You are on page 1of 13

Inside Social Network Analysis

Kate Ehrlich 1 and Inga Carboni 2

Introduction

A m a n a g e m e n t co n s u lt i n g f i r m h o p e s t o w in a l u c ra t iv e co n t r a ct w it h a l ar ge
internat io nal financial institut ion. After weeks of intense preparation, the team
s e n d s o f f a p r o p o s a l . Sho r t ly t h e r e af t e r, t he y le arn t h at co n t rac t w a s g iv e n t o a
c o m p e t it o r w i t h w h o m t he c l i e nt h a d w o r k e d p r e v io u s ly . A lm os t s i x m onth s l a te r,
o n e o f t h e t ea m m em b e r s f i n d s out t ha t a no t he r gr o up a t t h e m a n ag e m e nt
c o n s u lt i n g f i rm h a d w o r k e d o n a p ro j e c t w i t h t h e p r o s p e ct iv e c li e n t a n d h a d g a i n e d
an in-depth knowledge of its business operat ions . W hy , as ked t he frust rat e d t e am
m e m be r , w as n ’ t t h i s c r i t ic a l k n o w l e dge s h a r e d w it h t h e t e am?

Th is i s ju st one example of the opp ortun iti es th at l arge comp anies can mi ss if they
f ai l t o u n der s t an d t h at s u c c e s s dep e n d s l es s on rep or tin g s t ructu r e an d mo r e on a n
in fo rm al w eb of cont a cts . In th e p as t, co mpan ie s th at enco unte re d a l o ss of bu sine s s
li ke the one d e scrib e d ab ov e, mi ght c on du ct a su rv ey an d int er vi ew empl o ye es to
d i sc o ve r wha t w en t wr on g. A n e w app ro a ch c al l e d s oci al n et wo rk an al ys i s o r S N A,
h a s be en g ainin g cur renc y am on g bu sine s s cons ult ant s as a n e ffe cti v e m etho d f o r
reveal ing the h idden connect ion s th at dri v e h ow w or k get s d on e .

I n t h e n e t wo r ke d o r g an i z at i on, i n div i du al su c ces s an d t h e su cc e ss of a t ea m d ep en d


l es s on repo rt ing st ru ctu re an d mor e on wh o yo u kn ow. Thi s p ape r p ro vi d es a b rie f
int r odu cti on t o the maj or concepts and m e asures in SNA and thei r appli c ati on t o
bu siness p roblems.

W hat i s a S oc ia l N etwo rk A na ly s i s?
A s o ci al n etw o rk an al y si s ex am ine s th e s tru ct u re of s oc ia l re lation ship s in a gr oup t o
un cover th e in form al connect ion s between p eopl e. In a con sultin g settin g, th ese
r el at i on shi p s a r e o f t en o n e s o f c om m u n i c at i on, a w a ren e ss , t ru st , and d e ci si on-
m ak ing. A s a n appr o ach t o l oo kin g a t th ese relati on ship s, SNA has b een around a
l on g tim e.

1
IBM T.J. Watson Research (katee@us.ibm.com)

2
Boston College (inga.carboni@bc.edu)

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 1
M o st n et wor k an al ys ts c it e Jo s eph M o r eno’ s int rodu cti on of th e tool s and meth ods of
soci ometry, in 1934 as th e year in which the fo rmal an al ysi s of soci al networks began.
SNA became much more p opul ar with rese archers in th e early 1970 s when advances
in computer t echn ol o gy m a de it p ossible to stu dy l a rge groups. Within th e l a st ten
years, SNA h a s ri sen to p rom inen ce in a nu mber of f iel ds, in clu din g organi zation al
beh avi or , anth rop ol o gy , so ci ol og y, a nd me di cine .

Most recentl y, SNA has become an imp ort an t too l f or o rg aniz at ion al con sult ant s
seekin g t o un derst and the connection bet ween p att ern s of int eracti ons and bu siness
o u t c om e s su ch a s j ob p e rf o rm an c e, j ob sa t i sf act i on, a dopt i o n o f n e w i d e a s o r
t echn ol o gies, li kelihood of inf o rm ati on gett ing sh ar ed , an d c r e ati on o f ne w i de as.

T o get h e r, t h e s e el em ent s can b e an al yze d t o re v ea l t h e s oc i al n et w ork an d d et ermi n e


i f it i s meetin g the bu siness n eeds of th e group. Th e ou tcom e of an SNA let s u s see
w h e re co l l ab or ati on i s b re ak i n g d ow n , wh er e t a l en t an d ex p e rt i s e c ou l d b e b et t e r
l everaged, where deci si on s are gettin g b o gged down, and where opportuniti es f or
d i f f u si on and i n n o v at i on ar e b e i n g l o st. Th e da t a giv e l ead e r s t h e p i ctu re t h e y n e e d
t o c r eat e a s et of r em ed i al act i on s f o r i n d i vi dua l s and l e ade r s t o i m p ro v e
p rodu cti vit y, eff ici en cy and innovation. Th ese action s in clu de m odif yin g rol e s and
resp onsibiliti es to foster m o re effect ive patterns of communi cati on, m ethods f or
imp r oving trust, bet ter u se of t echn ology t o re a ch othe r s, re - ali gnm ent of re w ard s
a n d i n c en t i ve p r o gr am s.

How is a Social Network Analysis different?


SNA di ff er s f r om c on venti on al appro a ch es t o bu sine s s p robl em s in one v er y
imp o rt ant way: SNA assum es that people are all interdepende nt. Th is assumpti on i s
r a di c all y d iff er ent f r om t r adi tion al r e se a r ch app ro a che s whi ch as sum e th at wha t
pe opl e d o, thin k, and fee l i s ind epen dent of who the y kn ow .

Th e f ocu s on int er d epen den ce m e an s th at SNA c an ask - - an d an s we r -- que sti on s


s u ch a s:

• I s S al es eff ec t i vel y com m u n i c at i n g w i t h Ma r k e t i n g t o sha re a n d c oo r din at e


in fo rm ati on a bou t the cu st ome r?
• Wh en two comp anies or o r gan iz atio ns me rg e, h ow c an man a gem ent us e
th e inf ormal network t o spread import ant messages?
• Ar e de ci si on s in a d ist ribut ed s oftw a re d ev el opm ent te am be ing ma de a nd
carried out effi ci entl y or are on e or more peopl e actin g as bottlen e cks?
• In R&D group, are th ere en ough p eople b rin ging in i deas f rom ou tside and
a r e t h os e i de a s b ei n g a ct ed u p on?

Many t r aditional stat istical techniques are b a sed on th e assumpt ion of in dep endence.
For thi s reason , traditional stat istics, su ch as comp arin g the mean s of t wo group s,
c ann ot b e co ndu ct ed on int er d epen dent d at a. T o de al with thi s p rob le m, SNA h as
d e vel op ed a s et of SNA- sp ec ifi c st ati st ic s such a s cent rality an d den sity th at provide
measu res of int erdependen ce

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 2
Why do a Social Network Analysis?
I f w e w ant to und er st and ho w a g r oup fun cti ons w e mi ght g o to an org a niz ati on ch ar t
t o f i n d t h e s en i o r p e opl e wh o ar e e m p o we re d t o m ak e d ec i sion s o r t o s ee h ow t h e
w o rk i s di vid e d up fun cti on all y. But in t h e e v olv i n g n et wo rk e d o r gan i z a t i on , t h i s
c ha rt i s n o lo ng e r an a dequ at e gui de to h ow the g r oup r ea ll y w o rk s. C on si de r the
f oll owin g diagram s wh ich rep resent th e producti on division of a l arge petrol eum
corp orati on (Cross et al. , 2001).

Upper man agement want ed t o know h ow thi s g ro u p w a s p r e p ar i n g t o sha r e i m p o rt ant


dri llin g kn owledge so they un dert ook an SNA. In th e organiz at ion al ch art on th e l eft,
we see th at J ones i s the seni or p erson in the group whil e Col e i s in a m ore juni or
r ol e. An SN A r ev e aled th at, in co ntr a st t o the fo rm al ch ar t, mi d- lev el m an a ge rs, and
C ol e in p a rti cul a r, w er e pl a yin g a p iv ot al role in th e group’ s com munication n et work.
N o t onl y wa s C ol e linke d t o m any pe opl e, m akin g him ve ry c ent ra l to th e g r oup, but
he was also th e onl y link b et ween the clu st er of p eop le at th e top wh o rep resented
p rodu cti on and the rest of th e group who were in volved in other di stin ct but cri tical
activiti es.

Th rough additi on al int erviews with key people in th e net work, in clu din g J ones, i t was
a p p ar en t t h a t J on e s h a d b e c om e r em ov e d f rom m an y of t h e d a y-t o- d ay w or ki n g s o f
th e gr oup. Hi s l ack of r e sp ons iv ene s s and p arti cip ati on in th e group oft en result ed in
del a ys whi ch imp a cted th e rest of th e group. W ithout an SNA, J one s’ l ack of ac tiv e
p articipation in th e group and Cole’ s imp ort ance may have gon e unnoti ced. Aft e r
con du ctin g th e SNA, upper man agement was abl e to consider a ran ge of opti ons,
in clu din g f orm alizin g Cole’s role.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 3
Business applications of SNA
SNA app lies t o a wi de ran ge of bu siness p rob lem s, in clu ding:

K no wle dg e Ma nag eme nt a nd Co l la bor at io n . SNAs can help locat e exp ert ise,
s e ed n ew co mmunit ie s o f p ra cti c e, d e vel op cr os s -fun cti on al kno wl ed ge- sh a rin g, and
imp r ov e strate gi c de ci si on- makin g ac r o ss l e ade r ship t e ams.

T ea m- bu i ldi ng . SNAs can con tribut e to the creati on of inn ovat ive team s an d
f acili tate p ost-merger int egrati on. SNAs c an re v ea l , f or ex amp l e, wh i c h i n d i v i du al s
are m ost li kely t o b e exp osed to n ew ideas.

Human Resources. SN A s c an i denti fy an d mo nit or th e ef fe ct s o f w or kf o rc e


d i v e rsi t y , on -bo a rdin g a n d r et en t i o n , and l e adership development . For in st ance, an
S N A c an rev e al wheth er o r n ot m en t or s a re cre at i n g r el at i o n sh ips b etween ment ees
a n d oth er emp l o ye es .

S a l e s a nd M a r k et i ng. SNAs can help track the adop tion of new p roduct s,
t echn ol o gies , and id ea s. Th ey c an al s o su g ge st c om munic ati on str at e gie s.

S trat eg y. S N A s c an su p p ort i n dus t ry ec o sy st e m an al y si s a s w el l a s p a r t n e r sh i p s a n d


alli an ces. They can pinp oint wh ich firms are linked to criti cal in dustry pl ayers and
which are not.

Conducting a Social Network Analysis


A soci al n etwork an al ysi s in a bu siness settin g h a s three imp ortan t element s.

• A g r o up . Th e fi r st step in c on duc tin g a s o cia l ne tw or k s a n aly si s i s t o


det ermine the group under study. In th e m an agem ent consultin g exampl e
describ e d at th e beginnin g of thi s arti cle, the group m a y be al l of the
c on sult ant s in th e p art ic ul ar p r ac tic e. A g roup can b e comp o se d o f pe opl e
w h o a re ded i c at e d t o a p ar t i cu l a r t a s k , su ch a s a s of t w a re d e vel op m e n t
t eam or it can be a col lect ion of p eople su ch as a commun it y, upper
m an a gem en t a c r os s dif fe r ent b u sin e s s unit s, or memb ers of a temporary
t a sk force. G roup s t ypicall y ran ge in siz e from 25 t o 200. Th e in dividu al
u n i t s i n a g ro u p a r e c om m onl y r e f e r re d t o a s a c t o r s o r n o d e s.
• I nt erac t ion s . An SNA l ooks at the rel ati onship s between actors. The
n atu re of the rel ati onship s we are int erested in will vary accordin g to our
reason for condu ctin g the stu d y. In th e case of th e man agement consult ing
comp an y, we mi ght be int erested in an alyzin g communi cation pattern s b y
l ookin g at which consult ant s int e ract with each other f or th e purp o se of
obt ainin g clien t-related in form ati on. Int e ra cti on s ar e a ls o r ef e rr ed to a s
th e lin ks or ti es bet ween p e ople. Th e p attern of int eractions in a group i s
called a soci al n etwork.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 4
• A t t r ib u t e s. Att ribut e data can help det ermine wheth er there are
s y st em ati c fa c to r s th at in fluen ce int er a cti on s be tw een p eopl e. Fo r
ex amp le, we often find th at p eopl e in one bu siness unit don’t routinely
sh are inf o rm ati on with peopl e in an other unit. Th e f actors th at mi ght
in fluen ce these int eracti on s ran ge from incentive p rograms th at m oti vate
peopl e t o spen d th ei r tim e with people in their own bu siness unit to
“ cu l t u r al ” di f f e r e n c e s , su ch a s l an gu a g e or wo r k e t h o s, t h at m ak e i t
d iff icu lt f or pe opl e t o co mmuni c ate e a sil y with o ne ano the r. In a stu dy o n
diversity, f or ex ampl e, att ribut es th at inf luen ce interaction s mi ght in clu de
ethni cit y, gen d er, an d attitu des about affi rm ati ve acti on. In th e
man a gement con sult ing case, rel evan t att ri b u t es m i ght i n clu de whe re
s o me one wo r k s ( e. g. co untr y, g eogr aphi c region), whi ch bu siness unit th ey
are in (e. g. sales, m arketin g, develo pment ), the ir le vel of s eni o rit y, an d
h ow l on g they h a ve b een with the comp any. They m a y also include
p e r s ona l i t y m e a su re s, su ch a s M ye rs - B riggs p erson al ity ty p e s or sc or e s on
aptitu de test s. Onl y att ribut es that are b elieved t o inf luen ce int e racti on s
are in clu ded in an SNA.

What can ties reveal?


Th e result of col lecting SNA dat a i s a m at rix showin g, in num erical form , the
ex isten ce, type, an d/or qu alit y of int eract i on s between p ai rs of people. In form ati on
a bou t the se k ind s of ti es i s c omm onl y c ol le cted thr ou gh int er vi ew s or su r ve y s, oft en
a d mini ste r ed onl ine. A t ypi ca l su rv e y mi ght lis t all of th e pe opl e in a w o rk t e am a nd
th en ask each individu al who th ey go to wh en th ey n ee d cli ent- r el ate d in fo rm ati on.
T h e r e sul t o f all t h e r e s p on s e s m i ght l o ok somethin g li ke the tab les bel ow whi ch
s ho w s, f o r e a ch p ai r of pe opl e, who g o es t o wh om f or info rm at ion.

B IL L J IM C A R OL PAM PAT

B IL L 0 0 0 1 1

J IM 0 0 0 1 0

C AR OL 0 1 0 1 0

P AM 0 0 0 0 1

P AT 0 0 1 0 0

In th is ex ampl e, you can see th at Bill, Jim and Carol all go to Pam f o r cli ent -rel ated
in fo rm ati on. P a m, ho wev e r, go e s onl y t o P at fo r in fo rm atio n an d Pat g o e s onl y to
C a r ol . T h u s , Car ol , who onl y one p e r s on t u rns t o di r ect l y f or i n f o rma t i on , m a y
actually b e th e sou rce of mu ch of th e cl ie n t - re l a t e d i n f o rm ati on sha re d i n t h e gro u p
( vi a P at an d P a m).
Inside Social Network Analysis
Ehrlich and Carboni
p a ge 5
Col lectin g inf o rm ati on abou t ties is not limit ed to su rveys. Thi s in form ati on can al so
be inferred from a number of exi stin g data s our c e s, such as em ai l ex ch an ge s, di rec t
o bs e rv ati ons o f g r oup int er a cti on, bil lab le h our s (i. e., who w o rk s on pr oj e ct s with
whom? ), p rofessi on al cit at ion s (i.e., wh o publ ishes with whom?), corporate b oard
i n t e rlo c ks ( i . e. , who se rv e s on a b oa r d wi t h who m ? ) , c on su l t i n g c on t r ac t s, p ac k ag e
t r ansmi ssion (i.e., wh o i s deli vering p ackages to wh om? ), o r ch ar it able donation s (i.e.,
w h o i s gi vin g m on e y t o w h om? ) . An y o f the se m etho d s c an gen er at e r el at ion ships
between t wo or more peopl e wh ich t o geth er can rep resent th e soci al network f o r
t h e wh ole gr ou p .

B e yon d simpl y l oo kin g a t wh o i s co nne cte d t o w hom an d w ho i s centr a l in the g ro up,


ti es can in dicate th e stren gth an d di recti o n of a rel ati on ship, as well as if th e
r el at ion ship i s dir e ct o r in di re ct. To d r a w th ese inferences it h elp s t o un derst and
some of th e con cep ts an d impli c ations of t ies.

A r e t ie s str o ng or wea k? St ron g ti e s, in dic at ed b y a hi ghe r numb er , a r e


characterized b y f requen t interacti on, f ee lin gs of closeness, and multipl e typ e s of
rel at ion ships. Fo r exampl e, a stron g tie may provi de you with emot ion al support,
j ob- r el ate d in fo rm ati on, a nd a p er so n to g o see y our fav or it e s ci -fi mo vi e s wi th. On
th e oth e r hand , it al so r equi r es a go o d d e al of en e rg y t o maint ain. Wea k t ie s m ay
n ot provi de as much soci al support but, si n ce they are easier t o maintain, you can
h av e m an y m or e of them. We ak tie s c an be crit ical for inn ovati on. For ex ample, a
r e se a r che r is mu ch more likel y to learn about a lin e o f r el e v an t re s ear ch i n an
o the rw is e un rel at ed fie ld f rom a c as u al acqu aint an ce th an fr om a g oo d f ri end . Thi s i s
be c au se g ood f rien d s ten d to h a ve a c c es s t o the sam e inf orm ati on whereas casu al
a cqu ain tan ce s t en d to o ff er n ew inf o rm ati on.

In bu sin e ss set tin gs, it is imp ortan t t o h a v e a go o d b al an c e b e t w een str on g and we a k


ti e s. T oo m an y s tr on g ti e s – a lw ay s g oin g t o th e s am e set of p eopl e f o r inf or ma ti on
or advice, m a y limi t access to import ant n ew in form ati on. In ou r fi rst example of th e
m an a gem ent con sult ing comp an y, the cl ient t e am h a d been int e ractin g m ainl y with
ex istin g trusted conn ecti on s wh o di dn’t know abou t the oth er t e am’ s rel at ion ship
with th e sam e cl ient becau se th e other t eam came f r om a diff erent p art of their
c om p an y.

A r e t ie s r ec i pr oca l? So me of our r el ati on ship s ar e n aturall y two-way. If I meet


with you, then you are p rob abl y al so meetin g with m e. H owever, m any rel ati on ship s
c an b e one - w a y and t h e d i r e cti on a l i t y c an p ro vid e addit ional inf o rm atio n. Fo r
ex amp le, I may respect you but that doesn’t me an th at y ou ne c es s a rily r e spe ct me .
In a n etwork di agram , the arrows in dicat e directi onalit y.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 6
B

C
Th e t i e b et w e en A an d
B is recip r ocal. Th e tie
be tw een A a nd C i s n ot
A r e cip ro c al ; A g o es t o C
but C d o es n ot got t o A

Wh en a tie goes in both directi ons we thin k o f i t a s b e i n g r e cip ro c al . I t c an b e


imp o rt ant to kn ow when a rel ati onship i s reciprocated. For in st ance, a group will
g en er al ly fun cti on bette r when a ke y de ci si on-m ak e r i s not onl y s ou ght aft e r fo r
in formati on but h e or she al so seeks inform ati on f r om the g r oup. In g en er al ,
recip rocated ties t en d t o be st rong er th an non -reciprocat ed t ies.

A r e t ie s d ir ec t or i n di r ec t ? A dir e ct conn ecti on i s th e co nne cti on be tw een t wo


pe opl e. A s w e s aw in the p et ro leum co mpan y ex amp le, Co le h a d m any di r ect
conn ection s in th e net work of peopl e in exp lorati on and p rodu cti on. H e was ce ntral
in th is n etwork. Cent ral peopl e h ave m o re in fluen c e in their n et work, t end to
r e ce i v e b et t e r p er f o rm an ce re vi ew s, an d t en d to be m ore sati sfi e d with th eir job s
th an pe opl e w ho a re l es s c ent ra l.

A p e r son c an al so b e i n d i r e ctl y c on n e ct ed t o an oth e r p e r so n t h r ou gh kn o wi n g a


“ f rien d of a f ri end. ” Tra c kin g th e numb er of conne ct ion s it t ak e s f or one p e rs on to
reach an other i s one way to m onitor th e fl ow of inf o rm ation an d opportun iti es in
network. A person who i s indirectl y lin ke d t o m an y top lev el p eop le, i s m or e li kel y
t o get p ro mo te d th an so me one who i s not .

Th e rel at ed, p roverbi al “sm all world probl em ”, ref e rs to the likelih ood th at two
peopl e wh o are n ot di rect ly conn ected t o ea ch other will be abl e t o com municate vi a
in di re ct conne ct ion s. Due to r and om conn e ctio ns b et we en gr oup s o f pe opl e, pe opl e
who appear un conn ected m ay, in f act, be just a f ew relation ship s apart.

T ie s c an rep re sen t a r an ge of re lat io nsh ip typ es . Fo r exampl e, a ti e c an indi c at e i f


o n e p er s on likes, t r us ts , r e sp e ct s, r ep o rt s t o , c om m u n i cat es w i t h, o r ge ts inform at ion
from ano t h er . Sim i l a rl y, t i e s c an rep re sen t sells to , buy s f ro m , d e l iv e r s t o , c o nt r ac t s
w i t h o r co l l ab o r at e s w it h rel ati on ship s. Th e presen ce of a network ti e in dicat e s th at a
rel at ion ship ex ists. The ab sen ce of a net w ork ti e indicat e s th at th e relati on ship does
n ot exi st .

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 7
= m al e

C = f em ale

B
A A ti e exi st s between A and B but not
be tw een A and C.

Th is di agram, drawn f rom a stu d y of p articipant s in a worksh op illu strates w ho l ik e s


w hom (Borgatti et al., 1992 ). It al so illu st rates a common p att ern ; n amel y, th at
p e opl e wh o s h ar e a p a rti cu l a r ch a ra c t e ri sti c or attribute, in th is case gen der,
f requ entl y int eract m ore with each other th an they do with peop l e wi th whom th ey
d o n ot sh are a p a rti cula r ch a ra ct eri sti c or att ribut e. In the di a g ram the male s are al l
int e ractin g with each oth er as are th e fe males with on ly two p oint s of cont act
between them.

Actions following a network analysis


An SNA u s ed in a con sult ing settin g i s p rim arily di agn osti c of th e group being
stu died. To imp r ove the fl ow of in form at ion, kn owl edge, or decision s in th ese
g r oup s the SNA sh oul d be a c compa nie d b y acti on s. Th ese action s m ay be th e
resp onsibilit y of in dividu al s, th e group, m an agement or the larger enterp ri se.

An in creasin gl y popul ar use of SNA i s t o feed b ack results t o the enti re n et work in a
st ructu r ed settin g in whi ch discu s sion can be produ cti vel y f acili tated. Viewin g the
SNA di agram s se em s to be p a rti cul arl y p o we rful. For ex ampl e, in th e c a se o f the
man agement con sult ing comp an y, the resu lt s sur f ac e d h i d den b a r ri e r s b e t w een sen i or
and junior consu lt ant s. Revealin g these pre viou sl y hid d en b ar ri er s c on st ruct iv el y w a s
a po we rfu l to ol fo r cr e atin g n ew opp ort un iti es f o r cross-group rel ati on ship
d e vel opm ent. An SNA c an al so sugge st w ay s of re st ru ctu rin g or ganiz ati on al charts o r
imp lem enting new processes.

Sim il arl y, an SNA c onduc te d on a s et of or g aniza ti on s c an le t le a de r s kn ow whe re


th ei r fi rm stands in the soci al n etwork. L e ad er s may th en m ak e st rate gi c de ci si on s
a bou t wh ere an d w ith w hom t o dev el op clo s er bu sine s s rel ati on s.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 8
Application of SNA: The case of the management consulting
company
Th e m an agem ent con sultin g company ment ion ed at the b eginnin g of this art icl e
needed help imp r ovin g it s con sult ant s’ awaren ess of rel ated proj ect s. Bef o re
embarkin g on p otent ially costl y organiz ati on al chan ges an d other init iati ves, the
ex e cuti ve l ea d e rsh ip el ec te d t o c ond uct an SNA t o fin d how inf orm at io n w a s ( or w a s
n ot) gettin g p a ssed on. Th e SNA was condu ct ed with a group of con sult ant s f r om
diff erent geographi es, practi ces, levels of seniorit y, and t enu re in th e comp an y.

Q u e st i on s i n clu d e d: “Ho w aw a re ar e you of t h e p r oje ct s do n e b y t h i s p e r s on i n t h e


p a st 12 month s? ” and “H ow oft en d o e s thi s per s on p ro vi de y ou wi th inf o rm ati on y ou
ne e d to de ve lop cli ent p rop o sa ls ?” B y l o okin g at wh o was cent ral in the awareness
and inf orm ati on -sh arin g networks, th e comp any coul d see which p eop le the t e am
t ended to go to for inform at ion. B y l o okin g at h ow oft en pe opl e f rom ea ch gr oup
int e ract ed with each oth er, the SNA reveal ed th at con sult ant s who h ad b een with the
comp an y the lon g est ten ded to get in form ati on f r om e a ch o the r ; n e we r emp l oy ee s
only turn ed t o each f or m ore inf o rm ati on if they were l o cated in cl ose ph ysi c al
p roxi mit y. Fo ll ow -up int er vi ew s re v ea le d th at tim e p ressu res l eft m emb ers of the
t ea m with fe w oppo rtunit ie s t o d ev el op rel ationsh ips with newe r or m or e di stant
e m p l o y ee s.

As a resul t of th e SNA, upper management t ook several acti ons. They in sti g ated a
m en t o r s yste m t o h el p n e w e m p l o ye e s, spon s or e d e ven t s se v er al t i m e s a y e ar t h at
b rou ght p eopl e to g ether f rom di ffer ent p a rt s of th e comp an y, and de ve lop ed an
a w a rd p r ogram f o r an y n e w e n g a ge m e n t s t h at were won as a resu lt of ex istin g
rel at ion ships f rom out side th e immedi at e team.

SNA: A final word…


SN A i s a n ew app r oa ch t o sol vin g tr a dit ion al pr obl em s in bu sine s s and m ana g ement.
It assume s th at pe opl e ar e int er c onne ct ed , that conn ec tio ns h ave real con sequen ce s
f o r p e rf or m a n c e and s at i sf a ct i on, an d t h at connect ion s can be st ru ctured t o optimi ze
in di vi dua l, gr oup, an d or g ani za tion al out com es. SNA responds t o the growin g
awaren ess th at som ethin g int an gible is in dan ge r of b ein g l o st as th e m a rk etpl ac e ( an d
t h e w or kp l ac e ) b e co m es i n cr e as i n gl y d i sp e rs e d.

Th e int erp erson al connect ion s th at we on c e t oo k f o r g r ant e d a re n o l o n g e r an


assu red p a rt of everyday lif e. SNA pl aces a premium on these rel ati on ship s and their
rol e in th e way we do bu siness.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 9
Appendix
SNA concepts
So ci al n et wo r k ana ly si s i s an e st abli she d subd isc iplin e o f ma ny a c ad emic fi eld s,
in clu din g man agem ent, so ci al p sycho lo gy, and so ci ol o gy . As a r esu lt, re se ar ch h as
al ready reveal ed rel ationsh ips that may be h elpfu l to p r acti tion e rs. Below are three
ex amp l e s of k e y SN A co n c ept s. To l e a rn m o re ab out S NA , ref e r t o t h e r ef er en c e s
li st ed at the en d of thi s Appen dix.

C e ntra l it y. Cent ralit y i s th e ext ent t o wh ich a person i s in th e cent er of a n et work.


Cent ral peopl e have more in fluen c e in th ei r network, ten d t o recei ve better
performance revi ews, an d ten d t o be more sati sf ied wi th their job s than peopl e wh o
a r e l es s cent r al. Me a sur e s of cent ra lit y in clu de d e g re e, betweenness an d cl o sen e ss
cent ralit y.

P e r son A i s t h e m o st
cent ral in the n et work

S oc ia l c a p ita l. So ci al c ap it al r ef ers t o the t otal sum of po tent ial o r a c tua l


r e s ou r ce s t h at a p e r son ac c rues as a result of int e rpersonal rel ati on ship s. Th is
t r ans la te s int o bein g abl e to g et f av o r s g r ant ed , gett ing r api d r e spon se to requ e st s,
o r h a vin g a cc e s s to influen ti al pe ople . So ci al c apit al m a y ari se fr om h av in g a d ens e
network of st ron g support or bein g in a p rivileged rol e with respect to oth e r peopl e,
r ef e rr ed to a s s tru ctu ral h ole (see below).

S tr uc t ura l h o l e s. A st ructural h ol e occurs when ever a person (a) h as a rel ati onship
w ith som e one wh o i s co nne cte d t o a s ep ar at e clu ster of peopl e and (b ) has n o other
d i r e ct o r i n d i r e ct c on n e ct i on with th e people in th at clu ster. In ef fect, the person i s
c onn ec te d to a di stin ct g r oup vi a one p e rs on.

A h as t h e m o s t
st ructu r al holes in th e
A network

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 1 0
A pe r son wh o ha s a l a rg e numb e r o f st ructu r al h ole s in h is o r h er n etw o rk i s l ik el y
t o b e ex p o se d t o m or e d i v e rs e i n f or m ati on an d opportunit ies th an a person who has
r el at iv el y few st ru ctu ra l h ole s in hi s o r h er n etwork. For thi s reason , a person with
a l arge number of st ructu r al holes is more li kely to b e p rom oted, develop inn ovati ve
p ro du ct s, an d t o g ai n p o w er an d i n f l u en ce b y actin g as a broker between otherwise
un conn ected groups.

Brokerage. A pe r son w ho conn e ct s t wo other wi s e unc onne ct ed p e opl e i s in a


p ositi on to m an a ge or “b roke r ” i n f o rm at i on f l o w. H o w h e o r she cho o s es t o do t h at
c an b e re vea le d th r ough an SNA. Fo r exa mple , Bi ll m ay b ro ke r th e r el at ion ship
be tw een s ale s rep s an d p ro du ct d eve lop er s b y recei vin g inform at ion from sales rep s
a nd p r ov idin g it t o p rod uct d ev el ope r s but n ot vi ce ve r s a. C on ve r se ly , B ill m a y
r e ce i v e an d p ro vi de i n f or m ati on t o b ot h sa l e s rep s an d p r od u ct d ev el op e r s.
Mo r eo ve r, B ill m a y be a s a le s r ep him s elf or he ma y b e a p ro du ct d eve lop er o r he
coul d b e in an unrel at ed group, such as di st ributi on. The di vi si on to whi ch he
belon g s h as imp licat ion s f o r the role h e pl ays (e.g., in si der, out si de r, e tc .). Br o ker a g e
diff ers from st ructu r al h oles in th at (a) it does n ot assum e th at group s are
un conn ec ted an d (b) it m or e cle a rly d e sc rib es th e fl ow of in fo rm ati on.

Common SNA statistics and measures


B e cau s e SNA r ej ect s the a s sumpti on of in dep en dent ob se rv a tion s un de rl yin g
t r adi tion al att ribut e-b a se d re s ea r ch, se v er al SNA - sp ec ifi c st ati sti c s h ave b een
d e vel op ed. So me of th e m o st c ommo n a r e li sted bel ow . A c omp r ehens i ve l ist of
SNA-speci fic stati sti cs can b e f ound in th e U CI n et ref e ren c e gui de ( see r ef e r en c es ,
below).

C e ntra l it y. T h e m ost c om m on m ea s u r e o f c en t r al i t y i s de g re e cen t ral i t y and i s a


simpl e t all y of th e number of p eople at tach ed to each person. A person with three
r e cip ro c al re la tion ship s, fo r ex ample , ha s a "d eg r ee " of 3. Oth er m e asu re s of
c en t r al i t y i n c l u d e b e t w eenn e s s an d c l o s ene s s.

Bet weenness cent ralit y in di cates the ext ent t o which a n ode lies on the sh ortest path
between every other pai r of no des. For ex ampl e, if Adam i s conn ected to Bill and
Bill i s connect ed to Cin dy but Adam i s n ot conne ct ed to C ind y, th en Ad am mu st go
“ t h rou gh ” B i l l t o “ re a ch” C i n d y b e ca u se B i l l i s b e t w e e n A da m and Cin d y. T h e p e r s on
with th e h ighest b et weenness cent ralit y i s th e person that others in the n et work
m o st c om m o n l y m u s t go t h r ou gh i n or de r t o r e a ch e a ch o t h e r.

A A i s hi gh in betweenn ess cent ralit y.


B
B is h igh in degree centralit y.

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 1 1
Closeness cen tral ity is th e number of lin ks th at a p e r son m u st go t h r ou gh i n or der t o
r e a ch e ve r yo n e el se i n t h e n et wo rk . I f A d am i s conn ec ted to Bill and Bil l i s
conn ected to Cin d y but Adam is n ot conn e cted t o C ind y, then Ad am mu st go thro ugh
t wo l inks (i.e., Adam -Bill an d Bil l-Cin dy) in or de r t o re ach C ind y. The pe r s on with
t h e h i ghe st c l o s ene s s cen t r al i t y s co r e i s t h e p e r s on wh o g o e s t h r ou gh t h e f e we st
numb er of ti es to reach everyone el se in the network.

D e n si ty. D en sit y captu re s h ow cl o sel y a g r oup or sub gro up i s knit . It is a


p rop orti on th at in di cates th e number of actu al ti es present in th e group rel ati ve t o
th e number of p ossib le ti es in the group (i . e., i f everyone had a relat ion ship with
everyone else in the group). Den sit y can b e cal cul ated within a group or b etween
t wo group s. Wh en cal cul atin g th e den sit y of an in di vidu al ’s net w ork, SNA looks at
h ow cl os el y c onn ec te d a pe r son ’s fr ien d s ar e to e a ch oth er .

C o he s i on . Th er e ar e s e ve r al m e as ur e s o f c ohe si on, in clu din g d ens ity . H ow ev er,


o n e co m m on m ea su re i s t h e a ve r a ge n u m b e r o f t i e s i t t ak es f or a p er so n i n t h e g ro u p
t o “reach ” another person in the group. If A da m i s conn ec te d t o Bi ll w ho i s
conn ected to Cin d y, then Adam i s at a d ist an ce of 2 fr om C ind y. The av e rage
di st ance f or th e group gi ves an indi cati on of the group’ s coh e sion .

S ubgr o up Identifica tion. S N A s can i dent i f y t h e n u m b e r o f clo s el y kni t sub g rou p s


or “cliqu es” in a network. Within a clique, every unit i s connected t o every other
u n i t . The s e s u b gr ou p s c an t h en b e a n aly z e d t o s e e i f t h e y s h ar e o ve rl a p p i n g
m embe r s. A net w or k th at cont ain s hi ghl y se g re g at ed sub gr oup s i s n ot a s w ell
int e gr at ed as a net wo rk in whi ch ind iv idu al s b el on g to s ev er a l o ve rl appin g sub gr oup s.

References
Borgatti, S.P., Bernard, H. R., and Pe lto, P. 1992. NSF Summer Institute on
Ethn ographic Research Meth ods. Av ail able from An al yti c Techn ol o gies
w w w. ana lyt ic te ch. co m

Brass, D. (2000 ). Achi evin g Success th rou gh So ci al Capit al: Tapp ing Hi dden Resources
i n you r Pe r so n a l an d B u s n e s s Net wo r k s. J os s e y - B a s s.

Cross, R. an d Parker, A (2004 ). Th e Hi dd en Power of So ci al Networks. Harvard


U n i ve rs i t y Pr e s s

Cross, R., Parker, A. , Prusak, L. & Borg atti, S.P. 2001. Knowin g Wh at We Know:
Supp ort ing Kn owl ed g e C r e ati on and Sha rin g in Soci al Networks. Organiz ati on al
Dyn ami cs 30 (2): 100 -120. [p df]

Cross, R. an d Pru sak, L. (2002 ). Th e Pe ople Wh o Ma ke Organiz at ion s Go—or St op.


H a rv a rd B u s in e s s R e v ie w 80 (6), pp 104 -112

Cross, R., Li edt k a, J. and Weiss, L. ( 2005). A Practi cal Guide To Soci al Ne tworks.
H a rv a rd B u s in e s s R e v ie w .

Hanneman , R. an d Ri ddle, M. (2005 ). In troduct ion to social network


methods. Ri versi de, CA: Universit y of Cal iforni a, Ri verside (publi shed in digit al
f o rm at http: //f acult y.ucr.edu/~h anneman /)
Inside Social Network Analysis
Ehrlich and Carboni
p a ge 1 2
K rackhardt, D an d Hanson , J. (1993 ). Info rmal n etworks: Th e comp an y behin d the
c h a rt . Harvard Busine ss Review, 71 (4 ), pp104-111

Borgatti, S. and Foster, P. (2003 ). Th e ne t w ork p a radigm in organiz ational research:


A r e vie w and t yp ol o gy . J o u r nal o f M a n a ge m e nt 29(6 ), 991 -1013.

Wasserman, S and Fau st, K. (1994). Social Network Analysis: Methods and Applicat io ns.
C a m b ri d ge : C a m b ri d ge U n i ve rs i t y Pr e s s

Fo r acc e ss to UCI Net so ft war e (the mo st comm onl y used software for SNA) and
a s s o ci ate d ar ti cle s: www . ana lyt ict ec h. com

Fo r in fo rm ati on abou t In Fl ow an al ys i s t ool s, a rti cl e s an d d ia g r am s:


http ://www.orgn et. com /in flow3.html

Inside Social Network Analysis


Ehrlich and Carboni
p a ge 1 3