social psychology quarterly the embedded self: a proofs embedded self.pdf · the embedded self: a...

The Embedded Self: ASocial Networks Approachto Identity Theory

Mark H. Walker1 and Freda B. Lynn1

Abstract

Despite the fact that key sociological theories of self and identity view the self as fundamen-tally rooted in networks of interpersonal relationships, empirical research investigatinghow personal network structure influences the self is conspicuously lacking. To address thisgap, we examine links between network structure and role identity salience. We identify twofeatures of personal networks that potentially affect how social ties shape identity salience:(1) proportion and strength of ties to role-based others (RBOs) and (2) embeddedness ofRBOs, or the breadth of access that a role-based group has to the rest of an individual’s net-work. Across three role identities (student, religious, and work), we find that our measure ofembeddedness predicts role identity salience but that the proportion and strength of ties donot. Thus, our study does not support the proposition that identity salience is a product ofan individual’s social and emotional attachment to role-based groups. Rather, our findingssuggest that a role identity becomes more salient as role-based others become more tightlywoven into an individual’s social fabric.

Keywords

identity, social networks, embeddedness, roles

According to most sociological theories ofself and identity, roles serve as the inter-face between social institutions and thethoughts, motivations, and behavior ofindividuals who occupy positions in thoseinstitutions. The effectiveness of this linkrests on culturally shared meanings andexpectations attached to role identities(Burke and Tully 1977; Heise 1979) andthat these expectations become internal-ized into individuals’ self-concept, therebyguiding the behavior of role occupants.The relationship, however, between rolesand behavior is complex, often resultingin only a loose fit between role expecta-tions and role performance (Stryker

1968; Turner 1962). In an attempt to rec-oncile this gap, theorists have drawnupon symbolic interactionist thought(Mead 1934), invoking the self as a medi-ating factor between role expectationsand individual behavior (e.g., McCalland Simmons 1978; Stryker 1968, 1980;Turner 1962). They argue that roleperformances match role expectations

1University of Iowa, Iowa City, IA, USA

Corresponding Author:Mark H. Walker, Department of Sociology,University of Iowa, Iowa City, IA 52242, USAEmail: [email protected]

Social Psychology QuarterlyXX(X) 1–29

! American Sociological Association 2013DOI: 10.1177/0190272513482929

http://spq.sagepub.com

insofar as roles are internalized andbecome part of the self-concept.

What, then, facilitates the internaliza-tion of role meanings? Previous workemphasizes social and emotional attach-ments to role-based groups (e.g., Stryker1980; Stryker and Burke 2000; Strykerand Serpe 1982) as well as the rewardsderived from investments in a role(McCall and Simmons 1978; Turner1978). In this study, we develop a newargument about how the extent to whichrole meanings are widely and routinelypresent in the social encounters of anindividual is potentially vital in shapingthe self-concept. The argument is tied toa concrete structural feature of personalsocial networks—the embeddedness ofrole-based others—that shapes the over-all prevalence of role meanings in oursocial worlds. In short, we argue thatthe structure of personal social networkscan affect which role meanings are pres-ent in social encounters by increasingthe likelihood that role-based others (i.e.,individuals with whom interaction isbased on role-occupancy) will be presentin any given situation.

An innovative vignette is used toengage respondents in a free recall taskof their personal networks. The vignettewas designed to elicit a large sample ofmeaningful social ties, from which weare able to identify the structural loca-tions of role-based others through a seriesof follow-up questions. In keeping withour structural theory of role internaliza-tion, our central hypothesis is that theoverall embeddedness of role-based othersin one’s personal social network increasesthe salience of the identity associatedwith that role, independent of the socialand emotional attachment to role-basedgroups. The results suggest that a deeperconsideration of network structure doesindeed improve our understanding ofrole internalization.

INTERPERSONAL TIES, SOCIAL ROLES,AND THE SELF

Sociological theories of self and identitygenerally view interpersonal socialrelations as an essential factor in theformation of self and role internalization(Cooley 1902; Mead 1934). Commonlycited sources of role internalizationinclude the number of significant othersan individual has role-based relationshipswith (Turner 1978), the amount of invest-ment in the role identity (McCall andSimmons 1978; Turner 1978), and therewards that are derived from role incum-bency and successful role performance(McCall and Simmons 1978; Rosenberg1979). Although many theories imply theimportance of interpersonal social ties,identity theory (IT; Stryker 1980) is per-haps the most explicit about the relation-ship between social ties and the self. Spe-cifically, IT argues that role expectationsare internalized into the self-conceptthrough commitments to role-basedgroups. Commitment is defined as socialand emotional attachment to a role-basedgroup, or the extent to which an individu-al’s relationships to others are contingentupon possessing a particular role andidentity:

Commitment, as a particularizedtranslation of ‘‘society,’’ focuses onsocial networks: the number of othersto whom one relates through occu-pancy of a given position [and] the‘‘importance’’ of others to whom onerelates through occupancy of a givenposition. . . . The concept of commit-ment can lead as deeply into socialnetworks as a theorist is prepared togo. (Stryker 1980:81)

Although this definition clearly recog-nizes the importance of social networksto self-structure, it nonetheless hasgalvanized a line of research in which

2 Social Psychology Quarterly XX(X)

commitment is typically measured not viaactual network ties but rather by queryingindividuals’ subjective assessments of thesocial and emotional loss they would incurwere they to no longer inhabit a given role(e.g., Serpe 1987; Stryker and Serpe 1982,1994). In fact, to date, social networkshave been largely relegated to metaphorin identity research, and the vast majorityof empirical studies lack a detailed investi-gation of network structure (see, however,McFarland and Pals [2005] for a notableexception).

Consequently, the features of socialnetworks that matter with respect to theformation of identities have received littleempirical and theoretical attention. Inthis study, we attempt to remedy thisissue by extending IT to examine thelink between personal networks and theself using formal network methods. Wethus build on the work of Stryker and col-leagues (Callero 1985; Stryker 1980;Stryker and Serpe 1982, 1994) by extend-ing the concept of commitment ‘‘deeper’’into social networks with an empiricalinvestigation of the relationship betweenthe structure of personal social networksand the self. Our formal networkapproach allows for the identification oftwo distinct features of personal socialnetworks that could be affecting howsocial ties shape the self-structure: (1)social and emotional attachment to role-based groups and (2) the embeddednessof role-based others in a person’s socialcircle. The attachment approach buildsdirectly on the notion that a particularrole identity becomes more salient ifone’s personal network is composed ofnumerous strong ties to others withwhom we enact that role (Stryker 1980;Turner 1978). Embeddedness, in contrast,refers to the idea that role identitysalience may depend independently onthe breadth of access (vis-a-vis socialties) that an individual’s role-based grouphas to the rest of his or her social world.

SOCIAL NETWORKS AND THESELF-STRUCTURE

Previous research on self-concept forma-tion tends to focus only on those relation-ships specific to a given role without con-sidering an individual’s overall fabric ofmultiple roles and relationships. Here, weargue that the systematic investigation ofthe structure and content of personal socialnetworks provides important informationin that it allows us to examine how individ-uals’ role-based groups function in relationto other role-based groups in the entire sys-tem that is their personal social network.1

The hallmark of social network analysisis to elicit the structure and content ofactual social ties (i.e., to go beyond the met-aphor) from which researchers can con-cretely analyze the structure of social rela-tions. The potential of this approach for thestudy of self and identity lies in the factthat our conceptualization of why socialties matter can be refined by takinga more rigorous approach to measuringthe ‘‘importance’’ of role-based others inpersonal social networks.

First, social network analysis providesa more precise measure of the strengthand number of ties to role-based otherscompared with the traditional methodsof measuring ties to role-based groups.Researchers who employ social networkanalysis to study personal social networksuse name generators to elicit the names ofpersons (alters) with whom the respon-dent (ego) has social ties and askthe respondent a number of questionsabout the alters in their personal socialnetwork (e.g., Marin and Hampton 2007;

1This approach aligns with the general viewespoused by identity theorists that the self is com-posed of multiple identities that are attached tomultiple roles through patterned ties to role-based others (RBOs), which in turn are embeddedin the larger network of social ties that compriseone’s personal social network (Owens, Robinson,and Smith-Lovin 2010).

The Embedded Self 3

McPherson, Smith-Lovin, and Brashears2006, 2008; Wellman 1979). The namegenerator used here (see Marsden [1990,2005] for a review of standard surveymethods for network data) elicits a nota-bly extensive personal social network(mean = 17.83 alters), catalogs thestrength of ego-alter ties in the personalsocial network, and elicits the tiesbetween individuals in respondents’ per-sonal social networks (alter-alter ties).

Second, name generators afford a moreinclusive and representative look atrespondents’ personal social networksthan methods traditionally used to mea-sure commitment. Rather than merely elic-iting ties to role-based others, the namegenerator used in this research wasdesigned to elicit meaningful social tiesbeyond respondents’ ‘‘inner circle’’ of socialcontacts, as well as to probe for additionalrole-based ties. Third, social network analy-sis offers a means of directly measuring theextent to which social groups are intercon-nected. Rather than examining role-basedgroups in isolation, a network approachallows us to ‘‘see’’ the ties connecting a givenrole-based group to ego’s overall social land-scape (i.e., extra-group ties), which we useas a measure of embeddedness.

For example, as depicted in network Aof Figure 1, the commitment index used inprevious research regarding the relation-ship between the self and the externalsocial structure (Stryker and Serpe1982) only gathers the ties that one hasto role-based others. That is, the tradi-tional approach focuses on the raw num-ber of role-based alters within an individ-ual’s personal network (e.g., in networkA, ego has three role-based alters).Although counting role-based others isnot problematic per se, it is possible thatcounts alone could be a misleading indica-tor of the presence of role-based others inego’s social world.

Consider, for example, networks Bthrough E, which illustrate the structure

of four personal social networks thatinclude role-based others in addition tonon–role-based others (hereafter referredto as RBOs and NRBOs, respectively).Networks B and C indicate the informa-tion added by gathering non–role-basedas well as role-based ties. Using tradi-tional methods of measuring commit-ment, networks B and C would be consid-ered equivalent with respect to thepresence of RBOs, since both egos havethree RBOs. However, by includingNRBOs, B and C are no longer equivalent;half of B’s personal network is comprisedof RBOs whereas only a quarter of C’spersonal network is comprised of RBOs.Furthermore, networks D and E of Figure1 illustrate the different ways in whichRBOs can be embedded in one’s personalsocial network. RBOs might be relativelyisolated from the rest of ego’s personalnetwork, as in network D, or they maybe deeply ‘‘entrenched’’ in the rest ofego’s personal social network, as in net-work E. Clearly, as illustrated by net-works B through E, studying RBOs alonecannot distinguish between importantways in which the network compositionof RBOs and NRBOs can differ acrossindividuals. Thus, a more comprehensiveapproach to understanding social net-works and identity salience involves gath-ering data on NRBOs as well as the tiesbetween RBOs and NRBOs.

In the following, we outline the proper-ties of personal social networks that canbe derived from identity theory’s concep-tion of commitment and propose howthese features of personal social networksmay operate in the internalization of rolemeanings into the self-structure. Next,we discuss how the structure of personalsocial networks can affect the breadth ofsocial encounters where role meaningsare relevant, which in turn can affectthe extent to which a role is internalizedinto the self-structure. Again, we focuson two features of personal networks


through which social networks mayimpact the self: (1) social ties as socialand emotional investments and (2) socialties as creating webs of embeddedness.

Social and Emotional Attachment toRole-Based Groups

One way that social networks might facil-itate the internalization of roles into the

self-concept is, as implied by identity the-ory (IT) and other theories of self (e.g.,Stryker 1968, 1980; Stryker and Burke2000; Stryker, Serpe, and Hunt 2005;Turner 1978), through the social andemotional attachments one has to role-based groups. This approach suggeststhat social ties are essentially invest-ments that we work to maintain by livingup to the self-meanings that are attached

Figure 1. Egocentric Network Examples

The Embedded Self 5

to those ties. Commitment, then, maybe thought to refer to the extent to whichwe are attached to role-based groups,either through the emotional closenessof social ties to RBOs or throughhow socially invested we are in a role-based group via the portion of one’ssocial contacts that are based on a partic-ular role. By this account, roles becomeinternalized as salient identities becauseof the social and emotional investmentsthat are ‘‘tied up’’ in role-based groups.Such an approach points to a bonding orbinding mechanism (Borgatti and Lopez-Kidwell 2011), wherein one’s social tiesto role-based others serve as social andemotional investments that ‘‘bind’’ themto the role-based group and in turn theself-meanings that are attached to thegroup.

Although social attachment is gener-ally conceptualized as the raw number ofties to RBOs (e.g., Callero 1985; Nutt-brock and Freudiger 1991; Serpe 1987;Stryker and Serpe 1982, 1994), Figure 1suggests that an alternative operationali-zation of social attachment may be war-ranted. Recall that commitment refers tothe loss that one would incur upon leav-ing a social role, and therefore forgo therelationships that they have to RBOs. Asdescribed earlier, networks B and C inFigure 1 would be considered equivalentusing the traditional approach to commit-ment given that both networks containthree RBOs. But networks B and Cclearly raise the issue of whether the pro-portion of ties to RBOs better suits theidea of ‘‘loss incurred.’’ Because person Bis situated in a 6-person social circle ver-sus a 12-person circle, person B standsto lose half of his or her social contactsby forgoing relationships to RBOswhereas the actor represented in networkC only stands to lose one-quarter of theirnetwork. Accordingly, we use a propor-tional measure of social attachment to

account for these potentially importantdifferences.2

Hypothesis 1: An increase in the propor-tion of ties to role-based others willbe associated with an increase in thesalience of the identity associatedwith that role.

Hypothesis 2: An increase in the strengthof ties to role-based others will beassociated with an increase in thesalience of the identity associatedwith that role.

The Embeddedness of Role-BasedOthers

A potential limitation of the attachmentapproach is that while it considers theextent to which one’s social relations arebased on a particular role identity, itdoes not specifically address the extentto which one’s social interaction is basedon a given role identity. We argue thatalthough these two concepts may beempirically related, they are analyticallydistinct in that they suggest different pro-cesses underlying the link between socialnetworks and role internalization.

2We also suggest that a proportional measureprovides a better theoretical fit with identitytheory’s view of self-structure. Since identity the-ory views the self-structure as being hierarchi-cally organized, identity salience is a relative con-struct, meaning that an identity’s position in thesalience hierarchy is contingent upon its level ofsalience relative to other identities. Thus, it seemsreasonable to assume that social attachment isa similarly relative construct. To illustrate, imag-ine someone whose network consists of only threeschool-based ties and three work-based ties (as inFigure 1B, assuming the black nodes representschool-based ties and the white nodes representwork-based ties). If this person were to form addi-tional relationships with people from each groupat the same rate, the proportional measure ofsocial attachment for each identity would remainunchanged. On the other hand, a count-basedapproach would assume that social attachmentfor both identities is increasing even though thesalience hierarchy would remain unchanged.


Focusing on identity-based interaction(rather than role-based ties) suggeststhat role-based others (RBOs) affect roleinternalization by altering ego’s socialenvironment, rather than through induc-ing group commitment. Indeed, many the-ories of self highlight the importance ofidentity-based interactions for the self-structure. For instance, McCall and Sim-mons (1978) argue that the self-structureis largely a product of role-support and sug-gest that the identities that can be reason-ably invoked in any given social encounterare constrained by the ‘‘opportunity struc-tures’’ available for self-verification. Oppor-tunity structures in this sense refer to thetypes of role performances that are likelyto be supported by one’s audience and arelargely determined by the identities ofthose present in the social encounter.Thus, since role performance is largelya product of one’s audience, we proposethat role internalization will be affectedby the proportion of social encounters thatconsist of the type of audience that invokesthe performance of a given role (i.e., socialencounters that contain RBOs).

Additionally, the social and emotionalattachment approach overlooks the essen-tial fact that the people we are connectedto may also be connected to each other.These alter-alter ties may matter for theway others in our social networks see usand, ultimately, the way we see ourselves.Indeed, extant research using social net-work analysis to study identity processesalludes to the importance of examiningself-meanings as being shaped by tiesbetween, rather than simply within,diverse social groups. For instance,McFarland and Pals (2005) show thatidentity imbalance (i.e., the extent towhich self-held identities do not matchthe identities one’s social contacts attachto them) is a dual function of (1) an individ-ual’s location within the structure of socialties and (2) the extent to which an individ-ual is tied to a diverse array of others fromdifferent ‘‘crowds.’’ They find that highschool students who inhabit ‘‘structural

holes’’ (i.e., students who have ties acrossvarious clusters of a peer network) aremore likely to report an imbalance withrespect to how they see themselves andtheir perceptions of how their peers seethem. Importantly, then, this researchhighlights how identity is affected byboth the overall structure of ego’s ties aswell as the differing identities of ego’ssocial contacts. For example, identityimbalance is less likely to occur if ego isembedded in a densely knit group of otherswho see themselves in the same way thatego does than if he or she is tied to otherswho are sparsely connected.

Building on these insights, we proposethat the overall fabric or structure ofa social network can inform the extentto which an individual’s interaction isbased on a particular role identity. In con-trast to the idea that actors can be sociallyand emotionally attached to a role-basedgroup, we ask whether the ties betweenthe role-based group and the rest ofego’s personal network can shape roleidentity salience and thus the internaliza-tion of a given role. There are two keyways in which the embeddedness ofRBOs—a structural feature of a personalnetwork—can affect the extent to whichone’s interaction is based on a particularrole identity. First, embeddednessincreases the probability that an RBOwill be present in ego’s social encounters.For example, if Jane, a college studentand member of a sorority, were to attendchurch weekly but keep her religiousfriends relatively separate from her soror-ity sisters and other friends (as in networkD of Figure 1), then she would likely onlysee her religious friends at church func-tions. Alternatively, if Jane’s religious-based alters were extensively tied to hersorority sisters (as in network E), then itis likely that her religious friends wouldbe present at sorority functions as well asother social outings she attends.

Second, having RBOs and non–role-based others (NRBOs) present in thesame social encounter can also facilitate

The Embedded Self 7

the diffusion of role meanings to others inego’s personal social network, due to thefact that observing and participating inrole performance sends a stronger andmore clear signal of ‘‘who Jane is’’ thanmerely knowing about role-incumbency.3

In addition, successful role performancesimply that the social actor negotiatesa definition of the situation wherein thosepresent in the social encounter come tosee (or are perceived to see) the actor ina way that reflects the actor’s self-heldrole meanings (Burke and Stets 2009).Thus, the very nature of role performancein the context of self-verification suggeststhat ego is actively engaged in trying toconvince NRBOs of his or her legitimacyas a role-occupant. As a result of this sus-tained, careful identity work, the copre-sence of RBOs and NRBOs in socialencounters serves the dual function ofbringing role-based meanings into thesocial encounter and facilitating the diffu-sion of role meanings to the NRBOsinvolved in the social encounter and shouldcontribute especially to identity salience.

The previous discussion underscoresan important difference between the wayembeddedness is conceptualized here (asan alter-alter phenomenon) and the wayit has been employed in previous research.Sociologists have long recognized theimportance of embeddedness for individualoutcomes (Granovetter 1985; Moody andWhite 2003; Portes and Sensenbrenner1993) but have focused primarily onembeddedness from the perspective ofego-alter relations. For example, a keytheme that emerges from the economicsociology literature on embeddedness ishow ties between firms reduce transactioncosts and uncertainty in market

transactions (Granovetter 1985) and long-standing relationships facilitate fine-grained information transfer in a waythat ‘‘arms-length’’ relations cannot (e.g.,Uzzi 1996, 1999). Other embeddednessresearch focuses on the link between nodeconnectivity and social cohesion (Moodyand White 2003); a student who is highlyembedded (i.e., densely intertwined) ina group of friends at school is expected toreport being more attached to her school.

In contrast to focusing on how individu-als are embedded in groups, we take whatis essentially an alter-based approach toembeddedness. That is, our conception ofsocial fabric is based on alter-alter ties asopposed to ego’s relations with variousalters. Such an approach is intended tocapture how ‘‘involved’’ or ‘‘active’’ certaingroup members are in an individual’soverall social life. We do not mean thisin the colloquial sense of ‘‘being involvedin someone’s life,’’ as a parent who par-ticipates in parent-teacher conferencesor supervises homework is involved inhis or her child’s life. Instead, ourapproach to embeddedness implies thata given friend or partner has the poten-tial to be more or less entrenched orinvolved in ego’s social circle (and conse-quently, more or less able to exert influ-ence or constrain ego’s identity forma-tion) to the extent that he or she isdirectly tied to the others ego knows.

An important feature of our theory ofembeddedness is that it describes howrole meanings can transcend beyond theorganization in which they are embedded(e.g., church, the workplace, school) andextend to social encounters outside theinstitutional setting. We argue that socialties connecting RBOs to NRBOs canbreak down the institutional boundariesthat can keep role meanings localized inthe relevant institutional setting. Forexample, our theory can account for howpeople outside of ego’s religious congrega-tion can come to associate religious

3This is consistent with Granovetter’s(1985:490) discussion of how information gath-ered from past dealings with an individual issuperior to information gathered from a thirdparty because of the fact that it is ‘‘richer, moredetailed, and known to be accurate.’’


meanings with ego, without viewing egoas the (sole) agent for this process.

In sum, we hypothesize that it is notsimply the proportion of RBOs that mat-ters for identity salience but the ‘‘reach’’those RBOs have into an individual’ssocial fabric. An increase in the breadthof ties a role-based group has to the restof ego’s social network likely increasesthe proportion of social settings for egothat contain self-relevant meanings asso-ciated with a given role; if RBOs are tiedto a large portion of ego’s social network,then they are likely to be invited to thesame parties, frequent the same restau-rants and bars, and attend many of thesame social functions as ego. This servesto alter the opportunity structures avail-able for identity verification, which inturn makes self-verification increasinglycontingent upon the successful enactmentof a given role identity. The drive to verifyidentities and the emotional gratificationone experiences through identity verifica-tion (Stets 2005) should result in theinternalization of the identity (i.e., anincrease in the salience of the identityassociated with that role).

Hypothesis 3: An increase in the embedd-edness of role-based others in ego’spersonal social network will be associ-ated with an increase in the salience ofthe identity associated with that role,net of social and emotional attach-ment to the role-based group.

In summary, our embeddednessapproach suggests that network flowand adaptation processes are at play inthe relationship between social networksand role internalization (Borgatti andLopez-Kidwell 2011). In contrast to theidea that social ties induce group commit-ment, our theory of embeddedness propo-ses that the embeddedness of role-basedothers affects the social environmentof an individual (i.e., the opportunity

structures for self-verification), which inturn causes individuals to adapt by con-structing interactional strategies for self-verification that include behavior thataligns with the relevant situational rolemeanings and expectations. Additionally,this approach acknowledges that theeffect of embeddedness on an individual’ssocial environment can be amplified bythe diffusion of role meanings throughoutthe individuals’ personal social network.

In the following we test the aforemen-tioned three hypotheses across three roleidentities using detailed personal networkdata from a sample of young adults (18–24years old). Following previous work inidentity theory (e.g., Burke and Reitzes1991; Stryker et al. 2005; Stryker andSerpe 1982, 1994), we examine the impactof personal social network structure on thesalience of the student, religious, and workidentities. The data we collect allow us tomeasure the proportion of ties to RBOs,the emotional strength of those ties, andthe embeddedness of role-based groupswithin respondents’ overall networks.

DATA AND MEASURES

Data on personal social networks werecollected via an online survey fora national sample of 198 young adultswho were between 18 and 24 years oldand childless.4 As shown in Table 1, thesample is mostly white (69.7 percent),female (55.05 percent), and roughly 22years old on average. About 30 percentof the respondents had a bachelor’s

4QualtricsTM hosted the survey and managedthe recruitment and compensation of partici-pants. Participants either signed up directlywith one of Qualtrics’ online vendors or theyresponded to an advertisement posted on selectwebsites (e.g., airline company websites for VIPmembers and restaurant club members). Thequestionnaire was completed online and partici-pants were compensated with cash-equivalentrewards (including airline miles, magazine sub-scriptions, and gift cards).

The Embedded Self 9

degree or higher and the average respon-dent had about two years of college educa-tion. Respondents tended to originatefrom middle-class families (mean paren-tal income = $66,020; SD = $52,133).The questionnaire is given in theappendix.5

Personal Social Networks

The name generator is the standardmethod of enumerating an individual’spersonal social circle (Marin and Hamp-ton 2007; Marsden 2011). Followinga recent push to elicit extensive personalsocial networks (e.g., Hogan, Carrasco,and Wellman 2007; Marin 2004; Marinand Hampton 2007; Mollenhorst, Volkerand Flap 2008a, 2008b), the name gener-ator used in this study (also see theappendix) is designed to elicit more thanjust the ‘‘inner circle’’ of social contactswho matter in the respondents’ lives(i.e., both weak and strong ties). Indeed,our goal is to elicit wide-ranging personalsocial networks that potentially includeties to individuals from multiple sectorsof social life. To this end, respondentsare presented with a vignette-style namegenerator plus a set of probes for addi-tional interaction partners (see theappendix for the survey protocol).

Eliciting network alters. Following informedconsent and a short series of questions onwork, school and church participation,respondents are presented with a vignette-style question about their social networksalong with visual instructions:

We want to learn about the people inyour life. Imagine that you have wonan award that you are very excitedabout. Those presenting the award toyou want to throw a party in yourhonor. They will pay for all of theexpenses related to the party, includ-ing food, drinks, travel, and housingfor guests. Who would you invite?

On the following screen, the question isrepeated and respondents are given 25blank text boxes where names can beentered. Respondents are instructed toinput only first names and last initialsand reminded that duplicates are notallowed (see the appendix for exact

Table 1. Demographic Characteristics(n = 198)

Variable

RacePercentage white 69.70Percentage nonwhite 31.30

SexPercentage male 45.95Percentage female 55.05

Parental income ($)Mean 66,020.41Median 50,000.00

EducationMean years 14.04Percentage four-year

degree or higher29.80

AgeMean 22.05

Hours of studying (n = 90)Mean 13.33

Religious attendanceMean 15.22

Hours of workper week (n = 134)Mean 34.54

5Although our sample is a convenience sam-ple, and not representative of the U.S. populationat large, we do not view this as a problem for theissue at hand. Since the purpose of this researchis to (1) examine an alternative, network-basedspecification of the effect of interpersonal socialrelations on the self and (2) examine the relativeeffects of attachment and embeddedness on roleinternalization, our ability to generalize toa larger population is not of great importance.Additionally, restricting our sampling frame toa relatively specific demographic populationallows us to tailor the identity salience items toinclude identities that are likely to be relevantto a large portion of the respondents.


wording). We refer to this as the respond-ent’s ‘‘guest list’’ of alters.

Since pilot data indicated that thisname-eliciting vignette tended to gener-ate strong ties (Walker 2010), additionalprobes are used to identify alters withwhom the respondents were less closebut still had meaningful relationshipswith in their everyday lives. Thus, afterrespondents generated their guest listsand completed a set of name interpreters(see the following discussion), they areprompted to name any additional contactsthat they ‘‘interacted with on a regularbasis’’ but who were not currently on theguest list. For example, if respondentsreported attending religious services atall (i.e., all responses except for ‘‘never’’),they were asked to identify otherreligious-based contacts that were notyet on the guest list:

Can you think of other people whoattend the same religious services asyou that you see or talk to on a regularbasis who are not on this list? If yes,please name them below (regardlessof how close you are to them).

These questions were then followed bya set of name interpreters for these probe-based alters (see part 5 of the appendix).

Alter-alter relationships. To gather thestructure of alter-alter ties in respond-ents’ personal social networks, respond-ents were asked to indicate which of thealter pairs know each other ‘‘well enoughto stop and chat if they passed oneanother on the street’’ (check box if yes).This threshold of ‘‘knowing’’ is similar tothat used in the General Social Survey(GSS; 2006) ‘‘number known’’ moduleon social capital (see DiPrete et al.2011). The alter-alter questions aredivided into four categories: (1) guestdyads, (2) within-role contact dyads, (3)guest-contact dyads, and (4) between

role-contact dyads (see parts 6, 8, 9, and11, respectively, of the appendix).

Subgraphs and role-based groups. Oncethe personal social networks were gener-ated, role-based name interpreters (Mars-den 1990, 2005; McCallister and Fischer1978) were used to generate role-basedsubgraphs (see parts 3 and 5 of theappendix). These role-based subgraphscontained only the alters in the respond-ent’s personal social network who hada particular role-based relationship withthe respondent. For example, the reli-gious subgraph is generated by askingthe respondent to indicate which of hisor her alters attend the same religiousservices as they do. Following previouswork in IT, social ties based on the stu-dent, religious, and work roles are gener-ated (e.g., Burke and Reitzes 1991; Nutt-brock and Freudiger 1991; Stryker andSerpe 1982).

Identity Salience

Identity salience is defined as the proba-bility of enacting an identity in andacross social situations (Stryker andBurke 2000). Following previous workin identity theory (e.g., Merolla et al.2012; Serpe and Stryker 1987; Strykerand Serpe 1982, 1994), identity salienceis measured by asking respondents howlikely it is that they would invoke eachidentity in a number of different socialsituations. Specifically, they were askedhow likely they would be (1 = veryunlikely, 6 = very likely) to mentioneach aspect of self (i.e., work/career,school/academics, and religion or churchactivities) in four different social situa-tions: (1) meeting a roommate for thefirst time, (2) meeting someone at a party,(3) meeting a friend of a close friend, and(4) giving a short speech about oneself.The salience of each identity is measuredas the mean value of a given identity

The Embedded Self 11

across all four social situations. Thesalience items for the religious, student,and work identities each load on a singlefactor and are within acceptable levels ofscale reliability (a = .91, .84, and .82,respectively).

For the identity salience measures,respondents are asked to consider fiveidentities even though our analysis isfocused on explaining only the religious,work, and student identities. Of the fiveidentities (family, work, religion, friend,student) that have been most commonlystudied by identity theorists (Stets andBiga 2003; Stets et al. 2008; Stryker andSerpe 1982, 1994), only three (religion,work, and student) are relevant tothe scope of our theory (i.e., that embed-dedness should be relevant for identi-ties associated with institutions [e.g.,churches, workplaces, schools] given

that the reach of role-based others [socialties] can transcend these boundaries).6

Social and Emotional Attachment

Social attachment for each role-basedgroup is measured as the proportion ofego’s personal network who are role-based others (RBOs). For example, social

Figure 2. Network with High EmbeddednessNote: Ego is not included in this figure. Node size reflects strength of tie between ego and alter.

6We included these additional identities in oursurvey because it is useful to have respondentsrate multiple identities so that they can havea way to ‘‘anchor’’ their responses to other identi-ties. Although we—and others (e.g., Merolla et al.2012)—have abandoned the traditional rank-ordering approach to measuring identity saliencefor methodological reasons, we argue that givingrespondents a set list of common identities aidsin providing an adequate fit to the theoreticalconstruct of ‘‘salience hierarchy,’’ since it allowsthem to anchor their responses to the identitiesof theoretical interest relative to other identities.


attachment for the religious identity ismeasured as the number of alters thatego indicated as attending the same reli-gious services as they do (i.e., the size ofthe religious subgroup) divided by thetotal number of alters in the network. Inaddition, respondents were asked to indi-cate how close they felt to each alter, emo-tionally (not very close = 1, extremelyclose = 4). Emotional attachment toa given role-based group is measured asthe average strength of ties between egoand members of the role-based group.

Embeddedness of Role-Based Group

We conceptualize embeddedness ofrole-based others as the breadth of accessthe role-based group has into ego’soverall social fabric. Thus, we measure

embeddedness by calculating the propor-tion of ego’s personal social network thatis composed of either RBOs or those whoare socially tied to the role-based group.To illustrate, we display the personalsocial networks of two respondents (Fig-ures 2 and 3) with ego excluded. The net-works in Figures 2 and 3 have role-basedgroups that are proportionally similar insize (5/19 = .26 and 9/32 = .28, respec-tively) but differ dramatically in theirlevel of embeddedness ([5 1 13]/19 = .95and [9 1 1]/32 = .31, respectively). Thisis due to the fact that a vast majority ofthe non–role-based others (NRBOs) inFigure 2 are tied to a member of therole-based group (only one alter is isolatedfrom the role-based group), while in Fig-ure 3 the role-based group is nearly iso-lated from the rest of ego’s personal social

Figure 3. Network with Low EmbeddednessNote: Ego is not included in this figure. Node size reflects strength of tie between ego and alter.


network save for one NRBO bridge. Thismeasure thus reflects the ‘‘reach’’ of therole-based group in the respondent’ssocial network in that it captures the pro-portion of ego’s social landscape thatRBOs have access to.7 To be sure, ourmeasure of embeddedness is also empiri-cally distinct from social attachment.Table 2 shows that the correlationbetween embeddedness and social attach-ment ranges from low to moderate for thethree roles under study.

Role-Behavior

Role-behavior for each identity is alsoincluded in our analysis to address thepossibility that the relationship betweennetwork variables and identity salienceis spurious. For example, it may be thatthe frequency of attending religious serv-ices both increases how much one

identifies as a religious person and theextent to which one’s personal social net-work consists of religious people. Further-more, frequent attendance of religiousservices may, over time, increase theextent to which religious-based altersare connected to nonreligious alters inego’s personal network through processesof transitivity (Holland and Leinhardt1972, 1977); religious attendance couldbe driving both the salience of the reli-gious identity and connecting religion-based alters to the rest of ego’s personalnetwork. Traditionally, role-behavior isthought to be an outcome of identitysalience (Stryker 1980), but at timesrole-behavior can be constrained by exter-nal forces or may be a product of other sit-uational contingencies. For example, theamount of time one spends studying out-side of school may be due to a demandingmajor or a heavy course load, and thenumber of hours one works likely has todo with economic considerations andother role-obligations. Since we are pri-marily concerned with isolating the rela-tionship between network structure andidentity salience, we include measures ofrole-behavior in our models to rule it outas a possible confound.

Table 2. Correlation Matrices for Key Independent Variables

Emotional attachment Social attachment Embeddedness

Religious identityEmotional attachment 1Social attachment –.33 1Embeddedness .00 .46 1

Student identityEmotional attachment 1Social attachment .03 1Embeddedness .38 .34 1

Work identityEmotional attachment 1Social attachment .15 1Embeddedness .39 .55 1

Note: Emotional attachment refers to the average strength of ties to role-based others and socialattachment refers to the proportion of ties to role-based others.

7These calculations are made using only alter-alter ties. This measure is akin to a group-levelmeasure of centrality (excluding ego), whereonly nonredundant ties are considered. Nonre-dundancy in this case refers to the fact thatnon–role-based others (NRBOs) are viewed aseither tied to the role-based group or not (i.e.,the measure is not weighted by the number ofties a given NRBO has to the role-based group).


To measure religious role-behavior,respondents were asked to report on thefrequency with which they attend reli-gious services (never, less than oncea year, once or twice a year, several timesa year, about once a month, two to threetimes a month, almost every week, everyweek, several times a week).8 Our mea-sure of student role-behavior is respond-ents’ self-reports of the amount of hoursper week they spend outside of class doingschoolwork and our measure of work role-behavior was the reported number ofhours per week respondents work for pay.

ANALYTIC PLAN

In order to test the hypotheses outlinedpreviously, we estimate separate ordinaryleast squares (OLS) regressions predict-ing the religious, work, and student iden-tities. For the religious identity, threemodels are estimated: model 1 is a base-line model estimating the extent to whichreligious identity salience is related torole-behavior and other control variables.Seven control variables are included (race[white/nonwhite], age, gender, years ofeducation,9 marital status [married/notmarried], parental income, and networksize) to address the possibility that demo-graphic factors may affect salience (e.g.,Benson, Donahue, and Erickson 1989;Wilson and Sherkat 1994) as well as net-work structure (Lee, Campbell, andMiller 1991; Lin and Dumin 1986; Moore1990). In model 2, we also include thesocial and emotional attachment varia-bles. Finally, model 3 estimates the effectof the embeddedness of RBOs on religiousidentity salience, net of demographiccharacteristics, role-behavior, and meas-ures of social and emotional attachment

to the role-based group. We then run thesame three models for both the studentand work identities.

Due to the scope of our theory andthe nature of the indices used for ournetwork-based analyses, we restrict oursample based on two conditions. First, ineach model we restrict our sample tothose who participate—at some level—ina social institution where the relevantrole is embedded. For the religious iden-tity, those who indicate that they neverattend religious services are not includedin the analysis predicting religious iden-tity salience. Likewise, only respondentswho indicate that they are currentlyenrolled in school (employed) are includedin the analysis predicting student (work)identity salience. These restrictionsreflect the fact that our theory of embedd-edness addresses how social structureincreases/decreases role identity saliencefor those whom the role is applicable; thetheory, however, is not well equipped todeal with why some choose to adopt a roleand others do not. Second, since the aver-age strength of ties to role-based otherscan only be calculated if there is at leastone RBO in their personal network, onlyrespondents with one or more role-basedtie are included in the analyses.10

RESULTS

Table 3 shows that the networks gener-ated using this method were quite large(mean = 17.83; SD = 9.91), moderatelydense (mean = .40; SD = .17), and moder-ately close (mean = 2.61 on a scale from 1to 4 for emotional closeness; SD = .52).This indicates that the name generator

8We collapse ‘‘almost every week,’’ ‘‘everyweek,’’ and ‘‘several times a week’’ into a singlecategory.

9For those currently enrolled in school, educa-tion indicates their expected degree.

10Analyses where respondents without anyRBOs are given a value of 0 for average strengthof ties provide substantively identical results. Wechose to present the models excluding theseobservations so as to avoid building anyunfounded assumptions regarding the strengthof ties to role-based others into our analyses.


used in this study tended to elicit rela-tively extensive personal networks withmoderate levels of connectivity and tiesthat range from emotionally intense toweak. For the average social network,slightly more than half of the ties arereported as ‘‘close’’ or ‘‘very close’’ (seeTable 3). Roughly 30 percent are reportedas ‘‘somewhat close,’’ and 16 percent are‘‘not at all close.’’

Beginning with the religious identity,Table 4 shows that religious-based sub-groups constitute a relatively small por-tion of personal networks but have higherlevels of density than personal networksas a whole. This is consistent with theclustering that one would expect of mean-ingful subgroups in personal networks.For people who attend religious services(n = 129), the religious subgroup tendsto consist of just under five alters (mean= 4.67; SD = 5.51) and constitutes abouta quarter of their personal social network(mean = .24; SD = .24). The religious sub-group tends to be quite dense (mean =.687; SD = .324) and respondents reportbeing, on average, ‘‘close’’ to their reli-gious contacts (mean = 3.01; SD = .71).Finally, respondents with religious sub-groups tended to display moderatelyhigh levels of embeddedness—on average57.9 percent of alters in respondents’ per-sonal social networks either attend thesame religious services as ego or are tiedto a religion-based other. To be sure, how-ever, there was nontrivial variation

across respondents in the extent ofreligion-based group embeddedness(SD = 37.8 percentage points).

With respect to the student identity,respondents who were enrolled in school(n = 90) had an average of just over threealters in their student subgroup, whichcomprised about 18 percent of their per-sonal social network (see Table 4). Thissubgroup tended to be slightly less densethan the religious subgroup (mean =.575; SD = .38), and the average strengthof ties to the student subgroup wasslightly lower than the religious subgroup(mean = 2.87; SD = .78). Finally, similar tothe religious subgroup, embeddedness forthe student subgroup was 52.4 percenton average (SD = 34.5 percentage points).Of the three identities, the work subgrouptended to be the smallest (mean = 2.36)and displayed the lowest amounts of close-ness and embeddedness (see Table 4).

The results of the OLS regression pre-dicting religious identity salience arereported in Table 5. Model 1 indicatesthat controlling for other demographicvariables, frequency of religious atten-dance and marital status are the onlycontrol variables that significantly affectreligious identity salience. Controllingfor other variables in the model, individ-uals who reported attending religiousservices ‘‘nearly every week’’ or morehad a 1.331 more salient religious iden-tity than those who attended ‘‘less thanonce a year’’ (p \ .001). Additionally,married respondents tended to havemore salient religious identities thannonmarried respondents (p \ .05),although this relationship becomes non-significant when social and emotionalattachment variables are included inthe model (see model 2).

Model 2 indicates the impact of thesocial and emotional attachment varia-bles, showing that controlling for reli-gious attendance and other control varia-bles, the strength and proportion of ties to

Table 3. Network Characteristics (n = 198)

VariableMean or

percentage

Size 17.83Density 0.40Closeness 2.61

Percentage very close 23.61Percentage close 29.80Percentage somewhat close 30.58Percentage not at all close 16.00


religion-based others do not significantlyimpact the salience of the religious iden-tity, providing a lack of support forHypotheses 1 and 2. This suggests thatsimply being tied to a proportionally largenumber of religious others is not enoughto increase the salience of the religiousidentity, nor is being emotionally close toreligion-based others.

Model 3 shows that as hypothesized,the extent to which religious-based altersare connected to the respondent’s overallpersonal social network significantlyincreases the salience of the religiousidentity (p = .035), net of demographicvariables, frequency of religious atten-dance, and social and emotional attach-ment. This indicates that controlling forother variables in the model, a .10increase in the embeddedness ofreligious-based others is associated withan increase of .12 in religious identitysalience, providing support for theembeddedness hypothesis (Hypothesis3). To illustrate, we would expect a net-work with embeddedness levels corre-sponding to that in Figure 2 (.95) tohave a religious identity salience that is.77 higher than a network with embedd-edness levels that correspond to those inFigure 3 (.31).

To test Hypotheses 1 through 3 for thestudent and work identities, we run mod-els analogous to those in Table 5. How-ever, for the sake of brevity we only

discuss the coefficients for social andemotional attachment and embedded-ness. As with the religious identity,model 2 of Tables 6 and 7 shows thatsocial and emotional attachment varia-bles have no statistically significanteffect on the salience of student andwork identities (see Tables 6 and 7,respectively). However, the embedded-ness of role-based others is associatedwith an increase in the salience of thestudent identity (p = .040) as well asthe work identity (p = .028), providingadditional support for Hypothesis 3.

Overall, the results suggest a consis-tent relationship between the embedded-ness of role-based others (RBOs) inrespondents’ personal networks and iden-tity salience. For the student, work, andreligious identities, the breadth of tiesbetween RBOs and the rest of ego’s per-sonal social network is associated withan increase in identity salience, net ofrole-behavior, the proportion of ties toRBOs and the strength of those ties, aswell as other demographic variables. Twofindings that are consistent across all threesets of models are especially noteworthy: (1)neither the proportion of ties to RBOsnor the strength of those ties has aneffect on identity salience, and (2) theembeddedness of the role-based group ispositively related to identity salience,independent of the proportion of ties toRBOs and strength of those ties.

Table 4. Subgroup Characteristics and Identity Salience

Subgraph Religion (n = 129) Student (n = 90) Work (n = 134)

Size 4.67 3.31 2.36Density .69a .58b .67c

Average closeness 3.01d 2.87e 2.07f

Prop. RB others .24 .18 .21Embeddedness .58 .52 .40Identity salience 2.73 4.36 4.49

an = 81. bn = 57. cn = 79. dn = 99. en = 76. fn = 92.


Although the embeddedness coeffi-cients are relatively modest in size, thereare a number of features of our analyticstrategy—such as the fact that we controlfor role-behavior as well as the proportionand strength of ties—that likely result inconservative estimates. Overall, thesefindings do suggest that somethingbeyond the dyadic relationships betweenego and RBOs is affecting identity

salience. The fact that we observe thisrelationship independent of the propor-tion and strength of ties to RBOs suggeststhat the reach that a role-based group hasinto respondents’ social fabric is moreconsequential for identity salience thansocial and emotional attachment toRBOs. Furthermore, that the social andemotional attachment variables had nostatistically significant relationship with

Table 5. Ordinary Least Squares Regression Predicting Religious Identity Salience (N = 99)

Variable Model 1 Model 2 Model 3

Female .386 .483 .467(.258) (.259) (.253)

Age .060 .084 .097(.080) (.080) (.078)

Parental income .001 .002 .001(.002) (.002) (.002)

Education –.107 –.107 –.092(.067) (.068) (.067)

Married .793* .545 .519(.387) (.399) (.391)

White –.144 –.279 –.336(.290) (.293) (.288)

Religious attendance(less than once a year = baseline)About once or twice a year –.556 –.653 –.662

(.429) (.427) (.418)Several times a year –.033 –.062 –.021

(.431) (.426) (.417)About once a month 1.112* 1.116* 1.227*

(.551) (.543) (.534)Two to three times a month .536 .440 .435

(.629) (.623) (.610)Nearly every week 1 1.331*** 1.126** 1.137**

(.370) (.377) (.369)Network size –.016 –.017 –.014

(.013) (.013) (.013)Emotional attachment –.230 –.314

(.184) (.184)Social attachment .847 .177

(.605) (.669)Embeddedness of RB group 1.201*

(.559)Constant 3.258 3.276 2.344

(1.701) (1.747) (1.765)R2 .412 .442 .472

*p \ .05. **p \ .01. ***p \ .001 (two-tailed tests).


religious, work, and student identitysalience is a notable one, since thisresearch is the first of which we are awarethat elicits the number and strength ofties of RBOs individually.

DISCUSSION

The purpose of this study was to dig moredeeply into the role of personal networksin shaping an individual’s identity.Although social psychologists have previ-ously recognized the theoretical impor-tance of social networks on identityformation, few researchers have movedbeyond the metaphor of networks whenstudying the origins of self-structure. Inthis article, we elicited comprehensiveego-network structures, using these todistinguish between two complex features

of personal networks that should beessential to the internalization of rolesinto the self-concept. Whereas our attach-ment dimension can be viewed as a formalnetwork operationalization of socialand emotional attachment to role-basedothers, we raised the additional questionof whether the embeddedness or ‘‘reach’’of role-based others (as opposed to justthe proportion and closeness of role-basedothers) could uniquely affect the forma-tion of self-structure.

The results indicate that independentof the strength and proportion of ties toRBOs and the quantity of time spent ina given role, the extent to which RBOsare tied to the rest of one’s network is asso-ciated with the salience of the religious,work, and student identities. The factthat the relationship between the

Table 6. Ordinary Least Squares Regression Predicting Student Identity Salience (n = 76)


Female .547** .556** .536*(.204) (.209) (.204)

Age –.007 –.006 –.015(.064) (.065) (.064)

Parental income –.001 –.001 .000(.002) (.002) (.002)

Education .029 .029 .048(.052) (.053) (.052)

Married .027 .050 .016(.336) (.345) (.337)

White –.345 –.333 –.356(.214) (.220) (.215)

Hours of studying .021 .021 .019(.012) (.013) (.012)

Network size .024* .026* .020(.010) (.011) (.011)

Emotional attachment .048 –.086(.130) (.142)

Social attachment .122 –.447(.626) (.668)

Embeddedness of role-based group 0.878*(0.420)

Constant 3.563* 3.355* 3.312*(1.398) (1.508) (1.470)

R2 .258 .260 .307

*p \ .05. **p \ .01 (two-tailed tests).


embeddedness of RBOs and identitysalience is robust across all three identitiesstrongly suggests that this is a generalsocial process involving personal networksand the self. Importantly, the three identi-ties studied here are composed of widelydiffering cultural meanings and areenacted within distinctive institutionalcontexts. Thus, embeddedness processesdo not appear to be context specific. How-ever, further research is needed to testwhether the embeddedness mechanismapplies to other institutional contexts notexamined here.

Meanwhile, we consistently found thatstrength and proportion of ties to RBOswere not significantly associated withreligious, student, and work identity

salience, which is noteworthy given thetheoretical importance of social and emo-tional attachment for identity theory(e.g., Burke and Stets 2009; Stryker1980). We should be clear, however, thateven though these results seemingly failto support what identity theorists predictwith respect to social and emotionalattachment and salience, our analysis isnot in any way a direct test of identitytheory (IT) itself. Rather, our point isthat by using a formal network approach,(1) IT’s conceptions of commitment can beoperationalized in concrete network termsand that (2) other aspects of networkstructure turn out to be more importantthan what is implied by social and emo-tional attachment. In sum, our findings

Table 7. Ordinary Least Squares Regression Predicting Work Identity Salience (n = 92)


Female .702*** .660*** .744***(.192) (.193) (.192)

Age .102 .101 .123*(.059) (.058) (.058)

Parental income .001 .000 .000(.002) (.002) (.002)

Education .036 .043 .058(.050) (.050) (.049)

Married –.307 –.343 –.267(.324) (.323) (.317)

White .114 .126 .086(.232) (.231) (.226)

Hours of studying .003 .004 .005(.008) (.008) (.008)

Network size .019 .027* .031**(.010) (.011) (.011)

Emotional attachment .112 0.012(.108) (.114)

Social attachment .872 .079(.647) (.723)

Embeddedness of RB group .863*(.385)

Constant .872 .224 –.653(1.437) (1.477) (1.493)

R2 .258 .285 .327

*p \ .05. **p \ .01. ***p \ .001 (two-tailed tests).


suggest that ties between RBOs andNRBOs are important for determiningrole identity salience and that commit-ment-based processes, when formalizedinto aspects of network structure, are not.

We are aware of two basic limitationswith this analysis. First, given thefact that this research draws from cross-sectional data, we have no straightfor-ward way of establishing direction ofcausality. In turn, a possible alternativeto the claims we made here is that thesalience of a given identity affects howone forms role-based social ties. Forexample, individuals with salient identi-ties may be more likely to foster andencourage relationships between peoplein their role-based group and others atlarge in their personal social network.While previous research has indeed founda reciprocal link between role-based tiesand identity salience, the direction of theeffect seems to operate largely from theformer to the latter rather than vice-versa(Stryker and Serpe 1982). To be sure,however, the issue of causal directionalitydoes not strictly relate to our key finding,which is that the link between embedded-ness and identity salience is observed netof proportion and strength of ties to role-based others as well as frequency of role-behavior. In other words, if identitysalience drives the formation of ties, onewould expect that identity salience wouldbe associated with both the proportion ofties and embeddedness.

Second, while our method for generat-ing personal networks seems to havebeen successful in eliciting an extensivesample of social contacts, it has not beensubjected to the same degree of methodo-logical scrutiny as other commonly usedname generators. Thus, the validity of ourname generator is still largely an openquestion. However, additional analysesindicated that the networks generatedusing our approach display similar pat-terns of network composition to those

generated using validated name genera-tors (e.g., the ‘‘discuss important matters’’name generator used in the 1985 and2010 General Social Survey networks mod-ules).11 Additionally, post-session inter-views conducted during a pilot study usingour name generator revealed no systematicbias (aside from tie strength) in terms ofwho was included on the ‘‘guest list.’’12

Another potential issue related to thesurvey instrument is the possibility ofrespondent fatigue. While we took everyeffort to set up the survey in a way thatmaximizes efficiency and reduces respon-dent burden, it is nonetheless still truethat respondents could receive up to1,225 alter-alter dyads to evaluate overthe course of the survey. However,

11To better understand the nature of the net-works elicited by our vignette name generator,we compare our results to those produced by tra-ditional name generators. Specifically, we com-pare our results (available upon request) to thedata from the networks module in the GeneralSocial Survey (GSS; 1985 and 2010); the NorthernCalifornia Community Survey (NCCS) survey con-ducted by Fischer (1977; see Fischer 1982); datafrom the Internet and American Life Project(IALP) conducted by Hampton (2008; see Hamp-ton, Sessions, and Her 2011); as well as an onlinesurvey conducted by Time-Sharing Experimentsin the Social Sciences (TESS) designed by Bra-shears (2010; see Brashears 2011). These five sur-veys are publically available and include at leastsome young adult respondents. To match the sam-ple used in our study, the summary statistics forthe GSS 1985 and NCCS 1977 are based only onthe 18- to 24-year-old childless respondents ineach of the three surveys. For the 2010 GSS, how-ever, we expanded the subsample to those 18 to 30years of age to achieve a reasonable sample size,and since there were no data available on parent-hood for the 2010 TESS and 2008 IALP studies, weinclude all 18- to 24-year-olds in the table.

12Respondents were asked how accurately theguest list reflected their personal social network(defined as the people who matter in their life).They were also asked if they could think of some-body or a group of people from their social net-work who they did not put on the guest list. Theonly common response was acquaintances or peo-ple who they are less close to.


additional analyses investigating poten-tial fatigue effects revealed little evidenceof fatigue-induced error.13

CONCLUSION: NETWORK MECHA-NISMS AND THE MICRO-MACRO LINK

Although previous work in identity the-ory has done much to enhance our under-standing of the link between social tiesand the self, the issue of how social tiesimpact the self-structure is still largelyunresolved. Thus, more work is needed tospecify and empirically examine the mech-anisms that are driving this relationship.This study distinguishes between anattachment approach (or bonding mecha-nism) and an embeddedness perspective,which instead highlights environmentaladaptation and diffusion mechanisms.Although more research is needed to fullyunderstand the different mechanisms thatare driving the relationship between social

networks and the self, our findings suggestthat social network analysis may providea useful tool in this endeavor.

Additionally, this research may pro-vide important clues regarding the linkbetween Stryker’s structural identity the-ory and micro-level theories of identity(e.g., Burke 1991; Burke and Stets 2009;Heise 1979, 2007), which describe howthe identities of those in social encountersproduce social behavior. Because of theirfocus on how the interplay between fea-tures of social encounters and identityaffect social behavior, the ways in whichsocial structure affects the meaningsthat are found in situations are generallyoutside the scope of these micro-level the-ories (Smith-Lovin 2007). Thus, whilethese theories have done much toadvance our understanding of the proxi-mate causes of behavior, we echo therecent call for a renewed focus on therelationship between distal, macro-struc-tural effects on self and social behavior(Smith-Lovin 2007; Stryker and Burke2000). Our research, which puts theoret-ical primacy on the ‘‘ecology of encoun-ters’’ where social (and symbolic) interac-tion actually occurs (Smith-Lovin 2007),provides one step forward in the broadagenda to clarify how the structure ofpersonal networks can affect the self.Future research is needed to more fullyexplore the interplay between structuralfeatures of social networks and opportu-nity structures for self-verification insocial interaction.

APPENDIX: SURVEY INSTRUMENT

The following is an abbreviated version of thequestions and instructions used in the online sur-vey. Items appear in the same order as the actualsurvey with skip patterns indicated in bold italics.Line breaks indicate changes in screens (i.e.,questions appearing between lines appeartogether on one screen). The section headersused here (e.g., part 1, part 2) are for organiza-tional purposes only and do not appear in theonline survey. Access to the actual online surveyis available upon request.

13To examine fatigue, we investigatedwhether the likelihood of a respondent reportingan alter-alter tie decreased as the survey pro-gressed. As described in the appendix, respond-ents were asked if pairs of alters would ‘‘stopand chat on the street’’ in four segments: part 6(guest-to-guest dyads), part 8 (within-rolecontact-to-contact dyads), part 9 (guest-to-contactdyads), and part 11 (between-role contact-to-contact dyads). Since fatigue should be most pro-nounced in part 11, the last segment of alter-alterquestions toward the very end of the survey, wefocused our efforts on understanding the corre-lates of high/low alter-alter density in part 11.Our findings (the details of which are availableupon request) indicate that (1) time to completionincreases steadily as number of alters increases(indicating that respondents are not simply click-ing through after a certain amount of time andenergy is invested), (2) there is no evidence ofwithin-page decrease in density, and (3) whilethere is marginal support for the hypothesisthat density decreases for part 11 as guest listsize increases, the magnitude of the effect is quitesmall. Furthermore, the mean time to completionfor this survey was 25 minutes and the mediantime was 19 minutes, which is substantiallyshorter than the average time it takes to completelarge-scale face-to-face surveys such as the Gen-eral Social Survey.


_______________________________________________________________________________________________PART 1. ROLE BEHAVIOR

[Five questions about role behavior (e.g., ‘‘How often do you attend religious services?’’), with the possi-bility of two additional follow-up questions.]_______________________________________________________________________________________________PART 2. VIGNETTE NAME GENERATOR

We are interested in learning about the people in your life. This survey will ask that you provide thenames (first name and last initial only) of some of your social contacts.

As you take this survey, keep in mind that:

! Your responses are completely anonymous and confidential. Your responses will be used for

research purposes only.

! We will not attempt to contact any of the individuals you name._______________________________________________________________________________________________

We want to learn about the people in your life. Imagine that you have won an award that you are veryexcited about. Those presenting the award to you want to throw a party in your honor. They will pay forall of the expenses related to the party, including food, drinks, travel, and housing for guests. Who wouldyou invite? [Graphic of guest list entry form]_______________________________________________________________________________________________

[INSTRUCTIONAL SLIDE: Duplicate names not allowed.]_______________________________________________________________________________________________Now it’s time to create your guest list.

Remember, you’ve just won an award that you are very excited about. Those presenting the award to youwant to throw a party in your honor. Imagine that you and your guests will NOT have to pay for any ofthe expenses related to the party—including food, drinks, travel, and housing costs. Think carefullyabout all the people you want to celebrate with. Who would you invite?

Please list only the first name and last initial for each person you want to invite. You can list up to 25people. Remember, no duplicates allowed.

*XHVW�� *XHVW�� *XHVW��

«� �*XHVW��

[Alters elicited in this segment are referred to hereafter as guest1, . . . , guestk.]_______________________________________________________________________________________________PART 3. NAME INTERPRETERS

Only if participants have at least one guest list entry:

From time to time, most people discuss important matters with other people. With whom on this list doyou discuss important matters? [Check box format with instructions]_______________________________________________________________________________________________

In the past six months, with whom on this list have you spent time socially?_______________________________________________________________________________________________Only if participants currently work for pay:

Who on this list is a coworker?_______________________________________________________________________________________________


Who on this list is a family member?

_______________________________________________________________________________________________

Only if participants are currently enrolled in school:

In the past six months, with whom on this list have you spent time doing schoolwork?_______________________________________________________________________________________________

Only if participants attend religious services (i.e., more than ‘‘Never’’):

Which of the people on your list attend the same religious services as you?_______________________________________________________________________________________________

PART 4. NAME GENERATOR PROBES

Thank you for your responses. Your guest list is displayed in the black box below.

In the next section, we are going to ask you to name people from different social groups that you interactwith on a regular basis, but did not put on your list._______________________________________________________________________________________________

Are there other family members in your life?

Can you think of other members of your family that you see or talk to on a regular basis who are not onthis list? If yes, please name them below (regardless of how close you are to them). [Respondent given fivetextboxes to enter names.]_______________________________________________________________________________________________

Only if participants currently work for pay:

Other coworkers in your life?

Can you think of other individuals at work that you see or talk to on a regular basis who are not on thislist? If yes, please name them below (regardless of how close you are to them). [Respondent given fivetextboxes to enter names.]_______________________________________________________________________________________________

Only if participants attend religious services (i.e,. more than ‘‘Never’’):

Other religious contacts in your life?

Can you think of other people who attend the same religious services as you that you see or talk to ona regular basis who are not on this list? If yes, please name them below (regardless of how close youare to them). [Respondent given five textboxes to enter names.]_______________________________________________________________________________________________

Other social contacts in your life?

Can you think of other individuals with whom you spend time socially on a regular basis who are not onthis list? If yes, please name them below (regardless of how close you are to them). [Respondent given fivetextboxes to enter names.]_______________________________________________________________________________________________

Only if participants are currently enrolled in school:

Other school contacts in your life?

Can you think of other people you’ve done schoolwork with that you see or talk to on a regular basis whoare not on this list? If yes, please name them below (regardless of how close you are to them). [Respondentgiven five textboxes to enter names.]

[All alters elicited from these probes are referred to hereafter as contact1, . . . , contactk.]_______________________________________________________________________________________________


PART 5. ADDITIONAL NAME INTERPRETERS

Please indicate which of these individuals are female. [List of all guests and contacts with check boxformat]_______________________________________________________________________________________________

Please indicate which of these people are a different race than you. [List of all guests and contacts withcheck box format]_______________________________________________________________________________________________

In many cases, people in your life fall into more than one social group. For example, if you work with yourbest friend, then s/he could be a coworker as well as somebody you spend time with socially. Please indi-cate if these individuals belong to multiple groups (check all that apply).

[Circles are pre-checked based on responses given in Part 4.]

FamilyMembers Coworkers

Attend SameReligious Services

Spent Timewith Socially

Did School WorkTogether

Contact 1 s s s s s

. s s s s s

Contact k s s s s s

How close are you emotionally with the people listed below?

Not at all close (1) Somewhat close (2) Close (3) Extremely Close (4)

Guest 1 s s s s

. s s s s

Guest k s s s s

Contact 1 s s s s

. s s s s

Contact k s s s s

PART 6. ALTER-ALTER TIES: GUEST DYADS

Only if participants named two or more guests:

Do these pairs know each other well enough to stop and chat if they passed one another on the street?

Yes, they would stop and chat.Guest 1 and Guest 2 s

. s

Guest 1 and Guest k s

. s

Guest k-1 and Guest k s

PART 7. SELF-RELATED EMOTIONS

[Three blocks of questions (19 total) regarding self-esteem and locus of control.]_______________________________________________________________________________________________

PART 8. ALTER-ALTER TIES: WITHIN-ROLE CONTACT DYADS

Only if participants named two or more contacts from each probe [family, work, religion, friend,student]:

Do these pairs know each other well enough to stop and chat if they passed on another on the street?


Yes, they would stop and chat.Family Contact 1 and Family Contact 2 s

. s

Family Contact 1 and Family Contact k s

. s

Family Contact k-1 and Family Contact k s

[Repeat for work, religion, friend, and student contacts.]

PART 9. ALTER-ALTER TIES: GUEST-CONTACT DYADS

Only if participants named at least one guest and at least one contact from each probe [family,work, religion, friend, student]:


Yes, they would stop and chat.Guest 1 and Family Contact 1 s

. s

Guest 1 and Family Contact k s

. s

Guest k and Family Contact k s

[Repeat for work, religion, friend, and student contacts.]

PART 10A. SALIENCE [location in survey randomly alternated with part 10B]

Now, we would like you to think about meeting people for the first time in various settings. You want totell them about yourself so that they will really know you, but can only tell them a few things about your-self. How likely is it is that you would talk about the items listed below?

First, think about meeting a roommate for the first time. How likely is it that you would talk about thethings listed below?

VeryUnlikely (1)

Unlikely(2)

SomewhatUnlikely (3)

SomewhatLikely (4)

Likely(5)

VeryLikely (6)

Religion or ChurchActivities (1)

s s s s s s

Family (2) s s s s s s

Social Activities (3) s s s s s s

School/Academics (4) s s s s s s

Work/Career (5) s s s s s s

Think about meeting someone at a party. How likely is it that you would talk about the things listedbelow? [Same choice set as above]_______________________________________________________________________________________________

Think about meeting a friend of a close friend. How likely is it that you would talk about the things listedbelow? [Same choice set as above]

Think about giving a short speech about oneself. How likely is it that you would discuss the items listedbelow? [Same choice set as above]_______________________________________________________________________________________________

PART 11. ALTER-ALTER TIES: BETWEEN-ROLE CONTACT DYADS

Only if participants named at least one contact from each probe [family, work, religion, friend,student]:



Yes, they would stop and chat.Family Contact 1 and Work Contact 1 s

. s

Family Contact 1 and Work Contact k s

. s

Family Contact k and Work Contact k s

[Repeat for Family-Religion, Family-Friend, Family-Student, Work-Religion, Work-Friend, Work-Student, Religion-Friend, Religion-Student, Friend-Student.]_______________________________________________________________________________________________

PART 10B. PSYCHOLOGICAL CENTRALITY [location in survey randomly alternated with part 10A]

[Three blocks of questions (15 total) regarding the psychological centrality of religion, family, socialactivities, academics/school, and work/career.]_______________________________________________________________________________________________

PART 12. DEMOGRAPHICS

[Seven questions about demographic characteristics with the possibility of up five follow-up questions.]

ACKNOWLEDGMENTS

We wish to thank Matthew A. Andersson, AlisonBianchi, Jennifer Glanville, Steve Hitlin,Anthony Paik, and Sheldon Stryker for valuablecomments and suggestions on various versionsof this article. An earlier version of this articlewas presented at the 2011 annual meeting ofthe American Sociological Association.

REFERENCES

Benson, Peter L., Michael J. Donahue, andJoseph A. Erickson. 1989. ‘‘Adolescenceand Religion: A Review of the Literaturefrom 1970–1986.’’ Research in the SocialScientific Study of Religion 1(1):153–81.

Borgatti, Stephen P. and Virginie Lopez-Kidwell. 2011. ‘‘Network Theory.’’ Pp. 40–51 in Handbook of Social Network Analysis,edited by P. J. Carrington and J. Scott. NewYork: Sage.

Brashears, Matthew E. 2010. ‘‘Anomia and theSacred Canopy: Testing a Network The-ory.’’ Social Networks 32(3):187–96.

Brashears, Matthew E. 2011. ‘‘Small Networksand High Isolation? A Reexamination ofAmerican Discussion Networks.’’ SocialNetworks 33:331–41.

Burke, Peter J. 1991. ‘‘Identity Processes andSocial Stress.’’ American SociologicalReview 56(6):836–49.

Burke, Peter J. and Donald C. Reitzes. 1991.‘‘An Identity Theory Approach to

Commitment.’’ Social Psychology Quarterly54(3):239–51.

Burke, Peter J. and Jan E. Stets. 2009. IdentityTheory. Oxford, UK: Oxford University Press.

Burke, Peter and Judy Tully. 1977. ‘‘The Mea-surement of Role/Identity.’’ Social Forces55(4):881–97.

Callero, Peter L. 1985. ‘‘Role identitySalience.’’ Social Psychology Quarterly48(3):203–15.

Cooley, Charles H. 1902. Human Nature andSocial Order. New York: Scribner.

DiPrete, Thomas A., Andrew Gelman, TylerMcCormick, Julien Teitler, and Tian Zheng.2011. ‘‘Segregation in Social Networks Basedon Acquaintanceship and Trust.’’ AmericanJournal of Sociology 116(4):1234–83.

Fischer, Claude S. 1982. To Dwell amongFriends: Personal Networks in Town andCity. Chicago: University of Chicago Press.

Granovetter, Mark S. 1985. ‘‘Economic Actionand Social Structure: The Problem ofEmbeddedness.’’ American Journal of Soci-ology 91(3):481–510.

Hampton, Keith N., Lauren F. Sessions, andEun Ja Her. 2011. ‘‘Core Networks, SocialIsolation, and New Media.’’ Information,Communication & Society 14(1):130–55.

Heise, David R. 1979. Understanding Events:Affect and the Construction of Social Action.New York: Cambridge University Press.

Heise, David R. 2007. The Expressive Order:Confirming Sentiments in Social Actions.New York: Springer.


Hogan, Bernie, Juan Antonio Carrasco, andBarry Wellman. 2007. ‘‘Visualizing Per-sonal Networks: Working with Partici-pant-Aided Sociograms.’’ Field Methods19(2):116–44.

Holland, Paul W. and Samuel Leinhardt. 1972.‘‘Reply: Some Evidence on the Transitivityof Positive Interpersonal Sentiment.’’American Journal of Sociology 77(6):1205–1209.

Holland, Paul W. and Samuel Leinhardt. 1977.‘‘A Dynamic Model for Social Networks.’’The Journal of Mathematical Sociology5(1):5–20.

Lee, Barrett A., Karen E. Campbell, and OscarMiller. 1991. ‘‘Racial Differences in UrbanNeighboring.’’ Sociological Forum6(3):525–50.

Lin, Nan and Mary Dumin. 1986. ‘‘Access toOccupations through Social Ties.’’ SocialNetworks 8(4):365–85.

Marin, Alexandra. 2004. ‘‘Are RespondentsMore Likely to List Alters with CertainCharacteristics?: Implications for NameGenerator Data.’’ Social Networks26(4):289–307.

Marin, Alexandra and Keith N. Hampton.2007. ‘‘Simplifying the Personal NetworkName Generator: Alternatives to Tradi-tional Multiple and Single Name Genera-tors.’’ Field Methods 19(2):163–93.

Marsden, Peter V. 1990. ‘‘Network Data andMeasurement.’’ Annual Review of Sociology16:435–63.

Marsden, Peter V. 2005 ‘‘Recent Developmentsin Network Measurement.’’ Pp. 8–30 inModels and Methods in Social NetworkAnalysis, edited by P. J. Carrington, J. P.Scott, and S. Wasserman. New York: Cam-bridge University Press.

Marsden, Peter V. 2011. ‘‘Survey Methods forNetwork Data.’’ Pp. 370–88 in The SAGEHandbook of Social Network Analysis, edi-ted by J. P. Scott and P. J. Carrington.Thousand Oaks, CA: Sage.

McCall, George J. and J. L. Simmons. 1978.Identities and Interactions: An Examina-tion of Human Associations in EverydayLife. New York: The Free Press.

McCallister, Lynne and Claude S. Fischer.1978. ‘‘A Procedure for Surveying PersonalNetworks.’’ Social Methods and Research7(2):131–48.

McFarland, Daniel and Heili Pals. 2005.‘‘Motives and Contexts of Identity Change.’’Social Psychology Quarterly 68(4):289–315.

McPherson, Miller, Lynn Smith-Lovin, andMatthew E. Brashears. 2006. ‘‘Social Isola-tion in America: Changes in Core Discus-sion Networks over Two Decades.’’ Ameri-can Sociological Review 71(3):353–75.

McPherson, Miller, Lynn Smith-Lovin, andMatthew E. Brashears. 2008. ‘‘Erratum:Social Isolation in America: Changes inCore Discussion Networks over Two Deca-des.’’ American Sociological Review 73:1022.

Mead, George Herbert. 1934. Mind, Self andSociety. Chicago: University of ChicagoPress.

Merolla, David M., Richard T. Serpe, SheldonStryker, and P. Schultz. 2012. ‘‘StructuralPrecursors to Identity Processes: The Roleof Proximate Social Structures.’’ Social Psy-chology Quarterly 75(2):149–72.

Mollenhorst, Gerald, Beate Volker, and HenkFlap. 2008a. ‘‘Social Contexts and CoreDiscussion Networks: Using a ChoiceApproach to Study Similarity in IntimateRelationships.’’ Social Forces 86(3):937–65.

Mollenhorst, Gerald, Beate Volker, and HenkFlap. 2008b. ‘‘Social Contexts and PersonalRelationships: The Effect of Meeting Oppor-tunities on Similarity for Relationships ofDifferent Strength.’’ Social Networks30(1):60–68.

Moody, James and Douglas R. White. 2003.‘‘Structural Cohesion and Embeddedness:A Hierarchical Concept of Social Groups.’’American Sociological Review 68(1):103–27.

Moore, Gwen. 1990. ‘‘Structural Determinantsof Men’s and Women’s Personal Networks.’’American Sociological Review 55(5):726–35.

Nuttbrock, Larry and Patricia Freudiger.1991. ‘‘Identity Salience and Motherhood:A Test of Stryker’s Theory.’’ Social Psychol-ogy Quarterly 54(2):146–57.

Owens, Timothy, Dawn Robinson, and LynnSmith-Lovin. 2010. ‘‘Three Faces of Iden-tity.’’ Annual Review of Sociology 36:477–99.

Portes, Alejandro and Julia Sensenbrenner.1993. ‘‘Embeddedness and Immigration:Notes on the Social Determinants of Eco-nomic Action.’’ American Journal of Sociol-ogy 98(6):1320–50.

Rosenberg, Morris. 1979. Conceiving the Self.New York: Basic Books.

Serpe, Richard T. 1987. ‘‘Stability and Changein Self: A Structural Symbolic Interaction-ist Explanation.’’ Social Psychology Quar-terly 50(1):44–55.

Serpe, Richard T. and Sheldon Stryker. 1987.‘‘The Construction of Self and the


Reconstruction of Social Relationships.’’Advances in Group Processes 4:41–66.

Smith-Lovin, Lynn. 2007. ‘‘The Strength ofWeak Identities: Social Structural Sourcesof Self, Situation and Emotion.’’ Social Psy-chology Quarterly 70(2):106–24.

Stets, Jan E. 2005. ‘‘Examining EmotionsinIdentity Theory.’’ Social Psychology Quar-terly 68(1):39–56.

Stets, Jan E. and Chris F. Biga. 2003. ‘‘Bring-ing Identity Theory into EnvironmentalSociology.’’ Sociological Theory 21(4):398–423.

Stets, Jan E., Michael J. Carter, Michael M.Harrod, Christine Cerven, and Seth Abru-tyn. 2008. ‘‘The Moral Identity, Status,Moral Emotions, and the NormativeOrder.’’ Pp. 227–51 in Social Structureand Emotion, edited by J. Clay-Warnerand D. T. Robinson. Oxford, UK: Elsevier.

Stryker, Sheldon. 1968. ‘‘Identity Salience andRole Performance.’’ Journal of Marriageand the Family 30(4):558–64.

Stryker, Sheldon. 1980. Symbolic Interaction-ism: A Social Structural Version. MenloPark, CA: Benjamin Cummings.

Stryker, Sheldon and Peter Burke. 2000. ‘‘ThePast, Present, and Future of an IdentityTheory.’’ Social Psychology Quarterly63(4):284–97.

Stryker, Sheldon and Richard T. Serpe. 1982.‘‘Commitment, Identity Salience, and RoleBehavior: Theory and Research Example.’’Pp. 199–218 in Personality, Roles andSocial Behavior, edited by W. Ickes and E.Knowles E. New York: Springer-Verlag.

Stryker, Sheldon and Richard T. Serpe. 1994.‘‘Identity Salience and Psychological Cen-trality: Equivalent, Overlapping, or Com-plementary Concepts?’’ Social PsychologyQuarterly 57(1):16–35.

Stryker, Sheldon, Richard T. Serpe, and Mat-thew O. Hunt. 2005. ‘‘Making Good ona Promise: The Impact of Larger SocialStructures on Commitments.’’ Advances inGroup Processes 22:93–123.

Turner, Ralph H. 1962. ‘‘Some Family Deter-minants of Ambition.’’ Sociology and SocialResearch 46(4):397–411.

Turner, Ralph H. 1978. ‘‘The Role and thePerson.’’ American Journal of Sociology84(1):1–23.

Uzzi, Brian. 1996. ‘‘The Sources and Conse-quences of Embeddedness for the EconomicPerformance of Organizations: The Net-work Effect.’’ American Sociological Review61(4):674–98.

Uzzi, Brian. 1999. ‘‘Embeddedness in the Mak-ing of Financial Capital: How Social Rela-tions and Networks Benefit Firms SeekingFinancing.’’ American Sociological Review64(4):481–505.

Walker, Mark H. 2010. ‘‘The Embedded Self:Investigating the Relationship betweenNetwork Structure and Self-Structure.’’MA thesis, University of Iowa, Iowa City,IA.

Wellman, Barry. 1979. ‘‘The Community Ques-tion: The Intimate Networks of EastYorkers.’’ American Journal of Sociology14(3):1201–31.

Wilson, John and Darren E. Sherkat 1994.‘‘Returning to the Fold.’’ Journal for theScientific Study of Religion 33(2):148–61.

BIOS

Mark H. Walker is a graduate student inthe department of sociology at the Univer-sity of Iowa. His research interestsinclude self and identity, social networks,and the measurement of personal socialnetworks. His dissertation investigatesthe interplay between identity conflict,personal network structure, and psycho-logical well-being.

Freda B. Lynn is an assistant professorof sociology at the University of Iowa.Her research interests include networks,status dynamics, and inequality. She iscurrently engaged in a network-basedstudy of citation patterns funded by theNational Science Foundation. Her recentwork on status appears in the AmericanJournal of Sociology and the AnnualReview of Sociology.


social psychology quarterly the embedded self: a proofs embedded self.pdf · the embedded self: a...

Documents