Social Science Research 45 (2014) 117–130

Contents lists available at ScienceDirect

Social Science Research

journal homepage: www.elsevier .com/locate /ssresearch

Identities in flux: Cognitive network activation in timesof change

0049-089X/$ - see front matter � 2014 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.ssresearch.2014.01.001

⇑ Corresponding authors.E-mail addresses: menon.53@osu.edu (T. Menon), ned-smith@kellogg.northwestern.edu (E.B. Smith).

1 Both authors contributed equally.

Tanya Menon a,⇑,1, Edward Bishop Smith b,⇑,1

a Fisher College of Business, The Ohio State University, 700 Fisher Hall, 2100 Neil Ave., Columbus, OH 43210, United Statesb Northwestern University Kellogg School of Management, 2001 Sheridan Road, Rm 358, Evanston, IL 60208, United States

a r t i c l e i n f o

Article history:Received 18 August 2012Revised 5 January 2014Accepted 6 January 2014Available online 11 January 2014

Keywords:IdentityCognitive network activationChange

a b s t r a c t

Using a dynamic cognitive model, we experimentally test two competing hypotheses thatlink identity and cognitive network activation during times of change. On one hand, affirm-ing people’s sense of power might give them confidence to think beyond the densest sub-sections of their social networks. Alternatively, if such power affirmations conflict withpeople’s more stable status characteristics, this could create tension, deterring people fromconsidering their networks’ diversity. We test these competing hypotheses experimentallyby priming people at varying levels of status with power (high/low) and asking them toreport their social networks. We show that confirming identity—not affirming power—cognitively prepares people to broaden their social networks when the world is changingaround them. The emotional signature of having a confirmed identity is feeling comfortableand in control, which mediates network activation. We suggest that stable, confirmedidentities are the foundation from which people can exhibit greater networkresponsiveness.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

According to White (2008: p. 1), ‘‘Identities spring up out of efforts at control in turbulent context[s].’’ This paper providesevidence of a reverse causal process whereby people’s images of their social worlds can also spring from their sense of self.Specifically, we propose that people’s identities—i.e., their schematic knowledge about who they are (Markus, 1977)—bothemerge from social relations (see Fuchs, 2001; Podolny, 1993; Bothner et al., 2010; Smith, 2011) and affect people’s percep-tions of who they are connected to. As people’s identity states shift (Smith, 2013; Benoit-Smullyan, 1944; Marks, 1977;McAdam, 1982) so too does their mental representation of their social networks.

We argue that people’s identities affect whether they envision their social networks as broad (i.e. sparse and relationallyheterogeneous) or narrow (i.e. dense and redundant), independent of the size and density of their actual or potential socialnetworks. This paper thus marks an exploration of the psychological antecedents that allow people to explore the diversity oftheir networks. Like White, turbulent contexts, and social change specifically, play an important role in our analysis. Whilechange creates an impetus to search for new information (Cyert and March, 1963; Stinchcombe, 1990; Weick, 1995, cited inGreve and Taylor, 2000), perhaps triggering people to dynamically activate diverse subsections of their networks, peoplemust also feel comfortable enough to embrace the resulting diversity. If they do not, they might instead respond to changeby retreating to the densest, most homogeneous subsections of their networks.

118 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

We do not seek to fully adjudicate between the benefits of dense vs. expansive networks (Granovetter, 1973; Burt, 1980).While dense social cliques can offer safety and support, people may also benefit by exploring their surroundings and seekingout new sources of information. The latter behavior is particularly important, though not necessarily automatic, when nav-igating social change (Cyert and March, 1963; Stinchcombe, 1990; Weick, 1995; Greve and Taylor, 2000; Levinthal andMarch, 1993). Thus, using the condition of social change as a backdrop, our principal argument is that people must feel psy-chologically comfortable to engage the diversity that accompanies expanded search (Lawler, 2001). This tradeoff betweencomfort and search presents a problem, however: if people feel uncomfortable, they might respond to change by retreatingto the densest subsection of their social networks, thereby eliminating opportunities for exploration at the very moment thatexploration may prove most beneficial. Building on this premise, we test two competing hypotheses on the links betweenpower, identity, and social network perceptions.

On one hand, it is possible that affirming one’s power and agency would give people the confidence to network broadly.Indeed, many policy initiatives are constructed according to this view. To make the lives of the weakest members of societybetter, it seems intuitive that initiatives designed to empower those people would have positive consequences (Chakravarti,2006). The Government of India, for example, declared 2001 the empowerment of women year, and passed a national policyaround it (http://www.wcd.nic.in/empwomen.htm). Similar programs from community organizations in the US also rely onaffirming those without power and encouraging them to believe that they have the ability to change their situations (McAd-am, 1982). Accordingly, it is plausible that people who feel powerful should be more likely to ‘‘activate,’’ or call to mind,broad social networks.

On the other hand, having a stable sense of identity may be important for fostering the sorts of emotions—e.g., comfortand control—that are necessary to network broadly. Whereas power affirmations will produce such feelings for some people,people who are not used to feeling powerful (e.g., people with low socioeconomic status) might find it discomforting toimagine having power and exercising it (Ibarra, 1999; Rafaeli and Sutton, 1989). Thus, an alternative to the power affirma-tion hypothesis suggested above is this: confirming one’s identity (see Polzer et al., 2002; Milton and Westphal, 2005;Swann, 1983) promotes feelings of comfort and control that are necessary precursors to activating broad, diverse networks.It is not power, per se, but rather consistency between the temporary feeling of power and the more stable sense of self thatprepares people to engage diversity in the world around them.

2. Cognitive networks and social network activation

Rather than focusing on network properties as an objective external reality, cognitive approaches to studying networksfocus on how people mentally represent network properties (Carley, 1986; Kilduff et al., 2006; Krackhardt, 1987; Michaelsonand Contractor, 1992; Johnson and Orbach, 2002). Recent cognitive networks research additionally links people’s represen-tations of their networks to situational cues and psychological states (Smith et al., 2012) to illuminate the dynamic processesby which people’s perceptions of their networks can shift on a moment-to-moment basis. The present paper builds on argu-ments proposed in Smith, Menon, and Thompson, (2012) by (a) exploring identity shifts as a source of these dynamic cog-nitions about social networks, and (b) showing how psychological states (e.g. comfort) mediate between specific identitystates and network perceptions. How people see the self at a given moment directly impacts the way they feel and how theyenvision the social networks around them (Menon and Blount, 2003; Menon et al., 2006).

To develop our hypotheses, we draw on the tri-partite, nested representation of social networks proposed in Smith et al.(2012):

(1) Potential network. People’s full set of contacts. The potential network represents people’s budget constraint—all thecontacts who they might access are included in the potential network.

(2) Activated network. The core concern of the present paper is the activated network, or the cognitive representation ofpeople’s networks. As a subset of the potential network, which is an objective structure, the activated network is asubjective, cognitive structure (cf., Carley, 1986; Krackhardt, 1987). Cognitive activation occurs when a concept isprimed and becomes mentally available and accessible to perceivers (Higgins and Kruglanski, 1996). When appliedto network research, cognitive activation implies that situational priming can shape how network knowledge comesto mind (Smith et al., 2012).

(3) Mobilized network. This is the subset of the activated network from which people draw when they solicit resources. Incomparison to the activated network, which results from a private, mental activity, mobilization is a public, socialactivity through which people call on contacts and seek resources (McCarthy and Zald, 2001; Smith, 2005).

While the present research was not designed to definitively test the links between activation and mobilization, psycho-logical evidence does indicate that cognitive activation often facilitates behavioral responses. While activated knowledgestructures (e.g. primes, activated network contacts) could seem like mere fleeting suggestions, they can operate as goals thatcan endure until they are satiated or acted upon. Research has established the considerable endurance of primes (e.g. Woltzand Was, 2007 found no decrease in priming effects after 36 Stroop trials; Was (2010) found that priming effects persistedafter 24 h delays). One of the longest priming effects ever documented indicated that a single exposure to the American flagshifted support to Republicanism up to 8 months later (Carter et al., 2011). As it pertains to our paper, we simply suggest that

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 119

activating particular contacts makes them psychologically accessible, and therefore more likely to be mobilized (Higgins andKruglanski, 1996).2

The present perspective was designed to explore two key departures from past networks research. First, the model elicitsquestions of when and why particular subsections of people’s potential social networks are activated and mobilized butothers are not. We focus our attention on periods of change because navigating (and perhaps even benefitting from) suchsituations often requires people to search beyond the core of their social worlds. How people activate their networks canaffect whether they are able to mobilize resources, and thus, their ability to successfully navigate the changing environment.

Second, existing research that explores the antecedents of certain network properties tends to privilege stable, between-person sources of variation. For example, previous research indicates that socioeconomic status correlates with socialnetwork properties such as size and density (e.g., Mouw, 2002; McPherson et al., 2006). An alternative set of studies drawsconnections between personality and networks, finding that relatively stable factors rooted in personality (e.g. people withhigh neuroticism (i.e. negative affect) tend to inhabit less central network positions, Klein et al., 2004). Beyond the Big-5personality traits, individual differences in beliefs such as self-monitoring also have been extensively explored antecedentsto brokerage-type networks (Sasovova et al., 2010; Mehra et al., 2001; Flynn et al., 2006), as well as individual differences inindividualist vs. collaborative self-concept (Srivastava and Banaji, 2011).

Rather than focusing on individual differences that capture between-person variations, dynamic psychological forcessuch as threat and affirmation produce within-person variation in patterns of seeing (e.g., Smith et al., 2012) and valuing(Menon et al., 2006; Menon and Blount, 2003) networks. We argue that people’s ability to network broadly can be aboutmore than individual differences in having a broker-like personality or self-monitoring capabilities. We will use experimen-tation to randomly assign participants to specific situations which elicit particular cognitive states and which trigger within-person variation in people’s patterns of network activation. We propose that people’s dynamically shifting cognitive statesaffect their ability to envision the breadth of resources that surround them.

Before moving onto develop specific predictions, it is important to note that our goal is not to observe people deliberatelyactivating different networks for different situations (see Emirbayer and Goodwin, 1994 who consider various models ofagency and network formation). Rather than assessing the purposeful elicitation of social contacts, we are interested in iden-tifying how certain social and psychological conditions impact the way people envision their social networks. Given thatpeople’s responses to network surveys can vary temporally (see Marsden, 1990 for a review), the present research pinpointspsychological factors that affect these variations in network activation through experimental methods. Instead of being pur-poseful and strategic in networking behavior, we argue that people’s patterns of network activation can also arise from theirshifting psychological states, which help explain when and why people exploit (and sometimes squander) the resourcesavailable to them via their social networks.

2.1. Identity-based underpinnings of network activation

2.1.1. Power enables the activation of diverse networksAccount 1 is that affirming people’s power enables them to envision broad social networks. In our specific context, we

‘‘empower’’ people using one influential approach in psychological research: priming people to call to mind the self in pow-erful roles and positions. Scores of psychological studies establish that simply priming people to think about power (lack ofpower) promotes an action (inaction) orientation (Galinsky et al., 2003; Keltner et al., 2003). Specifically, the high-powermanipulation has been found to increase assertive behaviors such as taking social risks (Anderson and Galinsky, 2006;Galinsky et al., 2003), and approaching other people with regard to their usefulness or utility (Gruenfeld et al., 2008).Extending this research to a social networks context implies that empowerment may cause people to feel more confidentand enable them to imagine having more expansive social networks as well. This is the ‘‘power affirmation’’ hypothesis:making people feel powerful offers them the confidence that is necessary to activate diverse networks.3

While power may allow expansion, people facing threats that lead to perceived loss of control and power often exhibitconstriction and rigidity as well (Staw et al., 1981), and have been shown to mobilize more limited and homophilous sectionsof their advice networks (e.g., McDonald and Westphal, 2003). The tendency to constrict under threat also varies, however, asa function of status (Smith et al., 2012). For instance, high-status people (who might have more confidence and control whencoping with threats) respond to job threats by activating more diverse networks, whereas lower status people respond to thesame threat by turning inwards. These differences emerged only when people were facing a threat—not when they were

2 Additionally, note that we do not assume that activation is a necessary prerequisite to mobilization. In addition to activation, another route to mobilizationis encountering contacts through social foci, i.e. locations where people congregate to socialize (Feld, 1981).

3 Note an alternative possibility that is raised in other research on power. Because the powerless exert less outcome control over the powerful, the powerfulmay be unmotivated to perceive the powerless in deliberative ways (Fiske, 1993). These heuristic, stereotyped perceptions could compromise and constricttheir network cognitions. By contrast, the powerless are dependent on the powerful, motivating them to pay attention (Boissevain; 1974; Fiske, 1993),potentially allowing for more accurate network perceptions. However, in contrast to these motivational hypotheses, other research finds that the powerful havemore accurate perceptions of advice networks (Krackhardt, 1990). While the present research does not resolve these issues by explicitly manipulating ormeasuring motivation, we control for it in another important way. Specifically, we show when and why power leads to heterogeneous network activation(when aligned with status, an effect mediated by comfort). By establishing moderation and mediation, we pinpoint the causal process, which cannot beexplained by motivational main effects associated with power.

120 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

asked to think about more comforting situations. Consequently, a prescriptive implication of the power affirmation hypoth-esis is that empowering lower status people might give them the confidence to network more broadly.

2.1.2. Identity confirmation enables the activation of diverse networksEmpowerment also has psychological implications for people’s identity. Account 2 suggests that empowerment might

create psychological discomfort for those who do not typically experience power in their environment.To motivate this account, consider one of the primary limitations of experimental manipulations of power—e.g., ‘‘think

about a time when you had power, (Galinsky et al., 2003).’’ Whereas such manipulations may be successful in cognitivelyactivating a person’s past experiences of power, they fail to account for more stable features of a person’s social standing.This limitation should present problems if people differ systematically in their ability to imagine having power and exercis-ing it, perhaps due to their identities.

Recent research has begun to explore these interactions between status and power, finding that dissociations betweenthe two produce discomfort (Blader and Chen, 2011; Fast et al., 2011). People whose identities are tied to lower-status roles,for instance, might experience discomfort when they imagine themselves with high-power (e.g., Ibarra, 1999; Kraus et al.,2009). Similarly, people who identify with high-status roles might be uncomfortable when asked to think about themselvesin low-power roles. The conflicting experience of being told one has power when used to having none, or alternatively, beingtold one has no power when used to having it, is disconcerting because it threatens to upend one’s sense of self. In both cases,people may be prone to feel ‘‘role conflicts’’ and ‘‘status inconsistencies’’ as they simultaneously play multiple, possibly con-flicting roles (Benoit-Smullyan, 1944; Hughes, 1945; Lenski, 1954, 1956; Marks, 1977). According to these theories, peoplewhose status is higher in one role than it is in another experience feelings of discomfort and cognitive dissonance (Ibarra andBarbulescu, 2010; Ibarra, 1999; Ebaugh, 1988; Goffman, 1961; Festinger, 1957), which can subsequently affect otherbehaviors.

While research on role conflicts emphasizes the cost of identity conflicts, research on self-verification offers a comple-mentary view that emphasizes the enabling potential of having a consistent identity (Milton and Westphal, 2005; Swann,1983). Focusing on verifying aspects of identity that derive from people’s positions in a social hierarchy, we offer an ‘‘iden-tity-confirmation’’ account: people with stable identities experience feelings of comfort and control that are necessary toventure beyond the densest subsections of their networks. Without this stable sense of identity, people may be uncomfort-able with thoughts of exploring the breadth of their social world (cognitively, at least, if not actually).

2.1.3. Social change and IdentityWe focus on social change as an exogenous trigger that both heightens the salience of identity and motivates network

dynamism (Erikson, 1970). Ours is not the first study to consider change and networks together. Previous research has ex-plored how particular network structures can facilitate responsiveness to change (Fleming et al., 2007), inhibit it (Burt,2005), and the conditions under which these effects emerge (Battilana and Casciaro, forthcoming).

Our key proposition is that identity constitutes a key psychological foundation from which network activation patternsemerge during times of change. Our experiment allows us to adjudicate between two competing accounts linking identity tocognitive network activation. The first hypothesis suggests that affirming the self—enhancing one’s power and agency—canprepare people to broaden their social worlds. The second suggests that stable and confirmed identities are a prerequisite foractivating heterogeneous networks.

2.1.4. Feelings of control mediate network activationThe two hypotheses suggest different psychological processes that constitute the basis for broad network activation. The

power-affirmation hypothesis predicts that power is necessary for people to view their networks as broad and prepare themto engage the resulting diversity. The identity-confirmation hypothesis, by comparison, implies that a stable sense of selftranslates into feelings of comfort and control that will enable people to (cognitively) venture beyond the dense core of theirsocial networks (e.g., Douglas, 1966; Averill, 1973). While both power affirmation and identity confirmation accounts areplausible and build on existing empirical and theoretical research, neither has been tested empirically with respect to net-work activation.

In sum, our model comprises three principal parameters: people’s environmental perceptions (i.e. environment as stableor changing); self-perceptions (i.e. perceptions of both place in a societal hierarchy and momentary experiences of havingpower); and feelings of comfort and control. We expect that these factors will affect how people contemplating change willactivate their networks.

3. Material and methods

We adjudicate between the ‘‘power affirmation’’ and ‘‘identity confirmation’’ hypotheses by priming people at varyinglevels of status with power (high vs. low) and then measuring their social networks using a standard recall-based networkname generator. Our primary dependent variables tap the network properties (i.e. density and the activated alters’ relation-ship with the ego) that the participants activated following the cognitive primes. Respondents’ socioeconomic status was ameasured variable. Power was manipulated. The interaction between status and power generates identity-confirmed

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 121

(high status and high power, low status and low power) and identity-conflicted (high status and low power, low status andhigh power) conditions. We measured participants’ psychological reactions to the status and power primes following themanipulation to assess the levels of comfort and control conferred by power or identity confirmation. We also primed peopleto think about no vs. significant social change, expecting the latter condition to trigger network responsiveness, and to betteralign our empirical investigation with practical implications about public policy and effective information search duringperiods of change (Greve and Taylor, 2000). We describe the specific experimental procedure below.

3.1. Participants

The participants in the study were 120 students at the Indian School of Business (70% male). The Indian context is a usefulone to test our competing hypotheses for several reasons. First, India is a context where status is a salient and stable featureof the self. India’s culture is rooted in the caste hierarchy, which traditionally summarized every aspect of a person’s identity,involving dimensions including occupation, ethnicity, and even personal character (Dumont, 1980). While we do not mea-sure caste (given the sensitive nature of such questions in the context), the vestiges of this historical institution are apparentin India’s high power distance values (Hofstede, 1980), which indoctrinate the acceptance of existing power relations.

Second, the relative rigidity of socioeconomic status in the Indian context allows us to meaningfully operationalize poweraffirmation and identity confirmation as distinct conditions. We measure social status as people’s a priori perception of theirposition in a stratified social hierarchy (Duncan, 1984). Power, by comparison, is a function of situations, and can fluctuateacross contexts (Blau, 1964; Pfeffer, 1992). Much like past research suggesting that ‘‘emotional discomfort arises when peo-ple are unable to draw a continuous link between old and new selves’’, or that ‘‘emotive dissonance. . .results from discrep-ancies between what people really feel and the images they feel compelled to convey in social interactions,’’ the identityconfirmation hypothesis suggests that inconsistencies between people’s status and power will engender feelings of discom-fort (Ibarra and Barbulescu, 2010, p. 135; see also Rafaeli and Sutton, 1989). Being low status but primed with high power, orhigh status and primed with low power, is directly analogous to prior conceptions of identity inconsistency (Blader andChen, 2011; Benoit-Smullyan, 1944; Fast et al., 2012; Hughes, 1945; Lenski, 1954, 1956; Marks, 1977). When people’s powerand status are aligned, however, our identity confirmation hypothesis suggests that they should experience feelings of com-fort and control that will be reflected in broader patterns of network recall. We test these claims below.

3.2. Procedures and materials

All participants answered a questionnaire which surveyed their sense of status, manipulated their sense of power, andasked them to report their social networks. Half of the participants were told to think about a stable society, and to createthe pre-conditions for network dynamism, the other half thought about a changing one.

Status. To gage participants’ social status, we asked: ‘‘When you think about your family, what would their percentilerank be in terms of India’s income distribution, 0–100? (0 = lowest; 100 = highest).’’ While this measure captures partici-pants’ perceived status, past research suggests that similar measures of status perceptions tend to be highly correlated withself-reported family income, (r(71) = .65, p = .00, Smith et al., 2012). Past research has also established the discriminant valid-ity of the perceptual status measure by assessing its correlations with other well-established measures in psychology. Per-ceived status was uncorrelated with participants’ political beliefs (e.g., ‘‘Which response best describes your politicalbeliefs?’’ very liberal, liberal, slightly liberal, moderate/middle-of-the-road, slightly conservative, conservative, very conservative,don’t know/not political) as well as the HEXACO personality dimensions (Honesty-Humility; Emotionality; Extraversion;Agreeableness; Conscientiousness, and Openness to experience) (Smith et al., 2012). This measure allowed us to gage peo-ple’s a priori sense of status.4

Power manipulation. We next sought to activate the momentary experience of power. We used Galinsky et al.’s (2003)power manipulation, which primes the experience of having or lacking power by asking participants to recall a situationwhere they had power or lacked it. Specifically, participants were randomly assigned to read and respond to either low(0) or high power (1) conditions:

Low power: ‘‘Please recall a particular incident in which someone else had power over you. By power, we mean a situationin which someone had control over your ability to get something you wanted, or was in a position to evaluate you. In 2–3sentences, please describe this situation in which you did not have power—what happened, how you felt, etc.’’

High power: ‘‘Please recall a particular incident in which you had power over another individual or individuals. By power,we mean a situation in which you controlled the ability of another person or persons to get something they wanted, or werein a position to evaluate those individuals. In 2–3 sentences, please describe this situation in which you had power—whathappened, how you felt, etc.’’

Identity confirmation and conflict. The interaction between status and power yielded identity-confirmed (High Status/High Power (HS/HP); Low Status/LP (LS/LP) and identity-conflicted (High Status/Low Power (HS/LP); Low Status/High Power(LS/LP)) conditions. Because this interaction term involves one continuous and one dichotomous variable, we are explicit in

4 In pilot testing, we attempted to manipulate social status (following Kraus et al., 2009). Given that social status is not merely a state of mind (a cognitivestructure) but also reflects ingrained, stable life experiences (especially within a society such as India with low social mobility), we were unable to replicatethese effects.

122 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

our analyses below about how confirmation and conflict are treated (i.e., first on a continuous scale and then by using piece-wise categories of the status variable to construct a discrete number of identity conditions that fit confirmed-conflictedcategories).

Social change. Anticipating that social change may trigger information search (see Greve and Taylor, 2000) and makeidentities salient (Erickson, 2000), we also asked participants to complete a second recall manipulation that focused themon either the presence or absence of change in society. Specifically, participants were randomly assigned to read and respondto either a low (0) or high change manipulation:

Low change: ‘‘Now, think about a time when you observed that the more Indian society changes, the more it stays thesame. In 2–3 sentences, describe how you’ve seen the same people maintaining their power in Indian society.’’

High change: ‘‘Now, think about a time when you observed dramatic changes in Indian society. Specifically, in 2–3 sen-tences, describe how you’ve seen new people rising in power in Indian society.’’

To reiterate, we view social change as an underlying cause that might motivate people to activate network relationshipsin different ways. The low-change condition operates as a control condition, i.e., people’s typical activation patterns whenthey lack motivation to respond dynamically. According to Greve and Taylor (2000) people who aren’t facing change are lesslikely to activate broad information search. Having a control condition also enables us to meaningfully differentiate the ef-fects of identity confirmation and identity conflict on network activation.

Mediators. Following the primes, participants reported their mood. We asked them: Please describe your mood at thismoment. They answered Turner et al.’s (1992) five-item scale and rated on a seven point scale the degree to which they feltcomfortable (1) vs. uncomfortable (7), calm (1) vs. shaky (7), secure (1) vs. tense (7), confident (1) vs. panicky (7) and relaxed (1)vs. frightened (7) at that particular moment (Alpha = .96). Given that control was also a key aspect of our hypotheses, we alsoasked them the extent to which they felt in control (1) or out of control (7).

Dependent variables. Next, we measured the key contacts participants activated in their social networks. We used anearly identical name generator as in the General Social Survey (GSS), where participants responded to the question: ‘‘Mostpeople discuss important matters with other people. Who are the people with whom you discuss matters important to you?List up to ten people in the spaces below (only use initials). You do not need to name ten.’’ Spaces were marked on the surveypage for up to ten contacts. We expanded the name generator from the GSS 5-person version to a 10-person version becausewe were concerned that the smaller sample might not capture the full range of contacts.

Importantly, we used a recall-oriented name generator because we did not want participants to actively elicit networks toserve a specific purpose. In other words, we did not ask that participants construct a social network that would be usefulwhen their power was high or low. Rather, our intention was to assess how certain psychological states might systematicallyskew people’s views about their social networks more generally.

We then followed the name generator with more detailed questions about the network contacts named. Specifically weasked participants about the nature of their relationship to each of the parties (e.g., were the contacts named friends, family,or advisors).

Finally, we presented each participant with a matrix in which to indicate the relationships among the contacts that werenamed. Instructions were: ‘‘Now, write the names of the same people you mentioned above in both the columns and therows of the matrix below. The next task is to describe the strength of relations between the people you named as contacts.You do this by marking cells in the matrix. Begin with the first person listed. Relations with the first person are listed in thefirst row.’’ Participants marked cells of the matrix with an N when the contact in the row and the contact in the column hadno relationship, an A when the same contacts had an acquaintance relationship, and a C when the participant perceived thecontacts to have a close relationship.

From the resulting matrix representation of each participant’s social network, we computed network density as the pri-mary dependent variable of interest (Freeman, 1979; Wasserman and Faust, 1994).5 The power affirmation hypothesis sug-gests that feeling powerful would lead people to activate less dense networks. The identity confirmation hypothesis suggeststhat high-status people primed with high power and low-status people primed with low power would activate less dense net-works. Again, we also expected change to provoke the hypothesized network dynamism. Finally, it is important to note that ifpeople’s network activation varies in an experimental setting where conditions were randomly assigned, then it is possible toinfer a direct causal relationship.

4. Results

We performed a series of regressions with measured status, and the manipulation of power and social change as the inde-pendent variables and network density as the dependent variable. We present our results across three tables. All coefficientestimates are standardized to account for wide variation in the underlying distributions of our independent variables

5 We also computed a host of other network measures including network size, betweenness, and constraint. We opted to proceed with the simpler networkdensity measure due to the fact that network size varies between participants and is systematically correlated with both betweenness (r = 0.82) and constraint(r = 0.58), but not with density (r = 0.00). As we find no effect of our experimental manipulations on the size of participants’ activated networks, it wasimportant to use a dependent variable—i.e., network density—that both captured the internal structure of participants’ networks and was independent ofnetwork size. Approximately 44% of all respondents named the maximum number of possible network contacts. Future research might use name generatorsthat allow participants to generate even more contacts to allow for more variation in network size.

Table 1OLS regressions predicting network density: Testing the power-affirmation hypothesis.

Variables 1 2

Power, manipulated 0.028 �0.051(0.30) (0.40)

Social change, manipulated �0.227(1.74)

Power � Change 0.147(0.92)

N 120 120R2 0.00 0.03F 0.09 1.11

Note: All coefficients are standardized.� p < 0.05; �� p < 0.01.

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 123

(e.g., 0–100 for measured status vs. 0–1 for manipulated power). Table 1 presents results from models including power andchange only and is thus a test of the power affirmation hypothesis. Table 2 adds measured status to test the identityconfirmation hypothesis, revealing a significant three-way interaction between power, status, and change. Given that socialchange indeed triggers the network dynamism, Table 3 analyzes in more detail those respondents who thought aboutchange, with a series of regressions that test for empirical mediation to understand the specific drivers of Table 2’s significantthree-way interaction.

4.1. Network density

The regressions in Table 1 show no support for the power affirmation hypothesis. In fact, Model 1 indicates that the asso-ciation between power and network density may be positive—feeling powerful leads people to recall denser networks—thoughthis association does not approach statistical significance. Model 2 offers some evidence that the social change condition mayamplify the salience of our independent variables, but again indicates no support for the power affirmation hypothesis.

Models 3 and 4 in Table 2 are tests of the identity confirmation hypothesis. As noted above, we operationalize identityconfirmation (and conflict) as the interaction between respondents manipulated power and self-reported status. Model 3includes these two variables alone, without the interaction term. Model 4 introduces the interaction but does not differen-tiate between respondents across the two social change conditions. Although the sign on the interaction term is consistentwith the identity confirmation hypothesis, the coefficient in Model 4 is not statistically significant. Model 5 reveals that thepredicted effects in fact emerge when participants were also thinking about periods of social change. The significant three-way interaction indicates that high-status people primed with high power and low-status people primed with low power(i.e., people with confirmed identities) elicited networks that were less dense, on average, than respondents with conflictedidentities.6

4.2. Visualizing low, middle, and high status effects

Fig. 1 depicts the results of Model 5 visually by splitting the continuous-scale measure of status into low, middle, and highcategories. Bar heights indicate the mean network density of participants in each status group by power condition. Onlyrespondents in the significant social change condition are included. Three things are evident from the figure. First, iden-tity-confirmed respondents—that is, LS/LP, and HS/HP—reported networks that were less dense than identity-conflictedrespondents of equivalent status. Second, as our theory would predict, these effects were driven by high and low statusrespondents, not middle-status respondents. Third, the figure suggests that neither high-status nor low-status respondentsare uniquely responsible for the significance of the interaction in Model 2, though within-status group comparisons do sug-gest that the power manipulation had a somewhat greater effect among low-status respondents.

4.3. Perceiving networks differently or perceiving different network subsections?

We next sought to understand why participants’ activated networks varied in the predicted directions. There are two pos-sible explanations: First, respondents in the identity-confirmed condition may have activated the same set of contacts that

6 One question is whether identity-confirmed respondents recalled a less dense network or whether identity-conflicted respondents recalled a densernetwork. To address this question empirically, we can compare respondents in the change condition to a matched set of participants assigned to the no socialchange condition, which may be treated as a ‘‘baseline’’ condition reflecting people’s typical activation patterns when they lack motivation to responddynamically (Greve and Taylor, 2000). Among low-status respondents we found that there was no statistically significant difference in the network densitiesbetween the identity-conflicted (i.e., LS/HP) respondents and either of the low-status group respondents assigned to the no change condition (mean = 0.24 vs.0.25 and 0.21, respectively). By comparison, the average network density of those with confirmed identities (i.e., LS/LP) was statistically different than each ofthe other three groups (mean = 0.11 vs. means reported above). Results of identical tests using the high-status sample were similar. The identity-confirmedrespondents (i.e., HS/HP) reported networks with a mean density lower and significantly different than both groups in the no change condition (mean = 0.15 vs.0.22 and 0.21) and lower but not significantly different than the identity-conflicted participants in the change condition (i.e., HS/LP mean = 0.15 vs. 0.18).

Table 2OLS regressions predicting network density: Testing the identity-confirmation hypothesis.

3 4 5

Power, manipulated 0.017 0.302 �0.699(0.18) (0.69) (1.08)

Status, self report �0.112 �0.062 �0.217(1.22) (0.52) (1.42)

Status � Power �0.291 0.649(0.67) (1.01)

Social change, manipulated �1.113(1.95)

Status � Change 0.899(1.59)

Power � Change 1.73(2.26)*

Status � Power � Change �1.582(2.13)*

N 120 120 120R2 0.01 0.02 0.08F 0.79 0.67 1.45

Note: All coefficients are standardized.�� p < 0.01.* p < 0.05.

Table 3OLS regressions predicting network density: Tests for mediation by feelings of discomfort and lack of control.

Dependent variable Lack of control Density Density Density6 7 8 9

Power, manipulated 1.448 1.380 0.881(2.43)* (2.31)* (1.48)

Status, self report 0.382 0.175 0.044(2.09)* (0.96) (0.24)

Status � Power �1.344 �1.290 �0.827(2.28)* (2.18)* (1.41)

Lack of Control 0.395 0.345(3.22)** (2.67)**

R2 0.11 0.11 0.16 0.21N 58 58 58 58F 2.13 2.12 10.36 3.56% of total effect that is mediated 0.36

Note: All coefficients are standardized.* p < 0.05.** p < 0.01.

124 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

they would have in the identity-conflicted condition but, for one reason or another, failed to remember a tie existing be-tween two of more of those contacts. Second, respondents in the identity-confirmed condition may have substituted oneor more well-connected (from the vantage point of the ego network) network contacts for one or more less-well-connectednetwork contacts (e.g., someone who has fewer or no relationships with the other people named), resulting in less networkdensity overall.

Figs. 2 and 3 explore these two possibilities using two additional dependent variables; first, the probability that a respon-dent named someone he or she sees as an ‘‘advisor,’’ and second, the proportion of a respondent’s network contacts who areeither family or friends. Fig. 3 indicates that among high-status respondents, having a confirmed identity significantly raisesthe probability that a person will recall an advisor as being part of their network (mean = 0.44 vs. 0.19). There was no sig-nificant effect among low-status respondents. Fig. 4 indicates that identity-confirmed respondents in both high- and low-status groups named a smaller proportion of family and friends in their networks. These results are consistent with the ideathat identity confirmation precedes the cognitive activation of broader, more diverse social networks. Identity confirmedparticipants were both more likely to name an ‘‘advisor’’ to their network and named a smaller proportion of family andfriends. These results hold controlling for network size.

4.4. Sense of control

We next analyzed participants’ feelings of comfort and control. This analysis allowed us to confirm the centrality of statusconsistency. We had predicted that participants would feel more in control when facing consistency between temporarily

Fig. 1. Activating dense networks: Status and power interaction. Whereas our results section reports status as a continuous variable, we split status intothree equally-sized categories to visually represent the results here. Low-status respondents primed with low power and high-status respondents primedwith high power both recall networks that are significantly less dense, controlling for network size. Middle-status respondents exhibit no significantdifferences in network activation as a function of power priming.

Fig. 2. Activating advisors: Status and power interact in the change condition We again split status into three equally-sized categories to visually representthe results here. High-status respondents primed with high power recall networks that are significantly (p < .05) more likely to include someone viewed asan ‘‘advisor,’’ controlling for network size. Middle- and low-status respondents exhibit no significant differences in advisor activation as a function of thepower priming.

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 125

primed power and their socioeconomic status. To test this, we combined the 7-point ‘‘lack of control’’ response discussedabove with the 7-point discomfort scale in Turner et al. (1992) (alpha = 0.76). The resulting 2-factor variable, lack of control,is reverse coded, meaning that higher values indicate greater negative reactions to a sensed loss of control and feelings ofdiscomfort (mean = 5.43, sd = 2.64). Model 6 in Table 3 indicates that the Power � Status interaction had a negative and sig-nificant effect on participants’ reaction to control. For the sake of interpretation, only respondents in the social change con-dition are included in the models in Table 3. The sign of the coefficient on the interaction term in Model 6 confirms thatidentity-confirmed participants (HS/HP and LS/LP) exhibited greater control and comfort than did participants in the iden-tity-conflicted conditions (HS/LP and LS/HP).7

7 We tested the remaining items from Turner (1992) as well as other emotions including whether people felt sad, angry, frightened, or hopeless and found nosignificant effects.

Fig. 3. Activating family/friends: Status and power interact in the change condition We again split status into three equally-sized categories to visuallyrepresent the results here. Low-status respondents primed with low power (p < .05) and high-status respondents primed with high power (p < .10) bothrecall networks that have a lower proportion of family and friends named as alters, controlling for network size. Middle-status respondents exhibit nosignificant differences in friend/family activation as a function of the power priming.

Status x Power Density ofActivated Network

Lack of Control

-1.344*

-1.290 * (-0.827NS

)

0.395** (.345**)

Fig. 4. Lacking control mediates network activation. Note: Coefficients are standardized.

126 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

4.5. Mediation

Finally, we tested whether lack of control mediated the interactive effects of power and socioeconomic status on networkdensity. Mediation is an essential step as it establishes that the psychological processes we had pinpointed around control infact underlie network activation. Models 7–9 are the test for mediation. Model 7 replicates the significant interaction inmodel 5 using the reduced sample, again providing support for the identity confirmation hypothesis. Model 8 regresses net-work density on lack of control only and reveals a strong positive effect: feeling out of control was associated with reporting adenser network. Model 9 offers evidence of empirical mediation: when lack of control is added to model 7, the Power x Statusinteraction is no longer significant, B = �0.827, n.s., but lack of control is significant B = 0.345, p < .01. The proportion of thePower x Status interaction effect on network density that is mediated is approximately 36%, according to Sobel–Goodmantests for mediation (Z = �1.808, p < .1). Fig. 4 offers a visual illustration of the mediated model. Thus, our model directly linksfeelings of control, which are the product of stable identities, with the activation of diverse networks.

5. General discussion

According to Goffman (1961, p. 43), stress is the tension ‘‘between the world that spontaneously becomes real to the indi-vidual and the one in which he is obliged to dwell.’’ Consistent with this theorizing, our results show that in times of change,even if power primes spontaneously offer people a sense of power, identity confirmation offers people the comfort necessaryto activate broad, non-redundant network ties. Exploring the environment could enable responsiveness during periods ofsocial change (Cyert and March, 1963; Greve and Taylor, 2000; Stinchcombe, 1990; Weick, 1995), but concerns about iden-tity may determine whether people are psychologically prepared to do so.

This paper’s empirical evidence also supports the dynamic constructivist model of network activation, showing that,when faced with situational primes, people spontaneously activate different subsections of their networks (Smith et al.,2012). It is worth reiterating here that we did not ask people to report to whom they would reach out given their currentsituation, but rather to whom they typically speak about important matters. That our experimental manipulations affectedpeople’s recall in a theoretically consistent way offers credence to our focus on the activated network. In addition to net-works gradually changing with life course events, people’s perceptions of their networks shift on a moment-to-moment ba-sis as a function of their psychological states.

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 127

5.1. Identity and network activation

An important contribution of this study is to demonstrate how the prospect of macro-societal change creates ripples atthe micro (with respect to individuals’ identities and feelings) and meso-levels (as people activate and eventually mobilizetheir networks). Specifically, we showed that people’s networking behavior is contingent on their sense of identity. Thosewho experience stable identities feel comfortable, have a sense of control, and venture beyond the security of dense net-works. In this way we offer evidence that the common sociological view of identity and networks—that is, identities emergeas the product of social relationships (White, 2008)—may also work in reverse: people’s images of their social worlds emergefrom their present sense of self. Identity, according to this alternative view, is not simply the result (or correlate) of socialstructure, it is an important driver of it.

Whereas significant evidence has accumulated integrating individual psychological differences with networks (Flynnet al., 2006; Klein et al., 2004; Mehra et al., 2001; Oh and Kilduff, 2008), this project suggests integrating networks with socialpsychology as well. Whereas research questions that focus on individual differences explore how people with stable person-ality differences differ in their approach to networking, our social psychological perspective emphasizes how situationsguide people’s network responses. By manipulating situations (e.g. the experience of power and change), we can establishthe causal role that these social conditions play in affecting people’s network perceptions.

This paper additionally demonstrates that identity’s effects on network perceptions operate through people’s psycholog-ical experience of being comfortable and in control. While past research acknowledges and assumes links between controland specific network structures (Burt, 1992; White, 2008), to our knowledge, this paper represents the first evidence thatempirically links people’s sense of control to patterns of network activation. Our findings suggest that identity-confirmedparticipants (high status and high power, and low status and low power) were united by a common psychological processwhereby identity confirmation leads to particular feelings of control. We tested but did not find main effect differences be-tween the high- and low-status groups with respect to comfort and control, and found that these feelings did not mediate thenetwork response.

5.2. Further research

While one’s place within a social hierarchy is a critical basis of identity in the Indian context, we theorize that similarprocesses should emerge regardless of context or dimension of identity (e.g., gender, race, political affiliation, and occupa-tional status). We expect comparable patterns of network activation to emerge whenever one’s identity is confirmed or con-flicted. As indicated by our mediation analysis, when identity is confirmed and people feel comfortable, they have a greaterlikelihood of networking broadly. It is also possible that identity threat might produce a unique emotional signature (e.g.perhaps involving emotions such as anger, fear, or sadness, which were not relevant predictors in our data) that leads to dif-ferent patterns of network activation.

For instance, it is possible that people who experience ambiguities about their gender or racial identities might seek to re-confirm their identity by activating networks full of people who share that gender or ethnicity. In contrast to our focus on thepsychological antecedents of broad network activation, this argument suggests that network activation may also amount to acompensatory process. People experiencing identity threats may seek comfort in dense social cliques. Those experiencing nosuch threats feel buffered enough to explore diverse areas of their environment. The self may be the foundation from whichpeople construct their social worlds.

Another area for further development concerns the links between our perspective and role-based explanations of networkactivation. We have emphasized how role inconsistencies can evoke individual psychologies (e.g., identity, sense of control)that affect network perceptions. However, we do not find evidence for more general role-based alternatives, whereby peoplesimply adhere to external scripts of thought and action associated with their status. This is not meant to imply that suchexternal scripts are unimportant, however. When faced with school problems, for instance, middle-class parents draw onand mobilize support from a wider, more diverse set of individuals than lower class parents, possibly reflecting role-typicalbehavior in the face of threat and change (Horvat et al., 2003). For the present paper, note the interactive nature of our effectsprecludes these general role-based explanations (e.g. diversification is role-typical for high status people). Instead, we findthat people do not assume such roles when there is no change in the situation and that low status people can adopt similarpatterns in other situations. To more definitively test role-based explanations, future research might add other mediators(e.g., participants’ identification with particular roles, or conformity). One of the most interesting future possibilities ishow the psychologies we document (identity, consistency, and comfort) might help explain role content, emergence, andwhen and why people adhere to roles.

A related area for future research is the dynamic interplay between diversity and stability. Our research suggests that sta-ble identities allow for diverse social connections. However, the process of activating diverse ties may very well trigger back-lash (White, 2008), whereby people who have a significant amount of diversity represented in their networks eventuallyexperience destabilized identities. In response, people might ‘‘deactivate’’ some of their diverse ties. While our researchhas emphasized identity as a causal force underlying people’s construction of their networks, the reverse process of causa-tion whereby relationships destabilize identity is also likely to occur (cf., Battilana and Casciaro, forthcoming; Burt, 2005;Fleming et al., 2007).

128 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

Finally, note that although the present research does not directly test the conscious or unconscious nature of these pro-cesses, these processes at the level of identity are consistent with implicit processes (Srivastava and Banaji, 2011). Just aspeople might not explicitly think, ‘‘I’ve got a collaborative self concept, let me build a collaborative network’’ (Srivastavaand Banaji, 2011), we suspect that they are similarly unaware of how their affirmed and conflicted identities lay the ground-work for their network activation processes. While our experimental methods prime identities at a conscious level of aware-ness, future research could employ these implicit measures or subconscious priming (see Zemack-Rugar et al., 2007) to testthis process directly.

5.3. Policy implications

Our results offer an important caveat about the taken-for-granted positive consequences of empowerment. The experi-ence of empowerment forces a person to reevaluate his or her sense of self. If stability in one’s identity constitutes a neces-sary precursor to individual agency, then empowerment initiatives may in fact prove detrimental for low-status people,while paradoxically benefiting high-status people instead.

The natural—and troubling—question that arises from this evidence is whether we are in fact prescribing that low-statuspeople should be reminded of their low status, rather than striving to enhance their self-esteem through empowerment.Obviously, this prescription contradicts what we know about the psychological risks of negative labels (Rosenberg, 1979).Rather than actively reminding people of their low status, we suggest that people dealing with low-status populations mightinstead ensure that such groups feel comfortable with tensions that arise from identity shifts so that they are willing to ex-plore their relational worlds more broadly. Instead of simply telling people that they have power, we suggest that a bettercourse might be to find specific kinds of power that are consistent with their identities and help them envision themselveswithin these more powerful roles. As compared to simple confidence, we suggest that these network perceptions may offer amore precise link between psychological, cognitive aspects of change and institutional dynamics (e.g., willingness to reachout to others and organize, McAdam, 1982).

A second prescription that often emerges in the networks literature is about cultivating more ties or restructuring one’snetworks (e.g. building in more structural holes, Burt, 1992). An implication of our cognitive activation model is that the crit-ical issue might not be one of resource constraint but resource squandering. Rather than investing in building new ties, thecritical question may instead be about how to activate the appropriate subsections of one’s existing, potential network. Byfailing to activate certain ties, people squander the resources available in those ties (Smith et al., 2012). Levin et al. (2011)reach a similar conclusion in their analysis of people’s tendencies to underestimate the value of their dormant ties. In ourview, dormant ties may be cognitively inaccessible or people might feel uncomfortable in thinking about them and thereforefail to activate them. These arguments suggest that one way technologies such as Facebook and Linkedin can improve peo-ple’s outcomes is by facilitating the recall of diversity in people’s networks (see Wellman and Haythornthwaite, 2000).

Finally, further research might consider the functionality of network broadening and narrowing in times of social change.While we suggest that broad and diverse networks should offer people a diversified portfolio with which to confront the var-ied possibilities that accompany change, too much diversity might create confusion and overload. Indeed, the specific type ofsocial change might even dictate the types of networks that enable effective outcomes. Staw et al. (1981), for instance, notethat tightening controls/efficiency may be effective responses to resource scarcity—but maladaptive when scarcity is symp-tomatic of radical change. Future work could successfully adapt this important nuance to a social network context.

6. Conclusion

Both psychologists and sociologists have observed the complex interplay between the self and its relational set. On onehand, psychological perspectives have focused on the multiple selves that comprise our unique individual identities (Markusand Nurius, 1986). On the other hand, beginning with Simmel (1922/1955: p. 150), sociologists have noted ‘‘individuals be-came unique to the extent that they affiliated with many different non-overlapping groups.’’ Our research strives to disen-tangle the effects of intra-personal diversity at the level of identity from interpersonal diversity at the level of relationships,suggesting a paradoxical relationship between the two. When multiple images of the self lead to confusion and ambiguity,people might compensate by seeking stable cliques. By contrast, stable, confirmed identities enable people to explore diver-sity in their networks. Thus, in contrast to Mary Douglas’ argument that ‘‘purity is the enemy of change (1966: p. 163),’’ ourresearch suggests that purity can be a friend of change, when it offers a stable, unified view of self that allows people to com-fortably explore the world around them.

References

Anderson, C., Galinsky, A.D., 2006. Power, optimism, and risk-taking. Eur. J. Social Psychol. 36, 511–536.Averill, J.R., 1973. Personal control over aversive stimuli and its relationship to stress. Psychol. Bull. 80, 286–303.Battilana, J., Casciaro, T., 1990. Change agents, networks, and institutions: A contingency theory of organizational change. Acad. Manage. J. 55, 381–398.Benoit-Smullyan, E., 1944. Status, status types, and status interrelations. Am. Sociol. Rev. 9, 151–161.Blader, S.L., Chen, Y., 2011. What influences how higher status people respond to lower status others effects of procedural fairness, outcome favorability, and

concerns about status. Organ. Sci. 22 (4), 1040–1060.Blau, P.M., 1964. Exchange and power in social life. John Wiley and Sons, New York.

T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130 129

Boissevain, J., 1974. Friends of Friends: Networks, Manipulators and Coalitions. Basil Blackwell, Oxford, England.Bothner, M., Smith, E.B., White, H., 2010. A model of robust positions in social networks. Am. J. Sociol. 116 (3), 943–992.Burt, R., 1992. Structural Holes: The Social Structure of Competition. Harvard University Press, Cambridge, MA.Burt, R.S., 1980. Models of network structure. Annu. Rev. Sociol. 6, 79–141.Burt, R.S., 2005. Brokerage & Closure: An Introduction to Social Capital. Oxford University Press, NY.Carley, K.M., 1986. An approach for relating social structure to cognitive structure. J. Math. Sociol. 12, 137–189.Carter, T.J., Ferguson, M.J., Hassin, R.R., 2011. A single exposure to the American flag shifts support toward Republicanism up to 8 months later. Psychol. Sci.

22 (8), 1011–1018.Chakravarti, D., 2006. Voices unheard: the psychology of consumption in poverty and development. J. Consumer Psychol. 16 (4), 363–376.Cyert, R.M., March, J.G., 1963. A Behavioral Theory of the Firm. Prentice-Hall, Englewood Cliffs, NJ.Douglas, M., 1966. Purity and Danger: An Analysis of the Concepts of Pollution and Taboo. Routledge & Kegan Paul, London.Dumont, L., 1980. Homo hierarchicus: the caste system and its implications (Complete English edition, Revised). Nat. Human Soc. 540, 1970.Duncan, O.D., 1984. Notes of Social Measurement, Historical and Critical. Russell Sage Foundation, New York.Ebaugh, H.R., 1988. Becoming an Ex: The Process of Role Exit. University of Chicago Press, Chicago.Emirbayer, M., Goodwin, J., 1994. Network analysis, culture, and the problem of agency. Am. J. Sociol. 99, 1411–1454.Erikson, E.H., 1970. ‘‘Identity crisis’’ in perspective. In: Erikson, E.H. (Ed.), Life History and the Historical Moment. Norton, New York, p. 975.Fast, N.J., Halevy, N., Galinsky, A.D., 2011. The destructive nature of power without status. J. Exp. Soc. Psychol. 48, 391–394.Feld, S.L., 1981. The focused organization of social ties. Am. J. Sociol. 86, 1015–1035.Festinger, L., 1957. A Theory of Cognitive Dissonance. Stanford University Press, Stanford, CA.Fiske, S., 1993. Controlling other people: the impact of power on stereotyping. Am. Psychol. 48, 621–628.Fleming, L., Mingo, S., Chen, D., 2007. Collaborative brokerage, generative creativity, and creative success. Adm. Sci. Q. 52, 443–475.Flynn, F.J., Reagans, R.E., Amanatullah, E.T., Ames, D.R., 2006. Helping one’s way to the top: self-monitors achieve status by helping others and knowing who

helps whom. J. Pers. Soc. Psychol. 91, 112–1137.Freeman, L.C., 1979. Centrality in social networks: conceptual clarification. Social Netw. 1, 215–239.Fuchs, S., 2001. Against Essentialism. Harvard University Press, Cambridge, MA.Galinsky, A.D., Gruenfeld, D.H., Magee, J.C., 2003. From Power to Action. J. Pers. Soc. Psychol. 85, 453–466.Goffman, Erving, 1961. Encounters: Two Studies in the Sociology of Interaction. Bobbs-Merrill, Indianapolis.Government of India. National Policy for the Empowerment of Women, 2001. <http://www.wcd.nic.in/empwomen.htm> (retrieved 27.10.11).Granovetter, M.S., 1973. The Strength of Weak Ties. Am. J. Sociol. 78, 1360–1380.Greve, H.R., Taylor, A., 2000. Innovations as catalysts for organizational change: shifts in organizational cognition and search. Adm. Sci. Q. 45, 54–80.Gruenfeld, D.H., Inesi, M.E., Magee, J.C., Galinsky, A.D., 2008. Power and the objectification of social targets. J. Pers. Soc. Psychol. 95 (1), 111–127.Higgins, E.T., Kruglanski, A.E., 1996. Knowledge Activation: Accessibility, Applicability, and Salience. Guilford, New York.Hofstede, G., 1980. Culture’s Consequences: International Differences in Work-Related Values. Sage, Beverly Hills, CA.Hughes, E.C., 1945. Dilemmas and contradictions of status. Am. J. Sociol. 50, 353–359.Ibarra, H., 1999. Provisional selves: experimenting with image and identity in professional adaptation. Adm. Sci. Q. 44, 764–791.Ibarra, H., Barbulescu, R., 2010. Identity as narrative: prevalence, effectiveness, and consequences of narrative identity work in macro work role transitions.

Acad. Manage. Rev. 35, 135–154.Johnson, F., Orbach, M., 2002. Perceiving the political landscape: ego biases in cognitive political networks. Social Netw. 24, 291–310.Keltner, D., Gruenfeld, D.H., Anderson, C., 2003. Power, approach, and inhibition. Psychol. Rev. 110, 265–284.Kilduff, M., Tsai, W., Hanke, R., 2006. A paradigm too far? A dynamic stability reconsideration of the social network research program. Acad. Manage. Rev.

31, 1031–1048.Klein, K.J., Saltz, J.L., Mayer, D.M., 2004. How do they get there? An examination of the antecedents of centrality in team networks. Acad. Manage. J. 47, 953–

963.Krackhardt, D., 1987. Cognitive social structures. Social Netw. 9, 109–134.Kraus, M.W., Piff, P.K., Keltner, D., 2009. Social class, sense of control, and social explanation. J. Pers. Soc. Psychol. 97, 992–1004.Lawler, E., 2001. An affect theory of social exchange. Am. J. Sociol. 107 (2), 321–352.Lenski, G.E., 1954. Status crystallization: a non-vertical dimension of social status. Am. Sociol. Rev. 19, 405–413.Levin, D.Z., Walter, J., Murnighan, J.K., 2011. Dormant ties: the value of reconnecting. Organ. Sci. 22, 923–939.Levinthal, Daniel A., March, James G., 1993. The myopia of learning. Strateg. Manage. J. 14, 95–112.Marks, S.R., 1977. Multiple roles and role strain: some notes on human energy, time and commitment. Am. Sociol. Rev. 42, 921–936.Markus, H.R., 1977. Self-schemata and processing information about the self. J. Pers. Soc. Psychol. 35, 63–78.Markus, H., Nurius, P., 1986. Possible selves. Am. Psychol. 41, 954–969.Marsden, P., 1990. Network data and measurement. Annu. Rev. Sociol. 16, 435–463.McAdam, D., 1982. Political Process and the Development of Black Insurgency, 1930–1970. University of Chicago Press, Chicago.McCarthy, J.D., Zald, M.N., 2001. The enduring vitality of the resource mobilization theory of social movements. In: Turner, Jonathan H. (Ed.), Handbook of

Sociological Theory, pp. 533–565.McDonald, M., Westphal, J.D., 2003. Getting by with the advice of their friends: CEOs’ advice networks and firms’ strategic responses to poor performance.

Adm. Sci. Q. 48, 1–32.McPherson, M., Smith-Lovin, L., Brashears, M., 2006. Social isolation in America: changes in core discussion networks over two decades. Am. Sociol. Rev. 71,

353–375.Mehra, A.M., Kilduff, D., Brass, D., 2001. The social networks of high and low self-monitors: implications for workplace performance. Adm. Sci. Q. 46 (1),

121–146.Menon, T., Blount, S., 2003. The messenger bias: how social relationships affect the valuation of knowledge. Res. Org. Behav. 25, 137–187.Menon, T., Thompson, L., Choi, H., 2006. Tainted knowledge versus tempting knowledge: why people avoid knowledge from internal rivals and seek

knowledge from external rivals. Manage. Sci. 52, 1129–1144.Michaelson, A., Contractor, N., 1992. Structural position and perceived similarity. Social Psychol. Q. 55 (3), 300–310.Milton, L.P., Westphal, J.D., 2005. Identity confirmation networks and cooperation in workgroups. Acad. Manage. J. 48, 191–212.Mouw, T., 2002. Are black workers missing the connection? The effect of spatial distance and employee referrals on interfirm racial segregation.

Demography 39 (3), 507–528.Oh, H., Kilduff, M., 2008. The ripple effect of personality on social structure: self-monitoring origins of network brokerage. J. Appl. Psychol. 93, 1155–1164.Pfeffer, J., 1992. Managing with Power. Harvard University Press, Harvard.Podolny, Joel., 1993. A status-based model of market competition. Am. J. Sociol. 98, 829–872.Polzer, J., Milton, L.P., Swann Jr., W.B., 2002. Capitalizing on diversity: interpersonal congruence in small work groups. Adm. Sci. Q. 47 (2), 296–324.Rafaeli, A., Sutton, R.I., 1989. The expression of emotion in organizational life. Res. Org. Behav. 11, 1–43.Rosenberg, M., 1979. Conceiving the Self. Basic Books, New York.Sasovova, Z., Mehra, A., Borgatti, S.P., Schippers, M.C., 2010. Network churn: the effects of self-monitoring personality on brokerage dynamics. Adm. Sci. Q.

55, 639–670.Simmel, Georg, 1922 [1955]. Conflict and the Web of Group Affiliations (Translated and edited by Kurt Wolff). Free Press, Glencoe, IL.Smith, E.B., 2011. Identities as lenses: how organizational identity affects audiences’ evaluation of organizational performance. Adm. Sci. Q. 56 (1), 61–94.

130 T. Menon, E.B. Smith / Social Science Research 45 (2014) 117–130

Smith, E.B., 2013. From role conflict to evaluation discordance. How do conflicting performance evaluations affect risk taking in multiple audience contexts?Research in organiaztional sciences. In: Svyantek, D., Mahoney, K. (Eds.), Received Wisdom, Kernels of Truth, and Boundary Conditions in OrganizationalStudies. Information Age Publishing, Charlotte, NC.

Smith, E.B., Menon, T., Thompson, L., 2012. Status differences in the cognitive activation of social networks. Organ. Sci. 23, 67–82.Srivastava, S.B., Banaji, M.R., 2011. Culture, cognition, and collaborative networks in organizations. Am. Sociol. Rev. 76 (2), 207–233.Staw, B., Sandelands, L., Dutton, J., 1981. Threat-rigidity effects in organizational behavior: a multi-level analysis. Adm. Sci. Q. 26, 501–524.Stinchcombe, A.L., 1990. Information and Organizations. University of California Press, Berkeley, CA.Swann Jr., W.B., 1983. Self-verification: bringing social reality into harmony with the self. In: Suls, J., Greenwald, A.G. (Eds.), Social Psychological

Perspectives on the Self, vol. 2. Erlbaum, Hillsdale, NJ, pp. 33–66.Was, C.A., 2010. The persistence of content-specific memory operations: priming effects following a 24-hour delay. Psychon. Bull. Rev. 17 (3), 362–368.Wasserman, S., Faust, K., 1994. Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge.Weick, K.E., 1995. Sensemaking in Organizations. Sage, Thousand Oaks, CA.Wellman, B., Haythornthwaite, C., 2000. The Internet in Everyday Life. Blackwell, Oxford.White, H.C., 2008. Identity and Control: How Social Formations Emerge, second ed. Princeton University Press, Princeton, NJ.Woltz, D.J., Was, C.A., 2007. Available but unattended conceptual information in working memory: temporarily active semantic content or persistent

memory for prior operations? J. Exp. Psychol. Learn. Mem. Cogn. 33, 155–168.Zemack-Rugar, Y., Bettman, J.R., Fitzsimons, G.J., 2007. The effects of nonconciously priming emotion concepts on behavior. J. Pers. Soc. Psychol. 93, 927–

939.