -
f er ences b e t we e n t he sexes i n t hi s r es pect .
Of t he s u p p o r t cl i que, 22.6% we r e t ypi c a l l y f e ma l e ki n, 33. 5% f e ma l e
non- ki n, 17.1% ma l e ki n, a nd 26. 8% ma l e non- ki n. The p r o p o r t i o n of
s u p p o r t sour ces t hat we r e ki n doe s not di f f er f r o m t he p r o p o r t i o n of t ot al
mo n t h l y cont act s (i.e., n e t wo r k si ze) t hat we r e ki n ( Wi l coxon ma t c h e d
pai r s t est s: f or me n, z = --0.18, P = 0.860; f or wo me n , z = - 0. 62, P = 0.536);
nor doe s t he p r o p o r t i o n of t he s u p p o r t cl i que t hat wa s of t he o p p o s i t e
sex di f f er f r o m t he p r o p o r t i o n i n t he t ot al n e t wo r k ( Wi l coxon t est s: f or
me n, z = -1. 172, P = 0.086; for wo me n , z -- - 1. 91, P = 0.056). Thi s s ug-
Kinship and Network Size 285
S u p p o r t c l i q u e s i z e
16
14
12
10
8
6
4
2
0
- 0- - WOmeN
[ ] [ ]
Men
9 9
9 9 9 ...... O
E 3 9 9 ' 9
9 /
/ 9 ..... o ~ o 9 [] 9 9
. . . . . . [ ] ' " [ ] ~ q L ~ ~ [ ] E ] [ ] 9 9
9 [ 3 [ 3 0 0 9
I [ ] ~ t I I i
5 10 15 2 0 2 5 3 0
Network si ze
3 5
Figure 6. Support clique size pl ot t ed agai nst total net work size.
gest s t hat , over al l , t he s u p p o r t cl i que is c hos e n on t he s a me bas i s as t he
wi d e r n e t wo r k of f r i ends, a nd it a p p e a r s t o be a mo r e or l ess r a n d o m
s a mp l e of t hat wi d e r ne t wor k. None t he l e s s , on a ve r a ge , 40% of s u p p o r t
cl i que me mb e r s we r e cl ose ki n: t he p r o p o r t i o n of ki n in t he s u p p o r t
cl i que wa s hi ghe r t han wo u l d be e xpe c t e d if t he y we r e d r a wn at r a n d o m
f r om an a c qua i nt a nc e s n e t wo r k of a b o u t 150 f or 2 4 / 3 3 me n a n d 5 4 / 6 6
wo me n (X2 = 6.82 a nd 26.73, r es pect i vel y; df = 1, P < 0.01 i n bot h cases).
As wi t h t ot al n e t wo r k si ze, b o t h me n a n d wo me n t e n d e d t o sel ect t he
t wo sexes of ki n wi t h a b o u t e qua l f r e q u e n c y as s u p p o r t s our c e s ( me a n
pe r c e nt a ge of ki n s u p p o r t s our c e s t hat we r e mal e: 49. 1% f or me n a n d
40.6% f or wo me n ) . Ho we v e r , a mo n g n o n - k i n s our ces of s u p p o r t , me n
a nd wo me n s h o we d s t r i ki ng pr e f e r e nc e s f or t hei r o wn sex ( me a n per -
cent of ma l e s a mo n g non- ki n s u p p o r t sour ces: 26. 6% f or wo me n a n d
82.5% f or me n, Ma n n - Wh i t n e y t est , z = 32.74, P < 0.001).
DI S CUS S I ON
We h a v e s h o wn t hat t he n u mb e r of i ndi vi dua l s c ont a c t e d on a r e gul a r
basi s (i.e., at l east once a mo n t h ) c o n f o r ms cl os el y t o t hat o b t a i n e d f r o m
e s t i ma t e s of t he si ze of " s y mp a t h y g r o u p s . " I n s el ect i ng t he me mb e r s of
286 Human Nature, Vol. 6, No. 3, 1995
this group, both sexes contact kin di sproport i onat el y more often than
they do non-kin; as a result, the number of kin available ultimately lim-
its the number of non-kin that can be i ncl uded in the network. Women
contact kin more often than men do, while both sexes exhibit a strong
t endency for contacts with their own sex to be more common than con-
tacts with the opposite sex. The inner clique of individuals from whom
support or advice might be sought tends to mirror these preferences
rather closely.
These results are generally in line with those report ed by both Rands
(1988) and Booth (1972) for Nort h Ameri can populations. Booth (1972)
noted that while there were no differences in net work size bet ween men
and women, there did seem to be a sex difference in social participation:
men were more socially active than women, but women mai nt ai ned
stronger emotional ties with their contacts and had more ties with kin
than men did.
Our results suggest that the sizes of both net works and support
cliques are bimodal. Although part of the difference bet ween small and
large networks can be attributed to the reproduct i ve status of individu-
als, it seems that there is some residual variation in net work size that is
due to differences in sociability. Al t hough net work size is known to vary
with life history stage (Larson and Bradney 1988), the possibility
remains that some of the variance in net work size is due to differences
in personality. We are currently exploring this possibility in more detail.
The preference for kin over non-kin seems to be in line with what
woul d be expected from the theory of kin selection (that individuals will
prefer to associate with and/ or be altruistic towards kin when all other
things are equal). In a now classic st udy of a worki ng class communi t y in
the east end of London during the 1950s, Young and Willmott (1957)
found a similar t endency for male and female net works to be distinct and
largely sex-specific. They also noted that kinship played a particularly
important role in female networks, with mot hers and daught ers formi ng
what amount to mutually supportive alliances. Bott (1971) also report ed
a t endency for maternal relatives to be more i mport ant than paternal rel-
atives in the social lives of London middle-class families. The present
study provides quantitative support for these largely informal studies; it
also suggests that these effects have remai ned stable despite the enormous
changes that have taken place in British society over the past half century.
These results appear to be at odds wi t h the view that human societies
are typically patrilocal (e.g., Foley and Lee 1989; Levi-Strauss 1969; Rod-
seth et al. 1991), such that in some cases women' s kinship bonds are
weakened or even severed. One possible reason why female kinship
bonds may become relatively more i mport ant in moder n industrial soci-
eties is that groups of males are no longer able to monopol i ze resources
Kinship and Network Size 287
or ot her s our ces of i n v e s t me n t t hat wo me n n e e d f or s ucces s f ul r e pr o-
duct i on. I n soci et i es wh e r e ma l e ki n- ba s e d al l i ances a l l ow me n t o
mo n o p o l i z e s uc h ser vi ces, wo me n ma y be f or ced t o c hoos e b e t we e n
t hes e s er vi ces a n d t hei r o wn ki ns hi p ties. I n t he a bs e nc e of mo n o p o l i z -
abl e ser vi ces, wo me n ma y f i nd t hat t hei r o wn ki n- ba s e d al l i ances ar e
mo r e va l ua bl e a n d me n ma y t he n be l ess i ncl i ned t o c ont i nue s e r vi c i ng
t hei r o wn ki ns hi p ne t wor ks .
The f i ndi ng t hat ki n do not p l a y a mo r e p r o mi n e n t r ol e i n t he s up-
por t cl i que t ha n t he y do in t he f r i e nds hi p n e t wo r k wa s , h o we v e r , une x-
pect ed. That t he y do not mi g h t refl ect t he f act t hat , i n mo d e r n i ndus t r i a l
soci et i es, i ndi vi dua l s of t en l i ve t oo f ar f r o m t hei r i mme d i a t e ki n t o be
abl e t o us e t h e m f or he l p in t i mes of i mmi n e n t crisis. Unf or t una t e l y, we
di d not as k i ndi vi dua l s wh e t h e r t hey l i ved n e a r t hei r ki n ( our c onc e r n
wa s s i mp l y wi t h wh e t h e r or not t he y contacted t hem) , so we ar e u n a b l e
t o d e t e r mi n e wh e t h e r t hos e wh o pr e f e r r e d f r i e nds as s our c e s of h e l p di d
so be c a us e t he y l acked n e a r b y ki n. Al t h o u g h t he e t h n o g r a p h i c l i t er at ur e
s ugge s t s t hat ki n ar e still wi d e l y s een as a p r i ma r y s our c e of uns t i nt i ng
s u p p o r t be c a us e " bl ood is t hi cker t ha n wa t e r " (see, f or e x a mp l e , Bot t
1971; Du n b a r et al. 1995; La r s on a nd Br a d n e y 1988), t he mobi l i t y t ypi cal
of mo d e r n s oci et y ma y ma k e it di f f i cul t f or i ndi vi dua l s t o be as i nt i ma t e
wi t h ge ogr a phi c a l l y di s t ant r el at i ves as t he y ar e wi t h unr e l a t e d f r i e nds
wh o m t hey see r egul ar l y. I nde e d, Bot t (1971) n o t e d a t e n d e n c y f or ki n-
s hi p t i es t o we a k e n wh e n r el at i ves mo v e d a wa y ( es peci al l y wh e n t he y
we r e pe r c e i ve d as d o i n g so i n or de r t o be t t e r t h e ms e l v e s soci al l y) .
We t hank Lilian Cameron, Barbara Forest, Jean Scott, Mary Spoors, and Lisa White
for help with distributing the questionnaires and the anonymous referees for their
helpful comment s.
Robin Dunbar, B.A., Ph.D., until recently a professor of biological anthropology at Univer-
sity College London, is professor of psychology at the University of Liverpool. His main re-
search interests concern mating systems and the evolution of mammalian social systems.
Matt Spoors, B.Sc., M.Sc., teaches at St Hugh' s School, Grantham (England).
REFERENCES
Bernard, H. R., P. D. Killworth, and L. Sailer
1982 I nf or mant Accuracy in Social Net wor k Data, V: An Experi ment al
At t empt to Predict Actual Communi cat i on from Recall Data. Social Science
Research 11:30-66.
1984 On the Validity of Retrospective Data: The Probl em of I nf or mant Accu-
racy. Annual Review of Anthropology 13:495-518..
288 Huma n Nat ure, Vol. 6, No. 3, 1995
Bert6, N.
1988 K' ekchi ' Horticultural Labor Exchange: Product i ve and Reproduct i ve
Implications. In Human Reproductive Behaviour, L. Betzig, M. Borgerhoff Mul-
der, and P.Turke, eds. Pp. 83-96. Cambri dge: Cambr i dge Uni versi t y Press.
Bliezner, R.
1988 Indi vi dual Devel opment and Int i mat e Relationships in Mi ddl e and
Late Adulthood. In Families and Social Networks, R. M. Milardo, ed. Pp.
147-167. Newbur y Park, California: Sage Publications.
Booth, A.
1972 Sex and Social Participation. American Sociology Review 37:183-192.
Bott, E.
1971 Family and Social Network, revised edition. London: Tavistock Publications.
Brown, B. B.
1981 A Lifespan Approach to Friendship: Age-related Di mensi ons of an
Ageless Relationship. In Research in the Interweave of Social Roles, Vol. 2:
Friendship, H. Z. Lopata and D. R. Maines, eds. Pp. 23-50. Greenwich, Con-
necticut: J. A. I.
Burt, R. S.
1982 Towards a Structural Theory of Action: Network Models of Social Structure,
Perceptions and Action. New York: Academi c Press.
Buys, C. J., and K. L. Larsen
1979 Human Sympat hy Groups. Psychology Reports 45:547-553.
Coleman, J. S.
1964 Introduction to Mathematical Sociology. London: Collier-Macmillan.
Dunbar, R. I. M.
1992 Time: A Hi dden Constraint on the Behavioural Ecology of Baboons.
Behavioural Ecology and Sociobiology 31:35-49.
1993 Coevol ut i on of Neocortex Size, Gr oup Size and Language in Humans.
Behavioral and Brain Sciences 16:681-735.
Dunbar, R. I. M., A. Clark, and N. L. Hur st
1995 Conflict and Cooperat i on among the Vikings: Cont i ngent Behavioural
Decisions. Ethology and Sociobiology, in press.
Firth, R., ed.
1956 Two Studies of Kinship in London. London: Athlone Press.
Fischer, C. S.
1982 To Dwell among Friends: Personal Networks in Town and City. Chicago:
University of Chicago Press.
Foley, R., and P. C. Lee
1989 Finite Social Space, Evol ut i onary Pat hways and Reconstructing
Homi ni d Behavior. Science 243:901-906.
Foot, H. C., A. J. Chapman, and J. R. Smith
1980 Friendship and Social Relations in Children. New York: Wiley.
Hames, R. B.
1979 Relatedness and Interaction among the Ye' kwana: a Prel i mi nary Anal y-
sis. In Evolutionary Biology and Human Social Behavior, N. A. Chagnon and
W. Irons, eds. Pp. 238-249. Nort h Scituate, Massachusetts: Duxbur y Press.
Kinship and Network Size 289
Hays, R. B., and D. Oxley
1986 Social Net wor k Devel opment and Functioning duri ng a Lifetime Tran-
sition. Journal of Personality and Social Psychology 50:305-313.
Hughes, A.
1988 Evolution and Human Kinship. Oxford: Oxford Uni versi t y Press.
Keesing, R. M.
1975 Kin Groups and Social Structure. New York: Holt, Rinehart and Winston.
Killworth, P. D., H. R. Bernard, and C. McCart y
1984 Measuri ng Patterns of Acquaintanceship. Current Anthropology
25:391-397.
Knoke, D., and J. H. Kuklinski
1982 Network Analysis. Beverly Hills, California: Sage.
Kudo, H., S. Bloom, and R. I. M. Dunbar
1995 Neocort ex as a Constraint on Social Net wor k Size in Primates. Sub-
mi t t ed to Behaviour.
Larson, R. W., and N. Bradney
1988 Precious Moment s wi t h Family Members and Friends. In Families and
Social Networks, R. M. Milardo, ed. Pp. 107-126. Newbur y Park, California:
Sage Publications.
Levinger, G., and A. C. Levinger
1986 The Temporal Course of Relationships and Devel opment . In Relation-
ships and Development, W. W. Har t up and Z. Rubin, eds. Pp. 111-133. Lon-
don: LEA Press.
Levi-Strauss, C.
1969 The Elementary Structures of Kinship. London: Eyre and Spottiswoode.
McCannell, K.
1988 Social Net works and the Transition to Mot herhood. In Families and
Social Networks, R. M. Milardo, ed. Pp. 83-106. Newbur y Park, California:
Sage Publications.
Milardo, R. M.
1988 Families and Social Net works: An Over vi ew of Theory and Met hodol -
ogy. In Families and Social Networks, R. M. Milardo, ed. Pp. 13-47. Newbur y
Park, California: Sage Publications.
Milardo, R., ed.
1988 Families and Social Networks. Newbur y Park, California: Sage Publications.
Mitchell, J. C.
1969 The Concept and Use of Social Net works. In Social Networks in Urban
Situations, J. C. Mitchell, ed. Pp. 1-50. Manchester, England: Manchest er
University Press
Mitchell, J. C., ed.
1969 Social Networks in Urban Situations. Manchester, England: Manchest er
University Press.
Rands, M.
1988 Changes in Social Net works following Marital Separation and Divorce.
In Families and Social Networks, R. M. Milardo, ed. Pp. 127-146. Newbur y
Park, California: Sage Publications.
290 Human Nature, Vol. 6, No. 3, 1995
Rodseth, L., R. W. Wrangham, A. M. Harrigan, and B. B. Smuts
1991 The Human Community as a Primate Society. Current Anthropology
32:221-255.
Thorne, B.
I986 Girls and Boys Toget her. . ~ But Mostly Apart: Gender Arrangements
in Elementary Schools. In Relationships and Development, W. W. Hartup and
Z. Rubin, eds. London: LEA Press.
Young, M., and R Willmott
1957 Family and Kinship in East London. London: Routledge and Kegan Paul.