Social networkers: Measuring and examining individual differences in propensity to connect with others

Social Networks 30 (2008) 283–296

Contents lists available at ScienceDirect

Social Networks

journa l homepage: www.e lsev ier .com/ locate /socnet

Social networkers: Measuring and examining individual differencesin propensity to connect with others

Peter Totterdell ∗, David Holman, Amy HukinInstitute of Work Psychology, University of Sheffield, Sheffield S10 2TN, England, United Kingdom

a r t i c l e i n f o

Keywords:Individual differenceEmployee networkExpertiseAdvice

a b s t r a c t

The research examined individual differences in people’s propensity to connect with others (PCO). A mea-sure of PCO, with components for making friends (strong ties), making acquaintances (weak ties), andjoining others (bridging ties), was developed and tested in two studies involving 144 undergraduates and197 health-care employees. PCO and its components were significantly positively associated with social

Well-being network characteristics (including size, betweenness centrality, and brokerage) and indicators of per-sonal adjustment including support received, attainment, well-being, influence, and suggestion-making.PCO had effects beyond those of major personality traits, and PCO components displayed distinctive

etwo

tmswosieaT

hIhikRhscRttoo

aotias

swrFctra

1

foe

0d

relationships with work n

Some people seem more inclined than others to make connec-ions with people they do not know. An individual’s propensity to

ake new connections may affect the extent and value of his or herocial network. The effect of individual characteristics on social net-orks is important because empirical studies of the consequences

f networks have discovered associations between the number,tructure, strength and content of network ties and a range ofndividual and organizational outcomes, including personal influ-nce, job performance, innovation, career success, satisfaction andffect (Brass, 1984; Flap et al., 1998; Forret and Dougherty, 2004;otterdell et al., 2004).

Research on social networks within and between organizationsas expanded greatly in recent years (Borgatti and Foster, 2003).

n relation to organizational behavior, social network researchas addressed a range of issues such as social capital (the value

ndividuals derive from their connections), group processes, andnowledge utilization (Brass et al., 2004; Gulati et al., 2002;aider and Krackhardt, 2002). Although social network researchas examined the influence of observable individual attributes,uch as gender, it has largely ignored the individual psychologi-al characteristics that may shape personal networks (Kalish andobins, 2006; Mehra et al., 2001). This is due, in part, to the fact
hat social network research has been principally concerned withhe structure and effects of relations between people, groups orrganizations (Brass et al., 2004; Tichy et al., 1979), rather thann psychological attributes of the individual. Nevertheless, such
∗ Corresponding author.E-mail address: [email protected] (P. Totterdell).

opmvaaea

378-8733/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.socnet.2008.04.003

rk characteristics.© 2008 Elsevier B.V. All rights reserved.

ttributes are likely to contribute to the formation and maintenancef ties between people within networks, and will thereby influencehe behavior of those networks. This conception, in which bothndividual agency and social structure determine action, offers anlternative to a strict structural perspective in which action derivesolely from the structure of social networks (Obstfeld, 2005).

We adopted this alternative conception in an effort to under-tand the nomological network of social networking. In particular,e examined how individuals’ propensity to connect with others

elates to their social network structure and personal adjustment.or this purpose, we developed an individual difference measureoncerning propensity to connect with others (PCO), and conductedwo studies in different settings to examine how this characteristicelates to a range of constructs pertaining to social networks anddjustment.

. Individual differences and social networks

Researchers have begun to examine the role of individual dif-erences in shaping social networks, particularly in relation torganizational networks. One approach to this topic has been toxamine whether the centrality of a person’s position within anrganizational network depends on particular traits of his or herersonality. For example, using Goldberg’s (1990) Big Five factorodel of personality, Klein and colleagues found that neurotic indi-

iduals were less likely to acquire central positions in friendshipnd advice networks (Klein et al., 2004). A number of studies havelso found a positive association between extraversion and thextent to which individuals engage in networking behaviors (Forretnd Dougherty, 2001; Lee and Tsang, 2001; Tziner et al., 2004;

http://www.sciencedirect.com/science/journal/03788733

http://www.elsevier.com/locate/socnet

mailto:[email protected]

dx.doi.org/10.1016/j.socnet.2008.04.003

2 Netw

Wpi(

sphtbabRdsfwci

raeapt(msicvt

fvn

taipieuMbpop

2

ctflriisbtsiotmw

84 P. Totterdell et al. / Social

anberg et al., 2000). Other individual differences that have beenositively associated with social network size include emotional

ntelligence (Austin et al., 2005) and self-monitoring orientationMehra et al., 2001).

Using a different method, Burt et al. (1998) derived the per-onality profile of “network entrepreneurs” by identifying theersonality characteristics of individuals who possessed structuraloles in their work networks (i.e., networks where the potentialo broker connections is high because there are few connectionsetween others in the network). Based on those personality char-cteristics, they concluded that network entrepreneurs prefer toe in authority, create excitement, and change things. Kalish andobins (2006) have also identified individual differences that pre-ispose people to structure their social environment by sustainingtructural holes or by seeking network closure. For instance, theyound that individuals who sustain structural holes in their net-orks tend to be more individualistic, whereas those who have

losed networks of strong ties tend to be more extraverted and lessndividualistic.

However, Becker (2004) has argued that results regardingelations between personality and network behavior have usu-lly been weak because personality does not have strong directffects on behavior. Becker proposes that the effects of person-lity on network building are likely to be mediated by moreroximal motivational antecedents, including attitudes, subjec-ive norms and perceived control over network building. Kadushin2002) has also emphasized the association between individual

otivations and network behavior. In particular, he argued thatafety and efficacy drives are rooted in network experiences dur-ng early childhood, and are, respectively, associated with havingohesive networks and networks with structural holes. These moti-ational perspectives are promising but have yet to be empirically
ested.
Another approach, and the one we have chosen to adoptor this investigation, involves identifying characteristics of indi-iduals that map more directly onto the formation of socialetworks, such as people’s desire and tendency to make connec-

wpbF

Fig. 1. Nomological network of propensity to connect with othe

orks 30 (2008) 283–296

ions with other people. Given that people form social networks inrange of contexts throughout their lifespan, differences between

ndividuals in their propensity to make connections may haveervasive personal and social effects, which extend beyond the

ndividual’s own network. Although previous researchers havexamined aspects of individual networking behavior, they havesually focused on people’s actual organizational interactions (e.g.,ichael and Yukl, 1993), the types of organizational networking

ehavior they utilize (e.g., Forret and Dougherty, 2004), or theirurpose in having connections (Ferris et al., 2005), rather thann their general disposition for making connections with othereople.

. Research hypotheses

We defined PCO as an individual’s orientation towards makingonnections with other people that is not specific to context andhat incorporates three related but distinct components: makingriendships, making acquaintances, and joining others. In the par-ance of social network research, these three components equate,espectively, with the formation of strong ties, weak ties, and bridg-ng ties in a network. Our proposed nomological network for PCOs shown in Fig. 1. Specifically, we expect people’s PCO to relate topecific aspects of their personality, to be expressed behaviorallyoth through how they appear to others and in the structure ofheir social networks, and to influence how they adjust to theirocial context. Our investigation tested seven hypotheses concern-ng aspects of these relations. Fig. 1 shows how the hypotheses mapn to the nomological network. The first three hypotheses wereested in relation to student networks within a university environ-

ent and the last four were tested in relation to employee networksithin a work environment.

An initial question concerning PCO that needed examining washether it is simply an alternative form of an already recognizedersonality trait. Earlier in this section we described how a num-er of research studies have found associations between the Bigive personality factors and network behavior. PCO seemed most

rs (PCO) indicating where the tested Hypotheses 1–7 fit.

Netwo

ltspoaktbct

Hws

l(Cpctc(a(

Ho

aeaahaep(e1aaapaft

Hta

PwPwoevr

imb

aw

Hrn

wsiab(tntewFspkisaGnnalosa

wbmthbwtfaiW

Hbbb

io(trfso

P. Totterdell et al. / Social

ikely to overlap with the extraversion factor of the Big Five becausehey both depend on having an outward sociable disposition. Inpite of this overlap, it was our belief that PCO would have greaterredictive validity than extraversion in accounting for the sizef a person’s friendship network because it is more focused onperson’s orientation towards forming network ties. This is in

eeping with a general trend away from studying broad traitsoward studying more specific behavioral orientations becauseroad traits are often too diffuse to capture specific behavioral out-omes (Kalish and Robins, 2006). Hence our first hypothesis washat:

ypothesis 1. PCO will be positively related to extraversion, butill be a better predictor of friendship network size than extraver-

ion.

People’s attempts to make connections with others seem moreikely to succeed if the affective characteristics of what they expressverbally and nonverbally) are appraised as positive by others.ross et al. (2002), for example, found that individuals who wereerceived to be more energizing by their colleagues were moreentral in their organization’s energy network and performed bet-er in their job. In the circumplex model of affect, energy is mostlosely aligned with markers of positive affect such as enthusiasmRemington et al., 2000) and we therefore proposed that individu-ls with PCO would appear to others as having positive affectivitye.g., enthusiasm).

ypothesis 2. Individuals with greater PCO will be perceived bythers as having greater positive affectivity than other people.

Expressive displays are usually regarded as more authentic,nd are therefore more effective, when affective experiences andxpressions are aligned (Diefendorff et al., 2005). Hence, as wells expressing positive affect, individuals with greater PCO maylso be more inclined to experience it. They are also likely toave more opportunities for obtaining rewarding positions withinnetwork, and hence accrue other advantages (Burt, 2004). For

xample, previous research has found that the size and reach ofeople’s networks are related to the social support they receiveHanson and Ostergren, 1987), their positive affect (Totterdellt al., 2004), their well-being (Cattell, 2001; Cohen and Wills,985), their social adjustment (Hays and Oxley, 1986; Riggio etl., 1993), and their performance (Flap et al., 1998). Individu-ls with greater PCO should therefore exhibit greater personaldjustment. In particular, they should receive more emotional sup-ort, experience greater affective well-being, and show greaterttunement to the demands of their specific environment in theorm of social adjustment and attainment. Thus, we hypothesizedhat:

ypothesis 3. PCO will be positively related to received emo-ional social support, affective well-being (enthusiastic, relaxed),cademic adjustment, and academic attainment.

Our next three hypotheses concerned how the components ofCO would relate to characteristics of social networks within aork context. Specifically, we expected that each component of

CO (making friends, making acquaintances, and joining others)ould give rise to specific characteristics within a particular type

f employee network. Furthermore, we anticipated that the influ-nce of PCO would be additional to the influence of other structuralariables such as gender, organizational tenure, and leadership
ole.
With respect to propensity to make friends, it seemed likely thatndividuals who are more inclined to make friends would attract

ore ties in a friendship network. Friendship networks containoth strong and weak relationships but they are primarily directed

wpteO

rks 30 (2008) 283–296 285

t the ties that arise from the solidarity of close affiliations. Hence,e hypothesized that:

ypothesis 4. Propensity to make friends will be positivelyelated to the size of an employee’s friendship network (i.e., theumber of employees who report them as friends).

In contrast, instrumental actions tend to be accomplished viaeak ties. Advice relations can arise from both instrumental and

olidarity contacts because there may be a personal cost involvedn asking for advice (Flap and Volker, 2001; Klein et al., 2004), butn extensive advice network seems most likely to be facilitatedy a propensity to make acquaintances. A tenet of Granovetter’s1973) strength of weak ties theory is that weak ties are more likelyhan strong ties to provide bridges to other segments of the socialetwork. People seeking advice may therefore be likelier to obtainhe information they do not have through their weak ties. How-ver, propensity to make acquaintances may need to be combinedith relevant expertise in order to attract others seeking advice.

or example, Borgatti and Cross (2003) found that the likelihood ofeeking information from another person depended on the valuelaced on the person’s knowledge, as well as on access to thatnowledge. In the same vein, consumer research has identified thenfluential role of individuals known as market mavens, who act as aignificant source of advice for others because they actively acquirend share market knowledge (Feick and Price, 1987). According toladwell (2001), mavens generally provide the message and con-ectors (i.e., individuals with high PCO) spread it within a socialetwork, but occasionally a person has the characteristics of bothconnector and a maven and a connector is therefore particu-

arly influential in spreading ideas and behaviors. A combinationf individual propensity to make acquaintances and work expertisehould therefore be positively associated with the size of a person’sdvice network.

The strength of weak ties argument means that individualsith propensity to make acquaintances should also have greater

etweenness centrality in an advice network. Again, however, theyay attract more ties from others if they have relevant exper-

ise themselves. We therefore anticipated that individuals wouldave greater betweenness centrality in advice networks if they hadoth a greater propensity for making acquaintances and greaterork expertise than other employees. Weak ties can also provide

he necessary conduits for information passing into a work grouprom employees who are external to that group, and hence wenticipated that employees who had greater propensity for mak-ng acquaintances would also have greater gatekeeper brokerage.

e hypothesized that:

ypothesis 5. Propensity to make acquaintances will, when com-ined with work expertise, be positively related to the size andetweenness centrality of an employee’s advice network, and wille positively related to advice gatekeeper brokerage.

Individuals who are inclined toward joining people by introduc-ng disconnected others or by facilitating action between connectedthers have recently been described as having a tertius iungens“third who joins”) orientation (Obstfeld, 2005). Obstfeld noteshat the activity of joining people across bridges has received littleesearch attention, even though such activity may be a foundationor social skill and the root of collective action. A person’s propen-ity for joining others should correspond to the likelihood that her she serves as a bridging link between other people in a net-
ork. This may be especially apparent in an advice network wheneople seek advice from someone who either brings them into con-act with a person with the information they need or knows wholse advises people on that issue (Krackhardt and Kilduff, 2002).ur expectation was that propensity to join others would have

2 Netw

dneoe2rt

Hwcr

emstedvfa

ppe1fwvtssHto

toartrotit(iathetepst

Haas

3

oowaPtibfhtn

PstazTststowpnwtr

4

4

4

wfpof(

4

sfitaotamoh


irect associations with advice network size and advice between-ess centrality, but the associations would be enhanced by workxpertise. Individuals who perform a gatekeeping role within anrganization provide a special kind of bridge that for many oth-rs serves as their primary link to external information (Obstfeld,005). Hence, we also expected that propensity to join others wouldelate to advice gatekeeper brokerage. We therefore hypothesizedhat:

ypothesis 6. Propensity to join others will, when combined withork expertise, be positively related to the size and betweenness

entrality of an employee’s advice network, and will be positivelyelated to advice gatekeeper brokerage.

Our final hypothesis concerned how PCO would relate tomployees’ feelings and involvement at work. In particular, we esti-ated that if PCO confers the aforementioned (as well as other)

tructural advantages to employees, then it should also contributeo the affective well-being they experience from their job andnhance their involvement within the organization. As previouslyescribed, the size of people’s social network has been related toarious aspects of well-being, including positive affect and satis-action at work (Cohen and Wills, 1985; Flap et al., 1998; Kawachind Berkman, 2001; Totterdell et al., 2004).

Structural theorists have argued that individual influence (orower) in an organization is determined by an employee’s networkosition because it is the network of relationships among employ-es that governs the control and distribution of resources (Brass,984; Ibarra, 1993; Thompson, 2005). In support of this, it has beenound that employees’ influence is related to their centrality inorkflow and interaction networks, as well as to other structural

ariables including being in a position of authority, organizationalenure and gender (Brass, 1984, 1985; Ibarra, 1993). It has also beenhown that proactive employees pursue initiatives by developingocial networks that increase their influence (Thompson, 2005).ence, an individual’s influence on decisions within an organiza-

ion will depend partly on his or her ability to develop a networkf relationships, which will be facilitated by their PCO.

Another indicator of employees’ influence in organizations isheir involvement in innovation (Ibarra, 1993). Obstfeld (2005)bserved that organizational innovation is often a process of cre-ting new social connections between people and the ideas andesources they carry, and hence depends on individuals’ orien-ations towards action in the network. In accordance with thiseasoning, he found that individuals who had a tertius iungensrientation had higher levels of involvement in innovation. Weherefore expected that PCO would also be related to innovationnvolvement, especially the PCO component for joining others. Inerms of what kind of innovation involvement can be expected, Burt2004) argued that individuals who have connections across groupsn an organization have more options to select and synthesize fromnd are therefore more likely to suggest new ideas. As evidence ofhis, he found that employees whose networks spanned structuraloles in an organization were more likely to express ideas that werevaluated as valuable. However, Burt could find little evidence thathe ideas were acted upon, and therefore concluded that employ-es suggested ideas to display competence rather than change workractices. Consequently, we expected that PCO would relate to theuggestion of ideas but not to the implementation of ideas. Weherefore hypothesized that:

ypothesis 7. PCO will be positively related to job-relatedffective well-being (feeling enthusiastic and feeling relaxed)nd organizational involvement (decision-making influence anduggestion-making).

ttps

orks 30 (2008) 283–296

. Overview of studies

Two studies were conducted to test these hypotheses. The aimsf the first study were to (a) evaluate the psychometric propertiesf PCO, (b) examine how it relates to personality traits, social net-orks, and appearance to others, and (c) assess whether it helps

ccount for personal adjustment. In other words – assuming thatCO could be identified – we wanted to know whether it is an orien-ation that is distinct from personality traits, whether it translatesnto the networks that individuals form, whether it has detectableehavioral manifestations, and whether it has likely consequencesor the individuals themselves. This involved testing the first threeypotheses. To achieve this we studied undergraduate students sohat PCO could be examined in individuals who had recently formedew social networks.

The aims of the second study were to extend understanding ofCO by: (a) retesting its psychometric properties in an employedample, (b) examining how it relates to the structural parame-ers of individuals’ personal networks in a work context, and (c)ssessing whether it relates to job-related well-being and organi-ational involvement. This involved testing the last four hypotheses.o achieve this we sought networks that individuals were able tohape for themselves (that is, where they had choice in the connec-ions they made) and networks that arose from a shared contexto that networks could be compared. Hence, we chose to studyhe friendship and advice networks of employees within a singlerganization, i.e., who employees were friends with and who theyent to for work advice within the organization. To preclude theossibility that discovered associations between PCO and personaletworks could be due to individual differences in self-report style,e used the friendship and advice nominations of others within

he organizational network (i.e., in-ties rather than out-ties) andelated those nominations to self-reported PCO.

. Study 1

.1. Method

.1.1. ParticipantsThe sample consisted of 144 undergraduate students. There

ere 104 females and 40 males in the sample, and their ages rangedrom 18 to 34, with a mean age of 19.82 years (S.D. = 1.77). Of thearticipants, 84 were in their first year at university, 20 in their sec-nd year, and 40 in their third year. The participants were studyingor a variety of degrees but the majority were psychology studentsn = 96).

.1.2. ProcedureThe volunteers were recruited in two waves of data collection,

eparated by 1 year. There were 64 participants recruited in therst wave and 80 in the second. All volunteers were informed thathey would be participating in a study examining social networkst university and all completed a questionnaire containing a batteryf self-report measures. The measures we used from the ques-ionnaires were presented in both waves of data collection, exceptffective appearance and emotional social support which were onlyeasured in the first wave and academic attainment which was

nly measured in the second wave. These three measures thereforead a reduced sample size.

The first wave of data collection involved forming interac-ion groups prior to questionnaire completion in order to obtainhird-party ratings of affective appearance. To form these groups,articipants chose a time slot when they could take part in thetudy. A minimum of three participants and a maximum of four

Netwo

patTtdutrpot

44itejmwtdwc

4tstaefeBtcsmaitntwms

4wpi

pwttiaAfWct

fo(t

4cgprcgtpi“eworaTwor

4weaw1h“ae

4v(mW2fpir(


articipants were allowed per time slot, and no friends took partt the same time. The participants were seated in pairs. One par-icipant had to wait if there were only three people in the group.he pairs randomly picked one of five cards on which was written aopic (food, holidays, hobbies, home, and animals) that they were toiscuss for 2 min. These particular topics were chosen as affectivelynbiased experiences that would be common to all participants. Athe end of the 2 min, each participant privately completed someatings concerning the other person’s affective appearance. Thisrocedure was repeated until the participants had rated each of thethers in their group. Pairs selected topic cards until they found aopic that neither person had previously discussed.

.1.3. Measures

.1.3.1. Propensity to connect with others. The scale consisted of ninetems developed for the purpose of the study.1 PCO was designedo measure three components of propensity to connect with oth-rs: making friendships (˛ = .85), making acquaintances (˛ = .65), andoining others (˛ = .75). Each component had three items, which

easured the extent to which participants have such ties, the easeith which they make such ties, and their attraction to making such

ies. Each item had a 5-point response scale ranging from 1 (does notescribe me very well) to 5 (describes me very well). The responsesere averaged to give a score for overall PCO (˛ = .85), as well as

omponent scores. The full set of items is shown in Appendix A.

.1.3.2. Personal networks. In the first wave of data collection, par-icipants were asked to report the number of people who they hadome type of social interaction with in the previous 2 weeks thathey considered to be close friends from (a) their course, (b) theirccommodation, and (c) other places at university. These threestimates were summed to measure the size of their universityriendship network. This method has been used in other studies tostimate the sizes of different parts of individuals’ networks (e.g.,erkman and Syme, 1979), but has the drawback that specific con-acts do not have to be recalled. So, in the second wave of dataollection the measure for deriving participant’s social networktructure was adapted from Burt’s (1992) network survey instru-ent. This required participants to record the initials of people

t university with whom they had discussed a matter of personalnterest in the last 4 weeks. Participants then indicated which ofhose contacts they considered to be close friends. Size of friendshipetwork was calculated by counting those contacts. A two-tailed-test showed no significant difference, t(142) = .37, n.s., in the net-ork sizes produced by the first method (M = 11.13) and secondethod (M = 10.90). A single measure was therefore used for friend-

hip network size.

.1.3.3. Personality traits. The personality factors of the Big Fiveere measured using the Mini-Marker Set (Saucier, 1994). Partici-ants used a 9-point response scale, which ranged from 1 (extremely

naccurate) to 9 (extremely accurate), to rate how accurately each of

1 Gladwell (2001) described a method for identifying individuals who have a highropensity for making connections with others—which involves presenting peopleith a list of 250 surnames chosen at random from a phone directory and asking

hem to add up the number of people they know who have those names. In informalests, he found that individuals who score highly on this measure have the behav-oral attributes of connectors and appear to play an important role in helping ideasnd behaviors spread rapidly and extensively through very large social networks.lthough this technique is appealing, it is likely to be susceptible to bias arising

rom variations in the frequency of different surnames in different communities.e therefore chose to develop a self-report measure of an individual’s propensity to

onnect with others that was independent of situation (i.e. non-context specific) andhat could be easily administered within a battery of other questionnaire measures.

stics

4usrmma

4wp

rks 30 (2008) 283–296 287

orty traits described themselves. The scores for the eight itemsn each factor were averaged to produce measures of extraversion˛ = .83), agreeableness (˛ = .80), conscientiousness (˛ = .86), emo-ional stability (˛ = .84), and openness (˛ = .74).

.1.3.4. Affective appearance. Participants in the first wave of dataollection were rated by the other members of their interactionroup (see Section 4.1.2) concerning the extent to which the partici-ant displayed certain affective characteristics. The group membersated the extent to which they thought participants possessed eachharacteristic on a 5-point response scale from 1 (not at all) to 5 (areat deal). The terms enthusiastic and relaxed were chosen fromhe measure for affective well-being to reflect the appearance ofositive affect and (low) negative affect, respectively. The specific

tems were “how enthusiastic do you think this person is?” andhow relaxed do you think this person is”. The item “how knowl-dgeable do you think this person is?” was also included to assesshether knowledge was more salient than affect in the appearance

f individuals with propensity to connect. The scores from differentaters were averaged to produce a single score for appeared enthusi-stic, appeared relaxed, and appeared knowledgeable per participant.o enhance reliability, scores were only retained for participantsho received two or more ratings for each characteristic, which

ccurred for 58 participants (32 had three raters and 26 had twoaters).

.1.3.5. Emotional social support. This was measured in the firstave of data collection using the five-item subscale for daily

motional support from the social support questionnaire for trans-ctions (Suurmeijer et al., 1995). Participants rated the frequencyith which they received emotional support on a 4-point scale from(seldom or never) to 4 (often). Example items were “Does it ever

appen to you that people show their understanding for you?” andDoes it ever happen to you that people are willing to lend youfriendly ear?” Responses were averaged to produce a score for

motional social support received (˛ = .86).

.1.3.6. Affective well-being. This was measured using an adaptedersion of the 12-item job-related affect scale developed by Warr1990) and later modified by Sevastos et al. (1992). The items

easure the anxiety–comfort and depression–enthusiasm axes ofarr’s circumplex model of job-related well-being (Warr, 1990,

002). Participants were asked to indicate how much they hadelt each of 12 feeling states over the past month (whereas peo-le are asked how much their job has made them feel those states

n the original job-related version of the measure). Answers wereecorded on a 5-point response scale ranging from 1 (not at all) to 5a great deal). The three items measuring depression were reversecored, and combined with the three items measuring enthusiasmo produce a six-item measure of feeling enthusiastic (˛ = .81). Sim-larly, the three items measuring anxiety were reverse scored andombined with the three items measuring comfort to produce aix-item measure of feeling relaxed (˛ = .86).

.1.3.7. Academic adjustment to university. This was measuredsing a scale developed by Kaya and Weber (2003). The scale con-isted of six items, for example “I feel good about my degree” and “Iegularly attend my classes”. Participants rated their level of adjust-ent on a 7-point response scale ranging from 1 (does not apply toe at all) to 7 (applies to me very much). The responses were aver-

ged to produce an overall score for academic adjustment (˛ = .84).

.1.3.8. Academic attainment. This was measured in the secondave of data collection. First year students recorded how they haderformed on their first three pieces of course work (they had not

2 Networks 30 (2008) 283–296

ytbfaa

4

b(wfi

4

is(htrtiwcatEatcm

wp(lraagpfifstftitr

aao(lvf

as(

dev

iati

ons,

and

corr

elat

ion

sbe

twee

nSt

ud

y1

vari

able

s

MS.

D.

12

34

56

78

910

1112

1314

1516

17

nsi

tyto

con

nec

t3.

560.

65–

mak

efr

ien

ds

3.88

0.83

.86

–m

ake

acqu

ain

tan

ces

3.61

0.77

.86

.66**

–jo

inot

her

s3.

190.

76.7

6.4

4**.4

7**–

twor

ksi

ze10

.97

4.42

.25**

.27**

.24**

.12

–6.

041.

24.5

7**.5

4**.5

2**.3

4**.1

3–

s7.

050.

91.0

1.1

5−.

04.1

1−.

05−.

02–

snes

s6.

191.

29.0

1.1

0−.

09.0

1−.

08−.

04.3

3**–

bili

ty5.

341.

29.1

7*.1

9*.1

6.0

8.0

1.1

2.3

1**.1

0–

6.40

0.97

.05

−.02

.02

.13

.02

−.08

.25**

.14

−.04

–th

usi

asti

c3.

710.

78.2

9*.2

4.2

7*.2

3−.

05.2

3.1

3.0

8.0

8.0

9–

laxe

d3.

620.

58.0

4−.

05.0

2.1

4−.

18−.

10.1

4.2

3.1

1−.

10.2

5–

owle

dge

able

3.90

0.47

−.05

−.07

−.06

.01

.07

−.06

.14

.14

−.09

.05

.31*

.39**

–ci

alsu

pp

ort

rece

ived

3.14

0.54

.38**

.40**

.33**

.22

.17

.29*

−.17

−.03

−.01

−.13

.26*

−.10

.04

–ll

-bei

ng:

enth

usi

asti

c3.

770.

56.2

5**.3

0**.2

1*.0

9.1

5.2

7**.0

2.1

0.4

4**−.

06.1

7.0

6.1

1.2

5*–

ll-b

ein

g:re

laxe

d3.

500.

61.2

7**.2

9**.2

2**.1

6.0

8.2

7**.0

6−.

01.5

0**.1

7*−.

03.0

5−.

06.2

9*.7

0**–

just

men

tto

un

iver

sity

4.87

1.03

.21*

.22**

.17*

.14

.13

.09

.16

.19

.08

.15

−.05

−.03

.03

.11

.41**

.24**

–ta

inm

ent

4.77

0.71

.21

.30**

.16

.04

−.02

.11

−.07

−.05

.03

.15

††

††

.12

.11

.41**

cept

vari

able

s11

–13

wh

ere

N=

58,v

aria

ble

14w

her

eN

=6

4,an

dva

riab

le18

wh

ere

N=

80.

tail

ed).

volv

edd

ata

from

dif

fere

nt

par

tici

pan

tsso

rco

uld

not

beca

lcu

late

d.


et sat exams), while second and third year students recorded howhey had performed on their four most recent exams. Scores wereased on the bands in the university’s marking scheme and rangedrom 1 to 6 with 6 being the highest level of attainment. The aver-ge score for each participant was used as the measure of academicttainment.

.2. Results

Table 1 shows the means, standard deviations, and correlationsetween the study variables. Of the variables collected in two wavesPCO, Big 5, well-being, adjustment), only the relaxed dimension ofell-being was significantly different, t(142) = 2.15, p < .05, in therst wave, M = 3.62, compared to the second wave, M = 3.41.

.2.1. Psychometric properties of PCOTo assess the psychometric properties of PCO, we first examined

ts factorial validity by conducting a confirmatory factor analy-is using structural equation models implemented in LISREL 8.7Joreskog and Sorbom, 2004). We hypothesized that the scale wouldave a hierarchical structure, with the three components forminghree first-order factors nested within a single second-order factorepresenting overall PCO. In line with accepted practice, we alsoested two other plausible models for comparison. First, we exam-ned a model that had a single factor for the whole scale. Second

e examined a model that had three alternative factors. Since eachomponent of the scale had the same three indicators (extent, easend attraction of forming component-specific tie), it was possiblehat the indicators would form better factors than the components.xtent, ease and attraction of tie formation were therefore tested aslternative factors. For each model, individual items were allowedo load on only one factor and the latent variables were allowed toorrelate. The results of the CFA, using maximum likelihood esti-ates from LISREL 8.7, are shown in Table 2.Multiple indicators of model fit were used. The first indicator

as the Chi-square ratio (�2/d.f.), which should be ≤5 if the pro-osed model adequately fits the data. In line with Hu and Bentler’s1998) recommendations for testing small sample maximum like-ihood models, we also used the standardized root mean squareesidual (SRMR) supplemented by the incremental fit index (IFI)nd the comparative fit index (CFI). Hu and Bentler recommendedn upper bound of good fit of .08 for SRMR and a lower bound ofood fit of .95 for IFI and CFI. Our hypothesized hierarchical modelrovided a good fit to the data based on all four fit indices, andtted better than the single factor model or the alternative three-

actor model. A Chi-square difference test of the two nested modelshowed that the hierarchical model fitted significantly better thanhe single factor model (��2(3) = 74.93, p < .01). All hypothesizedactor loadings in the hierarchical model were greater than or equalo .52. The internal consistency reliabilities of the overall scale andts three components all exceeded .6 (see Section 4.1), and wereherefore satisfactory. Correlations between the three componentsanged between .44 and .66 (see Table 1).

In support of the concurrent validity of PCO, we found that over-ll PCO and two of its components – making friends and makingcquaintances – were significantly positively correlated with sizef university friendship network (see Table 1). Gender, year of studycoded as dummy variables for each year), and age were not corre-ated with overall PCO or its components, and partialling out theseariables did not change the significance of the correlations with
riendship network size.
Concerning relationships between PCO and the Big Five person-lity factors, we found that overall PCO and its components were allignificantly positively correlated (see Table 1) with extraversionp < .01). Of the other personality factors, only emotional stabil- Ta

ble

1M

ean

s,st

and

ard

Var

iabl

e

1.O

vera

llp

rop

e2.

Prop

ensi

tyto

3.Pr

open

sity

to4.

Prop

ensi

tyto

5.Fr

ien

dsh

ipn

e6.

Extr

aver

sion

7.A

gree

able

nes

8.C

onsc

ien

tiou

9.Em

otio

nal

sta

10.O

pen

nes

s11

.Ap

pea

red

en12

.Ap

pea

red

re13

.Ap

pea

red

kn14

.Em

otio

nal

so15

.Aff

ecti

vew

e16

.Aff

ecti

vew

e17

.Aca

dem

icad

18.A

cad

emic

at

Not

e.N

=14

4,ex

*p

<.0

5.**

p<

.01

(tw

o-†

Var

iabl

esin

P. Totterdell et al. / Social Netwo

Table 2Confirmatory factor analysis fit statistics for propensity to connect with others scale

Model d.f. �2 IFI CFI SRMR

Study 1 sample3 first-order factors and 1

second-order factor24 56.31** .96 .96 .06

1 factor 27 131.24** .89 .89 .093 alternative factors 24 120.99** .90 .90 .10

Study 2 sample3 first-order factors and 1

second-order factor24 76.65** .97 .97 .04

1 factor 27 238.78** .90 .90 .073 alternative factors 24 219.26** .91 .91 .07

Nm

iPgri

4a

tewfaaˇWss

gtsgaca2

4a

iwlwtdmt(w

4

ftC

sPtSstpwnaaifˇb

5

5

5

eawa1sa

rwToytMtmwr

ttstwN

55As(

5dwtw

ote. IFI: incremental fit index; CFI: comparative fit index; SRMR: standardized rootean square residual.

** p < .01.

ty was significantly correlated with PCO (specifically with overallCO and with the making friends component). Partialling out age,ender, and year of study did not change the significance of theseesults. The discriminant and predictive validity of PCO were exam-ned by testing Hypotheses 1–3.

.2.2. Relationships between PCO and extraversion, positiveffectivity

To examine whether PCO was a better predictor of network sizehan extraversion, as predicted by Hypothesis 1, both variables werentered simultaneously in a regression model with friendship net-ork size as the dependent variable, and age, gender and dummies

or year of study entered as control variables. Overall PCO remainedsignificant predictor of network size, ˇ = .23, t(131) = 2.27, p < .05,nd extraversion was not a significant predictor of network size,= −.01, t(131) = −0.09, n.s. These results supported Hypothesis 1.hen emotional stability was also added to the models, the only

ignificant relationship was overall PCO as a predictor of networkize, ˇ = .24, t(130) = 2.27, p < .05.

With respect to Hypothesis 2, concerning how individuals withreater PCO appear to others, the correlations (in Table 1) showhat individuals who had greater overall PCO (and greater propen-ity to make acquaintances) were perceived by others as havingreater positive affectivity, as measured by ratings of their enthusi-sm following brief social interactions. Individuals’ overall PCO andomponents were not related to how relaxed or how knowledge-ble they appeared to others. These results supported Hypothesis.

.2.3. Relationships between PCO and indicators of personaldjustment

We examined how PCO related to personal adjustment throughts relations with received emotional social support, affective

ell-being, academic adjustment and academic attainment (corre-ations reported in Table 1). In support of Hypothesis 3, overall PCOas significantly positively correlated with: participants’ percep-

ion of the amount of emotional social support they receive, bothimensions of their affective well-being, and their academic adjust-ent. Corresponding significant relationships were also found for

he making friends and making acquaintances components of PCOsee Table 1). In addition, the making friends component of PCOas significantly positively correlated with academic attainment.

.2.4. Supplementary analysesHypotheses 2 and 3 were tested further using regression models

or each of the dependent variables in order to assess the poten-ial influence of control variables, network size, and personality.ontrol variables for gender, age and (where appropriate) year of

pTw

e

rks 30 (2008) 283–296 289

tudy were entered in the models prior to entering overall PCO.CO became a significant predictor of academic attainment, ˇ = .27,(74) = 2.38, p < .05, but was unchanged for the other variables.ize of friendship network was then entered into each model. Theignificance of PCO was unchanged in each case which suggestshat the effects of PCO were not just due to the size of partici-ants’ friendship networks. Extraversion and emotional stabilityere then entered in place of network size. PCO remained a sig-ificant predictor of emotional support, ˇ = .40, t(52) = 2.18, p < .05,cademic adjustment, ˇ = .25, t(130) = 2.35, p < .05, and academicttainment, ˇ = .26, t(72) = 2.01, p < .05, but was no longer a signif-cant predictor of appearing enthusiastic, ˇ = .13, t(47) = 0.66, n.s.,eeling enthusiastic, ˇ = .06, t(130) = 0.65, n.s., and feeling relaxed,= .09, t(130) = 1.05, n.s. The effects of PCO therefore appeared toe independent of the personality traits, except in relation to affect.

. Study 2

.1. Method

.1.1. Research setting, participants, and procedureThe second study aimed to extend understanding of PCO by

xamining it within a work setting. The study was conducted indepartment of a health-care organization. The department dealtith customers’ claims and queries concerning health insurance

nd health care, and employees worked in teams of between 7 and8 members. The employees were invited to take part in an opinionurvey as part of a wider research program concerning job design,nd part of that survey questionnaire was used for this study.

The survey questionnaire was completed by 197 of 242 potentialespondents, which gave a response rate of 81%. The participantsere aged between 18 and 61 years (M = 33.56 years, S.D. = 9.87).

here were 145 females and 52 men in the sample. Tenure in therganization ranged from less than 1 year to 25 years (M = 6.64ears, S.D. = 4.57). There were 146 full-time workers and 42 part-ime workers (9 participants did not report their contracted hours).

ost participants worked in one of three main work roles: cus-omer advisor (n = 98), administrator (n = 48), or customer care

anager/team leader (n = 21). The remaining participants eitherorked in other miscellaneous roles (n = 17) or did not report their

ole (n = 13).The organization allowed all employees the opportunity and

ime to complete the survey questionnaire, which was adminis-ered by the research team during the course of two days at theite. The measures used for this study were all part of this ques-ionnaire, except the measure for innovation involvement whichas completed 8 months later (and had a reduced sample size of= 169 due to attrition).

.1.2. Measures

.1.2.1. Propensity to connect with others. This measure (shown inppendix A) was the same as the PCO scale used in Study 1. It mea-ured overall PCO (˛ = .91) and its three components: making friends˛ = .88), making acquaintances (˛ = .72), and joining others (˛ = .86).

.1.2.2. Work-related expertise. This scale consisted of three itemseveloped for the purpose of the study. Participants were asked tohat extent they “have an expert knowledge of work-related mat-

ers”, “solve work-related problems without help” and “performork-related tasks that others find difficult”. Each item had a 5-
oint response scale ranging from 1 (not at all) to 5 (a great deal).he three responses were averaged to produce an overall score forork-related expertise (˛ = .85).
To validate the measure it was administered to 18 full-timemployees in a call center who had volunteered to complete some

2 Netw

npy(wpmrkrost

5seftzpptt9tfspw

5tatt(

5sT1dssbes“(yphf

5

poattsnp

ciiTtktegawaTpgitar

5

5

CtTsaofincAgttt

5

bnftt

pnnTpeabsk

b�


ew questionnaire measures for use in future research. The sam-le included 9 females and 9 males, with a mean age of 24.67ears (S.D. = 6.30), and a mean organizational tenure of 3.48 yearsS.D. = 2.89). As well as completing the self-report measure ofork-related expertise (scale M = 3.33, S.D. = 0.80; ˛ = .82), thesearticipants also rated the extent to which each of their team-ates possessed work-related knowledge using a 5-point scale

anging from 1 (no or very little knowledge) to 5 (a great deal ofnowledge). The work-related knowledge of each participant wasated by a minimum of 2 and a maximum of 4 colleagues (numberf raters M = 2.67). The correlation between the self-report mea-ure and third-party measure was r(18) = 0.74, p < .01, which impliedhey measure similar constructs.

.1.2.3. Personal work networks. To derive the participants’ friend-hip and advice networks, they were asked to list the names ofveryone in the department whom they considered to be their workriends (friendship) and whom they consulted when they neededo know something about work (advice). Other studies of organi-ational networks have used similar open-response lists to deriveersonal networks (e.g., Brass, 1984; Mehra et al., 2001). Partici-ants were able to nominate as many people as they wished upo a maximum of 18 names. The maximum number of nomina-ions from any participant was 17 and 89% participants provided

or fewer nominations for both networks, which indicated thathe name limit was not unduly restrictive. Responses concerningriendship and advice ties were arranged into two 197 × 197 non-ymmetric binary matrices. For example, if person i nominatederson j, then cell Xij in the matrix was coded as 1, otherwise itas coded as 0.

.1.2.4. Job-related affective well-being. This was measured usinghe same 12-item scale used in Study 1, but the questions weresked in relation to how much the person’s job had made them feelhe 12 affective states during the past month. The scale assessed thewo dimensions of feeling enthusiastic (˛ = .88) and feeling relaxed˛ = .86).

.1.2.5. Organizational involvement. This was measured using twocales that assessed distinct aspects of organizational involvement.he first scale measured influence over decision-making (Parker,998). It had three items (˛ = .90), including “Can you influenceecisions about work procedures?”, and used a 5-point responsecale from 1 (not at all) to 5 (a great deal). The second scale mea-ured innovation involvement and was based on a measure devisedy Axtell et al. (2000) that distinguished between the ideas thatmployees suggest and the implementation of their ideas. Ideasuggested (˛ = .91) was measured with three items, for exampleMade proposals about doing things differently”. Ideas implemented˛ = .95) was also measured with three items, for example “Hadour proposals for doing things differently carried out”. Partici-ants were asked to report the extent to which these things hadappened in the previous 6 months using a 5-point response scale

rom 1 (not at all) to 5 (a very great extent).

.1.3. AnalysisUCINET 6 for Windows (Borgatti et al., 2002) was used to com-

ute the size, betweenness centrality and gatekeeper brokeragef the participants’ friendship and advice networks. The size ofparticipant’s network was measured by the number of people

o whom that person was directly connected. In-ties (i.e., rela-ions nominated by others in the network) were used to calculateize. Betweenness centrality was measured by the percentage ofetwork paths between all employees that passed through thearticipant. Non-directed ties (i.e., in- and out-ties) were used to

l(tsa

orks 30 (2008) 283–296

alculate betweenness centrality because directed ties are hard tonterpret for betweenness. The betweenness scores were normal-zed by dividing the scores by maximum possible betweenness.he scores for network size and betweenness centrality were logransformed because they were positively skewed. Gatekeeper bro-erage was measured by the number of network paths that passedhrough a participant from a source employee to a destinationmployee where the two employees belonged to different workroups and the participant was a member of the same work groups the destination employee. The employees were divided into threeork groups corresponding to the three main work roles (customer

dvisor, administrator, and customer care manager/team leader).he scores for gatekeeper brokerage were weighted in inverse pro-ortion to the number of other participants who had the sameatekeeping position, so as to give greater weight to the scores ofndividuals whose gatekeeping position was unusual compared toheir peers (Gould and Fernandez, 1989). Relations between PCOnd the other variables were then analyzed using correlation andegression procedures.

.2. Results

.2.1. Confirmatory factor analysis and reliability of PCOLISREL 8.7 (Joreskog and Sorbom, 2004) was used to conduct a

FA on PCO in order to see whether the hierarchical factor struc-ure found in Study 1 could be replicated in an employed sample.he fit statistics for the model are shown in Table 2. Our hypothe-ized hierarchical model provided a good fit to the data based onll four fit indices, and fitted better than the single factor modelr the alternative three-factor model (although these models alsotted reasonably well). A Chi-square difference test of the twoested models showed that the hierarchical model fitted signifi-antly better than the single factor model (��2(3) = 162.13, p < .01).ll hypothesized factor loadings in the hierarchical model werereater than or equal to .59. The internal consistency reliabilities ofhe overall scale and its three components all exceeded .7 (see Sec-ion 5.1.2.1), and were therefore satisfactory. Correlations betweenhe three components ranged between .60 and .73 (see Table 3).

.2.2. Normative characteristics of PCOTable 3 shows the means, standard deviations, and correlations

etween the study variables, including overall PCO and its compo-ents. Skewness (−.29, S.E. .17) and kurtosis (−.54, S.E. .35) values

or employees’ overall PCO were within the range for a normal dis-ribution. Overall PCO in this sample (M = 3.48) did not differ fromhat in Study 1 (M = 3.56), t(338) = .86, n.s.

Females scored higher than males on overall PCO and its com-onents, but none of the differences were significant. There wereo significant correlations between overall PCO (and its compo-ents) and age, organizational tenure, and work expertise (seeable 3), which suggest that PCO does not depend on experience orossession of domain-specific knowledge/skills. In contrast, workxpertise was significantly correlated with organizational tenurend with the size of an employee’s advice network (i.e., the num-er of people who consult the employee) but not with age, whichuggests that – unlike PCO – it does depend on domain-specificnowledge/skills.

A one-way analysis of variance showed that overall PCO differedetween participants in the three work roles, F(2, 162) = 3.30, p < .05,2 = .04. Post hoc LSD comparisons showed that managers/team

eaders (M = 3.88) had greater overall PCO than customer advisorsM = 3.47, p < .05) and administrators (M = 3.34, p < .05). To assesshis role influence in subsequent analyses, a new variable – leader-hip role – was created which had a value of 1 if the participant wasmanager or team leader, and 0 otherwise. Table 3 shows that hav-

P. Totterdell et al. / Social NetwoTa

ble

3M

ean

s,st

and

ard

dev

iati

ons,

and

corr

elat

ion

sbe

twee

nSt

ud

y2

vari

able

s

Var

iabl

eM

S.D

.1

23

45

67

89

1011

1213

1415

1617

18

1.O

vera

llPC

O3.

48

0.84

–2.

Prop

ensi

tyto

mak

efr

ien

ds

3.59

1.05

.89**

–3.

Prop

ensi

tyto

mak

eac

quai

nta

nce

s3.

49

0.89

.91**

.73*

–4.

Prop

ensi

tyto

join

oth

ers

3.37

0.91

.85**

.60**

.69**

–5.

Wor

k-re

late

dex

per

tise

3.73

0.73

.07

.01

.06

.11

–6.

Size

offr

ien

dsh

ipn

etw

orka

3.95

2.58

.20**

.26**

.13

.13

.13

–7.

Size

ofad

vice

net

wor

ka3.

203.

01.1

3.1

0.0

9.1

6*.3

4**.4

7**–

8.B

etw

een

nes

sce

ntr

alit

yin

frie

nd

ship

net

b1.

291.

62.2

1**.1

7*.2

0**.1

7*.0

8.4

5**.3

4**–

9.B

etw

een

nes

sce

ntr

alit

yin

advi

cen

etb

1.30

2.09

.12

.05

.09

.20**

.17*

.14

.54**

.33**

–10

.Gat

ekee

per

brok

erag

ein

frie

nd

ship

net

c0.

200.

80.0

3.0

4.0

2.0

1−.

05.0

6.0

9.1

8*.2

0**–

11.G

atek

eep

erbr

oker

age

inad

vice

net

c0.

100.

55.1

6*.1

1.1

7*.1

5*.0

1.0

2.2

0*.2

2**.3

7**.0

9–

12.A

ffec

tive

wel

l-be

ing:

enth

usi

asti

c3.

330.

84.1

0.1

0.0

9.0

9.0

9.1

2.1

1.0

7.0

4−.

06.0

6–

13.A

ffec

tive

wel

l-be

ing:

rela

xed

3.32

0.75

.21**

.25**

.19**

.12

.18*

.22**

.06

.21**

−.09

−.09

−.04

.66**

–14

.In

flu

ence

over

dec

isio

n-m

akin

g2.

320.

96.2

1**.1

5*.2

1**.2

0**.1

5*.1

4*.2

7**.1

3.2

4**.0

4.0

3.4

1**.3

2**–

15.I

dea

ssu

gges

ted

2.80

0.90

.20*

.14

.15

.25**

.27**

−.06

.21*

−.08

−.01

−.05

.03

.08

−.01

.15

–16

.Id

eas

imp

lem

ente

d1.

960.

99.0

7.0

2.0

7.1

0.1

3−.

01.2

6**−.

02.1

4.0

8−.

05.0

7−.

05.2

1*.5

9**–

17.A

ge(y

ears

)33

.60

9.87

−.11

−.12

−.08

−.09

.01

−.08

.14*

−.14

*.1

2−.

11.0

0.1

6*−.

11.0

2.0

8−.

05–

18.T

enu

rein

orga

niz

atio

n(y

ears

)6.

64

4.57

.00

−.03

−.02

.06

.29**

−.03

.30**

.02

.19**

.04

.06

.21**

.05

.27**

.08

.06

.38**

–19

.Lea

der

ship

role

.11

0.32

.16*

.08

.16*

.19*

.12

−.01

.35**

.01

.42**

.06

.26**

.21**

−.01

.47**

.16

.28**

.08

.30**

Not

e.N

=19

7.a

Var

iabl

es6

and

7u

sed

in-t

ies

and

wer

elo

gtr

ansf

orm

edp

rior

toco

rrel

atio

n.

bV

aria

bles

8an

d9

wer

elo

gtr

ansf

orm

edp

rior

toco

rrel

atio

n.

cV

aria

bles

10,1

1,15

,an

d16

had

are

du

ced

sam

ple

size

ofN

=16

7;an

d10

and

11w

ere

wei

ghte

din

inve

rse

pro

por

tion

toth

en

um

ber

ofp

arti

cip

ants

wh

oco

uld

hav

efu

lfill

edth

esa

me

gate

keep

ing

role

.*

p<

.05.

**p

<.0

1(t

wo-

tail

ed).

ipb

55aowiefkwt

5cstsnfattpyfaTdNwtf

5sitorabpsttmf

tapLaowfsg

5c

rks 30 (2008) 283–296 291

ng a leadership role was significantly correlated with overall PCO,ropensity to make acquaintances, and propensity to join others,ut not propensity to make friends.

.2.3. Relationships between employees’ PCO and their networks

.2.3.1. Overall PCO. Criterion validity for the PCO measure wasssessed by examining its relationships with various characteristicsf the friendship and advice networks that employees had formedithin the organization. Table 3 shows that overall PCO was signif-

cantly correlated with friendship network size (i.e., the number ofmployees in the sample who consider the participant to be theirriend), betweenness centrality in the friendship network, and gate-eeper brokerage in the advice network. However, the relationshipith gatekeeper brokerage was not significant after partialling out

he effects of gender, organizational tenure and leadership role.

.2.3.2. Propensity to make friends. Table 3 shows that this PCOomponent was significantly correlated with friendship networkize and friendship betweenness centrality. To test Hypothesis 4hat propensity to make friends would be related to size of friend-hip network, a regression analysis was conducted with friendshipetwork size as the dependent variable and propensity to make

riends, work expertise, and the product of friendship propensitynd work expertise entered as predictor variables. The predic-or variables were standardized, and control variables for gender,enure and leadership role were entered into the model prior to theredictor variables (as they were in subsequent regression anal-ses). There was a significant main effect of propensity to makeriends, ˇ = .22, p < .01, but not a main effect of expertise or an inter-ction between friendship propensity and expertise (see Table 4).his supported Hypothesis 4, and also suggested that participantsid not make friends on the basis of a person’s work expertise.o effects were found for friendship propensity when advice net-ork size was the dependent variable (see Table 4) which suggests

hat this component was – as it should be – more sensitive to theormation of friendship ties than advice ties.

.2.3.3. Propensity to make acquaintances. This component was notignificantly correlated with friendship or advice network size, butt was significantly correlated with friendship betweenness cen-rality and advice gatekeeper brokerage (see Table 3). In supportf Hypothesis 5 that propensity to make acquaintances would beelated to advice network size when combined with work expertise,regression analysis showed that there was a significant interactionetween propensity to make acquaintances and work expertise inredicting advice network size, ˇ = .14, p < .05 (see Table 4). Fig. 2hows that employees who had high propensity to make acquain-ances and high work expertise had the most colleagues goingo them for advice. In line with our expectations, propensity to

ake acquaintances did not interact with work expertise to predictriendship network size (see Table 4).

A regression analysis with advice betweenness centrality ashe dependent variable showed a significant interaction betweencquaintance propensity and work expertise, ˇ = .15, t(174) = 2.03,< .05, but not when leadership role was included in the model.ikewise, a regression analysis with advice gatekeeper broker-ge as the dependent variable showed a significant main effectf acquaintance propensity, ˇ = .18, t(156) = 2.27, p < .05, but nothen leadership role was included. Leadership role may there-

ore account for these associations. Hence, there was only qualified
upport for Hypothesis 5 in relation to betweenness centrality andatekeeper brokerage.
.2.3.4. Propensity to join others. This component was significantlyorrelated with friendship betweenness centrality and advice gate-

292 P. Totterdell et al. / Social Networks 30 (2008) 283–296

Table 4Regression models showing interactive effects of work expertise and components of propensity to connect with others (PCO) on network size in advice and friendshipnetworks

Predictor variables Dependent variables

Size of advice network Size of friendship network

ˇ t R2 ˇ t R2

Propensity to make friends (F) .08 1.15 .22 2.97**

Work expertise .24 3.43** .13 1.74F × E .05 0.71 −.10 −1.32Model summary .23** .10**

Propensity to make acquaintances (A) .02 0.24 .12 1.51Work expertise .25 3.57** .13 1.70A × E .14 2.07* −.02 −.32Model summary .24** .05

Propensity to join others (J) .03 0.44 .09 1.23Work expertise .25 3.51** .14 1.76J × E .15 2.18* .04 0.47Model summary .25** .05

Note. N = 177. Dependent variables were based on in-ties and were log transformed. Predictor variables were standardized. Control variables – gender, organizational tenureand leadership role – were entered prior to the predictor variables.

kiaHaspwaT

diwysbtqt

Fp

5o

ti

5ptasrtbnae

* p < .05.** p < .01.

eeper brokerage, but unlike friendship or acquaintance propensityt was also significantly correlated with advice network sizend advice betweenness centrality (see Table 3). In support ofypothesis 6 that propensity to join others would be related todvice network size when combined with work expertise, a regres-ion analysis showed that it interacted with work expertise inredicting advice network size, ˇ = .15, p < .05 (see Table 4). In lineith our expectations, propensity to join others did not inter-

ct with work expertise to predict friendship network size (seeable 4).

A regression analysis with advice betweenness centrality as theependent variable showed a significant interaction between join-

ng others and work expertise, ˇ = .15, t(173) = 2.03, p < .05, but nothen leadership role was included in the model. A regression anal-

sis with gatekeeper brokerage as the dependent variable showed aignificant main effect of joining others, ˇ = .15, t(156) = 1.90, p < .05,ut not when leadership role was included. Leadership role may
herefore account for these associations. Hence, there was onlyualified support for Hypothesis 6 in relation to betweenness cen-rality and gatekeeper brokerage.
ig. 2. Number of colleagues who go to participants for advice as a function of theirropensity to make acquaintances and work-related expertise.

(TgsPs

5alFnlnwaittraats

.2.4. Relationships between employees’ PCO and work-relatedutcomes

Finally, Hypothesis 7 which proposed that PCO would be posi-ively related to job-related affective well-being and organizationalnvolvement was tested.

.2.4.1. Job-related affective well-being. Overall PCO, friendshipropensity and acquaintance propensity were significantly posi-ively correlated with the feeling relaxed dimension of job-relatedffective well-being, but were not correlated with feeling enthu-iastic (see Table 3). Hypothesis 7 was therefore supported withespect to the relaxed dimension of well-being, but not with respecto the enthusiasm dimension. To assess whether the associationetween PCO and feeling relaxed was due to the size of participants’etworks, a regression analysis was conducted with feeling relaxeds the dependent variable. Gender, organizational tenure, and lead-rship role were entered in the model prior to entering overall PCO,log) size of friendship network, and (log) size of advice network.he analysis showed that employees who felt more relaxed hadreater overall PCO, ˇ = .19, t(171) = 2.55, p < .05, and larger friend-hip networks, ˇ = .21, t(171) = 2.49, p < .05. The association betweenCO and feeling relaxed was not therefore due solely to networkize.

.2.4.2. Organizational involvement. Table 3 shows that overall PCOnd propensity to join others were significantly positively corre-ated with influence over decision-making and ideas suggested.riendship propensity and acquaintance propensity were also sig-ificantly positively correlated with decision-making influence. In

ine with our expectations, PCO and its components were not sig-ificantly correlated with ideas implemented. A regression analysisith influence over decision-making as the dependent variable,

nd control variables for gender, tenure and leadership role enteredn the model, showed a significant effect of overall PCO, ˇ = .13,(176) = 2.04, p < .05, as well as significant effects for tenure, ˇ = .16,(176) = 2.34, p < .05 and leadership, ˇ = .40, t(176) = 5.85, p < .01. This
esult indicates that individuals who had longer tenure or hadleadership role had greater decision-making influence, but the
ssociation between PCO and influence was not dependent onhese variables. Entering (log) size of friendship network and (log)ize of advice network with overall PCO, showed that employees

Netwo

wPtawfd

wooccpiesPttfs

6

tomiptmoidaFpp

6

6

bitf(ocmtidbwoptltae

6

pwts1thEatha(m

6

aiiltnmennaotpicawwiawp

6

vssvStnsewstddn


ho had greater decision-making influence had greater overallCO, ˇ = .15, t(189) = 2.13, p < .05, and larger advice networks, ˇ = .25,(189) = 3.18, p < .01. In other words, advice network size was associ-ted with influence, but the association between PCO and influenceas not dependent on network size. Hypothesis 7 was there-

ore supported with respect to the association between PCO andecision-making influence.

With ideas suggested as the dependent variable, overall PCOas not significant, ˇ = .17, t(105) = 1.74, p = .09 in the presencef the control variables. Replacing overall PCO with the joiningthers component of PCO, showed a significant effect for thisomponent, ˇ = .21, t(105) = 2.20, p < .05, in the presence of theontrol variables. This means that individuals who had greaterropensity to join others were also more likely to suggest ideas

n the organization. A regression model examining the influ-nce of network size in the relationship between PCO and ideasuggestion, showed that employees who had greater overallCO, ˇ = .22, t(114) = 2.44, p < .05, larger advice networks, ˇ = .28,(114) = 2.90, p < .01, and smaller friendship networks, ˇ = −.20,(114) = 2.06, p < .05, suggested more ideas. Hypothesis 7 was there-ore supported with respect to the association between PCO anduggestion-making.

. Discussion

Research on social networks has mostly focused on the struc-ure and effects of relationships between individuals, rather thann how the attributes of individuals might contribute to the for-ation and structure of social networks. The two studies in this

nvestigation have focused on one such attribute, namely people’sropensity to connect with others (PCO). The results indicate thathis propensity is a measurable individual difference that can help

ake sense of: an individual’s position within an organizational orther social network; the kind of network ties that an individuals likely to form; the social signals that an individual is likely toisplay to others; and some of the personal and social outcomesn individual may experience as a consequence of his or her PCO.urthermore, the results suggest that PCO is not independent ofersonality but has explanatory power beyond that provided byersonality traits.

.1. Central findings

.1.1. Individual differences in PCOIn both studies, PCO was measured using a newly designed

rief self-report scale that had three sub-scales for assessing: anndividual’s propensity to make friends (strong ties), make acquain-ances (weak ties), and join others (bridging ties). The proposedactor structure was confirmed and replicated using two samplescomposed of undergraduates and employees from a health-carerganization), and the scale was found to have acceptable psy-hometric properties. As an individual difference construct, PCOay be developmentally stable because it did not vary in associa-

ion with age within samples and it produced similar mean scoresn samples with a mean age difference of 14 years. Possession ofomain skills did not appear to be a prerequisite to having PCOecause it was not related to a person’s work expertise. However,e found that managers and team leaders had greater PCO than

ther employees in the organization, which implies that having thisropensity may incline individuals to adopt or be adopted for cer-
ain roles within organizations. In particular, managers and teameaders were more inclined to form weak ties and bridging ties, andhese are precisely the kinds of tie that have been associated withcquiring power and influence in organizations (Brass, 1984; Brasst al., 2004).
ntt(a

rks 30 (2008) 283–296 293

.1.2. PCO, personality and appearance to othersAn individual with PCO is likely to be outwardly oriented in dis-

osition and therefore it was not surprising that Study 1 found PCOas strongly related to extraversion and to a lesser extent to emo-

ional stability. Nonetheless, PCO was a better predictor of networkize than these personality factors (which supported Hypothesis), and a better predictor of academic adjustment and attainmentoo. Individuals with higher PCO also appeared to other people asaving greater positive affectivity (which supported Hypothesis 2).xpression of positive affect may be one way in which individu-ls who want to make connections with other people can signalheir motivation to do so. In addition, research on emotional laboras established that displays of positive affect can help gain compli-nce, increase loyalty and enhance the positive evaluation of otherse.g., Pugh, 2001; Rafaeli and Sutton, 1990; Tsai, 2001), all of which

ay help cement ties between people.

.1.3. PCO and network characteristicsBoth studies found that individuals who had greater overall PCO

lso had larger friendship networks. The results showed strong pos-tive associations between the three components of PCO whichndicated that individuals who formed one kind of tie were alsoikely to form the others. Yet the components of PCO were never-heless differentially sensitive to the formation of different kinds ofetwork tie (as predicted by Hypotheses 4–6). Study 2 found thataking friends was directly related to the number of friendship ties

mployees had, but not to their number of advice ties. In contrast,umber of advice ties and betweenness centrality in the adviceetwork was related to the combined presence of work expertisend either propensity to make acquaintances or propensity to jointhers (but not propensity to make friends). This finding indicateshat a person’s propensity to connect sometimes needs to be com-lemented by other qualities in order to cement the ties he or she

s inclined to form. However, unlike acquaintance propensity, theomponent for joining others also showed direct associations withdvice network size and advice betweenness centrality. Individualsith greater propensity for making acquaintances or joining othersere also more likely to act as gatekeepers for information pass-

ng into their work-role group. However, these relationships wereccounted for by leadership role, which implies that individualsho have these propensities are more likely to have a gatekeepingosition because they are more likely to be in a leadership role.

.1.4. PCO and personal adjustmentThere was evidence from both studies that PCO may help indi-

iduals adjust and thrive in their social context. In Study 1, inupport of Hypothesis 3, PCO was positively related to emotionalupport received, affective well-being, academic adjustment to uni-ersity, and academic attainment on assessed work. Similarly intudy 2, in support of Hypothesis 7, PCO was positively related tohe calmness dimension of job-related affect, and to aspects of orga-izational involvement including decision-making influence anduggesting ideas. Unlike in Study 1, PCO was not related to thenthusiasm dimension of well-being in Study 2. Study 2 measuresell-being in relation to how participants’ jobs made them feel,

o perhaps there was something about their job that constrainedhe association. Concerning organizational involvement, the resultsemonstrated that individuals with high PCO reported having moreecision-making influence and suggesting more ideas in the orga-ization, irrespective of the size of their friendship and advice
etworks. Consistent with Burt’s (2004) analysis of network struc-ure and organizational innovation, PCO was not associated withhe implementation of ideas. However, innovation involvementideas suggestion and implementation) was measured 8 monthsfter PCO in the present investigation, so one possible explanation

2 Netw

iiniOa

6

rospcaihtrpbtpfan

wawe(vtt

sdosstdt(hh

trccmcnlffs

6

b

etinithBowtsw

ctOlhcpwtpafacbiai

nvhottbwto1

oiaaacaeptatcob


s that this period was sufficient to capture suggestion-making butnsufficient to capture implementation of ideas. The PCO compo-ent for joining others appeared to be the most active ingredient

n the association with suggestion-making, which accords withbstfeld’s (2005) link between having a tertius iungens orientationnd getting involved in organizational innovation.

.2. Limitations

A potential threat to the precision of our findings is that the accu-acy of people’s perception of their network structure can dependn their position within the network and on aspects of their per-onality and affectivity (Casciaro, 1998; Casciaro et al., 1999). Inarticular, Casciaro (1998) found that individuals who are moreentral in the friendship and advice networks of their organizationre more accurate in reporting the structure of their networks. Bynference, this might suggest that individuals with greater PCO willave been more accurate in reporting their network ties. Distor-ions arising from this source are more likely to have affected theesults of Study 1, because Study 2 used nominations from othereople in the network (in-ties) to calculate network size. However,etweenness centrality was measured using out-ties as well as in-ies. A perceptual account may therefore explain why acquaintanceropensity was related to betweenness centrality but not size in theriendship network. This account would argue that individuals withcquaintance propensity are more likely to perceive and thereforeominate others as friends.

It is also possible that individuals’ positions in their social net-orks were the cause rather than the consequence of their PCO,

lthough this seems a less parsimonious explanation in that itould then be necessary to explain how PCO arose from differ-

nt types of network (friendship, advice) from different life stagespre-career and career). We stated earlier that PCO may incline indi-iduals to adopt or be adopted for leadership roles. Alternatively,hose in leadership roles may have greater access to people andherefore score higher on PCO.

With respect to our measure of PCO, the components for friend-hip and acquaintance propensity were based on a commonly usedistinction between the strong ties of friendship and the weak tiesf acquaintance. However, different types of social network (friend-hip, advice, etc.) can contain both strong and weak ties. Our resultshow that the components map on to different types of network, buthat does not necessarily mean that they will distinguish betweenifferent strengths of tie within those networks. One possibility ishat the components will interact in accounting for tie strengthand possibly other outcomes). For example, individuals who haveigh friendship propensity and low acquaintance propensity mayave a predominance of strong ties in their social networks.

Another limitation of the present investigation stemmed fromhe briefness of interactions between participants in Study 1, whichestricted the basis on which participants’ affective appearanceould be judged by others. This brevity had the advantage that itaptured people’s first impressions of others, and first impressionsay be important in influencing whether or not people initiate

onnections with others. The disadvantage is that the results mayot capture how individuals with high PCO appear to others over a

onger time period involving repeated interactions. It was notable,or example, that although individuals with greater PCO reportedeeling calmer, they did not appear to others as calmer during firstocial encounters.

.3. Future research

A number of possibilities for future research are opened upy the findings of the two studies. First, the research could be

stOia

orks 30 (2008) 283–296

xtended to identify individuals who have other orientationsowards networks. Our focus has been on individuals who read-ly connect with others and who consequently have large socialetworks, but another focus would be to identify and examine

ndividuals who have a propensity to be members of networkshat are rich in connections (i.e., networks that are dense andave few structural holes). This would be in the same vein asurt et al.’s (1998) undertaking to identify the personality profilef individuals who possess structural holes in their work net-orks, but it would more directly examine people’s inclinations

owards having tightly knit networks. The characteristics and con-equences of this networking propensity could then be comparedith PCO.

Second, the research could be extended to examine how PCOombines with other individual characteristics to shape networkies and network-related outcomes (e.g., accrual of social capital).ur research has already established that individuals are more

ikely to be consulted within an organization if they have bothigh PCO and work expertise. Similarly, a combination of PCO andharisma (or affective communication) might have implications forerformance in another type of network (e.g., in a consumer net-ork). Other characteristics of individuals might also mean that

heir willingness to connect is unreciprocated by others. For exam-le, individuals who persistently express negative affect or whore indiscreet are unlikely to be attractive to others for providingriendship or help. Casciaro and Lobo (2005) found that employeesre less likely to seek help from colleagues they dislike even if thoseolleagues are very competent. Hence, individuals who have PCOut are perceived as dislikeable seem likely to suffer when their

nitiatives to connect are rejected. Similarly, there may also be neg-tive consequence for individuals exposed to the negative affect orndiscretions of prodigious connectors.

Third, the research could be extended to examine other types ofetwork connection. Study 2 looked at the characteristics of indi-iduals who were sought by others for advice, but it would also beelpful to understand the characteristics of individuals who seekut others for advice within networks. The study also looked athe extent to which individuals act as brokers in network pathshat pass into their work-role group, but there are other types ofrokerage roles to examine within networks including individualsho represent a group to those outside it, or who enter a group

o act as a consultant, or who act as an intermediary betweenther groups (Baker and Faulkner, 2002; Gould and Fernandez,989).

Fourth, the research could be extended to examine other typesf outcome for individuals with PCO. We studied organizationalnvolvement and well-being within an organization, but there maylso be implications for other types of behavior. For example, Wolffnd Moser (2005) found that the networking behaviors of individu-ls, in the form of maintaining internal and external organizationalontacts, were related to promotions and turnover. Networkingbility has also been related to the use of influence tactics (Ferrist al., 2005) and to supervisor ratings of initiative-taking and joberformance (Thompson, 2005). Burt et al. (1998) warn, however,hat measures of individual characteristics should not be used assubstitute for collecting network data. As evidence for this view,

hey demonstrated that a network which is associated with a traitan be advantageous to an individual even when the overt displayf that trait is not advantageous. We do not disagree with this point,ut our results show that equally network data should not be a sub-
titute for individual characteristics. For example, PCO was relatedo various outcomes independent of network size in both studies.ther network characteristics could potentially have accounted for
ts effects, but it may be difficult to identify all the relevant ones forny single analysis. PCO may also capture a propensity for acting

Netwo

dt

bPvebwoatw

6

sctnfatamaitiun

awttPhcs

oncaw(emo

A

UAa

A(

i

123

4

5

6

7

8

9

R

A

A

B

B

B

B

B

B

B

B

B

B

B

B

C

C


ynamically in a social network that is lost in a single snapshot ofhe state of a social network.

Fifth, and finally, the role of PCO in other social contexts shoulde explored. Relevant to this is the possibility that relations withCO may be different when social connections are formed throughirtual means. For example, extroversion may be less likely to influ-nce the expression of PCO in computer-mediated communicationsecause the connections can be formed away from a social settingithout physical proximity (Goby, 2006). People’s greater reliance

n virtual work networks and their increasing adoption of recre-tional social networking applications suggest that understandinghe role of PCO in relation to mediated communications behaviorill be an important issue to address.

.4. Practical implications

Research on individual differences in networks is at an earlytage but its implications for practical application within socialommunities can be considered. Most social communities are likelyo require a mix of individual networking orientations for theiretworks to function effectively. For instance, Oh et al. (2004)

ound that the effectiveness of groups is enhanced by havingmix of network configurations, including both bridging rela-

ionships and closure relationships (in which a person’s contactsre also connected to each other). Consequently, they recom-ended that large groups such as organizations should aim to

chieve a balance between individuals who are adept at creat-ng bridging ties and those who are adept at maintaining closureies. We have yet to identify the characteristics of individualsnclined towards closure ties, but PCO may help identify individ-als who will provide strong, weak and bridging ties within aetwork.

Although PCO has satisfactory psychometric characteristics andppears to be usable with both pre-career and career groups, weould not recommend it as a personnel selection tool for organiza-

ions. Not only would it be easy for an individual to fake scores onhe measure if they were motivated to do so, but it is unlikely thatCO is a uniformly desirable characteristic for any job. PCO mightowever be a useful diagnostic tool, which organizations and otherommunities could further develop to help them understand andhape their particular networks.

More generally, PCO could be a useful tool in the developmentf models for understanding the structure and behavior of socialetworks, such as exponential random graph models. New specifi-ations are emerging for these models that incorporate ideas suchs social circuit dependence, in which connections between net-ork nodes can be induced by connections between other nodes

Robins et al., 2007). The likelihood of such connections being gen-rated can be affected by the attributes of the nodes, and as suchay be partly determined by a person’s propensity to connect with

thers.

cknowledgements

The support of the Economic and Social Research Council (ESRC)K is gratefully acknowledged. We would also like to thank Carolynxtell, Jessica Burdon, Angela Davies, Anna Elderton, Jamie Rowan,nd Christine Sprigg for their help with the two studies.

ppendix A. Scale for propensity to connect with othersPCO)

For each of the following statements, please indicate how wellt describes you:

C

C

C

rks 30 (2008) 283–296 295

Does notdescribe mevery well

Somewhatdescribesme

Describesme verywell

. I have many friends. � � � � �. I make friends easily. � � � � �. I like to have manyfriends.

� � � � �

. I have manyacquaintances.

� � � � �

. I readily makeconnections with peopleI do not know.

� � � � �

. I like to know a lot ofpeople.

� � � � �

. I often put people intouch with the rightperson when they needsomething.

� � � � �

. I find it easy to bringindividuals together.

� � � � �

. I like being able toconnect people.

� � � � �

Components:Items 1–3: Propensity to make friends.Items 4–6: Propensity to make acquaintances.Items 7–9: Propensity to join others.

Scoring:Item responses are scored 1–5.Does not describe me very well = 1.Describes me very well = 5.

eferences

ustin, E.J., Saklofske, D.H., Egan, V., 2005. Personality, well-being and health corre-lates of trait emotional intelligence. Personality and Individual Differences 38,547–558.

xtell, C.M., Holman, D.J., Unsworth, K.L., Wall, T.D., Waterson, P.E., Harrington, E.,2000. Shopfloor innovation: facilitating the suggestion and implementation ofideas. Journal of Occupational and Organizational Psychology 73, 265–285.

aker, W.E., Faulkner, R.R., 2002. Interorganizational networks. In: Baum, J.A.C. (Ed.),Companion to Organizations. Blackwell, Oxford, pp. 520–540.

ecker, T.E., 2004. Building networks in organizations: a closer look at motivationand behavior. In: Paper presented at the Academy of Management Annual Meet-ing, New Orleans, August.

erkman, L.F., Syme, S.L., 1979. Social networks, host resistance, and mortality: anine-year follow-up study of Alameda county residents. American Journal ofEpidemiology 109, 186–204.

orgatti, S.P., Cross, R., 2003. A relational view of information seeking and learningin social networks. Management Science 49, 432–445.

orgatti, S.P., Everett, M.G., Freeman, L.C., 2002. UCINET 6 for Windows: Softwarefor Social Network Analysis. Analytic Technologies, Harvard, MA.

orgatti, S.P., Foster, P.C., 2003. The network paradigm in organizational research: areview and typology. Journal of Management 29, 991–1013.

rass, D.J., 1984. Being in the right place: a structural analysis of individual influencein an organization. Administrative Science Quarterly 29, 518–539.

rass, D.J., 1985. Men’s and women’s networks: a study of interaction patterns andinfluence in an organization. Academy of Management Journal 28, 327–343.

rass, D.J., Galaskiewicz, J., Greve, H.R., Tsai, W., 2004. Taking stock of networks andorganizations: a multilevel perspective. Academy of Management Journal 47,795–817.

urt, R.S., 1992. Structural Holes: The Social Structure of Competition. Harvard Uni-versity Press, Cambridge.

urt, R.S., 2004. Structural holes and good ideas. American Journal of Sociology 110,349–399.

urt, R.S., Janotta, J.E., Mahoney, J.T., 1998. Personality correlates of structural holes.Social Networks 20, 63–87.

asciaro, T., 1998. Seeing things clearly: social structure, personality, and accuracyin social network perception. Social Networks 20, 331–351.

asciaro, T., Carley, K.M., Krackhardt, D., 1999. Positive affectivity and accuracy insocial network perception. Motivation and Emotion 23, 285–306.

asciaro, T., Lobo, M.S., 2005. Competent jerks, lovable fools, and the formation ofsocial networks. Harvard Business Review 83, 92–100.

attell, V., 2001. Poor people, poor places and poor health: the mediating role ofsocial networks and social capital. Social Science and Medicine 52, 1501–1516.

ohen, S., Wills, T.A., 1985. Stress, social support and the buffering hypothesis. Psy-chological Bulletin 98, 310–357.

2 Netw

C

D

F

F

F

F

F

F

G

G

G

G

G

G

H

H

H

I

J

K

K

K

K

K

K

L

M

M

O

O

P

P

R

R

R

R

R

S

S

S

T

T

T

T

T

W

W


ross, R., Baker, W., Parker, A., 2002. Charged Up: The Creation and Depletion ofEnergy in Social Networks. IBM Institute for Knowledge-Based Organizations,Cambridge, MA.

iefendorff, J.M., Croyle, M.H., Gosserand, R.H., 2005. The dimensionality andantecedents of emotional labor strategies. Journal of Vocational Behavior 66,339–357.

eick, L.F., Price, L.L., 1987. The market maven: a diffuser of marketplace information.Journal of Marketing 51, 83–97.

erris, G.R., Treadway, D.C., Kolodinsky, R.W., Hochwarter, W.A., Kacmar, C.J., Douglas,C., Frink, D.D., 2005. Development and validation of the political skill inventory.Journal of Management 31, 126–152.

lap, H., Bulder, B., Volker, B., 1998. Intra-organizational networks and performance:a review. Computational and Mathematical Organization Theory 4, 109–147.

lap, H., Volker, B., 2001. Goal specific social capital and job satisfaction. Effects ofdifferent types of network on instrumental and social aspects of work. SocialNetworks 23, 297–320.

orret, M.L., Dougherty, T.W., 2001. Correlates of networking behavior for managerialand professional employees. Group & Organization Management 26, 283–311.

orret, M.L., Dougherty, T.W., 2004. Networking behaviors and career outcomes:differences for men and women? Journal of Organizational Behavior 25, 419–437.

ladwell, M., 2001. The Tipping Point. How Little Things Can Make a Big Difference.Abacus, London.

oby, V.P., 2006. Personality and online/offline choices: MBTI profiles and favoredcommunication modes in a Singapore study. Cyberpsychology & Behavior 9,5–13.

oldberg, L.R., 1990. An alternative “description of personality”: the Big-Five factorstructure. Journal of Personality and Social Psychology 59, 1216–1229.

ould, R.V., Fernandez, R.M., 1989. Structures of mediation: a formal approach tobrokerage in transaction networks. Sociological Methodology 19, 89–126.

ranovetter, M.S., 1973. The strength of weak ties. American Journal of Sociology 78,1360–1380.

ulati, R., Dialdin, D., Wang, L., 2002. Organizational networks. In: Baum, J.A.C. (Ed.),Companion to Organizations. Blackwell, Oxford, pp. 281–303.

anson, B.S., Ostergren, P.O., 1987. Different social network and social support char-acteristics, nervous problems and insomnia: theoretical and methodologicalaspects on some results from the population study ‘men born in 1914’, Malmo,Sweden. Social Science Medicine 25, 849–859.

ays, R.B., Oxley, D., 1986. Social network development and functioning during a lifetransition. Journal of Personality and Social Psychology 50, 305–313.

u, L., Bentler, P.M., 1998. Fit indices in covariance structure modeling: sensitiv-ity to underparameterized model misspecification. Psychological Methods 3,424–453.

barra, H., 1993. Network centrality, power, and innovation involvement: determi-nants of technical and administrative roles. Academy of Management Journal36, 471–501.

oreskog, K.G., Sorbom, D., 2004. LISREL 8.7 (computer software). Scientific SoftwareInternational Inc., Lincolnwood, IL.

adushin, C., 2002. The motivational foundation of social networks. Social Networks24, 77–91.

alish, Y., Robins, G., 2006. Psychological predispositions and network structure: therelationship between individual predispositions, structural holes and networkclosure. Social Networks 28, 56–84.

awachi, I., Berkman, L.F., 2001. Social ties and mental health. Journal of UrbanHealth: Bulletin of the New York Academy of Medicine 78, 458–467.

aya, N., Weber, M.J., 2003. Privacy regulation and college adjustment: a comparisonof American and Turkish freshmen living in residence halls. College Student
Journal 37, 79–92.
lein, K.K., Lim, B.C., Saltz, J.L., Mayer, D.M., 2004. How do they get there? Anexamination of the antecedents of centrality in team networks. Academy ofManagement Journal 47, 952–963.

rackhardt, D., Kilduff, M., 2002. Structure, culture and Simmelian ties inentrepreneurial firms. Social Networks 24, 279–290.

W

W

orks 30 (2008) 283–296

ee, D.Y., Tsang, E.W.K., 2001. The effects of entrepreneurial personality, backgroundand network activities on venture growth. Journal of Management Studies 38,583–602.

ehra, A., Kilduff, M., Brass, D.J., 2001. The social networks of high and lowself-monitors: implications for workplace performance. Administrative ScienceQuarterly 46, 121–146.

ichael, J., Yukl, G., 1993. Managerial level and subunit function as determinantsof networking behavior in organizations. Group and Organization Management18, 328–351.

bstfeld, D., 2005. Social networks, the tertius iungens orientation, and involvementin innovation. Administrative Science Quarterly 50, 100–130.

h, H., Chung, M., Labianca, G., 2004. Group social capital and group effective-ness: the role of informal socializing ties. Academy of Management Journal 47,860–875.

arker, S.K., 1998. Enhancing role breadth self-efficacy: the roles of job enrich-ment and other organizational interventions. Journal of Applied Psychology 83,835–852.

ugh, S.D., 2001. Service with a smile: emotional contagion in the service encounter.Academy of Management Journal 44, 1018–1027.

afaeli, A., Sutton, R.I., 1990. Busy stores and demanding customers: how do theyaffect the display of positive emotion. Academy of Management Journal 33,623–637.

aider, H., Krackhardt, D.J., 2002. Intraorganizational networks. In: Baum, J.A.C. (Ed.),Companion to Organizations. Blackwell, Oxford.

emington, N.A., Fabrigar, L.R., Visser, P.S., 2000. Reexamining the circumplex modelof affect. Journal of Personality and Social Psychology 79, 286–300.

iggio, R.E., Watring, K.P., Throckmorton, B., 1993. Social skills, social support, andpsychosocial adjustment. Personality and Individual Differences 15, 275–280.

obins, G., Snijders, T., Wang, P., Handcock, M., Pattison, P., 2007. Recent devel-opments in exponential random graph (p*) models for social networks. SocialNetworks 29, 192–215.

aucier, G., 1994. Mini-markers: a brief version of Goldberg’s unipolar Big-Five mark-ers. Journal of Personality Assessment 63, 506–516.

evastos, P., Smith, L., Cordery, J.L., 1992. Evidence on the reliability and constructvalidity of Warr’s (1990) well-being and mental health measures. Journal ofOccupational and Organizational Psychology 65, 33–49.

uurmeijer, T.P.B.M., Doeglas, D.M., Briancon, S., Krijnen, W.P., Krol, B., Sanderman,R., Moum, T., Bjelle, A., Van Den Heuvel, W.J.A., 1995. The measurement of socialsupport in the ‘European research on incapacitating diseases and social support’:the development of the social support questionnaire for transactions (SSQT).Social Science and Medicine 40, 1221–1229.

hompson, J.A., 2005. Proactive personality and job performance: a social capitalperspective. Journal of Applied Psychology 90, 1011–1017.

ichy, N.M., Tushman, M.L., Fombrun, C., 1979. Social network analysis for organiza-tions. Academy of Management Review 4, 507–519.

otterdell, P., Wall, T., Holman, H., Epitropaki, O., 2004. Affect networks: a structuralanalysis of the relationship between work ties and job-related affect. Journal ofApplied Psychology 89, 854–867.

sai, W., 2001. Determinants and consequences of employee displayed positive emo-tions. Journal of Management 27, 497–512.

ziner, A., Vered, E., Ophir, L., 2004. Predictors of job search intensity among collegegraduates. Journal of Career Assessment 12, 332–344.

anberg, C.R., Kanfer, R., Banas, J.T., 2000. Predictors and outcomes of networkingintensity among unemployed job seekers. Journal of Applied Psychology 85,491–503.

arr, P., 1990. The measurement of well-being and other aspects of mental health.
Journal of Occupational Psychology 63, 193–210.
arr, P., 2002. The study of well-being, behaviour, and attitudes. In: Warr, P. (Ed.),Psychology at Work, 5th ed. Penguin Books, London, pp. 1–25.

olff, H., Moser, K., 2005. Differential effects of networking on career transitions. Atwo year prospective study. In: Paper presented at the XII Congress of EuropeanWork and Organizational Psychology, Istanbul, May.

Social networkers: Measuring and examining individual differences in propensity to connect with others

Documents