Lynn Smith-Lovin's "The Strength of Weak Identities" (2007)

Social Psychology Quarterly
The Strength of Weak Identities: Social Structural Sources of Self, Situation and Emotional Experience
Lynn Smith-Lovin
Social Psychology Quarterly 2007 70: 106 DOI: 10.1177/019027250707000203
The online version of this article can be found at:
Published by:
On behalf of:
American Sociological Association
Additional services and information for Social Psychology Quarterly can be found at: Email Alerts:
>> Version of Record - Jun 1, 2007 What is This?
Downloaded from by Wayne Mellinger on October 20, 2011

Social Psychology Quarterly
2007, Vol. 70, No. 2, 106–124
The Strength of Weak Identities: Social Structural Sources of Self, Situation and Emotional Experience*
Duke University
Modern societies are highly differentiated, with relatively uncorrelated socially salient dimensions and a preponderance of weak, unidimensional (as opposed to strong, multiplex) ties. What are the implications of a society with fewer strong ties and more weak ties for the self? What do these changes mean for our emotional experience in everyday life? I outline a structural view of self, situated identity, and emotion. It is an ecological theory in which interpersonal encounters are the link between the macro-level community structure and the micro-level experience of self-conception, identity performance, and emotion. In this ecol- ogy of encounters, multiple-identity enactments (especially of salient self-identities) are quite rare. But where they occur, they are important indicators of potential social change.
The central theme of this address is the impact of changing social structures on the self, its constituent identities, the ecology of interactions, and the resulting emo- tional experiences. As a structural symbolic interactionist, I argue that the person we become depends profoundly on the networks in which we are embedded. The actions we take and the emotions we experience depend on these networks. These networks are, in turn, shaped powerfully by the social settings that we occupy. Social change occurs, very rarely, when these forces operate to change the cultural meanings of identity labels.
Throughout my career, I have been inter- ested in how social settings influence the experiences of actors within them. My disser-
* This Cooley-Mead address was presented to the Social Psychology section of the American Sociological Association on August 13, 2006, at the Association’s annual meetings in Montreal. Address correspondence to Lynn Smith-Lovin, Department of Sociology, Box 90088, Duke University, Durham, NC 27708-0088; smithlov@ The National Science Foundation has sup- ported my research program throughout my career. Grants that supported specific findings mentioned in this address include National Science Foundation Grants SES- 9008951 and SES-0347699. I thank David R. Heise, Neil J. MacKinnon, Miller McPherson, Timothy J. Owens, Dawn T. Robinson, Sheldon Stryker, Allison K. Wisecup, and members of my Sociology 229 seminar at Duke University for comments on earlier drafts of this address. Allison K.Wisecup provided valuable help with the analy- ses reported here.
tation measured the affective meanings that cultures have for settings, and what actions operated to sustain those meanings (Smith- Lovin 1979, 1987b). I was fascinated by Goffman’s (1963) description of the expres- sive order of public places and by psychologist Roger Barker’s (Barker 1968; Barker and Wright 1951, 1959) careful description of the behavioral settings in a small Midwestern town. Goffman described how people were obligated to sustain the character of institu- tionalized settings. Barker and his colleagues recognized that what we do depends much more on where we are and who we are with than who we are. While many sociologists of the era were concentrating on personal atti- tudes and values, Barker noted that place determined the patterned elements of an inter- action, including its participants, actions, and possible outcomes.
These scholars had a sense of settings as constraining occupants’ activities, explaining much of the taken-for-granted variance in everyday interaction. They recognized that an actor’s socially structured environment deter- mined more about what he or she did than what was inside him or her. The person who was a quiet worshipper at church could be a boisterous extrovert at a party the night before. I still think that Goffman and Barker were correct in their assessment of settings’ impacts. In this address, I return to some of
Downloaded from by Wayne Mellinger on October 20, 2011
their themes. I develop a theory of how com- munity structure affects selves, interactions, and emotions.
I began this work six years ago in response to a review of Identity Theory research (Stryker and Burke 2000). The review argued that we (1) needed to further develop the conceptualization of the self as a set of multiple identities, and (2) needed to further specify how commitment to networks of relationships were related to internalized identities. The next year, I received an invita- tion to attend a conference honoring Sheldon Stryker (Burke, Owens, Serpe and Thoits 2003; Stryker 2003). This opportunity led me to think about the concepts of self and identi- ty as they related to networks of interaction. I developed an ecological theory of the structur- al conditions under which complex selves develop (Smith-Lovin 2003). I tried to predict when multiple identities are likely to operate in the same setting. In this address, I develop my new ideas on the topic with a little back- ground on why they have changed. I make six arguments:
  1. Current social systems—with their high differentiation, their relatively uncorre- lated social statuses, and their technolo- gies that free us from limitations of time and space– have resulted in complex selves but unidimensional (weak tie) relationships.
  2. Such a weak tie system will make selves less stable as well as more complex.
  3. Complex but unstable selves are more
    likely to be characterized by attributes and less likely to be characterized by institutionalized role identities.
  4. Simultaneous engagement with two or more self-identities is most likely to occur either when ruminating about dif- ferent parts of the self or in situations with multiple audiences.
  5. Mixed emotions are elicited by those rare situations where an actor simulta- neously enacts identities with very dif- ferent meanings.
  6. While rare, interactions in simulta- neously held self-identities and the mixed emotions that they evoke are important precursors of cultural changes in meaning.
I titled this address “the strength of weak identities,” with a bow to Granovetter’s (1973) classic paper that outlined the life-shaping impact of information carried by weak ties. Weak ties are the simple, less intense, more unidimensional ties that we are likely to have with those who are far from us in social space. My argument is that the structure of modern social systems leads us to spend much time in these weak-tie interactions. These interactions result in complex, fluid selves, but relatively simple interactional situations.
The ecology of our encounters depends more on the structure of our environment than on individual volition. To repeat the title of one of my own papers, “You are who you know” (Smith-Lovin and McPherson 1993). But who you know depends profoundly on the structure of organizations and institutions that surround you. Symbolic interactionists have often stressed the choices that we make among identity enactments (e.g., if “professor” is high in the salience hierarchy of our self- structure, we will start lecturing at the drop of a hat). I emphasize here that social structures around us often lead us to enact identities that are not central to our self-structure. Who wants to be a “traffic violator” when driving? Those “weak” identity enactments, while not central to our definitions of self, still influence our emotional lives in a profound way. The most common situation that evoked anger in the 1996 Module on Emotions in the General Social Survey was waiting in line at the gro- cery store.
In a sense, I argue for modifying Stryker’s (1980) classic statement of structural symbol- ic interaction: that society shapes self which then shapes social interaction. Society does shape selves. It also shapes interactions through the ecology of encounters. But in much of everyday life, selves do not dominate as the central mediators that they were in Stryker’s formulation. Instead, I propose below that the social environment (especially its network connections) shapes both the self and social interaction, and creates a somewhat spurious correlation between the two.
I hasten to qualify my revision of Stryker’s venerable statement. Stryker only asserted that selves influenced interaction in
Downloaded from by Wayne Mellinger on October 20, 2011
cases where choice of identity was possible. I am suggesting that the glass of “choice” is more than half empty—that the majority of encounters in actors’ lives are shaped not by self-structures, but by the social environments in which they are embedded. These environ- ments may be determined to some degree by past choices (e.g., the choice to marry, the choice to work at a given job). But, as sociol- ogists, we know that those “choices,” too, are influenced by many factors other than identity structures. As actors, our menu of opportuni- ties and the information accumulated from past network contacts strongly affect our cur- rent social position. So, as experienced in everyday life, selves are probably more impor- tant to how we think and feel about ourselves than to our interaction probabilities. We spend much time in identities that are not central to our self-structure. When we are enacting iden- tities that are central to the self-structure, it is more likely to be a function of the institution- al environments in which we are embedded than any immediate choices we make.
The fact that so many of our modern insti- tutional settings segregate us into single-iden- tity environments—we live in a Gesellschaft rather than a Gemeinschaft world—makes multiple identity enactments quite rare. One of the reasons that Affect Control Theory, as a theory of situations rather than selves, has been so successful in describing role behav- iors (Heise 1979, 2007; Smith-Lovin and Heise 1988; MacKinnon 1994) is that there are relatively few situations in which actors occupy multiple identities with strikingly dif- ferent meanings. That said, it is on those rare occasions when the ecology of encounters does lead to parts of the self being activated that are both simultaneously enacted and importantly different in meaning that, when experienced by multiple individuals, lead to cultural change. Understanding this mecha- nism of multiple-identity occupancy may show how micro-level interactions can lead to macro-level changes in cultural meanings (and the structures that they both generate and represent). I hope this perspective contributes to the much-neglected pathway from self to the reshaping of society.
The structural symbolic interaction upon which I build began with Sheldon Stryker’s (1980) linking of symbolic interactionist ideas with role theory. In his Identity Theory, Stryker rejected a symbolic interaction that focused only on creative, atypical behavioral productions in ill-defined, unconstraining behavioral settings. Instead, he concentrated on the stable, recurring interactions in our social system. By linking role behavior to the internalized meanings that roles had for indi- viduals, he provided the connection between social structure, meaning, and action that dri- ves structural symbolic interaction today.
Especially relevant is research conducted by Stryker and Richard Serpe (1982; Serpe 1987; Serpe and Stryker 1987, 1993) that developed an ecological understanding of the self. This work showed how the change in set- ting from high school to college led to two related processes. First, self-identities moti- vated students to seek out groups in the new environment that would allow the expression of salient, long-held roles. Individuals recreat- ed themselves in the new environment. Second, the social structure of the new envi- ronment had an impact on the selves that could be sustained in that setting. When groups were not available to reaffirm old role- identities, those identities withered and decreased in salience. This dynamic ecology of actor choice and social structural resources gives us a powerful picture of how selves and social environments shape one another.
We have not seen substantial development in this very productive line of work in the past three decades. The focus of work in structural symbolic interaction shifted to a control-sys- tem view of identity and action. This control system focused on one identity at a time and how it was maintained in interaction (Stryker and Burke 2000). Affect Control Theory led this movement with a mathematical model of the relationship between identity and action (Heise 1979, 2007; Smith-Lovin and Heise 1988; MacKinnon 1994; Robinson and Smith- Lovin 2006).
Downloaded from by Wayne Mellinger on October 20, 2011
Affect Control Theory had no model of the self. Social situations determined what identities would be enacted (although actors could seek out situations with a predisposition to enact an identity, as when a person enters a doctor’s office in a desire to become a Patient1). The theoretical argument focused on the behaviors and attributions that occur once actors define a situation and have determined what the relevant self-identities are within it. Affect Control Theory was a theory of social situations, not of the relationship between individuals and the social structures in which they were embedded.
Just as many of us turned to the interrela- tionship of emotion, identity, and action in the late 1980s (Smith-Lovin 1990; see review in Smith-Lovin 1995), it seems that it is time to put identity, self, and social structure back together again. I join MacKinnon and Heise (forthcoming) and Stryker’s recent work (Stryker, Serpe and Hunt 2005) in tackling the question of the social structural sources of self and social interaction. This structural empha- sis will connect our symbolic interactionism to important substantive issues like social movement participation (Heise 1998; Heise and Britt 2000; Stryker 2000) and the impact of identity occupancy on mental health (Thoits 2003; Simon 1995, 1998; MacKinnon and Golbournne 2006). At a more structural level, it allows us to link the impressive work on the dynamic evolution of social groups (e.g., McPherson 1983; McPherson and Ranger-Moore 1991) to the epidemiology of individual experience. Like Stryker (1980), I hope to reintegrate symbolic interactionist thought with the mainstream concerns of more macro-level thinkers.
In this address, I partition the broad issue of the multiple-identity self into a set of more specific theoretical questions. This dissection allows me to use some well-formed ideas from other research traditions to link social struc- ture and individuals into an ecology of encounters and identities. I present some
1 I will use the Affect Control Theory convention that capitalization indicates identity, behavior, and emotion labels that carry affective meaning within a culture that must be maintained by interactions.
data—from an experiential sampling study and from a nationally-representative survey– that illustrate some features of modern identi- ty structure. I then make some rather straight- forward arguments, based on Affect Control Theory, about how the rare, simultaneous experience of multiple identities with substan- tively different cultural meanings can produce emotions and actions that create social change.
I focus on five basic questions. First, what social structures determine available identi- ties? Second, how do these identities combine into selves as individuals internalize them through personal biography? Third, when do people occupy two or more identities within the same social situation? Fourth, how do those simultaneously held identities produce lines of social action and emotional response? Fifth, what does the experience of multiple identity occupancy imply for social change?
I use the term identity in a broader sense than Stryker’s (1980) definition. In his inte- gration of symbolic interaction and role theo- ry, Stryker (1980) focused on role-identities, the internalized meanings of roles for the indi- vidual. Instead, I adopt the more comprehen- sive image that Neil MacKinnon and David Heise (forthcoming) use in their new book on identity. MacKinnon and Heise define “cultur- al theories of people” as the set of categories that a culture provides for labeling types of people, as well as the logical implications among those categories. For example, the identity Supreme Court Justice implies Lawyer, which implies College Graduate, and eventually implies more abstract levels like Adult and Human. Notice that these cultural theories of people can change. Thirty years ago the same identity, Supreme Court Justice, might have implied the more abstract identity Man. While that might still be a part of the prototypical Justice, the strength of that impli- cation has softened in the past three decades.
In the theoretical argument here, I deal with identities that operate at the same level of the perceptual control system, rather than those that represent higher or lower levels of
Downloaded from by Wayne Mellinger on October 20, 2011
reference signal (see McClelland and Fararo 2006 for a more complete treatment of control systems in sociology). Control theorists know that multiple levels of control exist, with shifts at higher levels effectively resetting the refer- ence levels that are operative at lower levels. For example, if I come to view a Friend as a Rival for a romantic interest, I might reinter- pret earlier actions in this new framing, while not changing my view of what had physically occurred in those earlier interactions (a lower level of processing), nor changing my view of my Friend/Rival as a normal, human member of my social group (a higher level of judg- ment).
Here, I attempt to analyze the relation- ships among those identities whose meanings directly generate lines of action and emotion- al responses to the actions of others in inter- personal situations. An operational criterion is the nouns that people might use spontaneous- ly (or when asked) to name themselves or oth- ers within a situation. So, both Friend and Rival would be identities, because they are ways in which I could label someone with whom I have an interaction. They make sense within my culture, and the act of labeling someone communicates much to others in the culture who share those words and their social meanings.
Cultural identity labels include: (1) the role-identities indicating positions in the social structure, (2) the social identities indi- cating membership in groups, and (3) the cat- egory memberships that come from identifi- cation with some characteristic, trait, or attribute (Smith-Lovin 2003). For example, when we asked 38 members of the university community at Arizona about their identities in a 1995 experiential sampling study, they reported 328 distinct identities. These ranged from role-identities with clear role alters (e.g., Bartender, Landlord, Sister), to activity-based identities with ambiguous alters (e.g., Artist, Camper, Music lover), to social identities based on group membership (e.g., Church Member, Greek), to salient personal charac- teristics (e.g., African American, Responsible Person).
I argue that this wide range of social labels should be studied together because they
represent the ways that people think about themselves and others in situations. Cognitive labeling and affective meaning are inextrica- bly intertwined (Heise 1979; MacKinnon 1994). Labeling someone (including oneself) leads inevitably to control processes of identi- ty maintenance. Since identity labels carry cultural meaning and guide social action, I first ask what features of social structure determine the availability of potential identities.
What social structural features determine the “cultural theory of people” available to actors? All three types of identity—role-iden- tities, group memberships, and differentiating characteristics—have networks as their source. In the case of role-identities, a network relation with an alter defines a position within a social structure. That position has rights, responsibilities, and behavioral expectations vis a vis some other position (Merton 1957; Stryker 1980). In the case of group member- ship, the network tie is a connection to a named group of alters (Breiger 1974; McPherson 1983). In the case of personal- characteristic identities, interactions with peo- ple different from us create salient social cate- gorizations (Berger, Fisek, Norman, and Zelditch 1972). We only know we are intelli- gent if we compare ourselves with someone we think is less smart. Our social context (in particular, who we are in contact with) influ- ences the meaning of our category occupancy (Hogg and Abrams 1988). The meaning of our category membership (e.g., what it means to be British) is influenced by the context to which we compare ourselves. If the British compare themselves to Americans, tradition- alism and reserve may be most salient. If com- parison to new immigrant groups from older civilizations (e.g., the Middle East) is more proximate, distinctions like patriotism or tol- erance may be highlighted.
Given that network ties generate identity labels, we know that affective meanings (like status) will almost inevitably follow (Mark, Ridgeway, and Smith-Lovin unpublished). Once there is a noticeable difference among
Downloaded from by Wayne Mellinger on October 20, 2011
actors that allows people to be categorized, then the category label acquires status value (e.g., evaluative meaning or influence) within a culture with remarkable regularity.2 Even characteristics with no relationship to resources or other salient status distinctions will become imbued with evaluation and influence potential. Social systems can create evaluative meaning almost out of thin air. The smallest human groups have age and sex vari- ations, and these distinctions inexorably lead to “cultural theories of people,” with implica- tions about the relationships among categories and evaluative meaning.3
The clear dependence of identities on net- work relations is very useful, for there is a substantial literature about network features of social systems. The first principle upon which I draw is the relationship between size and dif- ferentiation (e.g., Mayhew, Levinger, McPherson, and James 1972; Mayhew 1974). In virtually any domain—from the entire social system to a voluntary association— larger size leads to increased internal differen- tiation. Relations between actors shift from Gemeinschaft to Gesellschaft. In larger sys- tems, we interact with those who are function- ally interrelated but different from us; in smaller systems, we interact with those who are similar. More differentiation means more role identities, more membership groups, and more salient distinctions among those who interact within the system.
Miller McPherson (2004) has made a closely related argument about the dimension- ality of the salient socio-demographic space, which he calls “Blau space.” He notes that there are few characteristics distinguishing individuals in small, technologically simple
2 More formally, evaluative differentiation of cate- gories is an attractor. If one is willing to make a couple of simple assumptions about the diffusion and loss of status beliefs, only very specific social conditions will prevent consensual status beliefs from forming.
3 Because I am an Affect Control Theorist, I expect that potency (powerfulness vs. weakness) and activity (liveli- ness vs. quietness) meanings would be created in the same way. However, the Mark, Ridgeway, and Smith-Lovin (forthcoming) paper is developed within the status con- struction paradigm and deals only with evaluative mean- ing.
societies (primarily age, sex and physical capabilities). As society grows in size and scope, the scale of the system acquires other dimensions, such as wealth and education, to organize social interaction. More importantly, McPherson (2004) argues that salient dimen- sions of social differentiation become less cor- related in large systems. This unfolding of the multi-dimensional social space leads not just to greater diversity in the system as a whole; it also allows the development of many more distinctive regions (niches) within the social system. McPherson (1983; McPherson and Rotolo 1996) has shown that these niches have profound implications for the shape and com- position of membership groups within the larger social structure. The composition of groups and their social environments, in turn, have profound implications for the network ties of their members (McPherson and Smith- Lovin 1981, 1986, 1987, 2002). Most theorists have focused on the information that such diverse ties bring (Granovetter 1983; McPherson and Smith-Lovin 1987). Here, I focus on the consequences for self-identities. Both the size-differentiation principle and the unfolding of Blau space in larger, technologi- cally advanced systems lead us to the same prediction.
Proposition 1: System size will be positively related to the number of identities in the system.
Having established the opportunity struc- ture for creating selves within society, we now turn to the question of how actors occupy and internalize identities to create a self-structure.

Now that I have described the resource space within which selves are formed, I can proceed to the next question: what determines the complexity and stability of selves? The differentiation in larger systems creates lower density of interactions among actors and greater segmentation of that interaction. This lower density of interactions has profound implications for the selves of social actors within the system.
Downloaded from by Wayne Mellinger on October 20, 2011
I develop a somewhat more nuanced set of predictions by using more of the ecological framework developed by those who study the interplay of population distributions, net- works, and social groups. McPherson (1983) developed an ecological theory that should apply to any social entity that (1) spreads through homophilous network contacts and (2) involves some level of competition for the time or energy of actors. We know that social systems are characterized by homophily, the increasing probability of interaction as actors become more similar on almost any character- istic, from physical distance to socio-demo- graphic features to information (McPherson, Smith-Lovin, and Cook 2001). Birds of a feather flock together. When social entities (like groups, cultural tastes, or occupations) compete for people’s energy in this homophilous resource space, they become localized in that space. Different kinds of peo- ple do different kinds of things. Entities as wide ranging as voluntary group member- ships, occupations, musical tastes, and reli- gious practices have been successfully ana- lyzed using this framework (Mark 1998, 1999; Rotolo and McPherson 2001; Chaves 2004).
An ecology of identity allows us to exam- ine the relationship between system-level characteristics and the range, diversity, and overlap of identity occupancy (Smith-Lovin 2003). For example, McPherson (2000) used simulations to analyze the relationship between the level of homophily in a system and some outcomes that are directly related to identity—the number of distinct groups, the heterogeneity within groups, and the member- ship overlap (the extent to which actors in the system are members of multiple groups simul- taneously). Remembering that “group” here can represent any social entity that spreads through homophilous networks and competes for time and energy, all of these features should be related to the self-complexity of actors within the system. In the simulations, a high level of homophily suppresses the extent to which groups overlap and the diversity of people within those groups. Effects on group size and the number of groups (net of system size) are minimal, primarily because homophily has countervailing direct and indi-
rect forces. The direct effect of homophily is to create more, smaller groups. But homophi- ly also has an effect on tie stability; homophilous ties are more likely to survive for longer periods. Recall that the effect of tie stability is to make groups less numerous and larger. When homophily’s direct effects on group size and number and its indirect effects through tie stability are taken into account, the net effect is near zero.
Therefore, any impact on the complexity of self structures from homophily comes from the overlap of groups or the diversity of groups, both of which should make for a more complex self. Homophily within a social sys- tem is likely to be created when socio-demo- graphic dimensions are more correlated, since homophily on multiple dimensions can be optimized simultaneously in such a system. Under such conditions, groups (and other social entities like communities that hold sim- ilar tastes, engage in similar activities, etc.) will tend to be small and less diverse, leading to a simpler self.
Proposition 2: Complexity of self structures will be positively related to system size.
Proposition 3: Complexity of self structures will be negatively related to the correlation of salient social distinctions within the social system.
Proposition 4: Complexity of self structures will be negatively related to the level of homophily in a social system.
This treatment of self structure is more structural and cultural than social psychologi- cal. I argue that this approach is useful in order to move us back to a serious considera- tion of the social structures in which individu- als are embedded and in which their selves are formed. Some niches in social space imply more complex selves than others. Without tak- ing the broader system-level phenomena into account, we risk viewing complex selves as something akin to a personality characteristic (an individual attribute). Instead, I view them here as a reflection of the social system and an individual’s location within it.
By focusing on the dependence of selves on network ties, I can generate some relative- ly straightforward predictions, based on what
Downloaded from by Wayne Mellinger on October 20, 2011
we know about the density and diversity of networks in different regions of social space. For example, we know from Peter Blau’s (1977) structural analysis that numerically smaller categories of people will have more out-group ties than those in larger categories (e.g., African Americans have more ties with Anglos than Anglos do with African Americans). Actors higher in the stratifica- tion system are more likely to have diverse networks that range further through the social system than those who are lower in the strati- fication system (Lin 2001). Each of these well-established empirical facts leads to a corresponding proposition about self com- plexity:
Proposition 5: Individuals occupying numerical- ly smaller categories will have more complex selves than individuals from numerically larger categories.
Proposition 6: Higher status actors will have more complex selves than lower status actors.4
Above, I have repeated my argument for an ecology of identities (from Smith-Lovin 2003), and have suggested a number of propositions about what such an ecology would look like. Hypotheses could use survey methods to locate respondents in social struc- tural space and to measure their self struc- tures. Further, these structural phenomena could be linked to self-concepts held by respondents in different social niches. I lack the space to develop these ideas fully, but here
4 Many readers are disturbed by this proposition, because it seems to imply that lower status people are inferior in some way. I have two reactions to that concern. First, like expectation states theory (Berger et al. 1972), the proposition is about position within the structure rather than any intrinsic characteristic of people. It is not racist or sexist to say that if people hold lower status posi- tion for which people have competence expectations, then this fact has implications for collective task interaction. Neither is it racist or sexist to say that occupying a posi- tion higher in a hierarchy gives one a more diverse set of network contacts. Second, we tend to assume that “com- plexity” is good. But, in this case, the contrast to com- plexity is not simplicity, but rather a gemeinshaft integra- tion of community and identity. This state is more often the focus of a mythic yearning than an object of scorn. It may not be a negative state for people or for social sys- tems.
is one example. Turner (1976) argued that society was undergoing a shift from the per- ception of self as an institutionally motivated actor toward a more impulsively, personally motivated one. Pescosolido and Rubin (2000) noted similar processes of self-perception in a postmodern world when actors were more often found at the center of a spoke structure of relationships, rather than in overlapping group affiliations. When people serve as a single bridge between different groups, they are more likely to perceive themselves in an individuated, autonomous way. They are less likely to be collective, institutional, and sub- merged in the social. I suggest that structural positions directly influence how people see themselves and their motivations.
Proposition 7: People with more complex selves will be more likely to self-describe in attribute terms and less likely to self-describe using group-membership or role-occupancy terms.
Now, I will turn to the domain of interaction to suggest how these system level properties will be reflected in actual social interactions.
Complexity of Situated Encounters
A complex self is a necessary condition for multiple identities to be enacted simulta- neously in a situation. Increasing self-com- plexity may make such multiple identity enactment more likely by chance. In my earli- er chapter on the ecology of identity, I made such a base-rate argument (Smith-Lovin 2003). I suggested that having a complex self would make one more likely to enact multiple identities in the context of a single situation. Some new data and a reconsideration of other theoretical ideas have changed my thinking. I believe that I had underestimated the extent to which modern interactions are segregated. This segregation leads complex selves to be played out in relatively simple single-identity interactions.
Two types of data, one on network ties and one on interactional encounters, made me reassess my base-rate argument. The first is a national survey of close confidant ties collect- ed with Miller McPherson in the 2004 General Social Survey. The fact that these close ties are rarer in 2004 than they were in a 1985 survey
Downloaded from by Wayne Mellinger on October 20, 2011
received much attention (McPherson, Smith- Lovin, and Brashears 2006). What strikes me about these data, however, is how few multi- plex (multiple relationship) ties we found.5 We know that these data measure very, very close ties (McPherson et al. 2006: 354–6). The respondents and the conf idants with whom they “discuss important matters” have known each other for an average of seven years and interactalmostfivetimesaweek(McPherson et al. 2006:360). Close ties are more likely than more distant ones to be multiplex. Indeed, multiplexity is one part of Granovetter’s (1974) definition of a close tie. Yet when we look at the very close tie of spouse in the 2004 data, we find that 58 per- cent of the spouses mentioned do not share any other structural link with their partners. These spouses are not also designated as coworker, comember of a voluntary group or other kind of relation. Of the 42 percent that did have some serious multiplexity in their relationship with their spouse (that is, other than labeling their spouse a “friend”), only 13 percent had more than two types of relations with their spouse. Of the 1467 respondents that we surveyed, only 33 had a spouse who was also a friend, advisor, coworker, and group comember—the kind of complex, mul- tiplex relationship that I have with my hus- band, Miller McPherson, and which many aca- demic couples assume is typical.
Non-kin ties were even simpler. Among non-kin relations, very few people in the 2004 data were neighbors and coworkers, or coworkers and comembers in a voluntary group. Only about six percent of the non-kin ties were multiplex in a seriously structural way, linking more than one institutional or group context.6 The unidimensional nature of these very close relationships led me to look back at experiential sampling data collected in 1995.
5 The data are publicly available at http://sda., for those who would like to explore this issue in greater depth.
6 Almost all non-kin ties were labeled as “friends,” but I do not consider the addition of this relationship label to make a tie multiplex. Instead, it seems almost synony- mous with the meaning of the question, which asked about “discussing important matters.”
The experiential sampling study mea- sured the self-structure of individuals in a tra- ditional way. We asked respondents to list up to 10 “more important” and 10 “less impor- tant” self-identities in a questionnaire. We then assessed the identities that they enacted within the context of situations sampled at random during an eight-day period. Here, I focus on two features of our respondents’ identities. First, of the 224 identities that respondents mentioned in the self-structure assessment, only 105 appeared in the situa- tions that were sampled in the study. Almost as many identities (104) appeared only as situat- ed identities, but did not appear in the self- structure. Clearly, respondents were spending much of their time (roughly 50 percent) in identities that were elicited by their situational contexts, but were not an expression of “who they were.” This picture of the interactional environment is somewhat at odds with our tra- ditional views of self-structure enactment, which would lead us to expect a large number of self-identities in interactional situations (Stryker 1980; Thoits 2003).
When we turn our view to the situation level, this tension is reinforced. Of the 578 sit- uations observed, 261 involved no self-struc- ture identities at all (see Table 1). The majori- ty of situations (378 of 578 or 65.4 percent) involved only a single identity. Over half of those (192 of the 378) involved a single iden- tity that was not central to the respondent’s self structure. While multiple-identity situa- tions were not rare (200 out of 578), they were as likely to involve two non-self-structure identities or a self-structure identity and a purely situational identity as they were to involve the combination or clash of two salient identities that were a part of the core self structure.
These empirical findings, generated by very different methodologies, lead me to reconsider my earlier view of an increasingly differentiated social system leading to increas- ingly complex social interactions. Multiplex (multiple identity) relationships do not seem extremely common, even in the context of very close ties. In fact, our experiential sam- pling data indicate few situations where multi- plex relationships lead to multiple-identity
Downloaded from by Wayne Mellinger on October 20, 2011
Table 1. Situations from an Experiential Sampling Study (N = 578)
Of 200 Mulitple Identity Situations
—69 had no self-structure identities
—77 had one self-structure identity and one purely situational identity —54 had two self-structure identities
Of 378 Single Identity Situations:
—181 involved self-structure identities —192 involved purely situational identities
interactions with a significant other. When multiple-identity situations did occur in our experiential sampling study, they seemed more likely (1) to result from unusual situa- tions with multiple audiences (in a group of several people), (2) as the result of enacting a situationally elicited non-salient identity while mentally occupying a salient self-structure identity, or (3) as the result of ruminating over non-interactional identity conflicts. For exam- ple, a Student reports worrying about a test while he is lunching as a Friend with some associates. A respondent reports worrying about which family member to stay with on an upcoming trip while she is interacting with strangers in a laundromat. Stryker’s point about the central importance of selves may have more force when predicting our elicita- tion of cognitive schemas (networks of infor- mation about our central self-identities) in thinking about our lives, rather than acting in them.
Complex selves did not seem to come from complex relationships. Nor did they lead to complex situations. Instead, they were mostly features of the mind—the process of ruminating about our lives, making plans, being “out of” the current situation and cog- nitively (or emotionally) in “another place” that is important to us.
Neil MacKinnon’s and David Heise’s (forthcoming) book begins with an extensive discussion of the postmodern self. In much of this literature, the self is a local, experientially based one. Postmodernists argue that we are now many people to many audiences, but not necessarily at the same time. While
MacKinnon and Heise (forthcoming) develop these ideas in an institutional context, Pescosolido and Rubin (2000) echo McPherson’s (1983, 2004) more structural argument in interpreting large-scale historic change in the relationships between self and identity. Pescosolido and Rubin argue that his- torical social network patterns progressed beyond an unfolding of salient dimensions to a postmodern society where individuals most- ly bridge structural holes, connecting nonoverlapping groups. In such a postmodern world, where most network ties are bridging ties, individuals would have complex selves but would seldom encounter situations in which multiple salient identities were relevant.
Pescosolido and Rubin suggest that the primary mechanism driving the postmodern “spoke” network structure, where individual actors act as bridging ties between otherwise unconnected groups, is the declining stability of ties. If ties are stable, long-term relations, the fact that most group memberships are recruited through network ties should lead one bridging tie to become many. Social groups will become cross-cutting social circles if ties persist long enough for new ties to build on the old connection. Indeed, McPherson’s (2000; see also McPherson and Smith-Lovin 2002) simulation studies of system-level para- meters and their effects on group structure show a strong positive relationship between tie stability and the membership overlap of groups. These simulations also show that tie stability fosters larger, more diverse groups that survive longer, although in smaller num- bers. The simulation results from these sys- tem-level relationships lead to my first situa- tion-level predictions. Ties that persist for longer periods of time are more likely to evoke multiple, overlapping group memberships
Downloaded from by Wayne Mellinger on October 20, 2011
and, net of other social forces, to result in the simultaneous operation of multiple identities.
Proposition 8: Interactions with alters with whom one has a longer history of interaction are more likely to involve multiple identities than interactions with those with whom one has a shorter term relationship.
Proposition 9: Interactions between alters with a multiplex relationship will more likely evoke multiple identity standards than interactions between alters who are connected by only a sin- gle relation.
An example from the experiential sam- pling study helps to illustrate this point. The respondent is an administrative assistant with three children. She is polled while in her uni- versity office, where she is “speaking with a soccer mother.” She occupies the identities of Coach, Mother, and Friend in the conversation (she volunteers the third identity, when the questionnaire only gives room for two). She is the coach of her neighborhood friend’s child’s soccer team, and all three of these identities are part of her self-structure (as recorded in the survey at the beginning of the study). Her embeddedness in this multiplex relationship, spawned in neighborhood, voluntary associa- tion, and family institutional contexts, has potential for discomfort. There are different expectations and cultural meanings for each role-identity.
But such multiplex ties are increasingly rare, as our recent paper indicates (McPherson et al. 2006). Instead, we are increasingly able to segregate our audiences. The implications of the increasing proportion of weak ties, and the scarcity of deep, embedded ties and cross- cutting social circles is actually quite pro- found. Consider the McPherson, Popielarz, and Drobnic (1992) research that shows how weak and strong ties influence our stability of memberships in groups. Strong ties shared with those in a group dramatically lengthen our stay in that group. Multiplex relationships tend to become more multiplex, as their shared information increases and people pull their friends into shared activities. Dense net- works inside a group make the meanings of identities shared within the group move
together (Keeton 1999). These processes increase the clarity of group boundaries.
Interacting in sparser, less interconnected networks of weak ties exposes us to new infor- mation, and makes us more likely both to leave current groups and to join new ones. Relating these patterns directly to identity, Robinson, Keeton, and Rogalin (2002) show that people who interact in less dense net- works adopt more identities with more diverse meanings in an experimental computer chat situation. If our world has a higher proportion of weak ties, we also have more complex, less stable selves—but not necessarily more com- plex situations.
Social situations must involve relation- ships that are multiplex or involve multiple actors who have different relationships in order to evoke multiple identities simultane- ously. While Pescosolido and Rubin (2000) argue that a postmodern world leads to increasingly segmented, segregated audiences for a limited slice of the self, having several people in a setting increases the chances that an actor will occupy a different identity for different alters in the situation.
Proposition 10: Interactions with more than one alter are more likely to evoke multiple identity standards than interactions between alters who are connected by only a single relation.
Again, a brief example from the experien- tial sampling study illustrates the process sug- gested by P10. A university administrative assistant finds herself in multiple identities at a meeting “working with a professor on the Minority Retention Committee to provide input as a minority and support as a staff per- son.” There are five people in the situation (2 males and 3 females—both males are white and the females are of various ethnicities). She interacts with all of the individuals present, “teaching” the others about the racial/ethnic climate on campus, but only knows and inter- acts with one of the men (her mentor, the pro- fessor) on a regular basis. She indicates the situationally relevant identities she occupies include Teacher (a self-structure identity) and Student (a non-self identity, elicited in this
Downloaded from by Wayne Mellinger on October 20, 2011
situation by her relationship to her former mentor, the professor).
Like the work setting above, most interac- tions occur in the context of larger institution- al foci. Therefore, the composition of these institutional settings has powerful effects on the direct connections among actors within those settings. McPherson and I have shown that most of the observed similarity among friendship ties formed in groups, for example, is due to the opportunity structure of the group, rather than individual choices made within the group (McPherson and Smith- Lovin 1986, 1987). Therefore, we can think of setting-level versions of P2, P3, and P4 to pre- dict when interactions will involve alters who differ in some socially-important characteris- tic (either roles or category memberships). People who interact in large, diverse groups or institutional settings with low internal correla- tions among social characteristics will be more likely to be embedded in interactions with diverse others, and to occupy multiple identities within those interactions. To state the most proximate prediction:
Proposition 11: Interactions involving actors who are dissimilar on socio-demographic char- acteristics will be more likely to invoke multiple identities than interactions among demographi- cally similar alters.
Note that P11 will be supported only net of the potentially powerful effects of tie stabil- ity in P8. Homophilous, multiplex, long-term relationships are more likely to be strong ties than weak ties. On the other hand, we would expect ties between dissimilar others to be weak ties. As such, the zero-order relationship between dissimilarity and multiple identity occupation (P11) might well be negative. But, controlling for tie duration (P8) and multi- plexity (P9), I expect dissimilarity to give rise to more category-membership personal identi- ties.
I have generated a series of predictions, derived mostly from social ecology and net- work theory, about when multiple indentities will be available in self structures and opera- tive within situations. I now move onto more familiar social psychological terrain to discuss
how multiple identities when simultaneously held will lead to lines of action in situations.
As Stryker and Burke (2000) noted, the “internal” branches of structural symbolic interaction—among which I would include Affect Control Theory—have mostly ignored the problem of multiple identities. They large- ly assume that one identity becomes central to an interaction and that actors operate to main- tain that identity. This emphasis may have been appropriate: most of our interactions involve unidimensional institutional contexts or place us in identities that are not central to our self-structure. Still, the multiple-identity enactment issue has been central enough for theorists to develop some intermediate solu- tions.
Affect Control theorists add qualifiers to identities (Averett and Heise 1987; Heise and Thomas 1989) to create amalgamated identity meanings, which actors maintain in a situa- tion. For example, socio-demographic charac- teristics can combine with role-identities to create composites like “a Female Doctor” or “a Latina CEO.” Moods and personality char- acteristics can be added in the same way, cre- ating combinations like “a Depressed Teacher” or “a Gregarious Clerk.” The mean- ings that a culture holds for these qualified identities can be predicted very accurately using the same empirical paradigm that Affect Control Theory uses to estimate affective meanings after an event (Smith-Lovin 1987a). Researchers assess the meanings of both the identity and the qualifying characteristic on the three dimensions of affective meaning that Affect Control Theory uses as its metric– evaluation, potency, and activity (Robinson and Smith-Lovin 2006; Heise 2007). The researcher assesses the meaning of the com- posite (qualifier + identity) using the same scales. Then the meaning of the composite is regressed on the meanings of the identity and qualifier measured in isolation to develop an empirically derived model of how the two combine to form a new fundamental meaning for the composite identity.
Downloaded from by Wayne Mellinger on October 20, 2011
Not surprisingly, when the attribute close- ly matches the identity in meaning, the identi- ty meaning is not changed significantly by the addition of a qualifier (e.g., a Rich CEO). In this case, the qualifying characteristic is already included in the prototypical meaning of the identity. When the qualif ier is signif i- cantly different in meaning (e.g., a Generous CEO), the resulting composite is a predictable function of the two distinct meanings (the meaning of being a Generous person and the meaning of occupying the position CEO). So, a Generous CEO seems nicer and more potent than the average CEO. INTERACT, the simu- lation program based on Affect Control Theory (available at ~socpsy/ACT/), does the calculations for researchers.
When identities are relatively close in meaning, they can be maintained simultane- ously by similar actions. Since it is the mean- ing in the three-dimensional (evaluation, potency, and activity) space that determines the actual processing of the event in Affect Control Theory, two identities that are very close in the meaning space are effectively the same identity. Consider, however, the relative- ly unusual case where two identities are quite different in meaning, but simultaneously evoked by the situation. In this case, actions that maintain one identity will be disruptive to the other. In the conceptual language of Affect Control Theory, “deflection” will result.
A connectionist model, which allows the parallel processing of multiple understandings of the situation, is consistent with the new challenges introduced by multiple identity standards. The distributed representations that are possible within the connectionist model (Kashima, Foddy, and Plakow 2001) are well suited to characterizing a multiple-identity self. Multiple aspects of the self can be acti- vated by a situation. Events can be perceived and processed simultaneously from the point of view of multiple identities (Smith-Lovin 2001).
If identities rather disparate in meaning are processed in parallel, maintenance in one will result in deflection for the other. Since deflection is experienced psychologically as a sense that the world is unpredictable or dis-
turbing, one would expect stress to result. There also might be a heightened probability of leaving the interaction. Our experiments show that people are more likely to select away from interactions with those who fail to confirm their identities (even when those identities are negative in evaluation) (Robinson and Smith-Lovin 1992).
Proposition 12: Interactions involving disparate identity meanings simultaneously held by one actor will create more stress than interactions involving a single identity standard.
Proposition 13: Actors will terminate interac- tions in which they simultaneously enact dis- parate identity meanings at a higher rate than interactions involving a single identity.
One phenomenon that this parallel multiple- identity processing could explain is the com- mon experience of mixed emotions. If our control models are correct, emotions are expe- rienced primarily as the result of the confir- mation or disconfirmation of identities. If an actor is occupying more than one identity simultaneously, and experiencing events from those multiple perspectives, it is natural that a mixture of emotions would result from events. For example, a directive action that would support the identity of Judge might produce negative deflection on the evaluation and pos- itive deflection on the potency dimension for the Woman who occupies that position. This might produce a mixture of feelings of being humble (the Judge) and being contemptuous (the Woman).
Proposition 14: Emotions experienced in inter- actions with multiple-identity meanings simulta- neously held by one actor will be more variable in their affective meaning than emotions experi- enced in interactions while occupying a single identity.
I note that in this Affect Control Theory based formulation, P12 and P14 are quite distinct propositions. In Affect Control Theory, deflec- tions can create emotions that are positive or negative, empowering or deflating, enlivening or quieting (Smith-Lovin and Robinson 2006; Robinson, Rogalin and Smith-Lovin 2005). The prediction in P14 is that our responses in situations where we are acting simultaneously in two or more identities will be characterized
Downloaded from by Wayne Mellinger on October 20, 2011
by mixes of quite different emotions, rather than a consistently positive or negative emo- tion. The sense of disquiet created by deflec- tion in Affect Control Theory is quite different (and analytically separate) from the emotions experienced as the result of events.
I will note that the connectionist repre- sentation of identity processing is also quite consistent with Affect Control Theory’s view of the relationship between individuals and the culture from which they derive identity mean- ings. Consider the view that each individual represents a variety of self-conceptions (iden- tities) within a parallel distributed processing system. The meanings associated with these self-conceptions are shared with other individ- uals and represented symbolically by cultural artifacts like books, films, and language use. This distributed cognition model captures sev- eral features that are central to Affect Control Theory and other sociological theories that grew out of Meadian symbolic interactionism. First, the model accurately represents the rela- tionship between the individual and the col- lective. Individual meanings are developed out of contact with society (in both personal and artifactual forms). Furthermore, individuals act as both learners, carriers, and (within lim- its) innovators of cultural meanings. Therefore, the ideas laid out here come back up to the system level at both ends. I call for an ecology of identity to discover the connec- tions among identities that are created by the complex selves that individuals create under varying social structures. Then, I suggest a connectionist model of individual processing that I think more accurately represents how individual actors operate as part of an inter- connected cultural system. They process mul- tiple parts of that system simultaneously in their own interactions. At the same time, they carry partial representations of a larger inter- connected cultural system in their self struc- tures.
One important feature of our return to the system level is the potential of individual experience for creating societal change. As sociological social psychologists, we are much
stronger on describing how society shapes selves than on describing how selves and interactions shape society. This bias is appro- priate, since it is the stability of social interac- tion that allows society to have its recogniz- able, extra-individual character. Still, social patterns do change, and we need to explain how these changes occur.
My suggestion is that cultural meanings associated with identity labels change when repeatedly deflected in the same direction. Major events that simultaneously affect large populations can cause such change. For exam- ple, MacKinnon and Luke (2002) find that a large number of false accusation and wrongful conviction cases in the Canadian news between 1981 and 1995 led to a decline in the evaluation and potency of criminal justice identities like Judge, Juror, Policeman, and Mountie among Canadian undergraduates. But another important mechanism is the co- occurrence of identities and the meaning maintenance problems that this co-occurrence implies. When an actor is faced with support- ing two inconsistent identity meanings simul- taneously, deflections of meaning are bound to occur. If many actors face the same problem routinely (e.g., because of changing organiza- tional demography), repeated deflections can create cultural change. When our cultural meanings change, patterns of action and emo- tional distributions change as well (under the assumptions of structural symbolic interac- tionism).
Many of the multiple-identity situations in our experiential sampling study are focused on work or family settings where people have relatively little control over their (multiple) audiences. These are institutional domains that have undergone major demographic changes in the last decades, as women have entered the labor force and families have been reshaped through divorce and non-marital childbearing.
For example, one situation with a large distance between two simultaneously occu- pied identities that we observed in the experi- ential sampling study occurred in the Minority Retention Committee meeting described above. The young Latina staff member describes her situationally relevant identities
Downloaded from by Wayne Mellinger on October 20, 2011
that she occupies as Teacher (a self-structure identity) and Student (a non-self identity). The situation causes considerable tension, because these identities are quite different (as rated by the respondent) on both the evaluation (good–bad) and potency (powerful–power- less) dimensions. She is feeling quite good and powerful in her role as an instructor in the Minority Retention effort and advisor to the others on racial sensitivity, but this role-iden- tity sits uneasily with the status relationship that she occupies with the other person who is leading the seminar with her (her former men- tor). If this situation leads only to local change (e.g., a shift in the staff person’s relationship to her mentor), it will have only local impor- tance. But if such situations occur over a large institutional scope, it might change the posi- tion of minority employees and their identities more generally (e.g., they have a type of expertise that has organizational usefulness).
Another situation with substantial dis- tance between two identities occurs in a fami- ly setting where a woman readies her family for church attendance. The distance between the Wife and Mother identities occurs almost entirely on the potency dimension and is sub- stantial. She feels quite powerless as a Wife, but relatively powerful as a Mother. The inter- esting thing about both of these situations is the inability of the actors to segregate audi- ences, given the multiple roles present within the institutional setting. I suggest that these situations are the most likely to lead to cultur- al change in a postmodern society. Mixed emotions, when experienced over a large num- ber of actors and interactions, signal this change.
If it is the tension between identity enact- ments that drives change, where does the change occur? Do Women become more pow- erful, or Judges less so? Do ethnic minorities receive higher evaluation and potency in employment settings, or do they get resegre- gated in new organizational roles? Do Wives become more powerful, or Mothers less so, when these role-identities can no longer be interactionally isolated?
A straightforward prediction is self- enhancement. People may change the mean- ing of the identity that moves them upward on
the evaluation, potency, or activity dimen- sions. As an Affect Control Theorist, I assume meaning maintenance is much stronger than self-enhancement. In addition, new evidence suggests that people are as willing to move self-conceptions in a negative direction as in a positive direction (Cast and Cantwell, forth- coming). Therefore, I am not convinced that self-enhancement will provide directionality to social change.
MacKinnon and Heise (forthcoming) focus on fundamental self-feelings (the funda- mental affective meaning of “myself as I real- ly am”) as the force that creates continuity in the complex, postmodern self. To extrapolate from their argument, one might change identi- ty meanings in a direction that is most consis- tent with self-meaning. If most people think well of themselves, this would effectively con- vert to a self-enhancement motive. If there is a wider distribution of self-meanings, this process would produce individual change in meaning but not accumulate to larger social, cultural change.
I tentatively propose a more structural solution. I build again on Stryker’s conception of commitment, the degree to which a self- identity is implicated in many relationships that are important to an actor. Some identities are more embedded in our cultural system than others. When many other role relation- ships depend on a focal identity, it will be much harder to change its meaning. When an identity is rich in connections (cognitive and affective) to other concepts within our cultur- al system, it will be difficult to move in affec- tive space. Too many other terms would have to move with it. When power structures have a strong vested interest in the maintenance of an identity’s meaning (e.g., the derogation of workers by capitalists), it will be more diffi- cult to change. When actors that are experi- encing local change are more in contact with one another, they are more likely to develop a subcultural understanding that can withstand the inertia of mainstream cultural meanings. This is the study of social movements. I applaud recent attempts by structural symbol- ic interactionists to enter this domain (Heise and Britt 2000; Stryker 2000)
Downloaded from by Wayne Mellinger on October 20, 2011
I argue here for an ecological theory of self and identity, where large-scale social structures (population distributions, correla- tion of social dimensions, homophily) influ- ence the availability and occupancy of identi- ties. These identities get incorporated into selves as they are enacted in networks of sta- ble, recurring relationships. But they also get elicited by situations, even when they are not part of a self-structure. The decline of stable, long-term, multiplex relationships over the evolution of human society, added to the increasingly differentiation and expansion of social space, has led to more complex selves but simpler situations. Our complex selves are available mostly in introspection, as we pon- der autobiographical narrative and conflicting role obligations. Weak ties pull us into and out of institutional and group settings at a higher rate. Selves are complex but fluid.
On the other hand, situations are largely simple. Our traditional models of meaning maintenance within the context of a well- defined situation do a remarkably good job of handling everyday interaction. Only institu- tions that restrict our ability to segregate audi- ences (like family and work) confront us with situations in which multiplex relationships, multiple interaction partners, or other features force the simultaneous occupancy of identities that have distinctly different meanings.
While rare, these complex situations are key for social change, if they are experienced by many people over a substantial time period. Since maintaining the meaning of one identity will necessary create systematic deflections in the other, we expect shifts in one or both cul- tural identity meanings. Physiological stress and mixed emotions will accompany this social change, as an individual-level manifes- tation of the interactional source of the cultur- al tension (Robinson, Smith-Lovin, and Rogalin 2004). This is the mechanism through which self and interaction can affect society.
Averett, Christine P. and David R. Heise. 1987. “Modif ied Social Identities: Amalgamations,
Attributions, and Emotions.” Journal of
Mathematical Sociology 13:103–132
Barker, Roger G. 1968.
Ecological Psychology: Concepts and Methods for Studying the Environment of Human Behavior. Stanford, CA:
Stanford University Press.
Barker, Roger. G. and Herbert F. Wright. 1951.
Boy’s Day. New York: Harper & Row.
Barker, Roger G. and Herbert F Wright. 1959.
Midwest and Its Children: The Psychological Ecology of an American Town. Evanston, IL:
Row, Peterson and Co.
Berger, Joseph, M. Hamit Fisek, Robert Z. Norman,

and Morris Zelditch Jr. 1972. Status Characteristics and Social Interaction.: An Expectation States Approach. New York: Elsevier.
Blau, Peter M. 1977 Inequality and Heterogeneity. New York: Free Press.
Breiger, Ronald L. 1974. “The Duality of Persons and Groups.” Social Forces 53: 181–89.
Burke, Peter J., Timothy J. Owens, Richard T. Serpe, and Peggy A. Thoits, eds. 2003. Advances in Identity Theory and Research. New York: Kluwer/Plenum.
Cast, Alicia D. and Allison M. Cantwell. Forthcoming. “Identity Change: Effects of Positive and Negative Feedback.” Social Psychology Quarterly.
Chaves, Mark. 2004. Congregations in America. Boston, MA: Harvard University Press.
Goffman, Erving. 1963. Behavior in Public Places: Notes on the Social Organization of Gatherings. Glencoe, IL: Free Press of Glencoe.
Granovetter, Mark S. 1973 “The Strength of Weak Ties.” American Journal of Sociology 78: 1360–80.
Heise, David R. 1979. Understanding Events: Affect and the Construction of Social Action. New York: Cambridge University Press.
Heise, David. R. 1998. “Conditions for Empathic Solidarity.” Pp. 197–211 in The Problem of Solidarity: Theories and Models, edited by Patrick Doreian and Thomas Fararo. Amsterdam: Gordon and Breach.
Heise, David R. 2007. Expressive Order: Confirming Sentiments in Social Actions. New York: Springer.
Heise, David R. and Lori Britt. 2000. “From Shame to Pride in Identity Politics.” Pp. 252–270 in Identity, Self and Social Movements, edited by Sheldon Stryker, Timothy J. Owens, and Roger W. White. Minneapolis, MN: University of Minnesota Press.
Heise, David R. and Lisa Thomas. 1989. “Predicting Impressions Created by Combinations of Emotion and Social Identity.” Social Psychology Quarterly 52: 141–148.
Hogg, Michael A. and D. Abrams. 1988. Social Identifications: A Social Psychology of
Downloaded from by Wayne Mellinger on October 20, 2011
Intergroup Relations and Group Processes. New
York: Routledge.
Kashima, Yoshihisa, Margaret Foddy, and M. Platow,

eds. 2001. Self and Identity: Personal, Social
and Symbolic. New York: Erlbaum.
Keeton, Shirley. 1999.
To Thine Own Self (And One’s Associates) Be True: The Strategic Management of Competing Identities Among Academic Women. Unpublished doctoral dissertation.
Louisiana State University, Baton Rouge, LA. Lin, Nan. 2001. Social Capital: A Theory of Social Structure and Action. New York: Cambridge
University Press.
MacKinnon, Neil J. 1994.
Symbolic Interactionism as
Affect Control. Albany, NY: State University of
New York Press.
MacKinnon, Neil J. and David R. Heise. Forthcoming.

Identities, Selves, and Social Institutions. Un-
published manuscript.
MacKinnon, Neil J. and Michelle M. Goulbourne.

2006. “The Affect Control Theory of Emotions: The Case of Depression.” Pp. 237–266 in Purpose, Meaning, and Action: Control Systems Theories in Sociology, edited by Kent A. McClelland and Thomas J. Fararo. New York: Palgrave Macmillan.
MacKinnon, Neil J. and Alison Luke. 2002. “Changes in Identity Attitudes as Reflections of Social and Cultural Change.” Canadian Journal of Sociology 27: 299–337.
Mark, Noah P. 1998. “Birds of a Feather Sing Together.” Social Forces 77: 455–85.
Mark, Noah P. 1999. “Beyond Individual Differences: Social Differentiation from First Principles” American Sociological Review 63: 309–30.
Mark, Noah P., Cecilia Ridgeway, and Lynn Smith- Lovin. Unpublished manuscript. “The Formation of Consensual Status Beliefs.”
Mayhew, Bruce H. 1974. “Baseline Models of Social Structure.” American Sociological Review 39: 137–43.
Mayhew, Bruce H., Roger L. Levinger, J. Miller McPherson and T.F. James. 1972. “System Size and Structural Differentiation in Formal Organizations: A Baseline Generator for Two Major Theoretical Propositions.” American Sociological Review 37: 629–633.
McClelland, Kent and Thomas Fararo. 2006. Purpose, Meaning, and Action: Control Systems Theories in Sociology. New York: Palgrave-MacMillan.
McPherson, Miller. 1983. “An Ecology of Affiliation.” American Sociological Review 48: 519–532.
McPherson, Miller. 2000. “Modeling Change in Fields of Organizations: Some Simulation Results.” Pp. 134–44 in Computational Modeling of Behavioral Processes in Organizations, edited by Daniel Ilgen and Charles Hulin. Washington, D.C.: American Psychological Association Press.
McPherson, Miller. 2004. “A Blau Space Primer:
Prolegomenon to an Ecology of Affiliation,”
Industrial and Corporate Change 13: 263–280. McPherson, Miller, Pamela A. Popielarz, and Sonja Drobnic. 1992. “Social Networks and Organiza- tional Dynamics.” American Sociological
Review 57:153–170.
McPherson, Miller and James Ranger-Moore. 1991.

“Evolution on a Dancing Landscape: Organizations and Networks in Dynamic Blau Space.” Social Forces 70: 19–42.
McPherson, Miller and Tomas Rotolo. 1996. “Testing a Dynamic Model of Social Composition: Diversity and Change in Voluntary Groups.” American Sociological Review 61: 179–202.
McPherson, Miller and Lynn Smith-Lovin. 1981 “Women and Weak Ties: Sex Differences in the Size of Voluntary Associations” American Journal of Sociology 87: 883–904.
McPherson, Miller and Lynn Smith-Lovin. 1986. “Sex Segregation in Voluntary Associations.” American Sociological Review 51: 61–79.
McPherson, Miller and Lynn Smith-Lovin. 1987. “Homophily in Voluntary Organizations: Status Distance and the Composition of Face-to-Face Groups.” American Sociological Review 52: 370–79.
McPherson, Miller and Lynn Smith-Lovin. 2002. “Cohesion and Membership Duration: Linking Groups, Relations, and Individuals in an Ecology of Affiliation.” Advances in Group Processes 19: 1–35.
McPherson, Miller, Lynn Smith-Lovin, and Matthew Brashears. 2006. “Social Isolation in America: Changes in Discussion Networks over Twenty Years.” American Sociological Review 71:353–75.
McPherson, Miller, Lynn Smith-Lovin, and James M. Cook. 2001. “Birds of a Feather: Homophily in Social Networks.” Annual Review of Sociology 27: 415–444.
Merton, Robert K. 1957. “The Role-set: Problems in Sociological Theory.” British Journal of Sociology 8: 106–120.
Pescosolido, Bernice A. and Beth A. Rubin. 2000. “The Web of Group Aff iliations Revisited: Social Life, Postmodernism and Sociology.” American Sociological Review 65: 52–76.
Robinson, Dawn T., Shirley A. Keeton, and Christabel Rogalin. 2002. “Network Density and the Organization of Social Identities.” Paper pre- sented at the Annual Meetings of the International Network for Social Network Analysis in New Orleans, LA.
Robinson, Dawn T., Christabel L. Rogalin, and Lynn Smith-Lovin. 2004. “Physiological Measures of Theoretical Concepts: Some Ideas for Linking Deflection and Emotion to Physical Responses During Interaction.” Advances in Group Processes 21:77–115.
Robinson, Dawn T. and Lynn Smith-Lovin. 1992.
Downloaded from by Wayne Mellinger on October 20, 2011
“Selective Interaction as a Strategy for Identity Maintenance: An Affect Control Model.” Social Psychology Quarterly 55: 12–28.
Robinson, Dawn T. and Lynn Smith-Lovin. 2006. “Affect Control Theory.” Pp. 137–64 in Contemporary Social Psychological Theories, edited by Peter J. Burke. Stanford, CA: Stanford University Press.
Rotolo, Thomas and Miller McPherson. 2001. “The System of Occupations: Modeling Occupations in Sociodemographic Space.” Social Forces 79: 1095–1130.
Serpe, Richard T. 1987. “Stability and Change in Self: A Structural Symbolic Interaction Explanation.” Social Psychology Quarterly 50: 44–55.
Serpe, Richard T. and Sheldon Stryker. 1987. “The Construction of Self and the Reconstruction of Social Relationships.” Advances in Group Processes 4: 41–82.
Serpe, Richard T. and Sheldon Stryker. 1993. “Prior Social Ties and Movement into New Relationships.” Advances in Group Processes 10: 283–304.
Simon, Robin W. 1995. “Gender, Multiple Roles, Role Meaning, and Mental Health.” Journal of Health and Social Behavior 36: 182–94.
Simon, Robin W. 1998. “Assessing Sex Differences in Vulnerability Among Employed Parents: The Importance of Marital Status.” Journal of Health and Social Behavior 39: 38–54.
Smith-Lovin, Lynn. 1979. “Behavioral Settings and Impressions Formed from Social Scenarios.” Social Psychology Quarterly 42:31–42.
———. 1987a. “Impressions from Events.” Journal of Mathematical Sociology 13: 35–70.
———. 1987b. “The Affective Control of Events within Settings.” Journal of Mathematical Sociology 13: 71–101.
———. 1990. “Emotion as the Confirmation and Disconfirmation of Identity: An Affect Control Model. Pp. 238–70 in Research Agendas in the Sociology of Emotions, edited by Theodore D. Kemper. Albany, NY: State University of New York Press.
———. 1995. “The Sociology of Affect and Emotion.” Pp. 118–48 in Sociological Perspectives on Social Psychology, edited by Karen S. Cook, Gary A. Fine, and James S. House. Boston, MA: Allyn and Bacon.
———. 2001. “Role-identities, Action and Emotion: Parallel Processing and the Production of Mixed Emotions.” Pp. 125–44 in Self and Identity: Personal, Social, and Symbolic, edited by Yoshihisa Kashima, Margaret Foddy, and Michael J. Platow. New York: Erlbaum.
———. 2003. “Self, Identity and Interaction in an Ecology of Identities.” Pp. 167–178 in Advances in Identity Theory and Research, edited by Peter
J. Burke, Timothy J. Owens, Richard T. Serpe and Peggy A. Thoits. New York: Kluwer/Plenum.
Smith-Lovin, Lynn and David. R.Heise. 1988.
Analyzing Social Interaction: Advances in Affect Control Theory. New York: Gordon and Breach Science Publishers.
Smith-Lovin, Lynn and Miller McPherson. 1993 “You Are Who You Know: A Network Perspective on Gender.” Pp. 223–41 in Theory on Gender/Feminism on Theory, edited by Paula England. New York: Aldine de Gruyter.
Smith-Lovin, Lynn and Dawn T. Robinson. 2006. “Control Theories of Identity, Action, and Emotion: In Search of Testable Differences between Affect Control Theory and Identity Control Theory.” Pp. 163–188 in Purpose, Meaning, and Action: Control System Theories in Sociology, edited by Kent McClelland and Thomas Fararo. New York: Palgrave MacMillan Press.
Stryker, Sheldon. 1980. Symbolic Interactionism: A Social Structural Version. Menlo Park, CA: Benjamin Cummings.
Stryker, Sheldon. 2000. “Identity competition: Key to Differential Social Movement Involvement.” Pp. 21–40 in Identity, Self, and Social Movement, edited by Sheldon Stryker, Timothy J. Owens, and R. White. Minneapolis, MN: University of Minnesota Press.
Stryker, Sheldon. 2003. “A Peek Back– and Then Ahead.” Pp. 217–23 in Advances in Identity Theory and Research, edited by Peter J. Burke, Timothy J. Owens, Richard T. Serpe, and Peggy A. Thoits. New York: Kluwer/Plenum.
Stryker, Sheldon and Peter J. Burke. 2000. “The Past, Present, and Future of an Identity Theory.” Social Psychology Quarterly 63: 284–97.
Stryker, Sheldon and Richard T. Serpe. 1982. “Commitment, Identity Salience and Role Behavior: A Theory and Research Example.” Pp. 199–218 in Personality, Roles, and Social Behavior, edited by W. Ickes and E.S. Knowles. New York: Spring-Verlag.
Stryker, Sheldon, Richard T. Serpe and Matthew O. Hunt. 2005. “Making Good on a Promise: The Impact of Larger Social Structures on Committments.” Advances in Group Processes 22: 93–123.
Thoits, Peggy A. 2003. “Personal Agency in the Accumulation of Multiple Role-identities.” Pp. 179–194 in Advances in Identity Theory and Research, edited by Peter J. Burke, Timothy J. Owens, Richard T. Serpe, and Peggy A. Thoits. New York: Kluwer/Plenum.
Turner, Ralph H. 1976. “The Real Self: From Institution to Impulse.” American Journal of Sociology 84: 1–23.
Downloaded from by Wayne Mellinger on October 20, 2011
Lynn Smith-Lovin is the Robert L. Wilson Professor of Arts and Sciences in the Department of Sociology at Duke University. Her research focuses on the relationship between social structure, identity, interpersonal action, and emotion. Current projects include a study of justice processes, identity disconfirmation, and emotion (in collaboration with Dawn T. Robinson and Jody Clay- Warner), and a study of the coevolution of networks and voluntary group association memberships (in collaboration with Miller McPherson). Both projects are supported by the National Science Foundation. She has served as President of the Southern Sociological Association, Vice-President of the American Sociological Association, and Chair of both the ASA Sections on Social Psychology and on Emotions. She received the ASA Section on Emotions Lifetime Achievement Award in 2002.